This PDF file contains the front matter associated with SPIE Proceedings Volume 13075, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

As the demand for autonomous driving and robot vision arises, semantic segmentation has developed rapidly. Semantic segmentation provides the means for machines to understand the environment. However, current methods face a balance issue between segmentation quality and computational resources. Our proposed edge-guided non-linear dynamic convolutional network (ENNet) achieves real-time and accurate semantic segmentation. The core of our network is a novel non-linear, dynamic combination (Nd-conv) that allows for a non-linear, dynamic combination (Nd-conv) of multiple convolution weights, improving the convolution's encoding ability while maintaining a low computational burden. The Non-linear Dynamic Convolutional Modules (NDCM) is introduced to enhance the segmentation accuracy of the network while maintaining efficiency. The Multi-stage Feature Fusion Module (MFFM) is also introduced to fuse low-level details and high-level semantic information to improve the segmentation accuracy of the network. Experiments conducted on a 1080Ti GPU show that our model achieves 74.6 mIOU on the Cityscapes dataset while being able to perform inference at 202 FPS. Our network design has achieved an optimal balance between speed and accuracy.

This paper mainly studies the joint separation scheme of Jingshan gravity dam, and calculates the dam body stress change from three schemes, namely, no transverse joint, transverse joint spacing of 60m and transverse joint spacing of 30m. The analysis results show that the dam body maximum tensile stress occurs on the upstream dam faces near the dam heel. The dam body maximum tensile stress is 2.16MPa, 1.97MPa and 1.62MPa respectively for three schemes, including no transverse joint, transverse joint spacing of 60m and transverse joint spacing of 30m. When the transverse joints are not divided, the actual tensile stress is greater than the C20 rockfill concrete tensile strength, and the dam concrete is more likely to crack. When the distance between transverse joints is 60m, only the tensile stress of individual points is greater than the concrete tensile strength, and the probability of dam concrete cracking is small; When the transverse joint spacing is 30m, the dam body actual tensile stress is less than the concrete tensile strength, and the probability of dam concrete cracking is small. It is suggested that the rockfill concrete gravity dam should be divided according to the transverse joint spacing of 30m when simple temperature control measures are taken.

To investigate the influence of nozzle contraction ratio (ratio of nozzle outlet area to front cross section of nozzle contraction section) on the propagation characteristics of subcritical kerosene Rotating Detonation Waves (RDW), experimental methods such as high-frequency pressure sensors and voltage stabilizing sensors were employed to analyze the establishment process of the RDW. By varying the nozzle contraction ratio, the effects of changes in the nozzle outlet area on the establishment and propagation of the RDW were analyzed. During the experiments, the air flow rate was maintained at 1500g/s, the kerosene mass flow rate was 100g/s, the total air temperature was 600K, and the equivalence ratio was set to one. The research results indicate that as the nozzle contraction ratio increases, the combustion pressure gradually decreases. The propagation speed of the RDW initially increases and then decreases, and the ignition time follows a similar trend. Although the combustion chamber pressure reaches its maximum at a nozzle contraction ratio of 0.13, the ignition time and propagation speed of the engine are significantly slowed down, which is not conducive to the initiation and propagation of the RDW.

The ultimate load bearing capacity of the nodes of the holding pole is one of the determining factors to measure the load bearing performance of the holding pole. In order to study the ultimate bearing capacity of the nodes connected with the pins of the holding pole, the overall bearing capacity of the holding pole under different working conditions was calculated and analyzed based on the Finite Element Method (FEM), the real bearing capacity information of the nodes was extracted, and the bearing capacity of the nodes was determined by the study. Simulation of the node force characteristics was carried out on this basis, and the node ultimate load capacity test was carried out using a five-axis linkage hydraulic loading device, and the reliability of the numerical model was verified through the comparative analysis of the test and simulation. The optimization of the pin shaft is improved by lightweighting, so that the component still has high safety performance after optimization.

One of the determinants in evaluating the load-bearing performance of a gin pole lies in the ultimate bearing capacity of its joints. In order to analyze the ultimate bearing capacity of welded joints in a floor-standing double-rocker gin pole when subjected to axial tension, a combination of finite element simulations and experimental studies is employed in this paper to investigate the ultimate bearing performance of welded joints in the gin pole under axial tension conditions. The effectiveness of the numerical model is validated through a comparative analysis of experimental and simulation results. The analysis conclusively demonstrates that the limb edges of the welded joints experience plastic deformation under the maximum axial tension condition. This conclusion serves as a reference for the subsequent design of gin pole joints.

This study utilizes learning analytics techniques, specifically factor analysis and clustering, to explore the commonalities and characteristics among different virtual simulation physics experiments. Additionally, it investigates the behavioral patterns of students in online experiments. The findings demonstrate that the integration of big data and learning analytics in education has revolutionized traditional teaching methods. By analyzing student data from experiments, educators can make more timely and effective adjustments to their teaching decisions, as well as implement scientific and efficient interventions to enhance students' learning experiences. Ultimately, this approach facilitates personalized teaching goals attainment.

When the wave source and receiver move relative to each other, the frequency received by the receiver will change relative to the frequency of the wave source, this is called doppler effect. This paper has summarized the principle of Doppler effect of mechanical wave and electromagnetic wave and introduced the application of Doppler effect in various fields. Doppler effect is widely applied in nuclear physics, astronomy, engineering technology, traffic management, medical diagnosis, etc. For example, it is applied for satellite velocity measurement, Doppler radar, doppler color ultrasonic diagnostic instruments, etc. Steven Chu of Stanford University won the Nobel Prize, he also adopted the Doppler effect in laser cooling. This paper has proposed a doppler effect verification device. It consists of ultrasonic power generator, ultrasonic transducer and digital oscilloscope. This device can visually demonstrate the Doppler effect and effectively verify the frequency variation phenomenon caused by the relative motion of the sound source and receiver.

With the continuous expansion of high-speed rail lines in China, high-speed trains are operating in increasingly complex and changing environments, facing challenges such as strong winds and heavy snow. When high-speed trains operate in high-altitude areas, they may collide, deposit, and adhere to key components, forming snow, directly reducing traction, shock absorption, braking and other performance, affecting high-speed trains’ comfortable and operational safety, and bringing many difficulties to railway construction and operation. This paper is based on wind tunnel experiments using the silk thread method to observe the streamline flow direction in the bogie area. Experiments have shown that key parts of the bogie are susceptible to impact and form snow, while snow is less likely to form on the leeward side and above the bogie.

This study aims to explore the interaction between low-temperature plasma and drug molecules and perform molecular dynamics simulations with the help of machine learning methods. We selected 200 drug samples for experiments, and the experimental environment was specific plasma conditions, including temperature, density, ionization degree and plasma type. By observing at successive time points, we record the rate at which drug molecules change, the types and number of chemical bonds affected, the amount of drug molecules remaining and the new compounds that may be produced. In addition, we considered other possible influencing factors such as pH, solvent type, relative humidity and temperature in the laboratory. Machine learning models are trained to predict the dynamics and reaction trends of drug molecules with plasma. Preliminary results show that machine learning methods demonstrate a high degree of accuracy and effectiveness in predicting and understanding complex interactions between plasmas and drugs.

Because of the intricate construction of wind equipment, particularly this irregularity and nonlinearity of the native air velocity magnitude seriously affects the stability of wind turbine output power. The conventional control techniques are difficult to meet the precise demands of the system. By introducing the fuzzy immune PID command strategy, this paper aims to embed the fuzzy control and immune-feedback algorithms into the traditional PID control framework in order to comprehend adjustment at any moment and optimize of the PID coefficient according to different wind speeds. This adjustment is designed to cope with the environmental conditions where the wind speed fluctuates drastically to ensure that the system can track the rated power efficiently. Finally, the Simulink simulation waveforms exhibition that fuzzy immune PID command possesses better stability and tracking adequacy than the conventional command pattern in step, flatus, and rushing winds.

Nowadays, environmental pollution, climate change and energy shortage have become widespread concerns in the world. With the rapid development of industrial production, more and more refractory organic matter enters the water environment. Traditional treatment methods have been unable to fully meet the requirements of wastewater treatment at the present stage. Studies in recent years have found that metal frame nanocomposites have a remarkable effect in the treatment of refractory wastewater. With the proposed goals of "carbon peak" and "carbon neutrality", lithium-ion batteries have been widely used in various fields of daily industrial production and life. The waste lithium-ion batteries have significant resource characteristics, and the valuable metal elements contained in them play an active role in the neighborhood of refractory wastewater treatment. In this paper, the recycled NCM material was prepared by the method of reusing the metal leaching solution of Ni, Co and Mn in waste batteries, which combined with the activation of persulfate to produce highly active sulfate free radicals and realized the efficient degradation of pollutants in organic wastewater.

TV guided missile is a kind of missile which is composed of TV camera and guidance system. In view of the problem that the target feature information is not obvious and the target and background are not distinguishable. The paper puts forward a method based on Top-Hat transformation to detect the sea level moving target, which can differentiate the target and the background effectively, and extract the typical characteristic value of the target. Simulation results prove the effectiveness of the proposed method.

In order to ensure the safety and reliability of the strength of the ship structure which is adsorbed during the automatic mooring operation, resistive strain gauges and an online strain monitoring system are arranged in the places where the stress of the ship structure is concentrated, so that the stress and deformation of the ship structure can be monitored and analyzed. The locations of the maximum values of stress and deformation of the ship structure are determined by the finite element method, and these locations are selected as the place of the resistive strain gauges. The measured values of the ship structure are obtained through the strain online monitoring system and compared with the results of the finite element calculation. Comparative analysis of the strain theoretical calculation results and test results of the ship structure shows that the theoretical data are in good consistency with the experimental data, which proves the correctness and effectiveness of the online strain monitoring system for the ship structure.

