With the rapid development of Large Language Models （LLMs）， dialogue assistants based on LLM have emerged as a new learning method for students. These assistants generate answers through interactive Q&A， helping students solve problems and improve learning efficiency. However， the existing conversational assistants ignore students’ personalized needs， failing to provide personalized answers for “tailored instruction”. To address this， a personalized conversational assistant framework based on student capability perception was proposed， which is consisted of two main modules： a capability perception module that analyzes students’ exercise records to explore the knowledge proficiency of the students， and a personalized answer generation module that creates personalized answers based on the capabilities of the students. Three implementation paradigms — instruction-based， data-driven， and agent-based ones were designed to explore the framework’s practical effects. In the instruction-based assistant， the inference capabilities of LLMs were used to explore knowledge proficiency of the students from students’ exercise records to help generate personalized answers； in the small model-driven assistant， a Deep Knowledge Tracing （DKT） model was employed to generate students’ knowledge proficiency； in the agent-based assistant， tools such as student capability perception， personalized detection， and answer correction were integrated using LLM agent method for assistance of answer generation. Comparison experiments using Chat General Language Model （ChatGLM） and GPT4o_mini demonstrate that LLMs applying all three paradigms can provide personalized answers for students， the accuracy of the agent-based paradigm is higher， indicating the superior student capability perception and personalized answer generation of this paradigm.

Aiming at the problems of federated spatial data isolation， spatial data indexing， and privacy of publishing spatial data， a Federated Spatial data Publishing （FSP） method based on dynamic quad-tree was proposed. Firstly， in each iteration of the FSP method， quad-tree replica was shared by the server with each client in the round， and each client encoded its own location data using the quad-tree replica， and discrete noise was generated through Polya distribution for locally perturbing the encoding results. Secondly， local masks were generated through LWE （Learning With Error） to encrypt the noisy results. Thirdly， the reported values from each client in the iteration were combined by the aggregator to perform secure aggregation and mask elimination. Then the aggregated results were sent to the server. The quad-tree structure was pruned by the server dynamically in a bottom-up way based on the collected encoding vectors and noise variance. Experimental results on four spatial datasets Beijing， Checkin， NYC， and Landmark show that the FSP method not only ensures client privacy， but also reduces the Mean Squared Error （MSE） in federated spatial data publication by 3.80%， 2.96%， 7.51% and 14.13% at a privacy budget of 1.8， respectively， compared to the existing better federated spatial data publication method AHH （Adaptive Hierarchical Histograms）. This indicates that the FSP method achieves higher precision than similar methods in federated spatial data publishing.

To address the issues of cross-media data fusion modeling and privacy protection in the hydropower industry， a dual vertical federated learning framework incorporating secret sharing technology was proposed. First， the participant nodes were stratified， with lower-tier nodes responsible for preliminary modeling， intermediate-tier nodes overseeing pre-model aggregation and optimization， and central nodes generating the final model. Then， in order to strengthen data privacy protection and prevent inference attacks， an intermediate parameter protection mechanism based on secret sharing technology was introduced， the communication data between the data owner and the model trainer was fragmented and divided， which ensured the covertness of the correspondence between the model parameters and the trainers， thereby increasing the complexity of inference attacks. Finally， in order to optimize the model aggregation process of federated learning， a node evaluation mechanism based on the disparity in information quantities was introduced， in which the node dissimilarity and data volume were assessed comprehensively. The weights of different nodes in model aggregation were finely adjusted， and the contribution of suspected malicious nodes was eliminated， thus optimizing the performance and convergence speed of the model. The real data of Guodian Dadu River Basin Hydropower Development Company Limited was selected for experiments. The results showed that： the intermediate parameter protection mechanism based on secret sharing technology was more stable during the convergence process and improves the convergence speed by approximately 14.6% compared to the differential privacy protection mechanism； by incorporating a node evaluation mechanism based on information disparity， the convergence speed was increased by approximately 13.5% compared to the federated averaging algorithm. It is verified that the proposed framework addresses the issue of cross-media data fusion modeling for hydropower data， and it possesses the advantages of data privacy protection and model convergence acceleration.

