This PDF file contains the front matter associated with SPIE Proceedings Volume 12594, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

This paper investigates an optimization problem corresponding to energy minimization of an in-band full-duplex relaying subject to a total data-bits constraint and a flight time constraint. In order to utilize light of sight channels, a fixed-wing unmanned aerial vehicle (UAV) is dispatched as an amplify-and-forward relay to forward messages for the source and destination terminals. To solve the optimization problem, a new received signal-to-noise ratio and a new lower-bound of the data size received by the destination-terminal are derived. Finally, the energy minimization problem is solved, giving the optimal flight speed and flight time of the UAV for energy efficiency. Computer simulation experiments are conducted and the results verified the correctness and the efficiency of the propositions of the paper.

This paper mainly discusses the influence of artificial lighting environment in a subway stations on pedestrians' visual and psychological perception, which is explained from two potential influencing factors - the relevant color temperature of light source and lighting mode. On this basis, nine lighting environments are subjectively and carefully analyzed by a questionnaire survey. The evaluation results show that the higher the subway ride frequency, the higher the subjective evaluation score; The change of lighting environment also has a significant impact on people perception. In the high color temperature environment, the lighting mode will not affect people perception of warmth. It is worth noting that the ceiling lighting score of the middle color temperature is the highest, which is the most suitable environmental mode designed in the subway station, and the middle color temperature is also the most relaxed color temperature level among the three color temperatures. The basic lighting with warm color temperature is not suitable for application in subway stations. By combining the relevant control technology, we have obtained a relatively suitable lighting scheme for subway stations.

A switchable wideband frequency selective rasorber/absorber based on PIN diodes is achieved in this paper. When the PIN diodes are switched ON and OFF, the designed structure switches the working states between an absorber and a rasorber, respectively. The proposed structure consists of non-switchable lossy layer and switchable bandpass frequency selective surface (FSS), which are separated by an air interlayer. In the state of rasorber, the 1 dB transmission band is obtained in 3.87-4.95 GHz with the minimum insertion loss of 0.28 dB at 4.31GHz. The structure presents an absorption band from 1.48 GHz to 3.37GHz with absorptivity up to 80% in absorbing state. In addition, an equivalent circuit model (ECM) is established to comprehend the operating mechanism of the switchable structure. Therefore, the designed switchable rasorber can be applied in the field of the smart radome.

Aiming at the problem of slow convergence rate of reinforcement learning (RL), an improved RL algorithm based on A* algorithm and velocity obstacle (VO) is proposed. Firstly, a node path obtained by A* algorithm is used to initialize a part of Q table. Then, in the process of path planning by RL, the VO model is applied to output the minimum steering angle required by the obstacle-avoiding flight. The action which is greater than the angle will be taken to interact with the environment. Finally, in order to satisfy UAV performance constraints, B-spline curve approach is applied for path smoothing. The experiment results show that, compared with the standard Q-learning algorithm, the path length of improved RL algorithm is reduced by 26.37%, the success rate of algorithm planning path is increased by 30.76%, and the convergence round is advanced by 88.66%.

SRAM occupies the main area of SoC, its power consumption performance become crucial. Low power SRAM is widely used in IOT device. Power-gating circuit can help SRAM hold low static power during non-read/write period, and multi-VDD using different voltage thresholds in different modules can help SRAM acquire low dynamic power. Read-Write assistant circuit composed by NMOS capacitor helps SRAM read and write faster than before to gain a fast timing performance, and it helps SRAM get the feature of low Vmin and DRV. This design in single port SRAM was tape-out by 14nm FinFET process, and it was verified by ADVANTEST V93000 test platform. The all test result of Vmin, current and data retention voltage meet the single port SRAM design requirements.

Intelligent patrol inspection of unattended substation can avoid the failure of distribution box, but a large number of distribution boxes bring some difficulty to the efficiency of intelligent patrol inspection of unattended substation. In order to improve the efficiency of intelligent inspection, the application of power ubiquitous Internet of things technology in intelligent inspection of unattended substation is proposed. Calculate the GPS coordinates of the substation according to the coordinates of the substation and the location of the substation, and complete the intelligent patrol path planning of the unattended substation through the kinematic constraints and obstacle avoidance constraints of the patrol path planning; Use the power ubiquitous Internet of things technology to obtain the three-dimensional spatial coordinates of the inspection path, and monitor the orientation of the UAV during inspection through the intelligent monitoring technology of the power ubiquitous Internet of things; The correction factor is introduced to correct the patrol inspection path of unattended substation; Using the minimum objective function formula of polynomial regression algorithm, the inspection track of unattended substation is optimized, and the intelligent inspection of unattended substation is realized. The experimental results show that applying the power ubiquitous Internet of things technology to the intelligent inspection of unattended substation can judge whether there is a fault in the distribution box in the unattended substation, improve the inspection efficiency to more than 95.3%, and meet the performance requirements of intelligent inspection of unattended substation.

The research of solar cells has gone through three generations, silicon solar cells, thin film solar cells and perovskite solar cells. There are a variety of characterization methods used to analyze the factors that affect the performance of solar cells. Thus far, there are some deficiencies in the research of algorithms of the characterization methods themselves. Three types of data: current density-voltage (J-V) curves, X-ray diffraction (XRD) spectrum and absorption spectrum are the most essential characterization methods. Here, the processing algorithms of these three kinds of characterization data are studied, intend to help researchers to solve the practical problems more efficiently, and improve the accuracy of experimental results. In this paper, the process of processing algorithm are summarized in the way of flow chart, and combined with the example to show in turn.

This paper describes the application status and existing problems of the nocturnal patrol technology of the overhead transmission lines in our country. A system for carrying night vision of unmanned aerial vehicle (UAV) on the nocturnal patrol of the overhead transmission lines is introduced in detail, and the design scheme of the system is presented. The specific contents are as follows: A set of transmission line UAV airborne night patrol platform is developed; Develop corresponding control system software and real-time image transmission system; Canny algorithm and improved adaptive Canny algorithm are used to detect the edge of the image returned by the night vision patrol system. The results show that this method has good intelligence and high identification, it is convenient for the staff to identify transmission line faults, and effectively improves the night patrol ability of overhead transmission lines.

Landscape sculpture can enhance the artistic effect of sculpture through lighting methods and lighting effects as a key symbol of a city. In this paper, a red sculpture, which is at the entrance of the commercial street in Xiaodong Road, Shenyang city, is taken as the research object. Lighting design and simulation are carried out by DIALux EVO software. Nine different lighting effects are obtained by changing the illumination index (100lx, 200lx, 300Lx) and the relevant color temperature (3000K, 4000K, 5000K). Through the way of subjective evaluation questionnaire, inviting 30 people to evaluate from the psychological feelings, explore the effect of good lighting design by using SPSS software analysis of different solutions for visitors how different psychological effects to determine the best lighting solution . The actual illumination of the surface of the sculpture deviates from the expected illumination, taking into account factors such as the installation distance of the luminaire and bad weather, so this paper designs an adaptive fuzzy PID controller to dimming control the lamps and optimize the illumination of the sculpture, and provides suggestions for future research on the lighting design of landscape sculpture.

To study the propagation of electromagnetic waves by the Lorentz media layer, this differential formula about TM waves propagation in lossy Lorentz media is obtained applying this time-domain finite difference method of the auxiliary differential equation (ADE). According to the formula, the waveform changes of the electromagnetic wave when the electromagnetic wave passes through the Lorentz thin layer of lossy dispersive medium, and the results indicate that the reflected effect of the Lorentz lossy dispersive medium layer on the plane electromagnetic wave is obvious.

With the increasing demand for the distribution network reliability, distribution network failure risk prediction has become the key to further improve power distribution reliability. Aiming at the uncertainty and randomness of distribution network faults, a fault risk prediction method of distribution network based on situational awareness is proposed. Through situational awareness and situational understanding of distribution network fault data, the optimal subset of fault characteristics is obtained by excluding 12 feature variables with high redundancy and weak correlation; The three elements of situation awareness, situation understanding and situation prediction are mapped to the distribution network fault risk prediction, and a prediction model based on the PSO-SVM algorithm is constructed. Through the numerical example analysis, the effectiveness and accuracy of the proposed method are verified, which provides technical support for risk prevention and control of distribution network.

Under the production mode of multi variety and small batch, different types of products on the mixed model assembly line involve different operation procedures, assembly time and required parts, and the time and quantity of material distribution are different. The quality of the material distribution scheme affects the production progress of the whole mixed model assembly. Based on the real-time production information, this paper combines the scheduling sequence of mixed model assembly line with the material distribution scheme, establishes the mathematical model of material distribution based on the scheduling sequence, controls the logistics of mixed model assembly line in the production process through the scheduling optimization results, realizes the timely and accurate distribution of materials, and maintains the efficient operation of the assembly line.

