This PDF file contains the front matter associated with SPIE Proceedings Volume 12253 including the Title Page, Copyright information, and Table of Contents.

High speed train automatic driving system (ATO) is one of the means to further enhance the efficiency, comfort and energy saving of high-speed trains. Its goal is to ensure the safety and precision parking of trains at the predetermined location, while ensuring the comfort of passengers during the parking process. Aiming at the high repeatability of the process of high-speed train parking in the station, and the high-speed train is a strong coupling complex nonlinear system, while avoiding the frequent switching of control output, an accurate parking control model based on Fuzzy PID iterative control is proposed. And the high-speed train is a strong coupling complex nonlinear system, while avoiding frequent switching of control output. The differential equation of the train braking model is obtained by the analysis of the train traction braking system, and then the learning parameters of the system gradient and convergence conditions are obtained. The fuzzy PID iterative control can effectively realize the correct parking of the train through the related simulation tests, and has significant robustness to the uncertainty of train parameters and the repeated uncertainty in the parking stage.

Aiming at the situation of local shading and multi-peak output characteristic P-U curve in the practical application of photovoltaic array, a hybrid algorithm is proposed, which uses the “three-step” strategy to quickly and accurately search the global maximum power point. The Hermite interpolation disturbance algorithm is used to select a large step size and search to the vicinity of the maximum power point through a small number of iterations. Then it is switched to the improved particle swarm optimization algorithm to find the maximum power point. When the result converges or reaches the maximum number of iterations, the voltage value near the global extremum point is output. Finally, the Hermite interpolation disturbance algorithm is carried out, and the small step length is selected to quickly and accurately stabilize at the global extremum point. This hybrid algorithm makes up for the shortcomings of a single algorithm. The advantages and disadvantages of particle swarm optimization algorithm and perturbation method complement each other, avoid falling into the local optimal value and reduce the convergence time, improve the speed and accuracy of the system, and evade the energy loss caused by oscillation. Simulink simulation verifies that the proposed hybrid algorithm has better tracking performance and response speed.

Three-level T-type grid-tied converter is the important device to connect renewable energy and grid. When the inductance parameters in converter model mismatch with those in controller, the conventional model predictive control (MPC) strategy causes grid-tied distortion and current errors. To solve the problem, an improved model-free prediction control strategy using Runge-Kutta algorithm (RKA) for three-level T-type grid-tied converter is proposed. The model-free predictive control strategy is analyzed based on ultra-local model of converter system. In order to calculate the lumped disturbance of the system, RKA is used in ultra-local model. Because the derived function that calculates parameters in RKA are unknown, Lagrange interpolation is performed to calculate the parameters. The ultra-local model of the converter system can be obtained by RKA and Lagrange algorithm. Compared with the conventional MPC strategy, the model-free current prediction control strategy based on RKA improves parameter robustness. Finally, simulation results verify the effectiveness of the proposed approach.

Task, information, and social networks have become more and more complex in collaborative command-and-control environments. How to appropriately describe, characterize, and assess the complexity of a command-and-control system (CCS) is critical to improving its overall efficacy. This study initially offered several concepts of collaborative networks from the perspective of human-system integration (HSI), such as command and control support (such as command and control software, displays and controls, and communication equipment) and personnel (teams). We then constructed a theoretical framework covering the complexity of a collaborative CCS, model-based characterization, and feature measurement. The theoretical model results show that this framework is effective in improving human factors in the design and evaluation of a collaborative CCS.

In order to improve the attitude control accuracy and anti-disturbance performance, this paper designs an attitude sliding mode controller based on the dilated state observer in order to improve the stability control of the quadrotor UAV with complex aerodynamics and susceptible to disturbances. First, establish the quadrotor dynamics equation using, and the expected value of the inner loop angle is obtained through PID position control and used as the input of the inner loop attitude control. Secondly, the inner loop attitude controller uses the expanded state observer to estimate the system input and output state information and the total internal and external disturbances in real time. Finally, the sliding mode control law is designed to compensate the estimated disturbance in real time to realize the attitude control, and the system stability is proved by the Lyapunov theory. The simulation results show that the designed controller has better active disturbance rejection capability and more accurate tracking capability.

Based on the theory of Active Disturbance Rejection Control, aiming at the problem of passive self-guided super-cavitating vehicle intercepting close-range maneuvering targets, a variable-structure terminal guidance law is designed by using the easily obtainable information of the line-of-sight angular rate, combined with the target's maneuver estimated by extended state observer. The guidance law is simulated and analyzed under a typical engagement scenario, with the results showing that it can achieve a better attack effect with little target's maneuver information. The profile of required acceleration command is smooth, which is suitable and necessary for the super-cavitating vehicles.

In this paper, an event triggered single network adaptive dynamic programming method is designed for multi-agent systems with input disturbances, and the optimal consistency control problem of the system is studied. When designing the controller, the coupling gain is multiplied by the analytical solution of the system cost function to construct a control strategy against the disturbance term. Then the input disturbance term is replaced by a neural network model, which is adjusted and restricted with the executive network. The optimal control strategy can minimize the cost function on the premise of the maximum input disturbance. The evaluation execution disturbance network shares the weight estimation rule of the evaluation network, and its update time is determined by the event trigger condition, which reasonably avoids unnecessary calculation in network learning. Simulation results show that this method can not only meet the expected results of the system, but also reduce the waste of information resources in the communication process.

With the continuous development of robot technology, the application scope of current industrial robots is becoming more and more extensive. However, the accuracy of industrial robots cannot meet the needs of high-end manufacturing. Therefore, it is necessary to improve the positioning accuracy of robots. In the process of this research, based on the kinematic error model of pose differential transformation and the kinematic error model of coordinate error transfer, the positioning accuracy of the robot is effectively improved. The kinematic parameter identification technology based on the BAS-PSO algorithm can guarantee the positioning accuracy of the robot to a large extent. After simulation experiments, it can be determined that the average comprehensive position error of the six-degree-of-freedom robot identified based on the BAS-PSO algorithm can be reduced from the original 0.302 mm and 0.210° to 0.081 mm and 0.042°. The average position error and average attitude error of the robot based on the positive kinematics model can be kept at 0.098 mm and 0.099°. Therefore, it can be determined that the robot positioning accuracy based on the BAS-PSO method has a significant improvement effect, and it has good identification accuracy and convergence speed.