The laser technology is a frontier and practical subject which has been widely studied and is developing rapidly in the recent years. Nano lasers, as an interdisciplinary subject of optics and nanotechnology, has brought new excellent properties to the lasers and is therefore attracting the researchers' attention. The miniaturization of the optical devices has always been one of the main goals pursued by the academic communities and the industry. The most immediate benefits of the reduced material scale are the ability to integrate more tightly with other devices and the lower power consumption. Furthermore, the nano lasers’ merits of great energy-efficiency, fast modulation speeds, and reduced lasing thresholds enables it to be applied in a variety of fields. Additionally, once the nano lasers are wavelength-scale in size, they can simultaneously localize electromagnetic fields in time, space, and frequency at optical frequencies, which is an innovative advancement. In this article, the inherent advantages and the recent research progress of the semiconductor nano lasers and plasmonic nano lasers will be introduced. Then the applications of nano lasers in integrated optical interconnects, sensing, and biological detecting are introduced and analyzed. The current scientific and engineering challenges and future development directions of nano lasers are also emphasized and discussed.

The structural improvement of high-pressure hydrogen storage tank is one of the hot research issues of new energy vehicles. The hydrogen storage tank consists of an aluminum alloy inner liner (inner layer) and a carbon fiber layer (outer layer). Due to the large difference in the specific strength and modulus of the two layers, residual stress exists in the structure after each pressure relief, which seriously affects the service life of the aluminum alloy liner. At the same time, hydrogen is in direct contact with the aluminum alloy liner, resulting in the problem of hydrogen embrittlement of the aluminum alloy liner. To solve the above problems, this paper proposes a method to improve the structure of hydrogen storage tank. Carbon fiber layer is laid on the inner layer of aluminum alloy inner liner, and a composite hydrogen storage structure is designed inside the carbon fiber layer to increase the hydrogen density inside the hydrogen storage tank. Finally, the superiority and effectiveness of the improved structure are verified by comparing with type IV hydrogen storage tank. The structure can effectively alleviate the residual stress inside the structure, avoid the direct contact between aluminum alloy and hydrogen, and further increase the safety of hydrogen storage tank.

In this paper, the generation technology of distribution network digital twin scene based on multi-module data fusion is studied. Firstly, multi-module data fusion is carried out in complex dynamic operating environment of distribution. Aiming at the deficiency of constructing digital twins from a single model source, various digital model resources in distribution link are collected and integrated to support the construction of hierarchical digital twin model of distribution network. The second step is to carry out the multi-source comprehensive reconstruction and entity hierarchical processing of the distribution network, establish the multi-source digital model of the universal distribution network equipment, and realize the global and local multi-level and multi-scale model display of the distribution network environment on the basis of comprehensive reconstruction; The third step is to configure the digital twin spatial mapping based on the distribution network's panoramic data fusion to realize the panoramic mapping between the real image, spatial coordinates and equipment monitoring data and the digital twin model, as well as multi-layer semantic information embedding such as equipment, region and job, forming the corresponding characteristics of multi-block, multi-level and multi-type digital twins. Finally, the distribution network panoramic digital twin model is dynamically updated to realize the projection of various real-time perception data to the digital twin, and the digital twin is updated synchronously and grown dynamically.

Compared with the traditional optical remote sensing satellites, SAR (Synthetic Aperture Radar) remote sensing satellites have a short-term high current pulse load, which has a special design requirement for satellite power supply and distribution systems. For low-cost SAR microsatellites, the demand for lightweight power supply and distribution design is more prominent. According to the principles of low cost and high reliability, this paper aims to meet the overall effectiveness and operational reliability of small and micro-SAR satellites. The key points of satellite power supply and distribution system design are analyzed, including power satisfaction, satellite power quality, and temperature of important components such as solar array and batteries. System-level analysis is conducted through topology optimization and isolated double bus configuration, which is verified by engineering in orbit and carried out to provide guarantee conditions for the normalized operation of microsatellite SAR loads.

Speech Recognition (SR) technology, as one of the core technologies of human-computer interaction, aims to enable computers to understand the process of converting speech signals into corresponding text or commands through natural language. With the exponential increase of internet information, the features of massive speech data have significant non-specific differences and noise interference. Common feature extraction and transformation methods are no longer sufficient to meet the current needs of model training and recognition. With the rapid growth of Machine Learning (ML), many researchers use Neural Networks (NN) to solve various problems in the SR field. This article designs a Deep Learning (DL) algorithm based on Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) for SR. Firstly, sample filtering, pre weighting, signal framing, and endpoint detection are performed on the speech signal. Secondly, the MFCC value of the preprocessed data is extracted. Finally, an NN model is trained and constructed, and the trained qualified model is used to complete the recognition of speech features. The experimental results show that the algorithm designed in this paper has a lower error rate for SR and stronger generalization ability, which is of great significance for the study of SR.

With the rapid proliferation of the internet and the continuous optimization of modern communication technologies, Voice over Internet Protocol (VoIP) has undeniably emerged as one of the predominant communication technologies. In the face of numerous known and potential network threats, VoIP endeavors to establish a comprehensive, lightweight, and highly robust security mechanism. In order to enhance the efficiency of analysis within the Quality of Service (QoS) management framework, this paper proposes an approach to evaluate communication quality and network security based on the primary communication protocol in VoIP, namely Session Initiation Protocol (SIP). Furthermore, a conceptual framework for constructing the model is presented. By classifying and extracting feature fields from SIP messages, it becomes possible to gain concrete insights into the status of the VoIP network. Additionally, a VoIP evaluation system is developed, which incorporates network administrator feedback and utilizes the constructed classifier for network assessment. This evaluation methodology holds significant practical significance for both the construction of VoIP security measures and the optimization of QoS mechanisms.

Taking the environmental landscape design of Jinghe Art and Culture Center as an object, this study explores innovative design methods and strategies by combining artificial intelligence and digital space technology. Through an in-depth analysis of the site characteristics and functional needs of the art and culture center, this study proposes a landscape design method based on artificial intelligence to provide a scientific basis for landscape design. At the same time, the study also explores the application of digital space technology in landscape design, including virtual reality technology and augmented reality technology, which can help designers and users to feel the design scheme more intuitively, thus improving the quality of design and user participation. This study combines artificial intelligence and digital space technology with environmental landscape design to provide new ideas and methods for the innovative design of Jinghe Art and Culture Center. Through scientific analysis and technical means, it can better satisfy people's needs for art and cultural experience and add unique charm and vitality to urban development.

The stability and reliability of computer communication networks are crucial for the proper functioning of modern society. However, network faults are inevitable, so efficient fault detection systems are needed to quickly identify and solve problems. This study aims to improve the performance of network fault detection by introducing improved neural network algorithms and applying deep learning techniques to this field. First, we studied in detail the various types of faults that may occur in computer communication networks and constructed the corresponding fault datasets. Then, we introduced improved neural network structures, including convolutional neural networks and recurrent neural networks, to better capture the spatio-temporal characteristics of network data. Experimental results show that the improved neural network algorithm exhibits excellent performance in network fault detection. Compared with traditional methods, our model achieves significant improvements in detection rate and accuracy. In addition, the model demonstrates adaptability to different fault types, including traditional faults and emerging network threats. In summary, this study provides an efficient and reliable solution for fault detection in computer communication networks by introducing an improved neural network algorithm. This approach is expected to play an important role in real network operations and improve network stability and security, thus providing more reliable support for information communication in modern society.

Power line carrier communication uses power transmission line as a medium to transmit information between data. Because of its popularity, it has been mined by many fields and commercialized. This paper combines carrier communication technology to monitor abnormal signals of power carrier and analyzes power carrier in real time through intelligent monitoring system. The main transmission paths of power carrier communication designed in this paper are data link and control link. When the power carrier and the signal transmitting base station are directly connected, the data link mode is adopted to improve the communication rate. However, when the communication between the power carrier and the signal transmitting base station is completed by the third party forwarding, the control link mode is adopted. For the functional test of the system, this paper uses PC to check the experimental data, and uses oscilloscope to observe the input and output waveforms, and compares the receiving and sending data in real time, which verifies that the system can realize high-reliability data transmission.

Aiming at the moving-coil permanent magnet synchronous linear motors currently used in the wired power supply and sliding contact power supply mode there are some wear and tear, carbon and sparks and thus limit the service life of linear motors as well as the scene of the problem, the use of ANSYS software design a wireless power supply type of removable transformers. Firstly, the magnetoresistive principle equivalent diagram of the removable transformer was established to analyze its working principle, and based on the simulation model of this removable transformer, the influence of the air gap on the no-load output voltage and no-load current of the removable transformer was investigated. Secondly, the effect of the dynamic stator length ratio on the output performance of the removable transformer is verified by simulation, and the best output performance is obtained when the dynamic stator length ratio is around 0.25 per unit volume. Finally, by increasing the teeth at the end of the contact surface, the output performance when per unit volume is effectively increased, which provides a reference for further optimization of the output performance of the removable transformer.

In order to reduce the cogging torque of IPMSM (Interior Permanent Magnet Synchronous Motor) symmetrical auxiliary slots are opened on the rotor surface. Firstly, the cogging torque is analyzed theoretically to find out the factors affecting the cogging torque of this motor. Conclusions are drawn from the finite element simulation analysis of the shape, position angle and radius of the auxiliary slot. Finally, the semi-circular auxiliary slot with position angle θ=38deg and radius r=0.3mm is used, and finally the waveforms of the cogging torque and no-load reverse potential of the motor are compared before and after cogging. The cogging torque can be weakened by 56.93% through the use of appropriate auxiliary slot parameters, as indicated by the simulation results, and the validity of the scheme is also confirmed.