At present, the structure of electronic system is more and more complex, and the function is more and more important. However, the degradation state of electronic products has no obvious external performance, and there are sudden and random faults. The development of fault diagnosis technology of electronic products is relatively slow. In view of the above problems, this paper proposes a complex electronic system health management method, which defines the modules such as state management, execution management, NVRAM manager and watchdog manager to define the functions and cross-linking relationships of each component, which realizes the health management of complex electronic system under the joint supervision of software application and hardware electronic equipment.

An empirical mode decomposition-extreme learning machine (EMD-ELM) wind power prediction method based on the hierarchical clustering method was proposed to solve the current problem of insufficient power prediction accuracy of wind power stations in this paper. This method uses the aggregation algorithm of hierarchical clustering to cluster the data with similar weather conditions, and uses the EMD method to decompose the power sequence of each group, which can obtain relatively stable data components, and finally uses the ELM method to predict and combine each component. Compared with the ELM wind power prediction model, the numerical simulation shows that the EMD-ELM wind power prediction model based on hierarchical clustering method makes the data characteristics of similar weather conditions more obvious, the value of each evaluation index is better and the model has higher prediction accuracy.

As an important technology and research direction to achieve AI, deep learning has been widely applied in the fields of computer vision, speech recognition, and natural language processing. How to effectively accelerate the computing power of deep learning has always been the focus of scientific research. Among various acceleration technologies, FPGA has the advantages of reconfiguration, high performance, small size, and low latency. As more and more FPGA-based neural network accelerators are developed, we notice there is a lack of a complete and detailed overview. In this paper, we give a comparative study of DNN accelerators on FPGA from the aspects of hardware structures, design ideas and optimization strategies. We further compare the performance of different acceleration technologies in different models and present the prospects of the FPGA accelerators for deep learning.

Wind turbine is the core equipment to realize the conversion of wind energy to electric energy in wind farms, and high reliability of wind turbine is of great significance to promote the application of typical wind power in new energy. With the passage of time, wind turbines run in bad working conditions for a long time, which leads to more faults of wind turbines in recent years. And it brings hidden dangers to the safe and stable operation of new energy grid-connected power systems. This paper expounds the common failures of key components of wind turbine, including converter, generator, variable pitch system, gearbox, yaw system and blade. Then it analyzes the causes from the aspects of design defects, complex operating conditions, improper maintenance and inadequate technical supervision. Finally it puts forward measures and suggestions to improve the reliability of wind turbine, and prospects the development direction of wind turbine reliability research.

This paper proposes a dual-polarized 1-bit digitally coding metasurface. The metasurface unit is composed of a single-layer substrate structure and a metal patch, and the pin tube is located at the junction of the middle section of the cross. By controlling the on-off of the diode, the topology of the metasurface unit can be changed, resulting in a 180-degree phase difference at 6.35GHz in the case of a perpendicular incident electromagnetic wave. Digital simulations show that this structure can still maintain good performance when incident at a large angle, and has a good application prospect on digitally encoded metasurfaces.

As an important way to achieve the "carbon peaking and carbon neutrality goals," the integrated energy system is worth further exploring its ability to respond on the demand side. In this paper, an electric-thermal integrated energy demand response based on a tiered carbon trading mechanism is proposed, and a low-temperature waste heat power generation device based on ORC is introduced to further solve the problem of insufficient demand-side response. Finally, the proposed model is verified with the lowest energy purchase cost, tiered carbon transaction cost and operation and maintenance cost. Through the example analysis, it can be seen that the integrated energy system considering the response of electric heat demand to formulate a reasonable carbon trading mechanism can achieve the synergy of economic and low-carbon system.

This paper presents a common emitter (CE) four-way combined W-band power amplifier in 0.13μm SiGe BiCMOS process. The PA is designed with four-stage CE structure and λ/4 impedance transformation networks for power combining. The four-way amplifier has a peak small signal gain of 22.6 dB at 92.6 GHz. The 3 dB bandwidth is 15GHz ranging from 83.2-98.2GHz. The PA has saturation output power (Psat) of 20-20.8 dBm at 90-98 GHz and power-added efficiency (PAE) of 7.8-11.6% and output referred 1dB compression point (OP1dB) of 18.2-19 dBm at 90-98 GHz.

As one of the clean energy utilization forms, wind power plays a positive role in promoting energy transformation and realizing the double-carbon goal, and wind power generation has been vigorously developed. Wind turbine is an important equipment in the process of wind power generation and wind farm, and its reliable operation should be ensured. In this paper, taking an actual wind farm as an example, the annual failure time of its direct-drive wind turbines is analyzed, and its failure types are sorted out. The causes of failures are analyzed, and the measures to improve the reliability are put forward to provide reference for the safe and stable operation of wind farms.

Multi-robot path planning (MRPP) enables the robot teams complete the tasks quickly and efficiently without collision. Based on the domestic and foreign research, the mainstream centralized algorithms for path planning have many shortages, so IL-A3C algorithm, which is based on the Dec-POMDP model, is proposed to solve these shortages. Then, the performance of IL-A3C is computed in terms of average path planning length, average path planning time, average collision probability and average planning success rate in different dimensions, and the simulation result shows that IL-A3C performs very well under low obstacle density and can be easily extended to a team of 128 robots. After that, IL-A3C is also compared with the centralized algorithm CBS and A3C algorithm, and the comparison proves that IL-A3C has higher success rate, stronger scalability and stability than CBS and A3C. The conclusion is that IL-A3C can be easily scaled to a large-scale robot team.

Fiber optic Fabry-Perot cavity hydrophones have the advantages of high sensitivity, simple structure and the resistance to electromagnetic interference, making them a research hotspot. Filling the Fabry-Perot cavity with water media would lead to a decrease in the reflectivity of the fiber end face, thus causing a decrease in the sensitivity of the fiber FP cavity hydrophone. In this work, the magnetron sputtering technology was utilized with alumina as the sputtering target. A thin film was deposited to the end face of the optical fiber to improve the reflectivity of the end face of the fiber. Coated fibers with varied film thicknesses were prepared and their reflectivity was characterized. The experimental results showed that the reflectivity of the end face of the optical fiber in water media can be increased by an order of magnitude to 2.61%, which is an excellent basis for the development of a highly sensitive optical fiber Fabry-Perot cavity hydrophone.

The eight-channel millimeter wave integrated receiver module is an important component of the passive millimeter wave security scan. In the actual module test, it is found that the measured value of the receiver module focus has a deviation of about 10mm from the theoretical value. This paper analyzes the effect of channel mutual coupling in the eight-channel millimeter wave integrated receiver module and finds that the inter-channel coupling will have a deviation of 3~5mm on the position accuracy of the module focus point, and identifies that the coupling in the eight-channel integrated receiver module as an important factor affecting the performance of the module.

As fossil energy is gradually exhausted and environmental pollution is becoming more and more serious, people are urged to change to the development and utilization of new energy. Wind power generation has been widely used in this context, and wind turbine is an indispensable equipment for wind power, which plays an important role in promoting the development of renewable energy. This paper statistics the fault frequency and duration of power failure of wind turbine in a wind farm from the point of view of practical wind farm project. On this basis, the various faults of components of wind turbine are summarized, the causes of faults are analyzed. Then the improvement measures are put forward, including improving the level and equipment quality of wind power generation devices, strengthening reliability management, optimizing overhaul and maintenance operations, and improving the management system of wind power enterprises. Finally, the future research direction is prospected.

In the process of signal acquisition by laser microphones, the high frequency components of speech are missing due to nonadditive distortion. In this paper, we proposed an end-to-end speech bandwidth extension (BWE) approach to recover narrow-band speech acquired by laser microphones. Our preliminary research showed that speech enhancement algorithms based on log-magnitude spectrogram in the frequency domain could not achieve satisfactory performance for this task. Therefore, we designed a speech BWE model in time domain, this model was modified by Wave-U-Net structure, we introduced the time convolution module (TCM), the dilation of convolution is helpful to increase receptive field, improves speech long-range correlation, at the same time introduced the multi-resolution loss function (L_MSTFT) instead of the mean square error (MSE), the time-domain Wave-U-Net method avoided the decoupling of magnitude and phase in the frequency domain. The results showed that the signal-to-noise ratio (SNR) of speech was improved significantly compared with approach in the frequency domain, and obtained elaborate high-frequency components than frequency-domain convolutional recurrent network (CRN). We chose laser speech to test the model in an actual scene, which further verifies the practicability of the structure through the speech spectrum analysis, and had better performance and generalization ability than the original Wave-U-Net model.