Chassis structures have a fundamental influence on the stability of self-balancing unicycles. This paper documents how using generative design can improve the robotic system’s stability and reduce the overall mass compared to the conventional designs. The chassis structure is divided into seven parts for stress analysis. According to the results from the stress analysis, loading conditions and constraints can be set for the generative studies. The ideal solutions were chosen based on the mass-safety factor graphs for the most optimized geometry. Stress analysis was then carried out on Inventor to verify the model’s reliability. To test the controllability of the generative unicycle, simulations were carried out in Simulink by comparing the generative design with the conventional design. The results showed that the generative chassis structure has a higher center of mass and lower mass, hence is easier to achieve balancing from an angular deviation.

In the traditional wireless power transfer (WPT) system, the angular misalignment between the transmitting coil (Tx) and receiving coil (Rx) will reduce the output power and efficiency of the system. To solve this problem, an angular misalignment insensitive WPT system based on planar-type coils and a 3-D rotating magnetic field is proposed in this paper. First, the general architecture of the system is given. Then, based on the LCC-S resonant compensation network and the circuit topology of a three-phase half-bridge inverter, the mathematical model of the system is established and the modulation method of the 3-D rotating magnetic field is given. Finally, the experimental verification of the feasibility and effectiveness of wireless energy transfer under full angular misalignment using a planar-type coil structure and a 3-D rotating magnetic field is presented.

The thread pool and the I/O multiplexing technology can effectively reduce the consumption of resources and enhance the concurrency of a server, which improves the capability of the system. Architects tend to use multiple threads to improve system concurrency, which has been proved to be effective in the past few years. However, the cost of creating, switching, and destroying threads remains to be an assignable problem when the system is having a concurrent peak. This paper designs and implements a server based on the technologies of the thread pool and epoll to figure out if the mixture of epoll and thread pool as optimization can help decrease thread operation costs. To deal with the system overhead, we introduce the thread pool technology based on an epoll web server to avoid the cost of frequent operations of creating or destroying threads. We build two sets of experiments to verify the effectiveness of our optimization, including a web server with thread pool and epoll as the experimental group, and a simple epoll web server without thread pool as the control group. Then, we test the system performance by altering the concurrency of each experimental and measuring the actual-time system overhead to describe the improvement we made to the multi-threaded server. The results show that the concurrent server with the technology of epoll and thread pool can achieve high performance and high concurrency.

The objective of this paper is to compare the performance between the anti-windup technique and the saturation degree coefficient skill for the input saturation problem. In this paper, two kinds of saturation adaptive control methods, which are saturation coefficient method and auxiliary system method, are proposed for the arneodo chaotic system with uncertain parameters and input saturation. The stability of closed-loop system under both control strategies is analyzed by using lyapunov theory. Then the comparison is made, and the result shows that both mthods have outstanding performance. But the scheme based on the saturation degree coefficient skill has better performance than the other scheme in terms of the root-mean-square of the tracking error and the root-mean-square of the control energy.

Switched Reluctance Motor (SRM) has developed rapidly and is widely used. At present, SRM has shown strong competitiveness in the fields of traction and transportation, electric vehicles, aerospace industry, household appliances, and servo control. However, due to its nonlinear and strong coupling system characteristics, its torque ripple, noise and vibration are severe, which limits the application range of SRM in practice. In order to suppress the torque ripple, this paper adopts the method of combining direct torque control (DTC) and fuzzy PI controller to design the fuzzy PI speed loop controller, and uses MATLAB/SIMULINK to carry out the direct rotation of the switched reluctance motor. Torque control and simulation research based on fuzzy PI controller. Comparing the simulation results, the direct torque control torque waveform based on fuzzy PI is better than direct torque control, which improves the dynamic and static characteristics of the motor and strengthens the stability of the motor.

For traditional typewriters, due to the low accuracy of the servo motor applied in the Z-axis, the pen may not reach the correct position. This project aims to improve the precision and robustness of the output of the traditional 3DOF typewriter by redesigning the gripping device based on fuzzy PID control. The desired positioning problem can be transformed into maintaining a constant force between the pen and the paper. In this way, to accomplish the fuzzy PID logic control, a twophase hybrid stepper motor is used to replace the servo motor, a transmission device based on a gear set is designed to amplify the motion, and a high-accuracy force feedback instrument is added to provide force feedback information. Design and virtual simulation experiments in micro-stepping operation are implemented in detail to track the transformed force problem in this paper. With mathematical models, the control problem exhibits dynamic uncertainties and nonlinearity. Some well-expected performances of the fuzzy PID controller are presented to show the success of improving the stability of the output.

Unmanned express vehicle is facing the problem of slow driving speed. In this paper, a new structure of unmanned express vehicle was built in Solidworks and simulated the moving motion in Adams to compare with those present structures of unmanned express vehicles. Comparing the vertical acceleration of the new structure and present structure by different speed and occasion. The root mean square of the new structure is smaller. The new structure of the unmanned express vehicle has a better stability and riding comfort than the present structures of unmanned express vehicles.

Taking the side bushing of dry-type SF6 gas insulated converter valve and DC through wall bushing as the research object, the field test technology and fault treatment scheme of two kinds of bushing are studied respectively. First of all, put forward the research basis and field test results of the insulation and equipment, and improve the reliability of the bushing according to the research results of the current insulation, test conditions and on-site commissioning, as well as the on-site test results, and improve the reliability of the bushing. Based on the analysis of different fault forms, fault properties and causes of the bushing on the side of the dry-type SF6 gas insulated converter valve and the DC through wall bushing, and combined with the relationship between the operating characteristics and the insulation state of the bushing. This paper analyzes incompleteness of the current test items on the performance assessment of the bushing in the field handover, pre-test and maintenance stages of the bushing, The field test methods and requirements for the dielectric loss spectrum curve, DC resistance, decomposition composition of SF6 gas, dissolved gas in oil and other items of dry-type SF6 gas insulated converter valve side bushing and DC wall bushing are put forward. According to the background interference law of local discharge in the valve hall of the current project converter station, the net insulation size of the valve hall and the technical level of DC source equipment, the improvement measures of on-site DC withstand voltage test of side bushing and DC wall bushing of dry SF6 gas insulated converter valve are put forward.

Aiming at the problem of difficult allocation of the dq-axis current term weight coefficients of the cost function in the traditional finite control set model predictive current control of a permanent magnet synchronous motor, a finite control set model predictive current control method based on a neural network to optimize the cost function is proposed. In this paper, the neural network is constructed to assign the weights of the cost function by judging the changes of the speed deviation and the speed change rate, and the method improves the dynamic performance of the system and reduces the influence of the unreasonable assignment of the dqaxis weight coefficients on the system. The proposed method is verified by simulation analysis.