Partial discharge detection, as an effective means for identifying insulation defects, plays a crucial role in the assessment and diagnosis of power equipment’s insulation. As a detection method that closely aligns with the intrinsic characteristics of partial discharge, the optical measurement method possesses inherent advantages such as strong anti-interference capability and rich information contained within the discharge spectrum. This paper presents an optical sensor 3D sensor array-based approach for locating partial discharges. The results demonstrate that when satisfying the distance relationship, light entering the sensor surface can be treated as parallel light. The response intensity of optical sensor sensors adheres to the cosine relation between the surface normal vector and incident light angle, resulting in minimal positioning error for discharges occurring in open spaces. However, different positioning accuracies are observed when distributing light sources at various positions within Gas Insulated Switcher (GIS) equipment due to occlusion from guide rods and housing; nevertheless, it still provides certain insights into light direction.

Since the capacity and voltage of a single lithium battery are very small, dozens of single batteries are usually connected in series and parallel to form a battery pack to provide sufficient energy for electric vehicles. However, with the popularity of electric vehicles, spontaneous combustion and explosion of lithium batteries due to faults in the batteries occur from time to time, seriously threatening the safety of people's lives and property. This paper takes the actual operation data of electric vehicles as the basis and uses the method of wavelet decomposition and reconstruction to reduce the noise of the data. Then through the improved Z-score model, the difference between a normal single battery and a faulty single is amplified, and the serial number of the faulty single battery is accurately identified, and the time of fault occurrence is determined. According to the diagnostic results, the method proposed in this paper, on the one hand, can identify the faulty cells in the early stage of the fault, on the other hand, it can reduce the fault false positive rate caused by the inconsistency of the cells.

Motor bearing fault is one of the common problems in motor equipment, and it is very important to accurately identify the type of fault to ensure the smooth operation of the equipment. However, the traditional fault diagnosis methods are often limited by noise interference and difficulty in feature extraction, which leads to low diagnostic accuracy. In order to improve the fault diagnosis rate of motor bearing, a method combining Bayesian optimization, PCA and BiLSTM is proposed to optimize the performance of the neural network model. First, the hyperparameters are optimized by Bayesian optimization, then the dimensionality is reduced by PCA algorithm to extract main features, filter noise and redundant information, and finally the fault features are classified and diagnosed by BiLSTM. Experimental results demonstrate that this method can effectively enhance the accuracy of motor bearing fault diagnosis and finally the accuracy of the test set reaches 99.07%, showing good robustness.

For the field of fiber optic prefabricated rod manufacturing, raw material evaporation tank pressure control system of nonlinearity, hysteresis and other issues, this paper proposes to combine PID control algorithm and fuzzy control algorithm, designed based on the adaptive fuzzy PID evaporation tank pressure control system. The simulation results in Matlab show that, through the traditional PID and adaptive fuzzy PID control algorithm comparison analysis, obtained adaptive fuzzy PID control system response speed, small overshooting, good stability and other characteristics, can well meet the requirements of the control of evaporation tank pressure system.

One of the essential technologies for wind energy utilization is adjusting the propeller's pitch. To address the challenges of PID-controlled variable pitch, ADRC is integrated into the variable pitch controller, resulting in the design of a speed-loop controlled variable pitch system. Beyond the rated wind speed, the system relies on speed as a feedback signal to maintain stable output power at the rated level. The parameters of the variable pitch controller are self-optimized using the Sparrow optimization algorithm. By utilizing MATLAB/Simulink simulation software, we conduct simulations on the variable pitch controller and its associated algorithm. The simulation results show that when the Sparrow algorithm is implemented in the controller, not only does optimization occur rapidly, but also the convergence speed is increased, further enhancing the system's anti-interference capabilities.

Aiming at the problem that the asynchronous motor used in the submersible LNG pump has reduced locked torque and long starting time at low temperature, the causes of this problem are analyzed theoretically. A scheme of using brass instead of cast aluminum cage rotor winding and reducing the cross-sectional area of cast aluminum rotor winding is proposed, and the two schemes are compared. The corresponding simulation calculation was carried out using Ansys Maxwell, and the dynamic process of motor starting and the rotor winding loss during stable operation were obtained. Theoretical analysis and simulation results show that the improved scheme is simple, practical and effective.

Most power companies are undergoing a critical period of digital transformation, and information systems are an important carrier of digital transformation. The success of digitalization depends on whether the support of information systems is comprehensive and reliable. The operation modes of information systems are diverse, and the functions of information systems are also closely related to power applications. The scheduling of data applications and computing resources in various aspects of power varies greatly, making it difficult to unify scheduling. There are also various types of computing power, such as traditional processors and graphics processors. In response to this issue, this article proposes a heterogeneous computing resource scheduling method for the operation of power information systems. Firstly, analyze the computing resources and arrange the tasks of the computing resources in a unified manner according to the requirements of computing power invocation and task requirements. Secondly, based on the classification of computing resource types in the process of power production and construction, a secondary compilation of widely distributed computing resources is carried out, with a focus on meeting the computing needs of important power nodes. Simultaneously considering the inclusion of cloud computing resources in the computing power resource scheduling system, cloud computing resources are allocated for decentralized small-scale computing power requests to improve computing power scheduling planning. To promote the efficient operation, management, and maintenance of information systems, provide strong support for the development of power production work, and standardize and improve the operation mode of information systems, so that the power information system has high timeliness and reliability.

A fault diagnosis expert system has been designed based on heavy-duty vehicles as the analysis object. This system is an intelligent fault diagnosis system designed based on vehicle maintenance information and expert knowledge, combined with the basic requirements of the expert system. Based on the regularity of fault occurrence and the particularity of fault phenomena, a combination of framework representation and production representation was used to represent knowledge. The process of establishing a fault diagnosis knowledge base and the implementation method of uncertain inference mechanism were analyzed. A fault diagnosis expert system for heavy-duty vehicles was designed using the CLIPS expert system development platform.

In the fault diagnosis technology of electric motors, the vibration signals during operation contain a large number valuable information. The identification of vibration signals can reflect the operating condition of a motor, thereby helping to repair faulty components. The purpose of this article is to establish a vibration-based engine-ball failure diagnostic device. Click in the dynamic input prompt perform default restrictions and given soft threshold denoising on the collected signals, and the denoising effects of the two methods will be compared. Then, signal processing is completed through the analysis method of time-domain feature extraction, which facilitates obtaining effective target feature data, reducing data dimensionality, and facilitating recognition. Finally, these feature parameters were put into the BP artificial neural network for training, and the results showed that the neural network achieved accurate recognition of different types of bearing vibration signals and completed online diagnosis of bearing faults.

Most of the pure electric vehicles currently on the market adopt a single motor single-axis drive configuration. This configuration can be shared with traditional internal combustion engine vehicles in terms of development platform, resulting in lower development costs and difficulty. However, when the power demand is low, the motor efficiency is relatively low, resulting in poorer overall vehicle efficiency. Compared with traditional single motor single-axis drive vehicles, dual motor dual-axis drive electric vehicles have better economy and performance. This article develops a vehicle control system for dual motor dual-axis drive electric vehicles, aiming to fully explore the potential of this configuration and improve vehicle performance. A simulation model based on MATLAB/Simulink and control strategy model were established. After HIL testing of the vehicle control unit, it was confirmed that the VCU function is reliable and meets the requirements.

The large-scale real-time data of the entire network provides complete and reliable basic data for global analysis and decision-making business scenarios, improves the accuracy of online analysis and calculation, and supports the integrated online analysis and decision-making of the large power grid. In the actual operation process, due to various reasons such as model parameter maintenance errors, abnormal equipment operation, and delayed startup of new equipment, data quality levels fluctuate, affecting the accuracy and real-time performance of calculation results. To further improve the quality of real-time data, balance analysis and calculation tools are used to accurately evaluate the quality of basic data in various power grids through real-time balance calculations and statistical analysis of calculation results for equipment such as power plants, topology buses, lines, and main transformers. This provides technical support for operation and maintenance personnel to promptly identify data quality issues and make targeted corrections.

With the continuous deepening of the research on active safety technology of intelligent vehicles, the assembly rate of Advanced Driver Assistance Systems (ADAS) has greatly increased. Active safety performance testing and evaluation for intelligent vehicles is becoming more and more important. However, considering the cost issue, the vehicle-in-the-loop platform test is preferred, and many domestic and international test standards only test and evaluate a single assisted driving function, which is not enough to comprehensively evaluate the strengths and weaknesses of the active safety functions of intelligent vehicles. Therefore, following the actual situation of road traffic in China, we propose the China New Car Assessment Program (C-NCAP) to comprehensively evaluate the active safety functions of intelligent vehicles. Based on the active safety test section of C-NCAP, this study utilizes RTK high-precision positioning test technology for testing to evaluate the performance of 23 vehicles equipped with ADAS. Based on the test results, the active safety system technology of domestic intelligent vehicles is analyzed and evaluated.

At high and low speeds, the back potential is not the same, but the flux linkage of the rotor is basically similar. If the flux link is used as the observation object, the negative effects of SNR and nonlinear factors at low speed can be avoided. Based on the principle of flux model, this paper builds a flux observer model. The test results show that it can run stably at low speed, has strong load capacity, smooth switching at low speed and high speed, and fast convergence at high speed.

Analyzed the current status of automotive waste heat utilization technology, mainly including two methods: engine waste heat refrigeration technology and engine waste heat power generation technology. Refrigeration technology is mainly achieved through three technologies: absorption refrigeration, adsorption refrigeration, and jet refrigeration. Power generation technology is mainly achieved through three technologies: semiconductor temperature difference power generation, turbine power generation, and Rankine cycle waste heat power generation, and analyzed the basic working principles of various technologies. And a comparison was made between the system parameters and structural characteristics of waste heat refrigeration technology and waste heat power generation technology, while a simple analysis was conducted on the technical requirements of the system structure. Finally, based on the common problems of the two technologies, key technologies to promote the development of waste heat utilization were proposed.