The combination of electromagnetic lens with wide-angle scanning capability and phased array can achieve high gain at low cost. At the same time, it can realize convenient scanning, so this combination has been studied extensively. In this paper, the metamaterial is used as the basic unit, and designing the phase distribution of the lens, finally the millimeter-wave electromagnetic lens with thin and wide angle is designed. The experimental results are consistent with the simulation results. The designed electromagnetic lens can realize 30° beam scanning by changing the feed position. In the future, the lens can be used in the territory of radar detection and communication to reduce the cost of antenna which has high frequency and high gain.

Considering traditional methods’ low accuracy and machine learning methods’ lack of interpretability, this paper proposed a pavement performance model for IRI prediction based on XGBoost, and introduced SHAP to enhance the interpretability of individual features of the model. The data used are from the America LTPP data. Firstly, data cleaning and preprocessing were conducted. Secondly, four prediction models were built based on classical algorithms, namely, LightGBM, XGBoost, SVM, and multiple linear regression. Then, by comparison, it was found that XGBoost performed better. Finally, parameter tuning for this model was performed, with the RMSE as 0.317, MAE as 0.219, and R² as 0.742. In addition, considering the prediction model’s lack of transparency, SHAP is utilized to perform the feature importance analysis and identify the main factors affecting the pavement performance, which can help the highway sector to improve the reliability of their subsequent prediction model analysis.

Several years have passed, DL methods have made a series of progress. However, within the field of road distress recognition, there are few works using recent methods, especially the transformer. Studying how modern models can help us in this field is important for us to figure out the direction we should move forward to. In this paper, we mainly research the performance of the modern DL model in asphalt road distress classification. GAPs is the dataset we used. We apply the recent SOTA of CNN model and transformer model to find out how far we have come. Totally, the improvement in these years is limited, 7.4% increment of f1 score for the validation set. Some problems like overfitting still need a better solution. And in terms of the dataset, characters like local similarity need to be considered when making a dataset and designing a model.

With the increasing emphasis on the application of renewable energy in the world, photovoltaic power generation has ushered in a rapid development stage. Photovoltaic power generation equipment is an essential part of photovoltaic power generation, and its reliability plays a positive role in the development of renewable energy. This paper makes statistics on the fault rate of the power generation equipment in a practical photovoltaic power station, including photovoltaic inverter, combiner box, photovoltaic modules, etc., sorts out the types of faults, analyzes the key factors that affect the fault of the equipment, and puts forward future research directions.

In the complex environment simultaneous localization and mapping (SLAM) system, the use of a single laser or visual sensor is susceptible to external factors, which affects the accuracy of mobile robot mapping. A SLAM mapping method based on vision, LIDAR and encoder fusion is designed. Firstly, LIDAR and encoder data are fused by extended Kalman filter (EKF) method. Then, the data obtained by the depth camera is converted into the virtual two-dimensional laser radar data. Through the adaptive weighted fusion algorithm, the EKF fusion results are weighted fusion with the camera processing data, and the fused data is used to construct the map. The experimental results of the actual environment show that the method of multi-sensor data fusion is better than single sensor in the accuracy of mapping, and can establish a map closer to the actual environment.

As an important subject in the construction of energy system, the new power system has become a hot topic in the current society and the country. With the increasing demand in energy, distributed generation technology has become a key research direction in the field of clean energy utilization in recent years due to its advantages of low investment, low pollution and flexible access mode. Therefore, distributed generation (DG) has also come into being, and it is of great theoretical significance to study the optimal configuration of DG access. In this paper, photovoltaic power generation characteristics, wind power generation characteristics, battery operating characteristics and load their uncontrollability are considered firstly, then the voltage stability mechanism of the "double-high" new-type power system is deeply analyzed on this basis so that the influence factors of voltage stability in the new power system are given. After that, an improved power flow model is established according to the characteristics of new energy nodes and the voltage characteristics of the load. Based on this model, the weak area of the system and new energy nodes can be quickly judged. by applying the eigenstructure analysis method, the minimum model sign value can be obtained so that the voltage stability of the system can be verified.

This work presents a new method to design the fractional-order linear phase filter based on the interior point algorithm. The transfer function of the fractional-order linear phase low-pass filter is derived from the general form of the fractional-order all-pole low-pass filter transfer function. The optimization problem of the phase-frequency characteristics of the fractional-order filter is modeled and solved. A detailed analysis of the stability and frequency domain of the optimized fractional-order linear-phase filters is presented. The active circuits for (1+α)-order filters are designed and simulated.

China has a large electricity consumption with a large proportion of people living in the country, and the electricity consumption and generation in diverse regions are not balanced. The forecast of electricity load and short-term energy consumption can be helpful in easing the long-term electricity consumption shortage. To optimize the power dispatching strategy of each province and city, we must pay attention to the short-term forecast of regional power load. In this paper, a new short-term power load forecasting HP-ARIMA-BP model is used to screen and count relevant power data from all over the country. Use LSTM to fill in the missing values of time series, with HP filter decomposition to decompose electricity consumption in Anhui Province, remove trend items and volatility interference in the data, apply ARIMA to predict the trend items, and use BP neural network to predict the volatility items. The final fit achieves high accuracy.

In today's agricultural information age, knowledge services for farmers cannot be provided without the Internet, and with the expansion of information data on the Internet, the field of agricultural planting technology has also generated a huge amount of digital information resources, among which professional knowledge presents redundancy, sparseness and diversity, requiring new means and methods to collect and store these big data and knowledge as well as to conduct knowledge Q&A in a timely manner. The aim of this paper is to study a question and answer system for agricultural cultivation technology based on knowledge graphs. Firstly, the knowledge graph model of agricultural cultivation technology is designed, and the existing deep text matching algorithm DSSM algorithm is selected. Based on the DSSM algorithm, a Bert+DSSM algorithm is proposed to improve the performance of the algorithm. The text matching algorithm, question and answer corpus and crop planting knowledge map are applied to the system to build an agricultural planting technology question and answer system, and the experimental results show that the relevant knowledge question and answer function of the system has a high accuracy rate.

With the advancement of technology, there are more and more information available and access to information. However, due to the clutter of official information, the public cannot obtain emergency knowledge and disaster information in a timely and effective manner. Early warning and popular science publicity of meteorological disasters through WeChat will be a new development direction. By analyzing the current status and defects of the disaster science popularization platform on the market, the functions of the disaster prevention science popularization system are analyzed and designed on the basis of sorting out and analyzing various disaster-related knowledge and protection information. Based on the official WeChat development platform, it adopts the Spring Cloud framework, develops using Java language, and uses databases and cloud functions for background data storage. The system can provide a wider platform for users to understand and browse disaster information, and facilitate the management of administrator background information.

Resource-rich millimeter wave (mmWave) contributes to the rapid development of 5G networks, but the problem of millimeter wave non-line-of-sight (NLOS) propagation has become an obstacle. A relay system is a key to solving this problem. One of the solutions is massive multiple-input multiple-output antenna technology (Massive MIMO), which is a crucial research direction to break through the mmWave propagation loss. In this paper, the focused beam shaping technology, antenna array design, and antenna module packaging technology of large-scale antenna technology are studied and analyzed, and the existing technologies are summarized and improved, including a beam scheduling algorithm based on a deep learning model, a foldable array antenna for millimeter wave non-line-of-view and thermal loss verification.

With the advent of the era of big data and artificial intelligence, the network has developed rapidly. Students have various hardware and platforms of more and more information, know how to mine from the massive data to useful information, and can predict the future development of things. So, we need to solve one of the urgent problems through the collection of big data, preprocessing, analysis to get the results we want, and then decision analysis and judgment. That is the focus of this article.

The precise and real-time forecasting of traffic is very important for the planning, control, and orientation of traffic in cities. However, forecasting circulation flows remain a troublesome issue because of the highly non-linear and complex nature of circulation systems. This research proposes a new temporal convolutional multi-attentive network-based traffic flow prediction model (TCMAN). TCMAN model captures the spatiotemporal features of traffic flow through temporal convolutional network (TCN) and codec. The codec includes several spatial-temporal attention blocks to simulate the effects of space-time factors on circulation conditions. The input traffic flow features are coded by the encoder and the output sequence is anticipated by the decoder. Finally, many experiments are carried out on the traffic datasets. Compared with the baseline method, TCMAN model has better prediction performance.

Automotive radar is the key component of driver assistance system. The main current requirements are high angular resolution, high target detection accuracy, low power consumption, and low cost. However, ambiguous target and low angle measurement accuracy will appear in the multi-target scenario. In order to solve these problems, TDM-MIMO (time division multiplexing multiple-input-multiple-output) radar is implemented to improve the angular resolution. We use the proposed TDM-MIMO LFMCW (linear frequency modulated continuous wave) echo model to derive the range, velocity and azimuth information of the target. And we present a 3D-FFT (three-dimensional fast Fourier transform) scheme to solve the problem of multi-target detection. In the simulation, we effectively solve the accuracy problem in multi-target detection without ambiguity, and the angular resolution is improved by about 14°.