In order to improve the position tracking accuracy and system control performance of permanent magnet synchronous motor in a complex environment, an improved backstepping sliding mode position tracking control method based on traditional sliding mode control is proposed. The sliding mode reaching law is optimized, the exponential reaching law is combined with the power reaching law, and the traditional symbolic function is replaced by the hyperbolic tangent function. According to the backstepping control principle, the appropriate Lyapunov function and virtual control quantity are obtained, and the position servo system controller is obtained by combining the sliding mode control. Aiming at the influence of parameter perturbation and external uncertain disturbance on the motor, the nonlinear disturbance observer is used to estimate it and feedback it to the backstepping sliding mode controller for compensation. Finally, the simulation in Matlab/Simulink shows that this method improved the response speed of position tracking, reduced the tracking error, and improved the anti-interference ability.

There exist problems such as long-time queuing, long-distance walking and complex process in traditional on-boarding mode, which cause longer time-consuming and lower competitiveness of civil aviation transportation. Aiming at this problem, Go-ahead Check-in mode and resource optimization allocation plan were proposed in order to effectively shorten on-boarding time. Firstly, the ‘put one equipment but carried more passengers’ target optimization model was established in the terminal for the purpose of putting minimum of the equipment quantity and shortening passengers’ onboarding time. Regarding uncertainty of each passenger’s on-boarding time, the volatility factor was introduced to improve the target optimization mode adaptability. Secondly, the improved NSGA-II algorithm was employed to solve the model and get the resource allocation plan in Go-ahead Check-in model. Finally, according to the analysis on the passengers’ departure data in a certain airport terminal and considering three conditions including low, medium, and high passenger flow, the average on-boarding time in the Go-ahead Check-in model has been respectively reduced by 23.80%, 7.34%, and 7.17% compared with the traditional mode. The result shows that the comprehensive performance index of the algorithm is improved.

By arranging displacement measuring points on the piston rod of the reciprocating compressor, the axis track diagram of the piston rod of the reciprocating compressor is obtained, and using the improved method of discrete point contour envelope of the axis track, the results are obtained under different air volume adjustment conditions of the reciprocating compressor. An envelope that is closer to the shape of the pivot position. The weights are calculated by the ReliefF method, and the RS features are obtained, and the feature analysis method after merging with the time-frequency domain features, obtains a very accurate fault feature identification method in three stages: normal piston rod, early fault, and fault deterioration.

In view of the complex operating environment of high-speed trains, this paper considers the additional resistance of complex lines and the influence of random external disturbances. Aiming at this problem, a sliding mode robust control algorithm for high-speed trains is proposed based on Hamilton-Jacobi Inequality (HJI) theory and radial basis function neural network (RBFNN). On the other hand, the introduced RBFNN can be used to reduce the dependence of the controller on train model parameters, and HJI theory can be used to ensure the anti-jamming ability of the system. The Lyapunov function proves that both the displacement tracking error and the velocity tracking error can be converged and stable under the method proposed in this paper. The parameters of the CRH380A high-speed train are used for simulation, and the given target speed and target displacement curve are tracked to verify the feasibility of the proposed control algorithm. The simulation results show that the proposed control algorithm has better tracking accuracy for a given speed and displacement than the traditional robust adaptive control method (TRAC) and can meet the requirements of punctual operation and fixed-point parking required by high-speed trains. It has a better control effect when dealing with complex road conditions changes and also has the more vital anti-interference ability for random external interference.

Aiming at the shortcomings of marine predators algorithm, such as insufficient operation accuracy and easy to fall into local optimization, an improved marine predators algorithm with hybrid-strategy is proposed. Chebyshev chaotic map is introduced in the initialization stage of the algorithm to homogenize the population distribution and reduce the randomness of the operation results. As the algorithm enters the later stage of iteration, the degree of individual assimilation gradually increases, and the Fish Aggregating Devices effect of marine predators algorithm is not enough to help the algorithm break away from local optimization. The Cauchy mutation strategy is introduced as the perturbation term to help the algorithm identify the local optimal solution more clearly. The experimental results show that the improved marine predators algorithm with hybrid-strategy improves the overall performance of the algorithm.

To improve the short-time forecasting accuracy of PM2.5 concentration, this paper presents a GA-SVR forecasting method based on support vector regression (SVR) and genetic algorithm (GA). GA can optimize the hyper-parameters of the SVR model to obtain higher forecasting accuracy. The inputs of the model contain air pollutant data, meteorological data and seasonal features. To confirm the effectiveness of the proposed method, models have been trained and tested based on two public data sets and compared to other machine learning methods.

Traditional active distribution networks generally use single-objective optimization algorithms for fault location. The objective function of the single-objective optimization model is weighted by each sub-objective function. Improper selection of weights will affect the location results, resulting in misjudgment or missed judgment. In this paper, a multiobjective optimization model suitable for fault location of active distribution network is proposed, and a non-dominant sorting genetic algorithm with elite strategy (NSGA-III) is used to solve the optimization problem. NSGA-III algorithm does not need to consider the weight problem, has good convergence and will not fall into local optimum. Finally, a modified IEEE33-nodes distribution network model is built by PSCAD, the three cases of information mistransmission of the intelligent terminal unit are considered, and the simulations are carried out from two aspects of single and multiple faults. It is verified that the active distribution network fault location method based on NSGA-III algorithm can still accurately locate fault segments with high reliability under single or multiple types of information misinformation.

In response to the national sustainable development strategy of energy utilization, good results have been achieved in the research of distributed power and microgrid technology in recent years, and more clean energy has been utilized to become a member of the microgrid. The output of the type power supply is easily affected by natural conditions. If it is not strictly regulated and directly input into the large power grid, it will bring adverse effects to the system.

Optimizing the support vector machine (SVM) classification model based on genetic algorithm (GA) can significantly improve the classification accuracy of traditional SVM. In order to solve the problem of abnormal power consumption classification in the public transformer area, this paper analyzes the classification methods based on SVM and GA-SVM. This paper extracts 15 eigenvalues, including parameters such as voltage, current, power, phase angle, power factor, etc., based on the research objects of 358 different users' electricity consumption information in multiple stations in Baoding, and divides the corresponding electricity consumption behaviors. There are 7 kinds of classification results, and then the data is classified based on SVM, and GA is used to find the optimal penalty factor C parameter and the g parameter in the kernel function, and the kernel function is the radial basis function (RBF). Finally, the above parameters are input into the GA-SVM model. The experimental results show that the trained model can quickly and accurately classify the abnormal power consumption behavior of the public transformer station area, and a new exploration of the abnormal behavior recognition algorithm in the public transformer station area has been carried out to ensure the power consumption quality of the public transformer station area. Further standardize the order of electricity supply.