Due to the inherent shortcomings of traditional PID control, such as manual adjustment of PID parameters and slow response speed, it is precisely because of these problems that the motor PID control cannot achieve the desired effect. A motor PID control algorithm based on GA optimization algorithm is proposed to address the inherent problems of traditional PID control. Establish GA optimized PID control simulation and fuzzy control PID simulation on the platform, compare the step response curves obtained by the two, and explain the differences between the two and traditional PID control, as well as the shortcomings of traditional PID control that have been improved.

Printed Circuit Board (PCB) is one of the core components of electronic equipment, ensuring the quality and reliability of PCB board is essential for product performance, life and safety. In view of the low detection accuracy and large number of model parameters in the traditional algorithm of PCB, this paper improves YOLOv5 model, using the ghost module instead of the standard convolution layer, to ameliorate the performance, make the model more lightweight. Employ coordinate attention mechanism module, improve the detection of PCB Small goals. Adopt the SIOU loss function, enhance the precision of the model. The empirical findings confirm that the improved YOLOv5 improves the average accuracy by 1.4% over a 47% reduction in parameter size.

With the rapid development of intelligent automobile technology, improving the safety and driving experience of driving at night has become an important research direction. The purpose of this study is to adjust the brightness of headlights in real time by using the sensor data of the surrounding environment of the vehicle, so as to provide lighting effects suitable for the current driving situation. Firstly, through data acquisition and sensor fusion technology, several sensors, including camera, radar and light sensor, are selected and deployed to obtain the information of the surrounding environment of the vehicle. Then, through environmental perception and analysis, the collected environmental data, including road conditions, traffic conditions and lighting conditions, are processed and analyzed to evaluate the lighting requirements of headlights. Based on the analysis results of environmental awareness, an intelligent lighting control strategy is designed. This strategy takes into account factors such as road curve, vehicle speed and vehicles in front, and provides lighting effects suitable for current driving conditions by adjusting the brightness of headlights in real time. Through experiments and simulation, the intelligent automobile headlamp intensity real-time adjustment system in this study shows good performance and effect. The system can automatically adjust the brightness of headlights according to the real-time environment and driving state, provide suitable lighting effect, and improve the safety and driving experience of driving at night.

Deep Learning (DL) can automatically extract useful features from massive data and conduct efficient pattern recognition. This technology provides a new method and perspective for solving many complex problems. Whether it is intelligent perception, decision control or fault diagnosis, DL has shown a broad application prospect. In this article, DL technology is combined with the research on the heat dissipation performance of intelligent networked automobile new energy battery. By constructing and training DL model, the heat dissipation performance of the battery is predicted and optimized, and the prediction results of the model are analyzed and discussed. By comparing the prediction accuracy, error and response time of different algorithms, the results show that the prediction accuracy of this algorithm is improved by 26.64%, the prediction error is reduced by more than 30%, and the response time is also obviously reduced. These results prove the effectiveness of the proposed algorithm in predicting the heat dissipation performance of new energy batteries. The optimization algorithm is of great significance for improving the energy utilization efficiency of intelligent networked vehicles, enhancing system stability, realizing intelligent energy management and promoting the development of autonomous driving technology of new energy batteries.

This paper designs a device for measuring the penetration depth of pile driving on construction sites in response to problems such as high subjectivity and ineffective recording in traditional penetration depth detection. The device uses a laser ranging sensor for non-contact measurement of the pile body, filters the vibration signal through a microprocessor, extracts the effective penetration depth, displays it on the human-computer interaction interface, records the penetration depth data via an SD card, and sends the data to a server via WiFi. This device replaces manual measurement, enabling more scientific and standardized measurement of penetration depth, timely notifying workers to stop hammering when the penetration depth meets the stopping criteria. Field experiments show that the accuracy of the device's hammering- stop prompt is 96.90%, and its detection has high standardization and meets user needs, improving construction efficiency.

To make the durability performance test of car window motors more closely resemble the actual state of use, an environment durability test method for car window motors based on position control is proposed, which derived from the distributed car window motor durability comprehensive performance testing apparatus developed in the laboratory. Utilizing a torque motor as the actuator, position signals and configuration parameters are set through upper and lower computer communication. The lower computer employs the LQR (Linear Quadratic Regulator) algorithm to adjust the controller parameters for position control. Simulation verification experiments demonstrate that the system responds quickly, with the testing's positioning accuracy controlled within ±2°. Simultaneously, based on random position control, the testing and control device can meet various motor durability test schemes, enhancing testing efficiency.

With the increasing demand for clean energy in society, the stable operation and efficient management of New Energy Power Generation System (NEPGS) has become a key issue. This research focuses on the real-time monitoring and fault diagnosis of fNEPGS based on Deep Convolutional Neural Network (DCNN), aiming at improving the reliability of the system, reducing downtime and improving maintenance efficiency. First of all, by constructing DCNN model, the real-time monitoring of NEPGS is realized in this study. The model can extract key features and dynamically adapt to the changes of system operation state by learning large-scale monitoring data. The experimental results show that DCNN performs well under normal working conditions and successfully captures the characteristic patterns under different working conditions. Secondly, this study applies DCNN to fault diagnosis, covering a variety of potential fault modes, such as broken teeth, tooth surface wear and so on. By introducing various fault samples into the training set, DCNN can identify and classify different fault types with high accuracy. This provides feasibility for providing real-time and accurate diagnosis when faults occur. The experimental results show that the system in this study has achieved ideal performance on the verification set, all samples have been correctly classified, and the diagnostic accuracy rate has reached 100%. This provides an advanced and feasible solution for health management and real-time monitoring of NEPGS.

With the continuous development of Intelligent Networked Vehicle (INV), electric vehicles have become one of the important trends in the automobile industry. In INV, battery management system is very important to ensure battery performance and life. In this study, the optimal application of Edge Computing (EC) in INV battery management is discussed, aiming at improving the real-time performance, energy efficiency and user experience of the system. Through data processing and decision-making in the vehicle, EC effectively reduces the dependence on the central cloud server, reduces the data transmission delay and improves the real-time performance of the system. A comprehensive mathematical model is established, which considers the dynamic characteristics, charging and discharging process and multiple optimization objectives of the battery. In order to solve this optimization problem, this paper adopts Adaptive Genetic Algorithm (AGA) to find the optimal solution in complex and nonlinear problems. Through the solution of the model, the feasible battery management strategies in the actual scene are obtained, which can balance the needs of users, maximize battery performance and prolong battery life. The research results show that with the support of vehicle EC, the battery management system can respond to different driving scenarios more intelligently, improve energy efficiency, prolong battery life and provide a better user experience. The conclusion is of positive significance for promoting the further development of INV technology and the continuous improvement of intelligent transportation system.

Ships are prone to localized structural damage under the adsorption of automatic mooring equipment. In order to reduce the damage of the ship structure during adsorption, a finite element model of the ship adsorption area should be established to analyze the structural strength and dynamic response of the ship adsorption area during the design of the ship. Apply the adsorption load to the ship adsorption part structure model and design three working conditions to dynamically simulate the adsorption process of the ship of the automatic mooring equipment. The structural strength and dynamic response of the ship during the adsorption process are analyzed. The results show that according to the maximum relative deformation and ultimate strength, the ship strength is ranked from low to high, and it is concluded that when the ship is adsorbed at berth, the more the ratio of the ship plate dimensions in the direction of the length and width tends to be equal, the better the stress is; the more intensively the reinforcement of the ship plate is, the better the stress is. Characteristics of adsorption change: when the ship structure comes into contact with the automatic mooring equipment, the collision force starts to rise rapidly and reaches the peak, which makes the ship structure undergo the maximum deformation, and then the force and deformation of the ship structure tends to stabilize and tends to a fixed value.

In response to the issues of insufficient depth and breadth of equipment monitoring and significant differences in various data collection, research is conducted on the data diversion and collection architecture technology for the full business scenario of substation equipment monitoring. A partitioned distributed and elastic collection framework is proposed, and on-demand data penetration and retrieval technology is studied to support remote transparent access of various businesses to substations and integrated comprehensive monitoring of substation main and auxiliary equipment.

This research delves into an innovative approach for diagnosing faults in aero-engine airways using a multimodal deep neural network. The method involves creating a sophisticated neural network model capable of precisely detecting and categorizing various airway faults within the engine. This is achieved by synergistically integrating data from multiple sensory modalities, such as sound and vibration, to enhance the accuracy of fault identification and classification. The study conducted in-depth analysis and optimisation in data preprocessing, feature extraction and model design to improve the performance of the diagnostic model. Experimental results show that the proposed multimodal deep neural network method exhibits high accuracy and reliability in the diagnosis of aero-engine airway faults, and has potential for practical application.

Electric power project management informatisation systems play an important role in improving the efficiency and accuracy of project management in the power industry, optimising resource allocation and enhancing data quality. This paper introduces the composition of the power project management informatisation system. It also points out the problems of project management informatisation in electric power enterprises. This paper also introduces the advantages and benefits of the power project management informatisation system, analyses the reasons why the system improves data quality, and discusses the development trends and challenges of the power project management informatisation system. Finally, it supports that by applying the power project management informatisation system in power enterprises, the power industry can better cope with the challenges in project management and achieve the goal of sustainable development.