Noise and interference are the main factors affecting high-frequency channels, and it is important to generate noise and interference that match the actual high-frequency channel characteristics when modeling high-frequency channels. To address the noise characteristics in wide-band high-frequency channels, we construct an Alpha-Stable distribution-based noise and interference model for wide-band high-frequency channels based on the analysis of Alpha-Stable distribution theory. Simulation results show that our model not only generates noise that is more consistent with the actual wide-band high-frequency channel but also effectively reduces the complexity of the current state-of-the-art wide-band high-frequency channel noise model. In order to better simulate the noise characteristics of the wide-band high-frequency channel, swept-frequency interference is added to our model, and the simulation results show that the wide-band high-frequency channel model with swept-frequency interference can better simulate the noise characteristics of the real wide-band high-frequency channel.

Aiming at the problem earthquake emergencies are hazardous, time-sensitive and difficult to control, so earthquake emergency plans play an important role in the emergency response to major emergencies. We conducted functional analysis and design of the earthquake emergency plan management system after analyzing the current state and flaws of earthquake plan management. Based on the PyCharm platform and B/S model, we combined the requirements for plan management in the emergency platform and used Django architecture to realize the system development using HTML, CSS, and JavaScr. The system can give the earthquake department the theoretical backing and technical assurance it needs to maintain and improve the current plans.

The aircraft's performance, such as the flight time and maneuverability, can be greatly enhanced by using carbon fiber in the composites, which makes these kinds of materials widely used in modern aircraft. The aircraft composites with carbon fiber have different transmission and reflection characteristics compared to the composites with metal, and the radiation pattern of aircraft antennae will be extremely complicated due to the carbon fiber with different types, proportions, growth methods, and thicknesses. However, most of the simulations and designs of aircraft antennas do not take into account the influence of different materials. In this paper, the radiation pattern of an L-band omnidirectional antenna in typical application scenarios is analyzed by simulation and experiment. The results show that the carbon fiber composite has a greater effect on the radiation pattern and the gain of the antenna drops more than to 15dBi in some directions. The results indicated that traditional aircraft antenna design does not consider the effect of different materials, so there is a significant simulation error, and the influence of carbon fiber materials should be considered in the early stage of link design.

The high-speed power line carrier (HPLC)+high-speed radio frequency (HRF) dual-mode communication technology can significantly improve the reliability and transmission efficiency of power communication systems. This paper first introduces the development status and application requirements of power communication technology. Then, five advantages of HPLC+HRF dual-mode communication technology are analyzed, including compatibility with existing HPLC technology, improvement of high frequency acquisition quality, enhancement of business carrying capacity, meeting minute level acquisition requirements and achieving accurate power outage reporting. Through pilot application, the high frequency acquisition capability, minute level acquisition capability and accurate power outage reporting capability of dual-mode communication technology were quantitatively verified. Finally, the application prospect of dual-mode communication technology in the new type power system in various fields is prospected.

Under the background of the booming industrial Internet, data acquisition and real-time calculation of industrial equipment have become the key issues of many industrial manufacturing enterprises. Therefore, this paper studies the design and implementation of web log analysis system. In this paper, a real-time industrial equipment big data acquisition and calculation system based on SparkStreaming, HBase and Hive is constructed. On the equipment big data acquisition technology, k-means algorithm in data mining technology is used to classify and integrate the data. Based on Spark framework, the parallel calculation of K-means algorithm is realized, which solves the problem of low efficiency of industrial big data analysis. Experiments show that the development of this system can effectively improve the efficiency of collecting and analyzing big data logs of industrial equipment, and further improve the timeliness and expansibility of transmission.

With the development of mobile network technology, network traffic has not only experienced exponential explosive growth, but also its application scenarios have become more and more extensive. It is a challenging proposition to find an efficient and accurate matching prediction model based on massive idiosyncratic data. This scheme proposes to introduce EMD modal decomposition to decompose the local feature components of data based on the time-scale features of the data itself, to do cluster analysis on the components by K-mean clustering algorithm, and then to model and predict the clustered local feature components by XGBoost model, so as to reduce the data dimensionality and prediction complexity. The results show that the modeling and prediction of the clustered local feature components using XGBoost model effectively improves the model prediction accuracy.

In view of the problem faced in Aviation Information Network (AIN) that the unbalanced load of platforms caused by the dynamically arrived service function chains (SFCs), we formulate the SFC migration problem into a multi-objective optimization model and propose a coalitional game based migration algorithm (CGM). In this paper, we take the aviation platform as the game player, the virtual network function (VNF) as the game commodity, and perform the migration of VNFs through the comparation and swap between different platforms to realize the efficient management of network resources. Experimental results show that the proposed algorithm has the advantages of low computational complexity and fast convergence rate, which can effectively reduce the network energy consumption and migration overhead.

As one of the potential technologies of 6G, THZ communication has a positive effect on data rate expansion, low latency, high reliability and seamless connection. Firstly, the technical challenges of THZ communication in motion platform are described. Then, a THZ self-tracking system is proposed to solve the problem. Finally, the technology is verified in the test site. The verification results show that the self-tracking technology can effectively enhance the continuous and stable chain building capability of the THZ communication link on the mobile platform.

Aiming at the problems that 3D optical measurement is easy to be affected by peripheral noise, difficult to calculate relative height and low measurement accuracy, a multi-frequency asynchronous phase shift algorithm based on time series changes was proposed. In order to improve the measurement accuracy and anti-interference ability of the system, five kinds of frequency grating projections were adopted. In order to improve the efficiency of phase undoing, the asynchronous phase shift method was used to extract the phase principal value. In view of the multi-frequency heterodyne phase solution method is prone to produce phase jump error, a phase jump error correction method based on threshold discrimination is proposed. The experimental results show that the improved algorithm has better measurement accuracy than the original algorithm, and compared with the traditional multi-frequency heterodyne method, it has strong anti-interference ability and higher measurement accuracy.

The complexity of combat environment makes it more and more difficult for traditional monostatic radar to track targets continuously. The radar network technology can effectively improve the overall tracking performance. Therefore, in the distributed radar network scenario, a joint sensor selection, bandwidth and dwell time resource optimal allocation algorithm is proposed for multi-target tracking. At first, the local state is estimated and corrected by the Covariance Intersection (CI) fusion strategy. Then, the algorithm aims to minimize the predicted Bayesian Cramer-Rao Lower Bound (BCRLB) of the worst target tracking mean square error. And a joint radar node selection, bandwidth and dwell time resource optimization model is established for multi-target tracking in distributed radar network. The radar node is selected with BCRLB as the metric criterion first. Then, the optimal model after sensor selection is carried out by using the cyclic minimum algorithm and the minimax algorithm. Simulation experiments show that compared with the uniform resource allocation and single-resource optimal allocation algorithms, the proposed algorithm can effectively improve the multi-target tracking accuracy under the limited resources.

High-throughput omics technologies have been widely used in biomedical research. While it’s still challenging for researchers without statistical foundations or programming skills to obtain biological insights from omics data, the published tools have some deficiencies in the environment configuration and analysis workflow definition. Here, we propose a one-stop analysis and visualization platform for omics data named MyOmics, which is freely available at http://myomics.bio-it.tech/. Its functions include data preprocessing, statistical analysis, functional enrichment analysis and visualization based on multi-index data frame structure. To evaluate the performance of MyOmics for omics studies of clinical cohorts, we re-analyzed a published Hepatocellular Carcinoma dataset and reproduced their main findings, which demonstrated that MyOmics is reliable and efficient to analyze omics data. Based on the Galaxy framework, MyOmics provides an easy-to-use solution for exploring omics data.

With the rapid development of communication technology and electric power industry, cables have been widely used. The demand for cables in various industries has risen sharply, so a variety of cables have appeared in the market. After a large number of cables are used, cable break failures occur frequently. In addition, most cables are installed in the special environment of high altitude or underground. Without professional testing equipment, line maintenance personnel can hardly find the fault location. Based on this, an efficient cable breakpoint location method is proposed, and the modulus maximum detection algorithm based on wavelet transform is studied to realize the cable breakpoint location. The wavelet threshold de-noising technology is used to remove the noise in the transmission signal and reflection signal collected by the system. By comparing the results of different wavelet decomposition, the optimal wavelet basis function is selected, and the modulus maximum sequence of the wavelet coefficients obtained after wavelet decomposition is calculated, then the time difference between the transmission pulse and the reflection pulse can be detected, and the location of the cable breakpoint can be further determined. The existing cables in the laboratory are tested for many times, and the actual test results are analyzed, which can meet the technical requirements of cable breakpoint detection.