As COVID-19 became a pandemic in the world, wearing a mask has become one of the best measures to prevent the spread of the epidemic, so face mask recognition in public places has become a very important part of controlling the epidemic. This paper mainly tests the performance of the OpenCV DNN preprocessing model (OpenCV DNN + SVM) based on the SVM algorithm model in the face mask recognition dataset. The dataset I use is from Kaggle called COVID Face Mask Detection Dataset. This dataset contains 503 face images with masks and 503 face images without masks. I test the performance of using OpenCV DNN + SVM and using only the SVM algorithm to evaluate this study by setting a control experimental group. In this study, it was found that using OpenCV DNN + SVM, the accuracy of ROI parameters and SVM parameters can reach 93.06% and F1score can also reach 93.06% without a lot of adjustment. The accuracy rate can only reach 68.31%, and the F1score reaches 68.31%. Findings suggest that the method using OpenCV DNN + SVM can achieve slightly better results in the COVID Face Mask Detection Dataset, and can perform better than only using the SVM algorithm. In addition, using OpenCV DNN preprocessing model based on the SVM algorithm plays an important role in feature extraction in face mask recognition. If the developer does enough parameters tuning, the accuracy will also increase.

Flight control signals of unmanned aerial vehicles (UAVs) are generally communicated by frequency hopping (FH) in spread spectrum mode. The detection method based on the back-end signal processing can not be widely used in frequency hopping signal detection and identification under different parameters, and has no practical value in engineering. In view of the above problems, this paper proposes a UAV communication technology based on image signal transmission, whose core content is sliding shift cyclic autocorrelation. The technology extracts the periodic characteristics of the internal cyclic prefix of OFDM signal, and establishes the characteristic database of multiple UAVs and Wi-Fi in advance. Compared with the traditional method, the recognition rate of the proposed method is 6. 74% higher than that of other methods, and the blind detection algorithm is 3. 1% better than that of other methods. It is verified that the proposed method can effectively prevent and solve the problems of frequency offset, system instruction update and surrounding electromagnetic interference. It contributes to the improvement of UAV signal transmission speed and performance.

Gun lunched projectile often has a challenge problem associated with the dynamic system is the modeling inaccuracies resulting from the rapid changes of atmospheric properties and aerodynamic characteristics at high flight velocity. A maximum likelihood method based on interior point algorithm to estimate the aerodynamic parameters for symmetric spinning projectiles is proposed in this paper. The proposed algorithm can improve the accuracy and efficiency compared with traditional maximum likelihood estimation, and reduce the dependence of initial conditions and keep the aerodynamic estimation results in a reasonable scope for a spinning projectile. Simulation results shows that the algorithm proposed in this paper can effectively identify the aerodynamic parameters with high precision and a quick convergence velocity and is therefore suitable for use in actual engineering.

In the actual trading process, investors can only give the best daily trading strategy based on the past price data of gold and bitcoin, then they need to predict and evaluate the trend of the investment items in the coming period and plan out the trading scheme in advance. We also draw on data from many investment questionnaires on websites such as Stock Market Analysis & Tools for Investors to give specific trading strategies. We choose the XGBoost regression price prediction model and enable the genetic algorithm to find the best learning rate parameters. The first 100 trading days of gold and bitcoin data are taken separately for learning training tests, and then the first 20 data are used to predict the price trend for the next five days, which is repeated every day. It provides more accurate prediction results based on the latest prices. An optimization model is established to increase the final investment value by judging the buying and selling indexes by whether the expected return exceeds the purchased commission.

To address the problems of slow convergence process and low accuracy of final convergence results in butterfly optimization algorithm (BOA), a symmetric augmentation optimized population and parameter dynamic adaptive butterfly algorithm (PEAPBOA) is proposed, which optimizes the initial population by symmetric augmentation combined with dominant population method to improve the speed of algorithm convergence, and then crosses and mutates the individuals with poor adaptation. The algorithm further introduces dynamically changing sensory modality c, power exponent a, dynamic switching probability p and position updating dynamic weights w1 and w2, so that the algorithm focuses on global search in the early stage and local search in the later stage, which improves the breadth of global search and the depth of local random search of the algorithm, thus improving the convergence speed and convergence accuracy of the algorithm. The test experiment verifies the effectiveness and superiority of the improved butterfly algorithm proposed in this paper.

Travelling Salesman Problem (TSP) is an NP-hard problem in combinational optimization important in operations research and theoretical computer science. The all-nearest-neighbor (NN) algorithm is a typical algorithm to solve TSP due to its brevity and effectiveness. However, as graphs become larger, the cubic growth of computational time makes sequential algorithms unfavorable. This paper presents two parallel all-nearest neighbor algorithms - the coarse-grained parallel model (PnnCG) and the integrated parallel model (PnnIN) to optimize the runtime of the sequential NN for TSP on large graphs. PnnCG employs a top-level parallelization to simultaneously search for the shortest tours within subgroups of the cities. PnnIN adds a lower level of parallelization based on PnnCG by again dividing the cities into subgroups when searching for the nearest city of the current one. To verify the effectiveness of PnnCG and PnnIN, we compare them with NN on ten TSP benchmark graphs whose sizes range from 48 to 11849. PnnCG reaches a speedup of 2-3 on graphs with a size from around 50 to 11500. PnnIN, though slower in smaller graphs, outperforms PnnCG on graphs with a size larger than 11500. Our analysis demonstrates that the proposed parallel models significantly decrease the runtime needed.