Deep learning algorithm has been widely used in the field of industrial equipment fault prediction, and it is favored because of its excellent performance on complex data sets. The purpose of this study is to develop a fault prediction model of inverter and high voltage motor based on deep learning algorithm and conduct detailed simulation research. In this study, Sparse Autoencoder Network (SAN) is used to predict the faults of inverter and high-voltage motor, so as to effectively capture the characteristics in different time scales. Through the training of large-scale data sets, our model can accurately predict the faults of inverter and high-voltage motor, including circuit faults, motor faults and sensor faults. The research results show that the SAN proposed in this paper can improve the diversity of data sets by increasing the number of samples through the deep learning data expansion model, on the other hand, the noise data generated by errors can enhance the data sets, and the prediction efficiency reaches 97.411%. This model is expected not only to improve the reliability and availability of industrial equipment, but also to reduce maintenance costs, reduce production downtime and improve production efficiency.

This paper firstly summarizes the traditional methods of power fault diagnosis in recent years. It is found that the traditional manual method is greatly affected by the work experience and professional ability of the inspectors, and the diagnosis time is long, and the accuracy rate is low. Then, an electrical equipment identification method based on East algorithm is proposed, and based on machine learning technology, the type of identification of electrical equipment in substations is deeply studied. According to the different characteristics of the network model and the characteristics of experimental data, a neural network based on GRU is constructed. The fault diagnosis model is based on the GRU model, and the experimental diagnosis analysis is carried out based on the GRU model.

This design is a 2.4G wireless remote-control car based on STM32, including motor drive module, steering gear module, ultrasonic obstacle avoidance module, OLED screen display module, key input module, camera acquisition module and 2.4G wireless transmission module. L298N motor drive module to realize the forward and backward movement of the trolley; SG90 steering gear can realize the steering function of trolley; HC-SR04 ultrasonic measuring distance achieves obstacle avoidance function. The wireless remote-control module has two remote control modes: NRF24L01 wireless transmission and ESP32WIFI. In this design, the rear wheel is powered by two motors, and each module and the main control unit STM32 single chip microcomputer complete the corresponding functions, which can control the car to move forward, backward, turn and stop, respectively, through the mobile phone APP or remote control. The two remote control methods in this design provide an extra guarantee for solving the problem that the robot is out of control, and the control is safer.

Since the 21st century, China's manufacturing industry has developed rapidly, and behind this rapid development, is the rapid development of machine processing equipment. The promotion of mechanical equipment is an important support to promote China from "manufacturing power" to "manufacturing power", and the quality of equipment is an important indicator of a country's industrial development level, is the key to improve a country's competitiveness. CNC machine tool belongs to one of the mechanical processing equipment, it is mainly through the mechanical principle, electrical control and hydraulic transmission principles to achieve the processing of parts. The overall accuracy of CNC machine tools is high, and the flexibility is good, and there are more applications in some important parts processing fields, such as parts processing of mechanical equipment, parts processing of large automation equipment, and even aerospace parts processing. Because the electrical control of the feed shaft will affect the quality of the whole equipment, so the study of the electrical control of the feed shaft and its fault diagnosis has important practical significance. The first part of the article is a review of the research background and status quo of CNC machine tools at home and abroad and determine the research objectives and purposes of this article. Secondly, it briefly introduces the structure and composition of CNC equipment involved in this article and analyzes the function of the electrical control system in CNC equipment. At the same time, some common faults of CNC machine motor are introduced briefly. The diagnosis and solution of electronic fault of a CNC machine tool are also briefly summarized.

In order to improve the visual effect of ceramic design, this paper analyzes the visual aesthetic feeling of ceramic design with intelligent computer color technology. Based on the expression of color emotion in color psychology, this paper divides ceramic color into mobile application color system with the color design of mobile application and puts forward a method of color visual style transfer about mobile application interface, so that designers can change the color visual style of mobile application interface at will. Through the emotional state quantification experiment, the paper discusses that different ceramic color design schemes are the internal mechanism of user's emotional change, constructs the quantitative relationship between emotional state and objective physiological characteristics under different ceramic color attribute inducing conditions, and puts forward a related spatial color transmission method based on intrinsic image decomposition. From the experimental results, it can be seen that the ceramic color art design system proposed in this paper has good effect and can effectively improve the simulation design efficiency of ceramic art.

In this paper, the application of Virtual Reality (VR) technology in college Chinese teaching is deeply studied, aiming at discussing and evaluating the influence of innovative teaching methods in VR environment on college Chinese subject. Taking the application of VR technology in the teaching of ancient literature in universities as a case, through empirical research and data analysis, we discussed the effect of new teaching methods in promoting students' learning enthusiasm, deep understanding and creativity. The empirical research results show that under the VR environment, students' study duration and the frequency of dialogue and interaction are significantly improved, and students feel the scenes of ancient literary works through VR, which stimulates their strong interest in literature. This not only improves students' learning enthusiasm, but also deepens their understanding of the cultural connotation of literary works. The VR environment creates an immersive learning experience, which makes the learning process more vivid and interesting, and improves students' practical application ability and comprehensive understanding of literary works. The innovative teaching method of college Chinese in VR environment brings new possibilities for Chinese education and provides useful reference and enlightenment for the future of subject education. This new way of learning is expected to cultivate Chinese professionals with more creativity and practical application ability, and inject more vitality into Chinese education.

This article introduces the widespread application of DL (Deep Learning) technology in VR (Virtual Reality), and then points out the challenges that exist in virtual environments and virtual character image generation tasks, such as the complexity of factors such as lighting, materials, textures, expressions, and actions. Furthermore, the main objective of this study was clarified, which is to achieve realistic virtual image generation through DL technology. The research questions include how to improve the realism and realism of virtual environments, as well as how to achieve realistic expression and action generation of virtual characters. These issues highlight the importance and potential applications of research. Subsequently, a brief introduction was given to the research methods, particularly the use of autoencoder models to learn potential representations of virtual environments and virtual characters, in order to achieve realistic image generation. The selection of these models highlights the crucial role of DL technology in image generation tasks.

Bird classification is essential in zoology, computer vision, and artificial intelligence. This paper combines the Inception V3 model with transfer learning to create an effective method for classifying bird images. In addition, this paper observes how the model is affected by the number of epochs. Some top convolutional layers are fine-tuned to perform better after training the fully connected layers. Meanwhile, training the model with more epochs to observe the influence of epochs on model loss and accuracy. This research is conducted on the "BIRDS 525 SPECIES" dataset, and the results indicate that the new model achieves good classification results with relatively few epochs. Fine-tuning the model further enhances its generalization ability, and it is observed that due to maintaining most weights unchanged in transfer learning, more epochs cause the model to converge without a decrease in accuracy, effectively addressing overfitting issues. The combination of transfer learning and convolutional neural networks allows the model to perform well on small datasets at a lower cost while improving overfitting issues, making it a promising approach for future applications on various platforms and small-scale devices.

Deep convolutional neural networks have made groundbreaking progress in various fields of computer vision. However, the high storage and processing expenses associated with building these networks limit their usefulness in low-powered devices like those found in the Internet of Things and smartphones. Network compression and acceleration techniques have emerged to address this issue by reducing redundancy in the network to compress network parameters and speed up computation. This paper proposes a small-sample convolutional neural network compression and acceleration algorithm. Although pruning algorithms can effectively identify redundancy in the network, they, like most network compression and acceleration algorithms, require a large amount of training data. However, sufficient training data is not always available due to privacy, copyright, and storage issues. To address this, this paper proposes a small-sample convolutional neural network compression and acceleration algorithm based on feature reuse and enhanced knowledge distillation. This algorithm only requires a small number of unlabeled samples to sufficiently compress the network. Specifically, the algorithm first establishes a compressed network prototype by locally reusing the main feature maps of the original network. Then, it further refines the established network prototype features by using an enhanced small amount of samples to detect the input-output structure of the original network from a global perspective. Extensive experiments conducted on benchmark datasets demonstrate the effectiveness of this algorithm.

With the popularization of mobile communication applications, image transmission plays an increasingly important role in communication. However, due to the limitation of communication bandwidth and storage resources, efficient image compression has become an urgent need. The purpose of this study is to design an innovative mobile communication image transmission compression algorithm by introducing the deep learning technology of Convolutional Neural Network (CNN). In this algorithm, through the feature learning of convolution level and the hierarchical structure of CNN, the abstract features in the image are gradually extracted. This feature extraction process can not only capture the texture and structure information of the image effectively, but also has the advantage of resisting the common information loss problem in traditional compression methods. In order to enhance the learning ability of the network, we introduce residual connection, which improves the network's long-range dependence and attention to important areas. Experimental results show that compared with traditional compression methods, the algorithm based on CNN can maintain higher image quality at the same compression rate. The algorithm is more universal and can meet the needs of mobile communication image transmission in different types and scenarios. In addition, the impact on transmission quality is relatively small, and it can maintain good visual effects even at high compression rate.

Reflective prediction model is an important means to control printing quality and reduce energy consumption. In order to improve the printing quality control technology and reduce the energy consumption of pre-press proofing, a new spectral reflectance model that can predict the dot printing quality is studied. The optical phenomenon of the light beam in the ink paper was analyzed. The optical parameters affecting the change of luminous flux were measured. The entire process of light propagation in the printed material was simulated. By comparing the simulated dot data with the optical microscope value, the effectiveness was evaluated. The experimental results show that the spectral reflectance of printing dot can be effectively obtained in the whole wavelength band from 400nm to 700nm of D65 light source. The method of simulating light beam propagation can meet the requirements of color printing quality control. This paper combines the optical properties of printing materials and computer simulation technology to study the spectral reflectance model of amplitude modulation dots. The spectral reflectance model can realize both quantitative and qualitative evaluation of color printing dots through one simulation, which broadens the quality prediction method and control technology.

