206 results about "Structure learning" patented technology

The invention relates to the field of new energy, and discloses a wind power generation power prediction method based on space-time diagram convolution and gating attention, and the method comprises the steps: obtaining the geographic position information, meteorological information and historical wind power generation power data of each fan in a wind power plant, and obtaining the normalized data; constructing a dynamic adjacency matrix based on the maximum information coefficient among the historical power data of each fan in the wind power plant, and generating a graph structure; a node set of the graph structure corresponds to each station in the wind power cluster, and an edge set is dynamically determined by a maximum information coefficient of historical power data between the stations; spatial feature extraction is carried out by using a graph convolutional network, and a graph structure learning module is introduced; and inputting the sequence output by the graph structure learning module into a gating circulation unit, introducing an Informer encoder based on a sparse attention mechanism, and generating a wind power prediction result of a future time step. According to the invention, high-precision prediction of the wind power generation power in a multi-fan scene is realized.

The invention relates to the technical field of electroencephalogram signal processing, and discloses an electroencephalogram emotion recognition method and system based on a deep neural network. The method comprises the following steps: collecting and preprocessing a multi-channel EEG signal; constructing a graph data structure, extracting multi-domain features by taking electroencephalogram channels as nodes, and constructing a self-adaptive dynamic adjacency matrix; constructing a graph convolution long and short-term memory network, learning spatial features by GNN, and extracting time dependence by LSTM; enhancing emotion capture by using a multi-scale time-frequency feature fusion method in combination with STF and CWT; constructing global topological information of an FCN brain extraction region in combination with brain network features; and outputting alertness and other emotion indexes by means of the classification model. According to the method, graph structure learning and time sequence modeling are combined, EEG signal emotion recognition is optimized, and personalized adaptation and emotion recognition accuracy is improved.

The invention discloses a large model tool calling hierarchical dynamic optimization method and system based on reinforcement learning, and provides a model training mode based on a hierarchical decoupling architecture, which is characterized in that a reward mechanism is adjusted to form a format + tool calling correctness reward, so that the reward efficiency is improved. The correctness rewards are decomposed into three-level verification of names, parameters and values, and formats and correctness reward weights are dynamically adjusted in the training process; thus, the model realizes progressive training from basic structure learning to complex strategy optimization, the generalization ability of the model is enhanced, and fine-grained feedback in the training process is also realized, so that the model can perform gradient updating aiming at specific errors, and the accuracy of the model is improved. The problems of low training efficiency and poor model output accuracy in the traditional technology are avoided; therefore, according to the method, the generalization ability, the training efficiency and the output accuracy of the model are improved, so that the method is very suitable for large-scale application and popularization.

The invention relates to the technical field of organization transformation management and data analysis, and discloses a knowledge graph-based influence prediction method and system, and the method comprises the steps: constructing a double-layer knowledge graph which integrates an organization formal hierarchy and an informal social network; processing the double-layer atlas structure by applying a hierarchical graph convolutional network, and calculating multi-dimensional influence features; analyzing based on the multi-dimensional influence features, capturing an organization structure evolution rule by applying a time-varying graph structure learning algorithm, and identifying key opinion leaders; the key opinion leader information is utilized, the organization influence propagation process is simulated based on the Agent technology, and different intervention strategy effects are evaluated; generating a multi-dimensional influence prediction result; compared with a traditional method with a single network structure, the method has the advantages that the group influence prediction accuracy is improved, the key opinion leader recognition accuracy is improved, and the time sequence prediction error is reduced.

The invention discloses an industrial Internet of Things time sequence self-supervision anomaly detection method and a monitoring and early warning system, and relates to the field of industrial Internet of Things, and the method comprises the steps: S1, constructing an anomaly detection model, and S2, obtaining a training data set; s3, training and optimizing an anomaly detection model; s4, acquiring to-be-detected data in real time; s5, performing anomaly detection analysis on the to-be-detected data, and outputting an anomaly detection result; through a time sequence and relation learning module, a dynamic graph topological structure learning module and an enhancement module, internal characteristics of a time sequence in a time domain and a space domain are deeply mined. The time sequence and relation learning module comprehensively captures a multi-scale time pattern, and the dynamic graph topological structure learning module eliminates dependence on a predefined graph structure; the enhancement module enhances the invariant representation under noise, and improves the recognition capability of the model to a normal mode; through wide experiments, the advancement of the method in detection performance is verified, and reliable support is provided for intelligent manufacturing and infrastructure diagnosis.