Aiming at the problem of insufficient representation ability of features extracted by the existing vehicle re-identification methods based on convolution Neural Network （CNN）， a vehicle re-identification method based on the combination of wavelet features and attention mechanism was proposed. Firstly， the single-layer wavelet module was embedded in the convolution module to replace the pooling layer for subsampling， thereby reducing the loss of fine-grained features. Secondly， a new local attention module named Feature Extraction Module （FEM） was put forward by combining Channel Attention （CA） mechanism and Pixel Attention （PA） mechanism， which was embedded into CNN to weight and strengthen the key information. Comparison experiments with the benchmark residual convolutional network ResNet-50 and ResNet-101 were conducted on VeRi dataset. Experimental results show that increasing the number of wavelet decomposition layers in ResNet-50 can improve mean Average Precision （mAP）. In the ablation experiment， although ResNet-50+Discrete Wavelet Transform （DWT） has the mAP reduced by 0.25 percentage points compared with ResNet-101， it has the number of parameters and computational complexity lower than those of ResNet-101， and has the mAP， Rank-1 and Rank-5 higher than those of ResNet-50 without DWT， verifying that the proposed model can effectively improve the accuracy of vehicle retrieval in vehicle re-identification.

Aiming at the problem that the nearest neighbor parameters need to be set manually in density peak clustering algorithm based on shared nearest neighbor， a density peak clustering algorithm based on adaptive nearest neighbor parameters was proposed. Firstly， the proposed nearest neighbor parameter search algorithm was used to automatically obtain the nearest neighbor parameters. Then， the clustering centers were selected through the decision diagram. Finally， according to the proposed allocation strategy of representative points， all sample points were clustered through allocating the representative points and the non-representative points sequentially. The clustering results of the proposed algorithm was compared with those of the six algorithms such as Shared-Nearest-Neighbor-based Clustering by fast search and find of Density Peaks （SNN?DPC）， Clustering by fast search and find of Density Peaks （DPC）， Affinity Propagation （AP）， Ordering Points To Identify the Clustering Structure （OPTICS）， Density-Based Spatial Clustering of Applications with Noise （DBSCAN）， and K-means on the synthetic datasets and UCI datasets. Experimental results show that， the proposed algorithm is better than the other six algorithms on the evaluation indicators such as Adjusted Mutual Information （AMI）， Adjusted Rand Index （ARI） and Fowlkes and Mallows Index （FMI）. The proposed algorithm can automatically obtain the effective nearest neighbor parameters， and can better allocate the sample points in the edge region of the cluster.

Aiming at the problem of big clustering error of the Sparse Subspace Clustering （SSC） methods， an SSC method based on random blocking was proposed. First， the original problem dataset was divided into several subsets randomly to construct several sub-problems. Then， after obtaining the coefficient matrices of several sub-problems by the sparse subspace Alternating Direction Method of Multipliers （ADMM） respectively， these coefficient matrices were expanded into coefficient matrices of the same size as the original problem and integrated into a coefficient matrix. Finally， a similarity matrix was calculated according to the coefficient matrix obtained by the integration， and the clustering result of the original problem was obtained by using the Spectral Clustering （SC） algorithm. The SSC method based on random blocking has the subspace clustering error reduced by 3.12 percentage points on average compared with the optional algorithm among SSC， Stochastic Sparse Subspace Clustering via Orthogonal Matching Pursuit with Consensus （S³COMP-C）， scalable Sparse Subspace Clustering by Orthogonal Matching Pursuit （SSCOMP）， SC and K-Means algorithms， and has all the mutual information， Rand index and entropy significantly better than comparison algorithms. Experimental results show that the SSC method based on random blocking can significantly reduce subspace clustering error， and improve the clustering performance.