In order to solve the problems of less prominent features in infrared and visible image fusion, an infrared and visible image fusion method based on compound decomposition and region feature ratios is proposed. Firstly, two original images are decomposed using nonsubsampled contourlet transform(NSCT) separately, and the low frequency coefficients and high frequency coefficients are obtained. Then the two-scale decomposition is used to decompose low frequency coefficients into low frequency base subband coefficients and low frequency detail subband coefficients. Secondly, for the low frequency basic and detail subband coefficients, the fusion rules of region feature ratios and absolute-maximum-choosing scheme are adopted respectively. High frequency subband coefficients are fused by region-energy-maximum scheme. Finally, the fused low frequency subband coefficients are obtained by two-scale inverse transform. The fused low frequency subband coefficients and fused high frequency subband coefficients are inversely transformed by NSCT to obtain the fused image. The experimental results show that the proposed method outperforms to conventional fusion methods in terms of objective evaluation and subject quality

Travelling Salesman Problem (TSP) returns the minimal length of traveling distance to travel through all the cities given by the data set and returning to the starting position, which allows the users to optimize their travel path to minimize total cost. Genetic algorithm (GA) is used to solve the problem due to its efficiency. However, as the sample size increases, the runtime of GA for TSP increases significantly. This paper introduces three new parallel GA (PGA) – Master-Slave model, Coarse-Grained model, and Combined PGA (MSCG)–to solve TSP to optimize performance and minimize runtime. The first two performs parallelization on a different level and Combined PGA is the combination of Master-Slave and Coarse- Grained models. The Master-slave model paralyzes the evolution process and divides the population into threads to enter the calculation. Coarse-grained separates the population even earlier, so before entering evolution, the populations are already divided into subsets. MSCG separates populations both before and after entering evolution to get the advantages of both approaches. To verify the effectiveness of the three proposed methods, we compare them with the non-parallelized GA. The results show that the master-slave method generally would produce on average 10% shorter route than the coarsegrained method but would have on average 40% higher time usage.

To respond to national policy requirements, reduce carbon emissions in the construction industry, and achieve the goal of "carbon peaking and carbon neutrality" as soon as possible. In this paper, a system dynamics model is proposed to predict and analyze carbon emissions in the construction industry in Liaoning Province. First, the influencing factors of carbon emissions in the construction industry in Liaoning Province are sorted out by searching relevant literature. Secondly, combined with the relevant data of the construction industry in Liaoning Province from 2009 to 2019 and the "14th Five-Year Plan" policy, set different simulation scenarios and use Vensim software to reasonably predict the carbon emissions of the construction industry from 2020 to 2030, and find out the impact of carbon emissions. The internal logical relationship of the main factors provides a theoretical basis and emission reduction path for reducing the energy consumption of the construction industry, reducing the total carbon emissions, and realizing the green and sustainable development of buildings.

With the rapid development of the social economy, various industries' water consumption is growing rapidly, and the sewage discharge is also increasing. This paper studies the modeling of the water quality of the Yangtze River. First of all, this paper uses the three-parameter comprehensive evaluation index to express the average pollution index of the observation stations of the Yangtze River in the past two years, makes a comprehensive evaluation of the water quality of the Yangtze River, obtains the comprehensive evaluation index, and analyzes the pollution situation of each region. Then the amount of permanganate and ammonia nitrogen polluted in the river section between the adjacent observation points on the mainstream (kg/ day) is used to evaluate the pollution status of the river section and analyze the seriously polluted areas. Finally, according to the grayscale principle, the gray prediction model is used to process the data, reduce the random influence of the original data, and predict the wastewater discharge of the Yangtze River in the next ten years.

Early identification of retinopathy is of great importance in reducing blindness in diabetic patients. With the rapid development of artificial intelligence, automatic diagnosis technology of diabetic retinopathy appears. Based on the opened image data of clinical patients with diabetic retinopathy, this article establishes feature engineering through feature selection, and derives an automatic diagnosis model of retinopathy by using three machine learning algorithms: GBDT, KNN and SVM. The model is verified by real data, while the accuracy and AUC value are used to evaluate the models established by the three algorithms. Both the accuracy from ten-fold cross-validation and AUC value of GBDT are the highest, which are 0.827 and 0.803 respectively. The results show that the automatic diagnosis model of retinopathy based on GBDT algorithm has the best performance and can be an invaluable aid to the clinical diagnosis of diabetic retinopathy.

Graph neural networks (GNNs) are approaches that extend deep learning neural networks on graph data. Research on graph neural networks has made tremendous progress today. Graph neural networks are usually categorized as spectral-based models and spatial-based models. The spectral-based method has been widely recognized by the academic community due to its solid theoretical foundation. However, the existing spectral-based models induced by the Laplacian matrix usually cannot achieve satisfactory results in experiments due to their insufficient expressive ability. We theoretically derive an unbiased Laplacian matrix based on biased random walks. As a graph shift operator, it is more general than unbiased Laplacian. Based on biased Laplacian, we propose a more powerful spectral-based graph neural network BiGNN. And it achieves better simulation results than traditional spectral-based graph neural networks on Cora, Citeseer and PubMed datasets.

In order to solve the difficulties of multi-scale feature extraction and weak representation in remote sensing image scene classification, a classification method based on multi-scale feature fusion (MFF) was proposed. The convolutional representation and fully connected features generated by the feature fusion of the MFF are used as high-level features to generate discriminative scene representations, which are then input into the softmax classifier to obtain the semantic labels of scenes. The existing convolutional neural network-based methods and MFF methods are tested on three widelyused datasets. The results show that the MFF method has higher overall accuracy than the existing convolutional neural network-based methods and can better meet the current demand for remote sensing image scene classification.

Recently, more and more server tasks are done using full automation, including grading tasks for students in the college courses, integrating tasks for programmers in big projects and server-based transactions, and visualization tasks for researchers in a data-dense topic. Using automation on servers provides a great possibility for reducing the burden on manual tasks. Although server tools like CI/CD for continuous integration and Hexo for automated blog deployment have been developed, they’re highly dedicated to certain functionalities and thus lack general usage. In this paper, we introduce a Golang-based automation framework that reacts to the events happening on GitHub in a multi-thread approach. This framework utilizes a queue to arrange the tasks submitted and execute each task with a thread in a preemptive manner. We then use the project GoAutoGrader to illustrate a specific implementation of this framework and its value in implementing high-freedom server applications. As Golang is developing in a rapid way because of its incredible parallel programming efficiency and a super-easy way to learn on the basis of C-like programming languages, we decide to develop this system in Golang.

The short-term load forecasting model based on convolutional neural network optimized by the chemical reaction optimization algorithm is proposed. Convolutional neural network is a feedforward neural network with deep learning ability. The adjacent two layers of neurons of convolutional neural network adopt sparse connection, and the neurons in the same layer share weights. The samples are pooled in the space for convolutional neural network. Due to these unique mechanisms, the data can not only keep the original features as high as possible and convert them into abstract features, but also reduce the complexity of the neural network and improve the prediction efficiency. The chemical reaction optimization algorithm is used to train the convolutional neural network. Chemical reaction optimization algorithm simulates the process of interaction between molecules in the process of chemical reaction and finally reaching a stable low energy state. Through the simulation using the actual load data of distribution network, simulation results show that the proposed model has satisfactory prediction effect.