The invention discloses a structured data self-learning method based on a graph neural network, and the method comprises the following steps: S1, analyzing structured data, extracting entity fields and relation fields, and constructing a structure candidate graph; s2, generating a node embedding feature matrix, and initializing and recording the adjacency relation of candidate edges; s3, constructing a graph neural network model, inputting node features and an adjacent matrix, and defining a task loss function; s4, evaluating the gradient contribution degree of edge connection by adopting a gradient sensitive sparse adjacency self-learning algorithm, and updating the graph structure representation; s5, introducing an embedded interpretability gradient backtracking mechanism, correcting an edge connection relation and enhancing interpretability; s6, training the graph neural network by using the corrected structure, and updating the node embedding and graph structure; and S7, outputting a final graph structure and an interpretability index, and generating a graph modeling visualization result. According to the method, efficient modeling and explanatory analysis of structured data are realized through a dynamic graph structure learning and gradient backtracking mechanism.

The invention relates to the technical field of data anomaly detection, in particular to an intelligent treatment anomaly detection method and system based on dynamic space-time hypergraph evolution. Self-adaptive modal decoupling and multi-view embedding are carried out based on the obtained original observation data to obtain initial node representation, and the initial node representation comprises variational modal-based signal decoupling and multi-view space-time embedding coding; performing dynamic evolution hypergraph structure learning based on the obtained initial node representation to obtain a deep feature tensor, including dynamic hyperedge generation based on metric learning and space-time hypergraph convolution evolution; performing multi-scale time sequence prototype memory prediction based on the deep feature tensor, including multi-scale time sequence feature extraction, prototype memory reading and reconstruction and future prediction of reconstructed features; according to the method, the problems that sudden anomalies are difficult to pre-judge and the depth model lacks interpretability are solved.

The invention discloses a power plant secondary circuit fault tracing method based on causal reasoning. The method comprises the following steps: collecting and preprocessing multi-source operation data of a power plant secondary circuit system; based on the secondary circuit schematic diagram, the relay protection configuration table and the signal connection relation, establishing a topology constraint set, and generating an intervention node set; constructing an enhanced causal Bayesian network model based on the topology constraint set and the intervention node set; based on the topology constraint set, carrying out structure learning and parameter estimation on the enhanced causal Bayesian network model; inputting the standardized operation data set into the trained enhanced causal Bayesian network model, and executing reverse causal reasoning; and performing consistency verification on the fault root cause candidate set, the propagation path set, relay protection logic and interlocking rules, and generating a fault traceability report. According to the method, the enhanced causal Bayesian network is adopted, so that the power plant secondary circuit fault can be accurately traced.

The invention discloses a diversified epiphyseal development map generation method and device based on a potential diffusion model. The method comprises the following steps: making a data set, covering and collecting an original X-ray film image, preprocessing data, labeling a skeleton maturity level and desensitizing the image; an original image is encoded to a potential space by using a variational auto-encoder, and the image detail generation capability is improved by optimizing a loss function; a fuzzy learning module is introduced in the de-noising stage of the diffusion model, and the diversity of generated skeleton features is enhanced through operations such as batch normalization and random disturbance; a progressive alignment strategy is adopted, skeleton basic structure learning is emphasized in the initial stage, a skeleton grade evaluation network is introduced in the later stage, and skeleton detail features are emphasized; optimizing the generation model by using the skeleton recognition model; and inputting development levels, ages and genders of 14 skeletons, and generating a corresponding left-hand skeleton X-ray film image. According to the method, diversified and high-quality skeletal development images are generated through a potential diffusion model framework in combination with VAE optimization, an FLM module and a progressive alignment strategy, skeletal development features are truly reflected, the problems of detail loss and insufficient diversity in the prior art are solved, the method is particularly excellent in performance when sparse data and long-tail distribution are processed, and the method is suitable for large-scale popularization and application. And the authenticity of the generated image and the practicability of medical diagnosis are improved.

The invention discloses a large language model semantic enhancement collaborative filtering recommendation method based on comparative learning, and belongs to the field of recommendation systems.The method comprises the steps that a user-item bigraph is constructed, user-item semantic information collaborative information is provided through GNN, user-item semantic information is extracted through cue words, a deterministic topology view enhancement strategy is adopted, and user-item semantic information collaborative filtering recommendation is achieved. According to the method, a semantic neighbor extension view is generated, a semantic neighbor reconstruction view is generated, collaboration and semantic information alignment are performed, double-view structure alignment is performed, and a loss function is integrally trained, so that the accuracy and the cold start capability of a recommendation system are remarkably improved through collaborative graph structure learning and semantic enhancement.