Dehazing rainy aerial images poses a challenging problem, as traditional dehazing methods have limited effectiveness under complex lighting conditions and various scenes. Therefore, an adaptive dehazing method for rainy aerial images based on deep learning is proposed. This method utilizes deep learning networks to learn the relationship between haze and foreground information in aerial images and removes the impact of haze by restoring clear images. To adapt to different lighting conditions and scenes, adaptive techniques are introduced in the network design and optimization process to achieve better adaptability and robustness. Experimental results demonstrate that this method has achieved significant improvements in dehazing quality and exhibits strong robustness across various scenes.

With the rapid development of science and technology, Virtual Reality (VR) technology based on image processing shows great potential and innovation in the field of interior design. The purpose of this paper is to explore the application of VR technology based on image processing in interior design system and analyze its potential impact on interior design field. Through the in-depth study of real-time rendering, dynamic layout adjustment, virtual and real space integration and other functions, we show the outstanding achievements of this technology in improving design efficiency, optimizing user experience and promoting design innovation. In the aspect of technological innovation, this paper emphasizes the influence of the continuous progress of image processing technology on the real-time rendering of interior design system. Through an example, we show that users can perceive the design effect in real time in the virtual environment and realize a more intuitive and immersive design experience. In addition, through dynamic layout adjustment, users can modify the indoor layout in real time through gesture recognition technology, thus participating in the design process more flexibly. The integration of virtual and real space is another key research point, and the deep integration of virtual furniture and design elements with real space is realized through VR technology. This enables users to personally experience the design effect in the actual environment and provides designers and customers with more realistic visual feedback. The research of VR technology based on image processing in interior design system not only provides technical innovation, but also brings more efficient, smarter and more interactive design experience to the interior design field, injecting new impetus into digital transformation and intelligent development.

The aim of this research is to explore methods for building architectural virtual scenes based on image processing techniques to meet the needs of the fields of architectural design, visualization and simulation. With the rapid development of digital technologies, the creation of architectural virtual scenes has become increasingly important as they provide a real-time, immersive way to present and evaluate architectural projects. Firstly, this study discusses the fundamentals of architectural virtual scenes, including the creation of virtual building models, the capture and fusion of real scenes, and the implementation of user interaction. Through image processing techniques, real architectural environments can be combined with virtual architectural elements to provide users with a highly realistic visual experience. Secondly, the study covers the application of image processing methods that help to process real scene images to make them suitable for embedding and interacting with virtual architectural models. This study highlights the potential of architectural virtual scenes for application in architectural design and visualization. With virtual scenes, architects and designers can make real-time design and layout adjustments, while all aspects of the architectural project can be visualized, including the exterior, structural and interior design. Finally, this study summarizes the importance of image processing in the construction of architectural virtual scenes, while exploring future research directions such as the development of augmented reality technology, improvements in virtual-reality fusion and user interactivity. Research in these areas will continue to drive innovation in architectural virtual scenes, providing more possibilities and flexibility in the field of architecture.

Optical Character Recognition (OCR) is a technology that integrates optical and computer technologies to convert printed text into machine-readable text. The process begins with the conversion of printed characters into a binary (black and white) image, achieved by detecting the light and dark patterns of each pixel in the document. Subsequently, OCR software applies recognition algorithms to this binary image to identify and convert the characters into digital text. This digital text is then formatted into a computer-readable format, enabling further processing and editing. This technology is instrumental in digitizing printed documents, automating data entry processes, and enabling text searches in scanned documents. In the daily information office, it is often necessary to recognize valid information in text images. This paper proposes a thresholding method for text optical character information translation recognition based on OCR technology, and uses a text image thresholding image processing method based on histogram analysis and the OTSU algorithm to segment the optical character information in text images. In the segmented foreground, a thresholding optical character information translation classifier based on OCR technology is used to realize the thresholding text optical character information translation recognition. It is verified that the proposed method can achieve high accuracy and efficiency in thresholding text optical character information translation recognition, and the recognition performance of the method is remarkable compared with other recognition methods.

This paper explores the potential of Generative Adversarial Networks (GANs) in icon generation for digital product design. Icons are essential visual elements in user interfaces, facilitating communication and enhancing user experience. Traditional icon design methods often require considerable manual effort and time. By harnessing the power of GANs, this study investigates how automated icon generation can streamline the design process and improve the quality and diversity of icons. This study concludes the building of the model for generating icons by integrating the generative adversarial network model with the icon design procedure itself. In order to solve the issue of creating icons automatically, this study conducts the following studies. 1) A Multi-Feature Icon Generation Network (MFIGN) is suggested, with its foundation in the conditional classification generative adversarial network. To guarantee that the icon produced by the model satisfies the supplied conditional feature, a multi-feature identification module has been introduced to the discriminator. 2) The icon generation model suggested in this paper outperforms existing methods on the icon dataset when it comes to the design of a wide range of icon feature expressions.

With the continuous progress of digital medicine and smart medical technology, the application of computer vision technology in medical image processing is also constantly developing and advancing. Image fusion technology in computer vision technology can complement the advantages of medical imaging through effective algorithms, discover useful and valuable information in medical diagnosis, and it is very effective in compensating for the shortcomings of medical image presentation technology and the lack of image information. The fusion of CT medical images and MRI medical images using computer vision technology can combine the advantages of CT images displaying clear bone information and high image resolution with the advantages of MRI images displaying clear soft structures, forming complementary advantages, thus making the display information of medical images more abundant. This article mainly utilizes wavelet transform algorithm to fuse CT and MRI anatomical imaging and proposes a medical image classification and diagnosis method based on local spatial sequence Recurrent Neural Network (RNN), extracting higher-level semantic features of images with better representativeness. This method extracts the local spatial sequence features of the samples and utilizes a temporal RNN to highly integrate and abstract the local spatial information. While obtaining high-level semantic features, it enhances the role of useful pixels in the local space and suppresses the influence of useless pixels. The experimental results show that the proposed method improves classification accuracy.

This paper studies the application of pose estimation algorithm in ballet movement analysis. In view of the complexity and diversity of ballet movements, traditional analysis methods often fail to accurately capture key movement and posture information. Therefore, in this study, a new pose estimation algorithm was proposed based on key point correlation modeling and full-view data expansion techniques, and a deep neural network structure was used to extract and classify ballet movements. Through the training network, the input ballet movement data is converted into key point coordinates and the exact position and pose information for each key point is predicted. Experimental results show that compared with traditional methods, the new pose estimation algorithm based on key point correlation modeling and full- view data expansion can more accurately capture the details and features of ballet movements, and the correct rate of pose analysis is increased by 7.76%. In summary, the pose estimation algorithm based on key point correlation modeling and full view data expansion technology provides an effective solution for ballet movement analysis. In the future, the algorithm is expected to play an important role in the fields of dance teaching, dance evaluation and art research.

Image recognition and classification have found extensive applications in the field of artificial intelligence. This study conducts a comparative study of conventional Convolutional Neural Network (CNN) and the Visual Geometry Group Network (VGG16) to explore the effectiveness of model classification. The aim is to comprehensively understand the interactions among different hierarchical image feature representations in scenery classification tasks, the models' performance in complex feature learning, and non-linear expression capabilities. The experiments conducted on the "Intel Image Classification" dataset demonstrate the superior accuracy of the pre-trained VGG16 model over the CNNs model, particularly in terms of feature learning and generalization capabilities. Moreover, this comparative analysis approach enhances the understanding of the characteristics and suitable scenarios of different network architectures. It guides the selection, design, and optimization of deep learning models for practical image classification applications. Therefore, the study makes a significant contribution to the field of scene image classification and offers practical implications. It establishes a foundation for future research directions that will emphasize innovation in models, integration of multimodal data, and improvement in robustness and interpretability, aiming to drive breakthroughs and applications of image recognition and classification technology across various fields.

One of the main goals of material design is to sift the proper materials with the properties we want. However, the traditional method, synthesizing and testing each material in laboratory, wastes time and energy, and the actual material we want is usually one in a million which makes it more difficult. Here, we develop a generative framework to give a guidance on material design with specific properties. Our framework is mainly drove by several variants of Generative Adversarial Networks (GANs) for material data generation. Our framework is trained with 86 perovskite-type material samples including their components information, and then we compared with various networks structures and algorithm, the result shows an acceptable accuracy of materials data generation which proved a possible method of inverse design of perovskite-type electrode of SOFC.

For the traditional algorithm to detect lithium battery defects, the missing rate is high and the speed is slow, an improved YOLOv7 algorithm was proposed. Firstly, CBAM attention mechanism is added to feature extraction part, which can enhance network's representation ability. Secondly, in the feature fusion part, ConvNeXt lightweight module was used to replace the original ELAN module to reduce the model's complexity. Finally, SPD module is added before the detection head at the output end to increase focus on smaller goals with surface defects of lithium batteries at the output end. The results show that the optimization algorithm can improve the accuracy and speed of the lithium battery. The proposed algorithm achieves a 92.7% detection accuracy, surpassing the original network by 2.1%.

In order to make the best use of solar energy and ensure that solar cells are always working close to the point of maximum output, A MPPT algorithm combining the fixed voltage method and the conduction enhancement method of the eddy current function is proposed to solve the problem of balancing accuracy and tracking speed in the current MPPT algorithm. The algorithm first adopts the fixed voltage method maximum power point monitoring quickly at startup, and then to increase the tracking speed and steady-state accuracy, the following stages use the method of optimizing the conductivity gain by the inverse tangent function. In this paper, the simulation results are compared with the traditional incremental conductance method in the same environment. The simulation results show that the tracking time and accuracy of the fixed voltage method and the improved variable step conductance increment method before and after the illumination change are 0.011s and 0.04 s and tracking accuracy is 0.3% and 0.1% after the illumination change are respectively. The tracking time and accuracy of the traditional conductance increment method before and after light change are 0.015s and 1.6%, respectively, after light change is 0.25s and 1.6%. Therefore, the control strategy of the conductance increment method optimized based on the fixed voltage method and the inverse tangent function has faster tracking speed and steady state accuracy.