To address the problem of privacy protection in satellite communication, according to the proposed new three-dimensional chaotic system, this paper achieves the concealment of picture information by one-dimensional index scrambling and encryption operation in QPSK modulation for 24-bit true color pictures and completes the reception and decryption of ciphertext picture data and recovery of picture information at the receiver side. Through the simulation of the encryption algorithm, it is proved that the encryption system has better encryption than the classic logistics chaos encryption system and has the ability to resist statistical analysis attack, differential attack and noise interference. The whole image encryption system is implemented on FPGA, which proves its great potential for future application in satellite communication.

WSN system can effectively meet the application of monitoring in terms of CNC status and structural health, but it is difficult to solve the problem of large power consumption. In this paper, we propose an optimization method that combines the cluster head selection strategy with the node dormancy strategy by modeling the energy consumption of CNC machine information in transmission. And it is demonstrated through simulation experiments that the proposed strategy survives more nodes after a certain number of iterations compared with the single use of cluster head selection strategy and node dormancy strategy. It is shown that it has the effect of saving node energy consumption, extending sensor lifetime and reducing production cost.

In China, 4G and 5G networks have been maturely applied to high-speed rail scenarios, and the user experience of networks on high-speed rail has been further improved. The critical issues of wireless communication coverage on high-speed rail have been solved with key technologies, including Doppler frequency offset fixed by shift correction and compensation algorithm, frequent switching of networks solved by the exclusive network coverage method and cell combination technology, and car body attenuation resolved by combining multiple frequency networks and prioritizing low frequency. This paper analyzed such key technologies and verified their effects on problem-solving via tests. With the popularization of high-speed rail, new problems have emerged in the communication field, such as complex networks with multiple formats and multiple frequencies, the influence induced by exclusive network coverage for high-speed rail, and the problem of insufficient capacity. This paper explored two topics, namely the breakpoint issue and capacity problem of the exclusive network when trains pass each other on high-speed rail, and proposed a solution of 4G and 5G fusion coverage to achieve continuous coverage at breakpoints and equalize capacity resources, which expanded the application of 5G on the basis of fully harnessing the 4G network.

This paper presents a new design of superformula ultra wideband with two notched bands conformal antenna for UAV. Because of its low profile, the conformal antenna on the aircraft can effectively reduce the impact of air resistance. The dielectric substrate is Rogers Duroid 5880, and the final size of the antenna is 30 mm × 40 mm. In order to avoid interference with WiMAX and WLAN bands, two notched bands are obtained by using ring and ring slots on the radiating element. The results show that the impedance bandwidth of the antenna is 2.7-14 GHz, and the double notch band is [3.2-3.9 GHz] and [4.7-5.8 GHz].

With the rapid development of power grid, power system security and stability will face more severe tests, and the stability analysis and control of power system research has received extensive attention. Aiming at the parameterization problem in transient stability analysis, a parameterized transient stability trajectory approximation method based on polynomial approximation method is proposed. In the proposed method, a polynomial approximation composed of polynomials and coefficients of a series of parameters is used to approximate the system variables, and the relationship between the variable parameters and the dynamic trajectory is established. Compared with the existing trajectory sensitivity methods, the proposed method greatly improves the accuracy under the condition of large parameter variation and strong nonlinear system, and has the global controllable accuracy characteristics within the variable parameter variation range.

Since network coding technology was proposed, researchers have been enriching the theory of network coding and its related research has gradually become a field of interest for scholars, but the application and research of network coding technology in cloud storage and digital signature are still relatively missing. This paper mainly studies the application and development of network coding technology in cloud storage and digital signature, firstly introduces the technical knowledge about network coding and reviews its development history, introduces the main ideas and technical points of network coding in detail, then discusses the application of network coding in the field of cloud storage and digital signature, researches and analyzes its development respectively, and finally puts forward its views and open issues, and gives an outlook on the application prospect of network coding technology.

In the fine-grained image classification task, the huge intra class variance determines that the classification of the task depends on coarse-grained and fine-grained information. However, there is still a lack of research on how to quickly and effectively fuse multi granularity features. Therefore, this paper proposes a fine-grained image classification network based on fast MMAL net and studies the impact on the classification accuracy and convergence speed of fine-grained classification network. The fast aolm module is used to predict the position of the object, and the fast APPM module is used to predict the information of the key areas of the object without the need for bounding boxes or labels. The method proposed in this paper achieves high accuracy on the three data sets of Stanford cars cub-200-2011, fgvc aircraft and Stanford cars respectively, which shows that this method can obtain excellent classification performance.

Nowadays, with the rapid development of Internet technology and the emergence of converged media, the rapid transmission of news and information and the ways of transmission are increasingly diversified, and the transmission of video information is gradually developing forward and entering the mainstream of society. Video communication has become an integral part of the new era. This paper proposes a conference mode, which is convenient for the video conference terminal to connect to the video conference software, and also convenient for later operation and maintenance. Our video conference terminal uses Polycom's hardware equipment, our software client uses Huawei cloud conference, and our management platform is managed by the intelligent edge platform of the Internet of Things. The conference system, which supports HD real-time video transmission, has lower delay when resharing the screen than the conference system developed by Tencent Conference. This system maintains stable transmission even in the case of serious renetwork packet loss.

In order to meet the antenna requirements of Beidou satellite service communication terminals, a design scheme of low profile planar spiral antenna is proposed. The antenna body is consisted of copper foil material, which use FR4 medium as support, in the case of limiting the overall volume of the antenna, the circular polarization radiation performance of the antenna is improved by loading a sine wave on the plane spiral shape, and the input impedance matching of the antenna is improved by three 90° bending feed terminals. The simulation results show that the antenna is in the operating frequency band of the Beidou satellite communication RDSS service, with a standing wave ratio of less than 2, a gain higher than 5 dBi, and an axis ratio of less than 3 dB, which has the characteristics of light weight, low profile and high gain, and has good radiation characteristics, which meets the radiation index of the Beidou RDSS satellite communication terminal antenna.

This paper investigates the output-based predictive control problem for networked control systems (NCSs) with delays and data loss using the dynamic event-triggered mechanism (DETM). And the output-based DETM is proposed to utilize the network resources more efficiently while ensuring the system performance. The combination of DETM and observer-based networked predictive control (NPC) approach is developed to actively compensate the network delays and data loss, which enables the closed-loop system to maintain the desired performance and stability. In addition, a sufficient condition is deduced to ensure the stability of the NPC system by adopting an augmented model approach. Finally, the validity of the design procedure is demonstrated by an example.

Tropospheric scattering communication has the characteristics of large single hop span, strong obstacle surmounting ability, high reliability, good anti-nuclear explosion and anti-interception ability, which can make up for the shortcomings of limited satellite communication channel resources, the shortage of microwave communication LOS transmission and the weakness of poor fiber mobility, and will play an important role in the future information system operations. Through three aspects of research, we have realized the time synchronization technology based on the troposcatter communication channel by using the two-way time comparison principle, including the data calculation of a single microcomputer, serial communication and LCD; FPGA data reception; Kalman filtering for differential signal processing. Finally, the demonstration results are displayed on an oscilloscope, reducing the time synchronization delay to microseconds.

The maximum node degree clustering method for multilayer mobile networks suffers from a high cluster head change rate. A trust-based maximum node degree clustering method for multilayer mobile networks is designed. Obtaining the highest node degree parameter, we calculate the energy consumption during state transitions, optimise multilayer mobile network routing protocols, adjust the value of the cost function for joining cluster heads, take direct interaction, send trust requests to third parties for verification, and design a clustering method based on trust degrees. Experimental results: The mean value of cluster head change rate of this designed multi-layer mobile network maximum node degree clustering method is 12.519%, which is 5.125% and 5.683% lower than the other two multi-layer mobile network maximum node degree clustering methods, respectively, indicating that the designed multi-layer mobile network maximum node degree clustering method is more stable after fully considering the factor of trust.

As the current situation that application offloading has not yet been widely used in 5G networks, this paper proposes a performance optimization technology for 5G User Equipment (UE) based on edge microservices live migration. If UE switches quickly between multiple 5G base stations, the edge microservice will actively migrate to the closest edge server, so as to provide continuous services. This technology solves the problem of maintaining UE service continuity after edge microservice migration. And the whole process only makes minor modifications to 5G core network functions, without the active participation of UE. The prototype test results show that this technology can be effectively compatible with existing 5G standard specifications, maintain the continuity of access to UE TCP connection type services, and improve average service access performance of UE by 12%, approximately.