The invention relates to an industrial internet multilayer causal motif abnormal propagation path identification method and system, and the method comprises the steps: firstly carrying out the construction and extraction of a multilayer high-order motif, extracting a motif unit which expresses the local high-order structure features through the construction of a semantic hierarchical graph structure in combination with a frequent sub-graph mining and cross-layer motif alignment mechanism, and carrying out the recognition of the abnormal propagation path of the multilayer causal motif. Stable and uniform multi-layer motif representation is formed; then, on the basis of the structural equation model, motif variables are regarded as endogenous variables of a causal model, a causal path between motifs is mined by introducing conditional mutual information and a Bayesian structure learning algorithm, an average causal effect is calculated to construct a causal consistency matrix, and causal community division is realized in combination with a weighted modularity optimization method; and finally, quantifying the dynamic change of a community causal structure by constructing a causal deviation graph between an expected causal graph and an observed causal graph, and assisting in identifying a causal-driven abnormal propagation path. According to the method and the system, accurate detection and causal traceability of equipment-level and subsystem-level abnormal modes in an industrial system can be realized.

The invention discloses an IOT (Internet of Things) equipment fault prediction method based on GraphRAG. The IOT equipment fault prediction method comprises the following steps: step 1, collecting and preprocessing multi-source heterogeneous data of various IoT equipment terminals; 2, constructing a graph structure based on the preprocessed data, a dynamic edge weight mechanism and a knowledge sub-graph; according to the graph structure, physical connection, functional dependence and communication topology between IOT devices are naturally expressed, so that the relevance between the operation states of the devices is fully captured. Sensor data, historical maintenance records and equipment configuration information are coded into node attributes and edge weights, so that the system can realize unified representation of multi-source heterogeneous data. The GraphRAG framework combines the structure learning ability of a graph neural network and the knowledge fusion characteristic of a retrieval enhancement generation mechanism, and shows stronger generalization ability in the recognition of a non-fault mode. When an abnormal signal occurs in a certain device, the model can trace the potential influence range through the state propagation path of the adjacent node, and the accuracy of early warning is improved.

The invention relates to the technical field of satellite internet, and discloses a low-orbit satellite phased array multi-beam interference modeling and suppression method and system, and the method comprises the steps: selecting a Kaiser window as a core filtering method, and achieving the optimization of beam characteristics through the dynamic adjustment of a shape parameter beta; generating an initial beam directional diagram based on a digital phase matching method, and multiplying the Kaiser window function coefficient by the excitation weight of the 64-array-element linear array element by element to realize spatial domain weighted filtering; the method comprises the following steps: constructing a training data set containing multi-scene interference characteristics, calculating a corresponding covariance matrix and an accurate inverse matrix thereof to form a sample pair, designing a deep neural network architecture, inputting a flattened covariance matrix vector, and learning a complex nonlinear mapping relation from the covariance matrix to the inverse matrix through a multi-layer full-connection structure; a mean square error is used as a loss function to constrain network output precision, and a multi-beam interference system model is constructed; according to the invention, stable and efficient communication of the low-orbit satellite system in a complex electromagnetic environment and under rapid channel change is ensured.

The invention provides a multi-view structure learning method based on multi-expert cooperation, and relates to the technical field of graph neural networks and multi-view learning, and the method comprises the steps: reconstructing initial multi-view data, and obtaining multi-view structure data; respectively training the single-view expert model and the shared expert model by utilizing a first loss function and a second loss function based on the multi-view structure data to obtain a trained single-view expert model and a trained shared expert model; constructing a collaborative decision model by using the trained single-view expert model and the trained shared expert model; training the collaborative decision model by using a third loss function to obtain a trained collaborative decision model; and analyzing the graph data by using the trained collaborative decision model to obtain a joint decision result, and completing learning of the multi-view structure. According to the method, the problems of large structural noise interference, isolated expert model information and insufficient node classification accuracy when an existing graph neural network processes multi-relation graph data are solved.