Aiming at the problems of slow labor efficiency and low precision of lithium battery defect detection, a new method based on shear wave threshold and K-means clustering segmentation was proposed. For the defect image, the ROI region of interest was first extracted to obtain the polar slice region containing only the defect parts. Then, the image denoising was carried out with the shear wave threshold and the histogram normalization was carried out with image enhancement processing to obtain the image with low noise and high contrast between the defect and the background. The defect was extracted by K-means clustering segmentation algorithm. Morphological corrosion and expansion operations were carried out to fill the defects. Finally, the prewitt edge detection operator was used to detect the defects and output the defect information. The experimental results show that the algorithm can detect metal leakage, damage, black spots and scratches very well, which is suitable for industrial production.

In recent years, with the increasing number of power users and the improvement of power consumption flexibility, how to cluster load curves effectively becomes more and more important. Therefore, for the problem that a single clustering analysis method cannot achieve accurate clustering of power load curves, an algorithm combining K-means and Agglomerative Hierarchical Clustering is applied, and the clustering effect is optimized by the Gap Statistic method. In this algorithm, the centroid is updated iteratively by using a mixture of the two methods to get the best clustering effect. Finally, the actual load data of a power grid in 2020 was selected for cluster analysis and compared with K-means clustering algorithm and agglomerative hierarchical clustering algorithm respectively, the advantages and effectiveness of the combined algorithm are demonstrated.

The distribution network is more important of the electric system that is directly connected to users, so whether the distribution network can operate stably is a very noteworthy issue. With the maturation of new energy generation grid connection technology, distributed power sources represented by photovoltaic installation are increasingly widely used in power systems. On the one hand, when a difficult occurs the distribution network system, distributed power sources can supply power of the load, thereby improving the safety of the distribution network; in the other words, due to the intermittency volatility distributed power sources, they also bring uncertain factors to the power system. However, the demand for power quality from power users is increasing, and there is a gap between the perceived reliability of electricity consumption by power users and the reliability indicators of distribution network power supply. With the many advantages of distributed power supply, such as economy, reliability, security, stability, etc., the traditional methods of distribution network reliability research are no longer suitable for the present time. Here, the reliability of distribution network with distributed power supply is taken as research object, the sequential Monte Carlo method is adopted, and the reliability assessment algorithm with distributed power supply is proposed. Finally, the reliability assessment of the main feeder F4 of IEEE RBTS BUS6 system algorithm on Matlab, which demonstrates that the reliability of distributed power supply can be improved when it accesses to distribution network.

In the process of microgrid operation, there are many unstable factors in the power generation mode such as wind and solar. To solve the problem of improving the processing capability of the micro-grid with cooling, heat and power supply system, and to cope with the power fluctuation in the process of wind and solar energy, this paper plans to carry out the research on the flexibility optimization dispatch problem of microgrid. During the whole operation of the micro-grid, the cooling, heat and electricity system has the characters of complementary supply and demand, as well as supplementary cooling and thermal load. This paper uses a variety of flexible resources on the energy supply side and demand side to establish the balance. The balance is co-ordination of supply and demand and analyzes the flexibility. Then, the model with the lowest operating cost, the lowest cost of environmental pollution control and the highest flexibility of the microgrid system is established. The example calculation shows that this optimized low temperature combined power generation system can improve the flexibility of the used system, and on this premise, the reducing the system of environmental pollution cost and the operation cost.

This paper mainly focuses on the optimal design of the heliostat field of the tower solar thermal power plant, and establishes the optical efficiency model and the layout scheme model to make the output power meet the demand, higher efficiency, and cost saving. Firstly, the atmospheric transmittance is obtained from the Euclidean distance and atmospheric transmittance formulae based on the solar mirror position, collector position and other parameters. The sun's altitude and azimuth are determined from the known geographic location data and the corresponding time point. Based on the solar altitude and azimuth angles and the above location parameters, the cosine loss can be directly solved by using the vectorial quantity product formula. The elevation angle and azimuth angle parameters of the heliostat are obtained by using the method of solving the normal direction angle of the plane reflector in three-dimensional space. Then establish the shadow shading efficiency calculation model and collector truncation model respectively and solve the shadow shading efficiency and collector truncation efficiency. Through the traversal algorithm, the optical efficiency and thermal power output per unit mirror area are obtained by summing and dividing by the number of mirrors sampled as well as the time value to obtain the monthly and yearly averages. The average annual optical efficiency was 49.1658% and the average annual thermal power output per unit area of mirror was 0.48 kw/m².

A fever is a prevalent sign of numerous illnesses, such as the COVID-19 that ravaged a few years ago or influenza. The demand for non-contact body temperature measurement technology is increasing with the development of technology. Infrared thermometers may be a better choice for screening body temperature. However, multiple variables can impact the thermometer's measuring accuracy, such as the measurement distance, forehead contamination, and measurement position. To overcome these factors, this paper designed and constructed a digital infrared thermometer. The thermometer is based on an Arduino board and comprises a MLX90614 sensor, a LCD1602 display screen, and a MH- FMG buzzer and so on. The forehead temperature of 20 participants of different ages and genders was measured using an infrared thermometer and the data obtained was analyzed by drawing a linear fitting function. The performance of the final assembled thermometer was calibrated and verified by experiments. The infrared thermometer might achieve accurate forehead temperature measurement within a short distance (error less than or equal to ±0.3°C), which could be similar to the performance of commercial thermometers. The authors provided a simple, low-cost, and efficient solution for the development of non-contact body temperature measurement technology, which could have important significance and value for preventing and controlling infectious diseases caused by fever and ensuring public health safety.

Realizing the interconnection, interaction, and precise regulation of adjustable resources in the source network, load storage, is the key to expanding the control mode of the new power system and establishing flexible, reliable, and economical control methods. This article focuses on the demand for interactive response of adjustable resources in the source network load storage and provides the architecture of the interactive control system for the source network load storage interconnection, as well as the integration and management of adjustable load information. A decision scheduling method for large-scale adjustable resource demand response is proposed to support the interactive response and optimization control of multiple types of resources. Finally, the effectiveness of the proposed method was verified through case analysis.

Compressed Air Energy Storage (CAES) systems have energy interfaces for multiple forms of energy, such as cold, heat, and electricity, and can therefore be combined with Combined Cooling, Heating, and Power (CCHP) microgrid applications, thus allowing both systems to fully utilize their respective advantages. However, the principle of multi energy coupling of CAES itself is complex, and the combination with CCHP microgrid makes the system architecture and technical implementation of the new system produce a lot of technical problems. Therefore, this report presents control strategies for CAES connected to the CCHP microgrid. Thus, the stable operation of the system can be achieved, and the supporting theory and practical application guidance can be provided to promote the development of CAES and combined CCHP microgrid. Firstly, the basic structure of CAES and CCHP microgrid is described, and thermodynamic models of key components are established. In the meantime, analyze the operation mechanism of internal combustion engine generator, wind turbine, photovoltaic and other energy supply components in CCHP microgrid system. The mathematical models suitable for optimal control are established. Through the above analysis model, it can provide a strong basis system characteristic analysis and optimal control. Secondly, the efficiency of the CAES system, the state and power of the storage, and the generation efficiency of the internal combustion generator are analyzed. Finally, a two-level optimization model is proposed, where the upper level is used to solve the optimal storage pressure initial value for the CAES system cycle. According to the characteristics of the compressed air energy storage system, NSGA-II is used to solve the output scheduling optimization problem in the lower level with economy, environmental protection and energy saving as the unobjective under the condition of determining the optimal storage initial value. The simulation results show that the proposed control strategy is reasonable and feasible.

In this study, a comprehensive electronic component performance optimisation system is constructed using an improved Genetic Algorithm (GA) and a reliability allocation process method based on a cost function is proposed to further improve the reliability design of electronic circuit systems. This method is based on the reliability design of electronic circuit systems, which considers factors such as component lifetime, reliability and failure rate. By optimising the reliability allocation, the study can achieve more reliable electronic circuit systems, reduce the risk of failure, increase the product lifetime, as well as reduce maintenance and repair costs. Not only does it improve the performance optimisation capability of the genetic algorithm, but it also provides an effective methodology for the reliability assignment process, which offers a more comprehensive solution for the design of electronic components and circuit systems. This will promote faster, more reliable and cost-effective product development in the field of electronics technology, helping to meet growing market demands and reliability standards.

To cater to the escalating demand for information transmission in today's digital society, prioritizing the development of high-speed wireless communication technologies is essential. Among these, Orthogonal Frequency Division Multiplexing (OFDM) has emerged as one of the most widely adopted multicarrier technologies in the communications arena. OFDM stands out with its high transmission rate and robust anti-interference capabilities. However, it has certain limitations, including low spectrum utilization and a lack of flexibility, which restrict the growth of communication technology. In response to these challenges, Filter Bank-based Multicarrier (FBMC) technology, a cornerstone of the fifth generation (5G) of mobile communication, presents itself as an effective solution to existing multicarrier technology issues. By leveraging Offset Quadrature Amplitude Modulation (OQAM), the FBMC-OQAM system boasts impressive resistance to subcarrier interference and high spectral efficiency. Although the FBMC-OQAM technique is still in its nascent stages, it holds promising potential. Its adoption can significantly enhance communication systems' performance by addressing interference and spectrum efficiency issues plaguing current technologies. Consequently, it is anticipated to become a cornerstone technology in mobile communication's future landscape. As we explore its full potential, the ongoing research in this area is expected to lead to breakthroughs that could redefine the way we think about wireless communications.