This paper presents a frequency-reconfigurable ultra-wideband tightly coupled antenna array. Different from the traditional ultra-wideband antenna, the tightly coupled antenna is used to expand the antenna bandwidth through the mutual coupling effect between the antenna elements. At the same time, it combines the technical characteristics of the reconfigurable antenna. The antenna can work in both high frequency and low frequency bands and can freely switch the working frequency band. At low frequency, the ultra-wide frequency band of 3.55-6.6GHz is realized by the tight coupling between antenna units. After the coupling part is disconnected, it is equivalent to a dipole antenna with a working frequency band of 14.2-14.6GHz in the high frequency band. The VSWR of the antenna is less than or equal to 2 in the two working frequency bands, and the size of the antenna unit is 0.14λ*0.27λ. By adding a reconfigurable unit (PIN diode) to the antenna arm of the tightly coupled antenna to control the disconnection and connection of the coupling part, switching between high frequency and low frequency working frequency bands is realized. The PIN diode is in a state of series resistance and inductance in the on state, so that the antenna arm is connected to the coupling part to enhance the bandwidth of the low-frequency operating frequency band, and in the off state, it presents a state of large resistance and capacitance in parallel. While changing the operating frequency, the influence of the coupling part on high frequencies is improved.

Most tasks in airborne scenarios are composed of neural networks (NN) and non-NN tasks. On the one hand, Deep neural networks (DNNs) have become the mainstream method of AI task processing. DNNs are accompanied by a large number of complex tensor calculations, as well as network delays and data privacy risks caused by using the cloud to process data. The combination of dedicated AI hardware accelerators and edge computing has become a new solution. On the other hand, dedicated AI hardware accelerators cannot efficiently handle non-NN class tasks (such as signal processing, control flow, etc.), so onboard systems need to design new hardware, complete compiler stacks, and runtime systems to handle non-NN class tasks. In order to effectively deal with multiple tasks in the airborne environment, the airborne computing system needs to establish a set of unified compilation and runtime systems, which can optimize multiple tasks and be deployed on the heterogeneous computing platform to achieve "one compilation, multiple deployment." This paper analyzes the development of compiler and runtime systems at home and abroad, and compares the advantages and disadvantages of different technologies. Finally, this paper discusses the application scenarios of the compilation system and runtime in the airborne environment and the technical directions that need to be breakthroughs in the future, so as to provide guidance for the development of the airborne intelligent compilation and runtime system.

Pseudo-random sequences are predictable and repeatable, and have statistical properties similar to random noise, so they are widely used in communication, radar, cryptography and other fields. This paper mainly takes m sequence commonly used in pseudo-random sequence as an example to study its production principle, implementation process and related characteristics, and uses MATLAB software for modeling and simulation. The simulation results show that the key to the generation of m sequence is to have appropriate feedback logic, that is, to find the original polynomial, and m sequence has good autocorrelation characteristics. Finally, its application in spread spectrum communication is given.

With the development of information technology, "information silo" problem is getting serious, which hinders the exchange of information between enterprises. Data integration is the key technology to solve this problem, but traditional data integration methods have problems such as weak scalability and differences in language representation. In this paper, a method is proposed to complete data integration by establishing ontology, data mapping and semantic inference, which effectively solves the differences in data structure, data source and language representation and has strong scalability. In order to verify the feasibility of the method, experiments are conducted in this study with data describing people, and the completion of ontology construction, data mapping and semantic reasoning illustrates that the experiments achieve the expected results. Finally, the ontology model of nucleic acid detection is constructed in this paper, and the integration of nucleic acid detection data is completed to verify the feasibility of engineering the method.

In the intelligent sorting stage, in order to improve the sorting speed, efficiency and accuracy, this paper takes the product area, radian, differentiation and surface characteristics as the research vector, using the objective identification method based on the minimum mean square delta rule neural network, intelligent identification of cutting samples, during leather cutting and processing. Considering the background light, the sorting system, the background difference method was used to obtain the accurate image coordinates of the motion work piece, in order to improve the sorting efficiency of leather cutting and processing.

Herbal medicine is an integral part of Chinese medicine, and Chinese medicine extract is the main way of herbal medicine application. Therefore, the extraction operation of Chinese medicine is an important unit operation in the production process of Chinese medicine. With the modernization of Chinese medicine, traditional Chinese medicine extraction equipment can not fully meet the needs of the development of Chinese medicine. Energy saving and emission reduction technology, integrated manufacturing technology, on-line monitoring technology and intelligent digitization are the main research directions of intelligent production of Chinese medicine. This paper studies the on-line quality monitoring and intelligent control system of Chinese medicine extraction process. Near infrared spectroscopy has the unique advantage of rapid analysis, which makes it play an important role in the development of rapid analysis and detection methods for Chinese medicine production process. Near infrared spectroscopy analysis platform includes field level, edge level and cloud computing level. Its core content is the design of near infrared analysis method for online monitoring. The main parts are extraction equipment, online sample pretreatment subsystem, online near-infrared spectrometer subsystem, online near-infrared spectrum quality analysis and intelligent control system main control subsystem, extraction workshop automatic control subsystem. On-line detection technology detects the key quality of raw materials, intermediate products and process in real time through process analysis instruments directly installed on the production line. This technology can feedback in real time to design, analyze and control the production process, which can better guide the production, reduce the batch difference of intermediates, ensure the stability of drug quality, and provide real-time production data for subsequent product quality backtracking.

Aiming at the optimization problem of operation dispatching of multi-energy systems including cold, heat, electricity and gas, in order to achieve the goal of energy saving and emission reduction, a microgrid optimization model including micro-sources such as fans, micro gas turbines, waste heat boilers, lithium bromide absorption chillers, etc. is established with the microgrid operation cost and new energy utilization rate as the optimization objectives. The improved multi-objective grey wolf algorithm (MOGWO) is used to solve the multi-objective model, which effectively compensates for the disadvantage that the grey wolf algorithm is easy to fall into local optimum. The simulation analysis shows that the algorithm has good solution speed and global search performance, which provides an optimal dispatching strategy for multi-energy systems, realizes flexible mobilization of microgrid, and achieves the effects of reducing operating costs and low grid-connected volatility.

Aiming at the timing synchronization problem in underwater acoustic single-carrier frequency domain equalization system (SC-FDE), the training sequence and new metric function is constructed by using the well-related Zadoff-Chu (ZC) sequence, and the genetic algorithm is used to optimize the combination structure, which solves the problems of peak platform and false peak interference in the commonly used synchronization methods. Simulation results show that the improved synchronization algorithm is more accurate at the timing point under low signal-to-noise ratio and has high synchronization accuracy.

Methods for classifying different human activities have been explored widely in recent years. In radar-based human activity classification, most deep learning-based techniques pay great attention to the frame-level structure. These approaches generally adopt short-time Fourier transform, or convolutional layers with a frame-level filter to process the raw radar data, while they seldom consider the overall efficiency. In this paper, we propose a sample-level convolutional neural network named SampleRadarNet. The proposed SampleRadarNet utilizes one-dimensional convolutional layers with small filters to learn more temporal information from the raw data, breaking through the frame-level model’s capacity limitation. In addition, the Squeeze-Inception-Mobile module is designed for the classification task, which involves the following three components: fire module, inception block, and depthwise convolution. The experimental results show that the proposed architecture can achieve better performance compared with the existing related methods in a seven-class radar-based human activity classification problem.

A new task scheduling scheme based on dispersed computing is proposed to address the problem of poor utilization of idle computing resources in the network by the traditional computing paradigm. The scheme designs a dispersed computing task scheduling model to reduce the load pressure on NCPs in a recursive manner by combining the idea of multiple hops. For the characteristics of geographic dispersion and network dynamics in the dispersed computing environment, the time delay is selected as the optimization objective, and the task scheduling decision is made in a distributed manner by combining it with the artificial fish swarm algorithm. At the same time, the scheduling algorithm is designed mainly by setting variable step size and field of view so that the algorithm can guarantee the convergence speed while the solution accuracy is improved. The experiments show that the algorithm meets the high requirements of dispersed computing on time delay and achieves the expected goal of the system.

Pedestrian re-identification is an integral part of surveillance video analysis, and the field of pedestrian re-identification has been growing rapidly in recent years with the development of deep learning. Most existing pedestrian re-identification methods are based on convolutional neural network deep learning models, and most studies consider learning representations from individual images, ignoring the potential interactions between them. In this paper, we propose a dual feature-enhanced pedestrian re-identification method based on Transformer architecture, which learns high discriminative features of pedestrians, suppresses the influence of irrelevant features on the network, and explicitly models the interactions between all input images to more fully learn the pedestrian feature representations from different perspectives, thus making the network more accurate and robust in overall recognition. Experiments with the model on four publicly available pedestrian re-identification datasets demonstrate that all evaluation metrics are higher than most of the current mainstream pedestrian re-identification models.