With the development of the economy, the Internet has become an indispensable part of people's lives. At the same time, due to the development of Internet of Things, big data, cloud computing and other technologies, the requirements of Internet users for communication systems are also increasing. Network application scenarios such as in-vehicle networking, ultra-high-definition video transmission, and real-time data transmission are emerging one after another. The breakthrough development of 5G communication technology has opened a chapter in the 5G era and has become the key to solving cellular communication technology. However, in the architecture of advanced networks It is still a big problem to solve the problem of load balancing and realize the reasonable scheduling of network resources. At the same time, it has also become a hot spot of current research. In the research of this paper, for the resource allocation and scheduling problem under the new SDN architecture in the core network, based on the network balancing algorithm, an algorithm based on reinforcement learning is designed, and the controller is designed. The scheduling system promotes the refined management of staff and can adapt to changes in flexible and effective market strategies.

The existing low-light image enhancement algorithms could not process brightness, contrast, color, and other details at the same time. Thus, a multi-scale low-light image enhancement network which fused with dense residual blocks is proposed. The image is first generated with feature-rich input through an input module, then the features are fed into a multi-scale backbone enhancement network with dense residual blocks, and finally a refinement module is used to enrich image details and remove halos. The experimental results show that the proposed method better improves the contrast and brightness, actualizes the color, enriches the texture details, and decreases the noise and artifacts in low-light images, which quantitatively and qualitatively demonstrated its advantages comparing with other mainstream methods.

The economic dispatching model of distribution network based on multi-verse optimizer algorithm considering reliability is proposed. The economic dispatching model of distribution network considering reliability mainly considers the two objectives of operation cost of distribution network and reliability cost of distribution network. The constraints of the model mainly consider the active power balance and the operation constraints of each device. The multi-verse optimizer algorithm is adopted to solve the economic dispatching model. The simulation results show that the scheduling effect can meet the requirements.

The advent of the Internet economy era, while bringing convenience to consumers lives, has brought new opportunities and challenges to the precise positioning of network marketing. Traditional marketing methods often use massive coverage advertising to promote products, reduce prices, discounts, and other promotional preferences to attract customers, which makes it difficult to accurately meet consumer demand, resulting in low advertising conversion rate, huge marketing costs and a serious waste of resources. This paper establishes a precision marketing model for the network community and obtains the characteristics of consumer behavior in the network community, accurate mining and analysis of the network community. Finally, through the experiment to test the effect of the precision marketing model for the network community, some suggestions on the construction and optimization of the precision marketing model and the network community are obtained to get the enlightenment of promoting the steady development of the sales industry and provide a reference for sales enterprises to improve the precision marketing mode and build and maintain their network community.

The combination of culture and tourism is becoming closer and closer, and in the process of tourism, how to make the public more convenient and accurate access to the relevant content of the culture accumulated for thousands of years is also a problem that all parties are trying to solve. In this paper, under the background of the widespread popularity of mobile terminals and the continuous enrichment of image and text data, the content-based image feature vector processing is carried out by convolutional neural network method, which is combined with the collected data set of cultural signs of tourist attractions to retrieve the main content of the image from identifying the main content. The applied system can be used as an image retrieval access of an image database of cultural attractions or cultural relics. The experimental results show that the average accuracy of the proposed algorithm is 77.6, which is better than other mainstream algorithms. As an important initial identification link, it can create conditions for the study of interest recommendation of cultural content and tourism path planning.

Distributed file systems are widely used in the cloud environment, which simplifies the process of accessing files in different machines. By using the distributed file system, commercial cloud storage like Google Drive, OneDrive, and iCloud offers convenient ways for users to access personal files. However, cloud storage has privacy issues, and the expense of purchasing these services is expensive. Therefore, a distributed file system for personal use that utilizes users’ vacant local disks is desired. This paper presents a simple distributed file system, the Individual Distributed File System (IDFS), that enables users to hold their data at their hands conveniently. IDFS allows users to access files that locates in personal devices in different network environments seamlessly. To this end, IDFS uses the wireless device and relay connection of the server. Instead of storing all files in centralized cloud servers in the public network, IDFS keeps files in personal devices, which is promising for keeping personal files private. To evaluate the effectiveness of IDFS, we design a prototype of IDFS and implement several experiments. The results show that IDFS makes an offline device-to-device file transfer quick and convenient, and it provides high file accessibility without consuming too much storage.

The 5G network slicing technology can greatly improve the network transmission efficiency of users and significantly reduce the transmission delay. Network Functions Virtualization (NFV) provides a new solution for more flexible network deployment and more efficient resource integration. The allocation of virtual resources is often realized through the deployment of Service Function Chain (SFC). This paper studies the problem of virtual network resource allocation under 5G network slicing. By designing an efficient SFC deployment model and VFN backup method, the virtual network resource allocation with high reliability and intelligent offload is realized.

With the development of artificial intelligence, the object detection model based on deep learning has also achieved great results. The detection model has also developed from the traditional manual extraction of features to the current neural network extraction. The classic single-stage detection model is based on YOLO series is representative. However, with constant research, it is discovered that the detection model based on deep neural network also inherits the shortcomings of neural network and is vulnerable to adversarial attacks. This paper proposes an optimized attack algorithm based on PGD, which realizes the adversarial attack on the YOLOv4 object detection model. Experiments have proved that this attack method in this paper reduces the mAP indicator from 87.61% to 0.12% on the VOC data set, and from 69.17% to 0.37% on the COCO data set. It has a certain improvement in the evaluation indicators PSNR and SSIM, and the attack effect Compared with the original PGD, the quality of the generated adversarial example is better.

In unmanned logistics end distribution, task distribution is a problem that must be solved. In this paper, we calculate the distribution range that can be covered by multiple UAVs, and make distributed interconnection communication and calculation between multiple intelligent bodies, i.e., between multiple UAVs and ground platforms. In the process of executing the task, the distribution result is adjusted by real-time replanning, so that the total distance in the distribution result is minimized. After the UAV takes off and starts to execute the distribution task, the objective function model is established based on the real-time location of the UAV, the location of the end point of the movement, the length of the path of the task execution, and the time collaboration of the multi-UAV execution task as constraints, and solved using the Improved Genetic Algorithm. Through the results of simulation experiments, we can see that the task allocation strategy used in this paper can perform resource calculation more efficiently and get optimized allocation results.