The invention belongs to the technical field of article recommendation, and provides an article recommendation method and system based on big language model enhanced graph representation learning, and the method comprises the steps: obtaining a user-article graph recommendation data set; extracting semantic knowledge features of the obtained graph recommendation data set; according to the extracted semantic knowledge features and an article recommendation model, completing article recommendation learning; wherein the article recommendation model adopts a big language model to enhance a graph representation learning model, embeds semantic similarity through the big language model, adopts an adaptive graph structure learning mechanism to identify a semantic boundary so as to determine structure information, and performs bidirectional knowledge interaction transmission on semantic and structure information; interactive noise is filtered and relieved in combination with information bottleneck regularization, optimization of an article recommendation model is carried out with the purpose of minimizing a comprehensive multi-objective loss function, and article recommendation is completed by calculating preference scores of articles.

The invention discloses a small sample remote sensing image classification method based on hierarchical spatial structure learning. The method comprises the following steps: firstly, extracting multi-scale features of a remote sensing image by using a ViT (Visual Transform) model, and capturing rich semantic information and spatial structure relationships in the image; secondly, constructing a graph structure based on spatial adjacency and attention weight to model a structured relationship between samples, and encoding graph node features through a graph convolutional network (GCN) so as to enhance the discrimination ability of the features in a structural semantic space; thirdly, a residual enhancement mechanism is introduced to fuse global semantic information, and the discrimination capability of graph embedding is improved; then, based on the structural similarity between the support set and the query set, performing classification decision, and realizing accurate classification under a small sample condition; and finally, carrying out joint optimization on the whole model by adopting a training strategy of a small sample meta learning task and a supervision loss function.

The invention provides an industrial process fault detection method based on a space-time causal diagram autoencoder, and the method comprises the steps: 1, carrying out the data preprocessing of the space-time process data of all process variables collected in the operation process of a target industrial process for the target industrial process; step 2, establishing a causal graph space-time auto-encoder CGSTAE; 3, executing a three-step causal graph structure learning algorithm to realize training of a causal graph space-time auto-encoder CGSTAE, wherein the training comprises three steps of pre-training, causal extraction and fine tuning; and step 4, obtaining a fault detection result based on hidden layer features of the causal graph space-time auto-encoder CGSTAE and residual data output by reconstruction. According to the method, effective process monitoring and fault detection are realized by constructing two statistical magnitudes in a feature space and a residual space. Compared with other methods, the fault detection method provided by the invention can improve the reliability and interpretability of industrial process monitoring.

The invention relates to the technical field of water quality parameter monitoring. The invention provides a water quality multi-parameter prediction method and model based on a space-time diagram convolutional network. On the time dimension, a mask code transformer module is designed, a pre-training task is self-supervised through a random mask to enable the model to extract trend features from long-period data, and the problem of response lag of a traditional model to a sudden change event is solved in combination with an accumulative effect of capturing water quality features through expansion causal convolution. And in the spatial dimension, constructing a dynamic graph structure learning module, and fusing the predefined adjacency matrix based on the station physical distance and the dynamic residual graph to generate a dynamic graph structure. Experimental results show that compared with other water quality space-time prediction models, the model has better prediction precision, and prediction R2 of all water quality indexes can reach 93% or above.

The invention provides a method for predicting miRNA-lncRNA-disease ternary correlation through deep tensor decomposition and hypergraph convolution, and relates to the technical field of miRNA-lncRNA-disease ternary correlation prediction. Comprising six steps of integration of multi-source heterogeneous data, generation of three-dimensional tensor representation, hypergraph convolution modeling high-order interaction, graph attention network feature refining, depth graph convolution network enhancement and correlation prediction. Node features in a graph attention self-adaptive refining similarity network are integrated, global structure learning is enhanced by adopting a depth graph convolutional network, and the combination can generate stable and information-rich embedding for final ternary correlation prediction, so that potential complex correlation among various biological entities such as diseases, genes and drugs can be accurately extracted, and the prediction accuracy is improved. The potential relation and mechanism between the biological entities are further disclosed, and comprehensive ternary correlation prediction is achieved.

The invention provides a multi-dimensional user portrait and adaptive graph fused public opinion evolution prediction method and system, and the method comprises the steps: carrying out the unified modeling of user language styles, personality features, social structures and theme preferences, and depicting the multi-dimensional individual features of a user; and a multi-dimensional user vector and a multi-relation social contact propagation path are used as input, neighbor co-occurrence coding and partitioning technologies are combined, node representation of time perception is obtained through Transform and an attention mechanism, and dynamic self-adaptive learning modeling of public opinion elements is realized. And finally, dynamic link prediction and dynamic node classification are completed based on the time perception representation, and a visualization result is output. According to the method, through multi-dimensional feature fusion, dynamic graph structure learning and time sequence prediction, the limitation of traditional static modeling is effectively broken through, the precision and interpretability of public opinion evolution prediction are improved, high technical innovation and application value are achieved, and the method can be used for public opinion monitoring, event prediction and public opinion evolution and risk early warning.