The trajectory probability hypothesis density (TPHD) filter is a promising multi-target tracking algorithm. However, in many practical applications, inaccurate measurement noise prior statistics can degrade the performance of TPHD filters. In this paper, we model the trajectory state in TPHD as the product of the Gaussian mixture with the inverse Wishart under the linear Gaussian model, and propose the IW-VB-TPHD filter to adaptively acquire the covariance of measurement noise. Compared with the IG-VB-TPHD filter, simulation results show that the IW-VB-TPHD filter has a better accuracy in target tracking.

To address the problem that the combination of conventional FDA and nonlinear frequency bias cannot achieve range-angle decoupling, a decoupled nonlinear frequency bias HL-FDA scheme is studied based on an accurate understanding of the range dimension dependence. By using the FDA-MIMO structure instead of the conventional uniform linear FDA structure, combined with the minimum variance distortion-free response (MVDR) adaptive beamforming algorithm, the interference outside the platform can be suppressed.

To address the current problems of high energy consumption and insufficient resource utilization in the scheduling process of cloud virtual machines, an optimization algorithm based on combined prediction for virtual machine placement was proposed. Firstly, a combined prediction model consisting of the BPNN, ARIMA model and LSSVM model was constructed in the load prediction phase to make the detection of overloaded hosts and underloaded hosts more accurate. Secondly, the classical PABFD algorithm was improved in the virtual machine placement phase by adding the constraint of remaining resource utilization of the destination host and proposed an optimal adaptation decreasing algorithm based on resource utilization. Finally, the results of simulation experiments in the CloudSim platform showed that the adoption of this virtual machine placement algorithm can effectively reduce energy consumption while guaranteeing the quality of service.

In the process of underwater target recognition, it is necessary to consider that the relative motion between the underwater vehicle noise source and the receiving hydrophone will produce Doppler effect. Doppler parameter estimation based on Doppler is conducive to the location and identification of low-frequency line spectrum noise sources and passive velocity measurement and ranging of underwater moving targets. Experiments show that underwater acoustic signals can be recognized effectively based on both WVD transform and Doppler estimation.

Pavement crack detection is a key technology used to judge whether the road is safe or not. Due to the complex and diverse background of cracks, the traditional crack detection algorithm is difficult to accurately detect cracks. In this paper, YOLOv5 algorithm with strong portability is used to detect road cracks. Mosaic data enhancement method is used to enrich the background of detection targets at the input end and improve the detection effect of small targets. The Backbone CSP structure divides the input into two branches, which greatly reduces the computational load while enhancing the learning performance of the entire convolutional neural network. The experimental results show that the model trained on the public dataset achieves 92% detection accuracy on the test set, and the detection time is 0.04s. The model is directly applied to the vehicle and pedestrian detection datasets, and the detection accuracy is improved by 2%, and the detection time is shortened to 0.028s, indicating that the model has good generalization performance. It can be used for crack detection and quality assessment in complex road scenarios.

Through data mining with different network security technologies, we analyze the security situation curve, grasp the dynamic changes of the network security situation, and make reasonable predictions of the security situation. In order to solve the shortcomings of existing research on network information transmission security situation awareness, this paper discusses the selection of situation assessment indicators, experimental environment and data sources based on the discussion of network information transmission situation awareness, Bayesian network reasoning, and LSTM neural network algorithm, and it discusses the design of situation awareness assessment model of network information transmission security based on Bayesian network algorithm and LSTM neural network algorithm prediction model. Finally, the two algorithms proposed in this paper are compared with other algorithms. It can be seen from the data results that the thinking accuracy, detection rate and recall rate of Yates network algorithm and LSTM neural network algorithm are more than 99%. Therefore, it is verified that the network information transmission security situation awareness algorithm based on intelligent computing has a certain use value.

Point cloud denoising is a key aspect of current point cloud data processing. Existing deep learning-based denoising methods rely on local feature information obtained from noisy point clouds as the basis for point displacement. However, in the presence of noise, the local features of points usually cannot accurately represent the underlying surface of the point cloud. We propose a new deep learning-based denoising method using the graph convolution operator, which can compute feature aggregation weights adaptively based on feature vectors to achieve hierarchical distinction between noise and the underlying surface. Meanwhile, in order to alleviate the phenomenon that the predicted gradient fluctuates more drastically during the iterative process, we introduce the momentum rise method in the gradient iteration process to improve the stability of the solution. We evaluated our approach on both toy datasets and real world datasets. Under different types and levels of noise disturbances, our method has better results compared to other models.

In this research, we propose an optimized robust Capon beamforming (RCB) algorithm, which is based on covariance matrix reconstruction along with steering vector (SV) estimation. It is crucial to assurance the robustness of the beamforming algorithm performs better. The proposed algorithm indicates an alternative method to roughly estimate the SV of the interference signal, thus we are able to construct the optimal Capon space spectrum. Secondly, the reconstructed interference signal covariance matrix is obtained by integrating the optimized Capon space power spectrum in the corresponding sector of the interference signal. Subsequently, we adopt the eigen decomposition of the sample covariance matrix (SCM) to obtain the noise covariance matrix, as well as complete the reconstruction of the interference-plus-noise covariance matrix (IPNCM). Finally, this paper optimizes the algorithm of uncertainty set constraint, as well as gets more accurate desired signal SV. The numerical simulation results obtained through two hundred Monte-Carlo experiments suggest that the proposed algorithm indicates higher robustness, better noise suppression ability, and higher output SINR, which is closer to the optimal output curve than several traditional RAB algorithms.

Because of the rapid development of computer technology, digital twin technology has been improved, in the background of industry 4.0 industrial production also uses digital twin technology to reform. Virtual scene simulation was conducted based on Unity3d technology, and the model was made by 3Ds Max software. Through the unique advantages of the virtual factory, the intelligent production system is built, and the traditional factory is transformed into a virtual factory for simulation, monitoring, decision-making, so as to apply it to the real new factory, so as to achieve cost saving, improve production efficiency, optimize the effect of production mode. This paper describes part of the development process of the smart factory software system.

In order to overcome the shortcomings of poor fault tolerance and complicated computation in blind identification of recursive systematic convolutional (RSC) codes, a blind identification method based on binary artificial bee colony algorithm is proposed. Firstly, according to the structure of RSC encoder, the product vector of output bit matrix and the coded polynomial vector is obtained. Then, the artificial bee colony algorithm for continuous decision variables is modified to be suitable for binary optimization problems, and the sum of the elements of the product vector is used as the fitness function of the binary artificial bee colony algorithm to optimize the coefficient vector of coded polynomial. Finally, the termination conditions of the algorithm are set according to the fitness value and the number of iterations. Furthermore, the termination condition of the algorithm is improved according to the characteristics of the coding polynomial. Thus, the RSC coded polynomial can be identified while the computation is reduced. The experimental results show that the algorithm can recognize RSC coded polynomials effectively, and the correct rate of recognition is over 98% for coded polynomials with different coding constraints when the bit error rate reaches 0.05.

In order to provide intelligence support for regional counter-terrorism work, this paper establishes a terrorist attack prediction model based on ARIMA time series algorithm, and the experimental results show that the model has good prediction ability for terrorist attacks, and the mean absolute percentage error (MAPE) of two experimental arithmetic cases are 13.4% and 10.5%, respectively, and the model can provide macroscopic decision support for regional counter-terrorism work. The model is then applied to terrorist attack data from 2000-2020 to address the counterterrorism situation in South Asia and further improve the security environment in South America.

As the most widely used pseudo-random number generation algorithm, the linear congruence algorithm has the characteristics of fast generation and can generate better random numbers in one-dimensional space. Random numbers are widely used in many fields, such as information security and communication, and it is important to study the generation of random sequences. However, the linear congruence method has some defects, for example, when the mode of the linear congruence method𝑚 takes 10, multiplier𝑎 takes 7, increment𝑐 takes 7 and the initial value𝑋₀ takes 7, the sequence obtained is: 7,6,9,0, 7,6,9,0, ..., which results in a sequence with a period of 4. In the linear congruence method𝑚 , the 𝑎 and 𝑐 are crucial to the size of the period of the whole sequence and will directly affect the security of the algorithm. In this paper, the conditions for the linear congruence method to reach the full cycle are given on the basis of previous work, and a rigorous mathematical proof is given.

