Satellite Edge Computing （SEC） improves computational efficiency for remote users by offloading tasks to satellite servers. However， SEC system performance may be affected by wireless transmission quality， since there is signal interference among User Equipments （UEs）. To this end， we proposed a User-Centric SEC （UCSEC） architecture to allocate Access Point （AP） clusters dynamically according to the real-time location of users and network conditions. To handle dynamic task offloading in UCSEC， we designed a Master-Slave multi-agent Proximal Policy Optimization with Gated Recurrent Unit （GRU） and Multi-head Attention mechanism （MSPPO_GMA） algorithm. The algorithm adopted a centralized control-distributed execution framework. The master agent， deployed on the SEC server， monitored global resource constraints and coordinated slave agents， while slave agents deployed on the UE executed task offloading and resource allocation according to network states. In MSPPO_GMA algorithm， GRU captured temporal correlations of network states， while multi-head attention focused on critical information of task offloading and resource allocation. Experimental results demonstrate that compared with MAPPO， MATD3 and other algorithms， MSPPO_GMA algorithm reduces task completion latency by 3.7% to 13.6% and UE energy consumption by 6.5% to 13.2%.

The difficulty， long delivery time， and high cost of last-mile delivery in rural areas make efficient and accurate last-mile delivery scheduling solutions particularly important. Aiming at the task allocation problem of multiple logistics Unmanned Aerial Vehicles （UAVs） in rural distribution scenarios， a multi-objective UAV task allocation model was established by considering the payload capacity of UAVs and the maximum flight distance of UAVs comprehensively， with the goal of minimizing the flight distance， dispatched quantity of UAVs and not violating time windows. Firstly， based on reinforcement learning， to address the problem of high dimensionality in task allocation， an encoder and attention mechanism were introduced to simplify the state space effectively. Secondly， the global-local search strategy was combined to explore the solution space while avoiding getting stuck in the local optimum， thereby improving the quality of the solution. Finally， further analysis was conducted on the parameter weight settings， and the optimal combination of weight coefficients for sub-objective functions was obtained through experiments. Simulation results show that compared to the Hybrid Q-learning network based Method （HQM）， Adaptive Large Neighborhood Search algorithm （ALNS）， Q-learning algorithm （Q-learning）， and Genetic Algorithm （GA） in terms of the obtained final path length， the proposed algorithm SG-HQM （Sine and Gaussian HQM） reduced it by 8.35%， 9.88%， 10.29%， and 12.48%， respectively.

Aiming at the mechanical system related to multi-component failure， a reliability evaluation method of multi-component system based on time-varying Copula function was proposed. Firstly， the nonlinear Wiener process was introduced to characterize the performance degradation process， and the Copula function was used to characterize the correlation between multiple component failures. Secondly， based on the evolutionary equation of the Copula function approximation of the Fourier series， the fitting effects of the Fourier series on common time-varying forms were verified by Monte Carlo （MC） simulation. In addition， the likelihood ratio statistic was used to test the existence of time-varying correlation， indicating the necessity of time-varying correlation research. The example analysis shows that compared with the static correlation model， the time-varying correlation model has the log-likelihood function value increased by 4.36%， and the Akaike Information Criterion （AIC） decreased by 3.81%， achieving more accurate reliability evaluation results.

In the reliability evaluation of many high-reliability and high-value products， product reliability often cannot be accurately evaluated due to the lack of objective test data. Aiming at this problem， a reliability evaluation method for high-reliability products based on improved evidence fusion was proposed in order to make full use of reliability information from different sources. Firstly， combining the characteristics of reliability engineering， the modified weight of each evidence was determined by the consistency of the evidence at credal level， pignistic level and the uncertainty of the evidence itself. Secondly， the optimal comprehensive weight was obtained by linear combination of each weight vector based on game theory. Finally， the Dempster’s combination rule was used to fuse the modified evidence， and the probability distribution of the product reliability index was obtained through the Pignistic probability transformation formula to complete the product reliability evaluation. The reliability evaluation results of one electronic device show that compared with the results of Jiang’s combination method and Yang’s combination method， which also consider multi-dimensional weight modification， the credibility of the conflict interval given by the proposed method is reduced by 69.6% and 54.6% respectively， and the credibility of the overall frame of discrimination given by the proposed method is reduced by 5.6% and 3.7% respectively. Therefore， in the application of reliability engineering， the performance of the proposed method in solving evidence conflict and reducing the uncertainty of fusion results is better than that of the comparison methods， and this method can fuse multi-source reliability information effectively and improve the credibility of the results of product reliability evaluation.