This device adopts advanced semiconductor refrigeration technology and intelligent battery management system, which has efficient energy utilization, reliable water production ability and superior environmental adaptability. The design concept of the device includes efficient energy use and water extraction. It is solar-powered and equipped with flexible solar panels that convert sunlight energy into electricity to drive the water-making process. At the same time, the smart battery management system accurately regulates the charging and discharging speed of the battery to ensure the stable operation of the water production process. Through the semiconductor refrigeration technology, the device can quickly condense water vapor into water droplets under the cooling effect of the refrigeration surface and collect into the device through the catheter. This innovative water production principle makes the device highly efficient and adaptable. Through the analysis of a series of experimental data, it is found that the device can produce an average of 200 ml of drinking water per minute, with high energy efficiency and excellent water production efficiency. Compared with the traditional way of obtaining water, the device has obvious advantages in terms of energy utilization, water production capacity and environmental adaptability.

In today's situation of increasing power load, in order to better promote the utilization of wind energy and further balance power production and market demand, an accurate, efficient and stable prediction model is needed to solve the complex problem of wind power generation. This study introduces a deep learning framework that employs a Bidirectional Long Short-Term Memory (B-LSTM) network. This network discerns the linear associations among various feature variables affecting the output power via the Pearson correlation coefficient. Additionally, it utilizes the Local Outlier Factor (LOF) to identify and filter out data points that are outliers relative to the core dataset. This study presents a dataset compilation from a Shanghai-based wind farm, integrating measurements of wind velocity, temperature, humidity, barometric pressure, and the actual power output from wind turbines at varying tower heights. The model was constructed within the Python execution environment, facilitating ultra-short-term forecasting. Following multiple simulations and debugging iterations, performance was assessed using metrics such as Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). A comparative analysis with alternative electricity forecasting techniques demonstrated the superiority of this approach.

As a kind of electronic equipment, the control and protection device for VSC-HVDC may fail under the influence of electromagnetic interference, which will bring security risks to VSC-HVDC system, so it is necessary to carry out the research on electromagnetic interference of the device. This paper summarizes the forms, types and sources of electromagnetic interference, points out that common-mode and conducted interference has the greatest impact on the device. The vulnerabilities of the device concerning EMI are analyzed, with simulations on cable paralleling and chassis shielding suggesting that analog small signals within the device are most susceptible to EMI. According to the existing weaknesses, the corresponding optimization design scheme is proposed, focusing on the propagation path of electromagnetic interference, the shielding measures and the improvement method of the forward loop of the analog-to-digital converter are given, and the selection ideas of operational amplifiers and filters are provided from the perspective of noise characteristics and signal bandwidth. This paper provides theoretical value for the analysis of electromagnetic interference weaknesses and the optimization design scheme of the device.

With the rapid development of modern logistics, retail and other fields, the bar code identification system plays a vital role in the automatic identification and tracking of items. This paper discusses the key application of photoelectric information technology in barcode identification system, introduces the technology of photoelectric sensor, image processing and pattern recognition, and its role in improving the efficiency and accuracy of barcode recognition. We prove the feasibility and superiority of photoelectric information technology in the barcode identification system.

With the rapid development of power system, the scale of power big data has become increasingly large, and the detection of power consumption anomalies has become an important part of power system management and maintenance. Therefore, this paper proposes a solution of power big data anomaly detection system based on deep learning network. In order to realize the efficient operation of the system, four modules of power consumption fluctuation calculation, embedded vector mapping, feature signal extraction and random forest anomaly discrimination are studied and designed. The experimental results show that the system has excellent performance in the accuracy of abnormal detection, real-time early warning and abnormal event recognition, and has high practical value, which is expected to provide a strong guarantee for the safe operation of the power system.

In light of the world's escalating energy crises, building a smart grid has gained significant attention in contemporary times. Power engineering technology is a crucial tool for building a smart grid since it can enhance energy utilization efficiency, lower energy consumption, and safeguard the environment. In order to strongly promote the sustainable growth of smart grid construction, the primary focus of this article is on the application of power engineering technology in smart grid construction. This essay provides an overview of the features and capabilities of smart grids, examines the role power engineering technology plays in their development, and examines how power engineering technology is applied in smart grid construction from the perspectives of information security, energy interconnection, distribution and consumption, and power generation and transportation.

With the rapid development of solar energy technology, machine learning based research on solar production prediction has become a key area to address the efficiency of solar power systems and sustainable energy planning. Using a variety of machine learning models, such as decision trees, support vector machines, and neural networks, this research delves deeply into the generation of solar energy. The experimental data covers different photovoltaic power plants, including polycrystalline solar panels and thin film solar cells, with a total generation capacity of 10 MW. The study firstly emphasizes the importance of model selection and performance evaluation, involving consideration of model complexity, computational efficiency and prediction accuracy. For the assessment of model performance, the Root Mean Square Error (RMSE) metric was used. The results show that the integrated model is able to produce accurate predictions on different PV plants, but the model performance may be affected by data clarity and quality. On the other hand, factors such as geographical location, PV plant characteristics and meteorological conditions also play a key role in model performance. The experimental results show that clear and accurate data are essential for improving model performance, while different meteorological conditions and geographical locations may lead to differences in model performance. In addition, the feasibility in terms of practical applications needs to be thoroughly considered in further research. This study provides important insights into machine learning approaches for solar energy production forecasting, which is expected to provide effective guidance for the planning and management of future renewable energy systems.

With the escalating energy crisis, there has been rapid development in the utilization of solar energy and photovoltaic power generation technologies. The enhancement of photovoltaic system efficiency has become the primary driving force for solar energy development. However, Because of the effects of sunlight and temperature magnitude on the environment, the generation of electricity using photovoltaic (PV) modules is highly stochastic. To optimize the photoelectric converting capacity of PV electricity production systems and make better solar energy utilization, it is of great practical significance to implement Maximum Power Point Tracking (MPPT) for PV cells. Accurately and rapidly tracking the maximum power point is crucial for the operation of PV power generation systems. Therefore, this paper, based on MPPT, focuses on analyzing the MPPT methods in the case of a single peak, including constant voltage tracking, incremental conductance method, perturbation observation method, fuzzy control method, etc., and analyzing their advantages and disadvantages. It also looks forward to control algorithms under partial shading conditions and future development trends.

Using a two-stage multiple Support Vector Machine (SVM) model as the foundation, this work proposes a unique forecasting approach focused on the medium-term electricity Market-Clearing Price (MCP) forecasting issue. The model aims to efficiently capture the complex nonlinear dynamic features of electricity market prices in order to provide accurate price forecasting information to electricity market participants and decision makers. Firstly, we introduce the core idea of the two-stage SVM, where the first stage SVM is used for the initial fitting of the data, while the second- stage SVM further fine-tunes the model to accommodate price fluctuations under different market conditions. Through this two-stage modelling approach, we are able to better address the non-linear and dynamic nature of electricity market prices. In this study, we pay special attention to two key SVM control parameters: the coefficients of the Gaussian radial basis kernel function (σ and the regularization constant (γ). The selection and optimization of these parameters is a key factor in the success of this model. We propose a method for parameter tuning using test data to ensure optimal performance of the SVM model. By optimizing the parameters, we can minimize the Mean square Absolute Error (MAE) of the SVM model and thus improve its accuracy in price forecasting. By applying it in medium-term electricity market price forecasting, we can better understand market trends and price fluctuations.

Electricity price prediction plays an important role in energy management and decision making. In this study, we use a deep learning model, specifically Long Short-Term Memory Network (LSTM), to predict electricity price data from different cities. We first pre-process the data, including data normalization and outlier processing, to improve the performance of the model. Subsequently, we constructed LSTM models with multi-layer structure for electricity price prediction in each city. In addition, in order to take into account the differences between cities, we introduced electricity load weights to more accurately synthesize the prediction results for each city. Finally, by applying the decision level fusion algorithm, the results show that the deep learning model performs well in electricity price prediction and provides strong support for energy management and policy making.

The precision of predicting the residual service life of lithium-ion batteries is of utmost vital to make sure their effective management and maintenance. Therefore, in order to increase the precision of predicting the residual service life of lithium batteries, this research puts forward a methodology that forecast the remaining useful life of lithium batteries through employing a hybrid model of CNN-BiLSTM and attention mechanism. firstly employing the Bayesian optimization algorithm for the purpose of the parameters optimization of the suggested model, and secondly, using the series connection of CNN and BiLSTM to catch the layered characteristic between same variables which can affect battery degradation, as well as the time dependence embedded into these characteristic, and using the attention mechanism to give more focus to the important features that impact the outcomes, so as to acquire the forecasted outcomes for the remaining lifespan of lithium batteries. The validation analysis performed on the NASA dataset illustrates that the proposed approach attains higher levels of accuracy in comparison to the predictive results obtained from alternative neural network methodologies.

As power engineering continues to evolve, projects related to networks of 10 kV and below are becoming increasingly critical. Effective management and thorough evaluation of temporary labor organizations are essential for the successful completion of these projects. This research offers a comprehensive overview of the unique characteristics and procedural complexities of such projects. It delves into the specific challenges faced in the organization of temporary labor for construction and proposes tailored solutions to address these issues. Through experimental designs and rigorous data analysis, the study establishes a clear correlation between various labor factors and the overall outcomes of the projects. Additionally, it explores the impact of integrating renewable energy sources, particularly photovoltaic systems, into these network projects. The findings of this study are aimed at providing valuable insights for effective project management in the evolving landscape of network-related projects, advocating for continuous improvement and innovation in this field. This research not only contributes to the current understanding of temporary labor dynamics in power engineering but also highlights the importance of integrating advanced technologies and sustainable practices in network project management.