Aiming at the problem that the traditional bilingual teaching resource recommendation technology does not take into account the differences of different learners' learning style, knowledge level and learning mode, and does not take into account the individuation. This paper proposes a bilingual teaching resource recommendation method for international trade practice based on knowledge graph and neural network. Firstly, we correlate the content to be learned with the resources, use the existing knowledge mapping technology, and then use Word2Vec to quantify its behavior and characteristics, and the recommendation model is embedded. Finally, BiLSTM and Attention are used to extract features, dig learners' implicit feedback information, and achieve the purpose of personalized recommendation for learners. Experimental results show that the proposed model is more effective than the traditional recommendation algorithm, and effectively improves the effect of bilingual teaching resources recommendation.

With the development and progress of cloud computing, big data, data analysis and mining, as well as the gradual promotion and application in the field of land resources, the discipline development and scientific research of land resource management have also received extensive attention. However, there is still room for development of the land resource management platform in the field of teaching assistance, and the scientific research development, data sharing and openness of land resources are still insufficient. It is an important and urgent challenge to realize the development and application of data. In this paper, through the analysis of the existing system architecture, an easy-to-expand research platform framework is established, using J2EE for architectural development, and modular integration of different functions, so that the intelligent land resource management research platform can better serve geography research workers in the field of research and majors in earth sciences, and provide reference and suggestions for land research work.

At present, there are many kinds of devices in distribution network. The devices have similar functions but cannot share their data with each other, causing a waste of equipment resources and high costs for operation and maintenance. To solve the problems above, we proposed a terminal design scheme with hardware of high versatility, software in the form of application (app) and functions based on modularization design. We developed 10 basic function apps and 20 business function apps, including the data center. And we designed Smart Fusion Terminal with the functions of topology identification, load identification and distributed energy management. We conducted Smart Fusion Terminal’s pilot application in 14 provinces and conducted test on terminal’s evaluation function for meter clock accuracy. The results show that Smart Fusion Terminal can realize all the monitoring and acquisition functions of the original devices with the success rate higher than 99%. By adopting expansion module and corresponding function app, it can also meet the diversified requirements in different distribution networks. The application of Smart Fusion Terminal can improve the flexibility and scalability of terminal construction, reduce the investment cost of power grid equipment and improve on-site maintenance efficiency.

Both of orthogonal time frequency space (OTFS) and non-orthogonal multi-access (NOMA) are wireless transmission technologies applicable to high mobility environment. According to the characteristics of the Internet of Vehicles, this paper introduces a cross-domain transmission scheme combining OTFS and NOMA. Based on downlink of OTFS-NOMA system, we deduce the expressions for the SINR of the signals. We propose a dichotomous iterative power allocation (DIPA) strategy and compare it with the traditional power allocation strategies. The simulation results show that the DIPA scheme has obvious improvement on sum rate (SR) of the system compared with the traditional power allocation schemes.

According to the encoding characteristics of MPEG-4 video standard and the encoding principle of MPEG-4, this paper proposes a video watermarking algorithm based on MPEG-4, which is suitable for copyright protection. The basic idea of the algorithm is as follows: firstly, the watermark information is obtained from the middle and low frequency range of the DCT coefficients after the brightness component is quantified, and then the watermark information is embedded into the DCT coefficients of all the brightness components of the video sequence I frame. Finally, the simulation experiment and result analysis are given. Simulation results show that the algorithm has good invisibility.

In order to solve the problems of large search space and low search efficiency in computer software defect automatic repair technology, a research method of computer software defect management and automatic repair technology was proposed. This paper points out the limitations and challenges of the current computer automatic repair technology by briefly describing the research on computer software defect management and computer automatic repair technology, analyzes three search spaces for automatic defect repair, and designs and implements the prototype tool GPSRepair in the existing technology. The experimental results on the Defects4j defect dataset show that GPSRepair has higher repair efficiency and accuracy than existing repair methods.

Spiking neural networks have the nature of high efficiency, energy saving, and bio-interpretability. They communicate through sparse and asynchronous spikes, so they have received extensive attention in the field of neuromorphic engineering and brain-like computing. At present, the commonly used encoding methods are mainly single-rate encoding and temporal encoding. However, rate encoding cannot make use of the time information in the spike train, which has high energy consumption. Temporal encoding limits the computing power of neurons and will produce dead neurons. Moreover, it is critical to find effective solutions that reduce network complexity and improve energy efficiency while maintaining high accuracy. Therefore, we propose a hybrid coding method based on rate coding and temporal coding to solve the limitation of single coding. We propose an adaptive online pruning strategy based on hybrid coding. In this pruning strategy, 100 neurons are pruned out of the 200-neuron network, which reduces the network size and obtains a more compact network structure. The memory capacity is reduced by 1.9×, the energy efficiency is increased by 2.4×, and the classification accuracy is reduced by less than 0.5%.

Attention mechanisms have been found to be effective for human gaze estimation. To address the problem that traditional attention has limited ability to extract higher-order contextual information in gaze estimation tasks, an ECA attention mechanism-based gaze estimation network is proposed, which aims to effectively exploit the channel relations of features through a global average pooling layer without dimensionality reduction, suppress some facial regions that do not contribute to gaze estimation, and activate subtle facial features that can improve gaze estimation. The model can take full advantage of the user's appearance, which helps to improve the accuracy of the gaze estimation model. In this paper, experiments are conducted on the MPIIGaze dataset, and the results show that the network based on the channel attention mechanism can reduce the estimation error, and the model proposed in this paper can achieve more accurate gaze estimation.

Instrument landing system (ILS) is the most widely used guidance system in precision approach landing for civil aircrafts, especially in low visibility. The ILS linearly transforms the angle deviation between the actual and nominal flight path into the form of ILS signals for the pilot to correct the path. The scale of ILS signal changes uniformly, while the actual safety changes nonlinear. The signal design with uniform scale change cannot reflect the dynamic change of real-time flight safety. To solve this problem, we adopt the method of reverse design and propose a signal design method based on Gaussian quantified safety according to the operational regulation of instrument landing for safety named CAT (Category). This method considers the changing rate of the safety with the deviation of instrument scale, so that the pilot can perceive the real-time safety while observing the ILS signal for safety. Furthermore, our proposed design will improve the safety during the approach and landing phase.

Graph Convolutional Networks (GCNs) are neural models that capture the dependence of graphs via message passing between the nodes of graphs. However, SGD-based GCNs training algorithms require a lot of computing resources, which limits the application scenarios of GCNs. Spiking neural networks (SNNs) are an efficient and energy-saving network with strong biological interpretation and great development prospects, helping GCNs work on mobile devices with strict power limits. However, the training effect of SNNs is still low compare to traditional CNN networks. In this paper, we combine federated learning (FL) structure with spiking GCN networks, and introduce asynchronous architecture to improve the training effect and improve the robustness of the system. The digital simulation experiment proves our conclusion.

Aiming at the high-order range walk and Doppler walk problems encountered by maneuvering targets in the long-time coherent accumulation process, a long-time coherent accumulation algorithm for variable acceleration maneuvering targets based on time reversing transform (TRT) is proposed. Firstly, the time series inversion transformation is performed on the echo signal after de-carrier frequency and pulse compression, which can correct the linear distance walking caused by the maneuvering target speed and the high-order distance walking and Doppler walking caused by the variable acceleration. Multi-linear function transformation is used to correct the second-order range walk and linear Doppler walk caused by acceleration. Finally, FFT and IFFT are used to achieve coherent accumulation of target energy. Simulation experiments show that the proposed algorithm is effective in correcting the moving objects with variable acceleration. At the same time, the detection performance of the algorithm is better than that of the PHAF algorithm, the RFT algorithm, the RFRFT algorithm and the RFRAF algorithm. It has lower computational complexity and does not require search for motion parameters.

This paper studies the realization of the virtual display of mobility tools based on Virtools. Through the three-dimensional modeling software 3ds max, the three-dimensional solid modeling of the transportation tool is realized; based on model optimization, the construction of the virtual display scene of the transportation tool is completed. Then, combined with the research on the characteristics of the product virtual display of the Virtools platform, the interactive design of the virtual display of the mobility tool is carried out, including the basic motion and the analysis of the motion relationship between the components, and the interactive control of the mobility tool is realized. Based on realizing the virtual movement of the mobility tool, the collision detection and the setting of the virtual camera in the display process were deeply studied. Based on the later integration, the dynamic demonstration of the virtual display of the mobility tool and the release of the works were realized.

The evaluation of the bird damage system is of great significance for the protection of bird damage prevention in high-voltage transmission lines. The entropy weight method is introduced into the TOPSIS evaluation model, and the weight of each evaluation index is calculated with information entropy to reduce the subjectivity of determining the weight. Then an improved TOPSIS bird damage system evaluation model based on the entropy weight method is established, The Jilin Province’s high-voltage transmission line 2021 annual bird damage evaluation was taken as an example, and the improved model was used to evaluate it. The results show that this model can weaken the impact of outliers, reduce errors caused by subjective empowerment, and effectively evaluate bird damage levels in real time.