Since the Semi-Airborne frequency domain receiving system adopts uav carrying the receiving system to realize motion measurement in the air, the receiving system will inevitably introduce interference noise, including UAV motion noise, environmental noise and human noise, etc. In order to improve the signal-to-noise ratio, the adaptive noise elimination technology is used to cancel the noise. The step size of traditional filter is a fixed value. Although the convergence speed is guaranteed, the steady-state error after convergence fluctuates greatly. In this paper, a new mathematical model of step factor and error signal is proposed by transforming sine function and introducing variable parameter R factor. The algorithm is applied to the Semi-Airborne electromagnetic receiving system, and the collected data are filtered by the algorithm, and the expected experimental data are obtained.

Episodic reinforcement learning (ERL) is a class of algorithms that use episodic memory for improving the performance and the sample efficiency of reinforcement learning (RL). Although it has achieved some success, existing ERL algorithms have to interact with the environment for many rounds to gain satisfying performance. In this paper, we propose the algorithm episodic memory by forcing state representations (EMSR) to improve the performance and sample efficiency of ERL. Specifically, our algorithm uses a transition model to predict the hidden state representations of the agent’s multiple future steps for augmenting reward maximization, which can help the agent learn quickly. In this way, our method can achieve better performance and higher sample efficiency than previous state-of-the-art algorithms. Experimental results demonstrate the superiority of our method.

In order to improve the positioning accuracy of the boom end of the dual-arm rock drilling rig, first, the boom structure was simplified. A kinematic model from the end of the boom to the geodetic coordinate system was established. After that, the flexible deformation analysis and calculation of the main boom and the propelling beam were carried out. A parameter error model that couples the positioning conversion error and the flexible deformation error of the car body was established. A self-organizing genetic algorithm was proposed. The probability of mutation changed periodically. The mutation probability changed periodically to avoid falling into the local optimum. The fitness function comprehensively considered the errors of the root of the main boom, the front end of the main boom and the end of the drill boom to identify the optimal joint parameter errors. Finally, a total station was used to observe the key position coordinates of the drill boom and collect experimental data to verify the effectiveness of the proposed algorithm. The research results show that: compared with the average position error before error compensation, the root error of the main boom is reduced by 21.06%, the front end error of the main boom is reduced by 86.41%, and the end error of the boom is reduced by 80.43%, which effectively reduces the positioning error of the whole boom.

Facing the demand for key equipment of UHV DC transmission projects in China, and aiming at the technical difficulties of ±1000kV SF6 gas insulated through wall bushing, this project carries out the core technology research and device development of ±1000kV SF6 gas insulated through wall bushing. Through the research of this project, the problems in the design, manufacture and test during the development of ±1000kV DC through wall bushing, including the homogenization design of internal and external electric field, mechanical strength and current carrying capacity of bushing, put forward the design scheme of anti corona, anti pollution and external insulation of bushing, and master the performance of epoxy resin material for DC bushing and the electrical characteristics of epoxy resin insulator, formulate process flow, test scheme and technical standards. Finally, the prototype of ±1000kV DC SF6 gas insulated through wall bushing passing the type test is developed to meet the engineering application conditions and strive to realize the engineering application. The research of this project will solve one of the key difficulties restricting the development of ±1000kV UHVDC project in China, promote the manufacturing level of UHVDC key equipment in China to rank among international advanced ranks, and improve China's high-tech research and development capacity and the international competitiveness of the industry.

The elbow joint is prone to stiffness and adhesion after trauma or surgery. High-energy trauma can easily lead to loss of mobility of the elbow joint. Mild trauma can also cause stiffness in the elbow joint. In recent years, despite the remarkable progress made in the treatment of trauma to the elbow joint and surrounding tissues, postoperative elbow joint contractures are still very common. The improved elbow orthosis can provide a portable rehabilitation environment for the elbow joint after the operation, which is not affected by the environment, can be used for rehabilitation exercises. Bioimpedance spectroscopy (BIS) can quickly and accurately obtain mechanism information through the analysis of bioelectric signals, and has the characteristics of highspeed, portability, and non-invasiveness. Therefore, this paper proposes a technical solution for elbow joint orthosis based on the combination of bioelectrical impedance spectroscopy and GRNN neural network. A feedback type elbow joint control system based on GRNN network is proposed, which realizes the control strategy of converting the patient's elbow joint pathological information into elbow joint orthosis control information. The data obtained in the experiment were processed by the improved Cole-Cole model, partial least squares method, and improved VMD-HHT model. As the elbow joint continues to heal, the internal changes are the first to produce a large amount of extracellular fluid. And the conclusion that the wound can be healed through cell division. And input the information into the GRNN network for training, and finally adjust the force applied by the elbow joint orthosis to the patient's elbow joint through the training results, and achieve good results.

Mobile robots are widely used in various fields, such as unmanned aerial vehicles (UAV), unmanned ground vehicles (UGV), unmanned underwater vehicles (UUV). It is because the single robot is only able to complete relatively simple tasks (while being helpless for complex and large-scale tasks) that a multi-robot system has emerged. The cooperation between multiple robots will not only promote the efficient completion of tasks and enhance the reliability of the system but advances cooperation to complete additional complex tasks. For a multi-robot system, the path planning and control of the robot is the basis for the successful completion of a particular task, which is simultaneously the key technology that needs to be solved. The application of warehousing robots can greatly improve the efficiency of e-commerce warehousing and logistics, as well as alleviating current situations of short supply in the warehousing and logistics industry. This article provides a multi-robot cooperative motion control method to solve the collision avoidance problem while the warehouse robot performs tasks.

1100kV UHV AC bushing is the capacitive structural bushing with the highest voltage level in China. Its aging performance under the combined action of high voltage and high current needs to be studied and analyzed by on-site quantitative test. In order to fully simulate the actual operation environment of UHV bushing, the full-size 1100kV UHV bushing is used as the long-term electrothermal joint assessment object. Therefore, the development of all working condition aging platform for UHV capacitive structure has become the focus of research. The development results of the aging platform show that the UHV oil-air casing adopts the "vertical" structure installation type, and the UHV oil-gas casing adopts the "horizontal" structure installation type. The two types of casing are connected in series through the central conductor, which can ensure that the casing can bear the same voltage and current value. The UHV capacitive bushing is installed on the aging platform, the radial and axial electric field intensity inside the core meets the control requirements of 3kV/mm, and the electric field intensity on the surface of key fittings inside the oil filled tank is lower than the control value of 20kV/mm. In this paper, the electric heating joint test and assessment scheme platform of UHV voltage grade capacitive bushing is proposed for the first time, and the strategic scheme of optimal layout and reasonable voltage sharing of two UHV grade capacitive structural bushings in the limited space is given, which comprehensively assesses the performance of 1000kV AC bushing (Oil-SF6, oil air), so as to provide strong support for the engineering application of the 1000kV AC bushing. It provides theoretical and practical basis for realizing the real localization and on-site operation and maintenance of 1000kV AC bushing.