Bilevel programming involves two optimization problems located at upper-level （leader） and lower-level （follower）. The constraint domain of the leader is determined by the follower implicitly， the leader objective dominates in a bilevel optimization procedure， and the follower objective must be optimized with respect of the follower variables. The hierarchical structure of the bilevel optimization problem causes large computational complexity. Especially， the frequent computations of the follower can accumulate a large amount of computational cost. In order to solve this kind of problem effectively， an evolutionary algorithm based on approximation technique was developed. Firstly， a multi-population co-evolution approach was applied， and the crossover and the mutation operators were used respectively to balance the exploitation and exploration capabilities of the algorithm. Secondly， based on the sensitivity analysis theory， an approximation evaluation method for new individuals was designed to reduce the computation frequency of the follower carried out by the algorithm. The demonstration results of the approximate effect of a numerical example show that most positions of the approximate offspring individuals and the exact offspring individual are mostly coincident. In addition， the results on 10 common examples show that the proposed algorithm can find better optimal solutions than the multi-valued mapping algorithm. CPU time comparison shows that the approximate technique improves the speed of finding the optimal solution effectively， thereby reducing the running time. Therefore， the effectiveness of the approximate technique adopted by the algorithm is demonstrated.

Two key problems, endianness and memory capacity limit appear to be obstacles when electric enterprises implement a function consistency model for embedded smart meter software via disassembly technique, thus affecting the overall performance of the model. To solve these problems, a in-depth analysis was conducted combined with internal features of embedded smart meter and hardware architecture theory. Two algorithms named Code Double Inverse Preprocessing Algorithm (CDIPA) and Segmented Disassembling Algorithm (SDA) were proposed. CIDPA was used to generate adjusted binary code, together with raw binary as two inputs of disassembly. Thus endianness problem was solved by choosing the result more adaptable to hardware environment. SDA was adopted to decrease size of input binary so as to disassemble more times in limited memory. The experimental results show that CDIPA and SDA can effectively resolve the problems mentioned above and show up favorable robustness and portability.

In view of the current problems that neural network ensemble generates individual networks with low difference degree and its poor generalization ability, a method of evolutionary ensemble of neural network based on niche technique was proposed. Niche technique's good performance was used in increasing population diversity and improving local search capability of evolution, and the similarity degree's sharing function among individuals was adopted to adjust individual network's fitness. Then the individuals were selected according to the new adjusted fitness to get individual network with diversity. Theoretical analysis and experimental results show that this method can generate individual network with great difference degree and can improve the generalization ability and calculation accuracy of neural network ensemble system.

The key to body movement emotion recognition lies in extracting emotional features existed in human body movements. To solve the problems of poor emotional feature learning capability and difficulty in improving emotion recognition accuracy in existing models， a body movement emotion recognition method based on Emotional Latent Space Learning （ELSL） and Contrastive Language-Image Pre-training （CLIP） model was proposed. Firstly， CLIP model was introduced to improve the emotional feature learning capability of the model. Secondly， for the fine-grained multi-label emotion classification task， ELSL method was proposed. By learning discriminative mappings from emotional latent space to various subspaces， the subtle differences between emotion categories and the feature information beneficial to the classification of each emotion category in various emotional subspaces. Experiments were carried out on real-world open scenarios-oriented Body Language Dataset （BoLD） The results demonstrate that the proposed method makes use of the advantages of CLIP model and latent space learning in feature learning effectively， leading to significant performance improvement. In specific， compared to Movement Analysis Network （MANet）， the proposed method has a 1.08 percentage points increase in mean Average Precision （mAP） and a 1.32 percentage points improvement in mean Area Under Receiver Operating Characteristic Curve （mRA）.