Unmanned surface vehicles (USV) has been widely used in the fields of surface quality monitoring and dangerous water exploration, among which the detection and segmentation of obstacles is an important condition for USV to realize autonomous obstacle avoidance. Recently, most of the segmentation methods of water surface obstacles are based on CNN, but due to the limited receptive field of CNN, most of the methods do not achieve satisfactory results in the segmentation of obstacle edge details, sea-sky segmentation and surface reflection. To solve the above problems, we propose a water surface obstacle segmentation network WSSS with sea-sky-line detection as the prior knowledge for feature fusion, Swin Transformer as the dominant backbone and feature pyramid network (FPN) as the feature fusion network. In this network, sea-sky-line detection is used as a preprocessing module to increase the receptive field through the self-attention mechanism and sliding window, and feature extraction is enhanced by FPN, which makes the network segmentation of obstacle edge and water Sky more detailed and reduces the influence caused by reflection and weather. Extensive experiments on the USV datasets MaSTr1325 and Modd2 show that the proposed method WSSS outperforms the current state-of-the-art methods for water surface obstacle segmentation.

Three-dimensional (3D) printing is now becoming a huge discipline and has innovated the manufacturing method in biomedical engineering over these years. By applying this 3D printing technique, more biomedical products (such as surgical implants, orthopedics products and anatomical models) can be produced more accurately, efficiently, and economically. In this study, we will simulate the 3D printing process of one sample of bone tissue and another sample of anatomical model in virtual environment. We found that the mechanical properties is still largely determined by the choice of biomaterials and will not change significantly through the 3D printing process. By comparing printing time, yield strength of the products and other mechanical properties, we found that the material is still the determining factor of the mechanical properties of the product while different nozzle design and multi-material combinations can also influence the biomedical 3D printing process. The nozzle diameter will determine the 3D printing speed and the smoothness of surface texture. By choosing the material combinations that have better chemical affinity, the biomedical 3D printing products have better mechanical properties and less likely to fail. By applying this study, we can develop methods to optimize biomedical 3D printing process through picking different materials and different nozzle design.

Complex networks are widely applied for describing the topological structures of complex systems. Besides the exploration of the topologies for complex networks, the dynamics on the complex networks receives much attention from researchers in the field of complex systems, the techniques of computer simulations are used to intuitively understand the emergent behaviors of systems consisting of large number of interacting agents. In this paper, we aim to design to a platform for visualization of dynamics on complex networks: By setting parameters such as network topology, types of dynamics and running time, the platform can visualize agent-based micro-characters as well as system-based emergent micro-behavior for evolutionary game dynamics, opinion dynamics and infection dynamics on complex networks. In addition, it also has functions such as logic control operation, configuration of initial network state, and export of state data. Compared with other networks modeling and simulation platforms, the platform has the characteristics of multithreading, scalability and cross-platform. Especially, it can also enhance the computing efficiency by techniques of GPU parallel computation.

Disasters can have a significant impact on air quality and thus pose a threat to the health of people in the affected areas. There are no post-disaster air purification systems or specialized technologies currently. In this paper, a post-disaster air purification system is designed based on the mechanism of lung purification of mammals. It uses two-stage purification, which mainly purifies particulate matter in the air, such as PM2.5 and PM10. The first stage is coarse purification, which simulates the villi fibers in the nasal cavity to block particulate matter. The second stage simulates macrophages in the alveoli to achieve efficient purification of particulate matter. The post-disaster air purification system adopts intelligent control and introduces an innovative mechanical structure design with a special adsorption structure and guiding structure, which can effectively increase the adsorption area of particulate matter in the air and its adsorption capacity. Also, it has a simple structure and can be disassembled and assembled with less effort, which meets the special emergency needs after the disaster.

In this work, a simple single underwater glider motion model is established. By using the multi-body system theory, the active and inertia force are unified into the generalized force, and the dynamic equation of the multi-underwater glider system is described. The maintenance and transformation of formation are discussed based on potential field method. Based on the continuous potential field function and by introducing the virtual leader, a method of constructing stable formation with arbitrary shape based on mesh-shaped formation is given. Finally, based on the virtual rigid body structure constructed by virtual leader, a fast formation construction and transformation method is realized, and the feasibility of the method is verified by simulation experiments.

With the growth of civil aviation transportation, runway capacity has become the main factor restricting the growth of air traffic flow. By using the ASDE-X operating data of CTU Airport, the runway occupancy time, interval buffer and landing time interval data of inbound flights are extracted, and the influence of these factors on runway capacity under different combinations of aircraft types is compared. The results show that when heavy aircraft is used as the front plane, the runway occupancy time is the main factor of runway capacity. The runway capacity under different runway occupancy time is simulated. With the decrease of runway occupancy time, the runway capacity increases, but finally tends to a fixed value.

This paper presents a method of haze removal and computer-generated holographic display of degraded images in coal mine. Firstly, the image enhancement of underground coal mine is realized by using the dark channel prior haze removal algorithm, which greatly weakens the shielding of coal dust and water mist in the roadway environment. Next, using the computer-generated hologram algorithm based on angular spectrum diffraction, the phase only hologram is generated with the haze removal image as the input. The peak signal-to-noise ratio (PSNR) of the reconstructed image of the red channel is 65.47dB, the PSNR of the reconstructed image of the green channel is 64.98dB, the PSNR of the reconstructed image of the blue channel is 65.78dB, and the average PSNR is 65.41dB. The simulation results show that high-quality reconstructed image can be obtained by combining dark channel prior and computer-generated hologram, and the image enhancement of underground coal mine is realized.

The construction of shield tunnel cause the disruption on the urban cable tunnels nearby, resulting in the emergence of structural diseases in cable tunnels. Numerical simulation is used to derive the mechanical response law on the cable tunnel during shield tunnel construction. Different affecting aspects are explored, such as geographical location relationship, cable tunnel section size, and spacing, and finally the safety of cable tunnel is analyzed.