The invention relates to an Internet of Things time sequence root cause analysis method based on a dynamic causal diagram, and belongs to the technical field of Internet of Things. The method comprises the following steps: embedding a fine-tuning large language model by utilizing an Internet of Things knowledge graph, embedding, splicing and constructing a graph structure through entities and relationships, and combining text word embedding and mask prediction task optimization model; on the basis of knowledge graph subgraph construction, a triple is converted into a natural language to be input into a large model to generate a causal hypothesis; assumptions are converted into causal constraints, dynamic causal graph structure learning is carried out in combination with a Bayesian information criterion scoring function, and conditional probabilities of father nodes and historical values are modeled; performing parameter learning by adopting kernel density estimation, and quantifying time hysteresis among the features; feature probability distribution is predicted based on sliding window observation data, accumulative error contribution is calculated through asymmetric Shapley values, and causal ancestor features are preferentially sorted to determine root causes. Real-time and effective root cause analysis of the Internet of Things system is realized, and the system has the capability of intelligently solving faults.

The invention relates to the field of data processing and information technology operation and maintenance, and discloses an intelligent operation and maintenance decision-making method and system based on multivariate heterogeneous data fusion. Comprising the following steps: constructing a dynamic causal graph reflecting a system state and an environmental factor causal relationship based on a causal structure learning algorithm; constructing a multi-agent game decision model based on the dynamic causal diagram, and respectively setting a global operation and maintenance target and a component local demand as a leader strategy and a follower strategy of the game; and solving game equilibrium by using multi-agent reinforcement learning, and outputting an optimal joint operation and maintenance decision instruction. The system is composed of a data acquisition preprocessing module, a multi-modal feature fusion module, a causal structure learning module, a game decision solving module and an execution monitoring module. According to the method, the causal inference and the game theory are fused, so that the causal logic can be accurately extracted, the decision robustness and interpretability are improved, and the global optimal collaborative configuration is realized.

The invention relates to a fine-grained target real-time image segmentation method and system based on a dynamic state modeling network, and belongs to the technical field of intelligent image processing. The method comprises the following steps: extracting multi-scale detail features by using a lightweight backbone network; through a dual-scale two-dimensional selective scanning module, the features are divided into a thin branch and a thick branch, and local scanning and global scanning are executed respectively; a dynamic cross-scale feature selection and aggregation module is adopted, redundancy is suppressed through reweighting and statistical filtering, and key target responses are highlighted; at a decoding end, local details and global semantics are fused through jump connection and an edge extractor; and finally, introducing a form-guided pseudo label hierarchical supervision strategy, and improving the structure learning ability of the model by using a coarse-to-fine morphological prior. According to the method, the segmentation precision, the boundary integrity and the tiny target recall rate of the fine-grained target under the scenes of ore separation, industrial defect detection, pavement crack recognition and the like are remarkably improved.

The invention discloses a sports market demand prediction method and device based on deep learning and a medium, and relates to the technical field of market demand prediction, and the method comprises the steps: collecting sports demand data, and carrying out the preprocessing; performing relation mining and graph structure learning on the preprocessed sports demand data through a graph attention space-time network to generate a macroscopic demand potential energy graph; performing potential area identification on the macroscopic demand potential energy diagram by adopting a pre-trained sports market simulation model, and outputting local demand prediction data; performing weighted fusion and error correction on the macroscopic demand potential energy map and the local demand prediction data, and outputting a sports demand prediction score; and making a sports market demand strategy according to the sports demand prediction score and the multi-granularity demand prediction report, and transmitting the sports market demand strategy to an enterprise manager through an enterprise decision support interface. According to the method, multi-level accurate prediction and decision support of sports market demands are realized through dual-mechanism cooperation of the graph attention space-time network and the space-time convolution.

The invention discloses a GNN-MAPPO task unloading method based on dominant value noise addition, which is characterized in that an MLP is changed into a GNN on the basis of the existing MAPPO framework, a multi-agent system can be directly modeled into a graph structure, an interaction relationship among multiple agents can be better established, Gaussian noise is added on the dominant value, the exploration capability of a model is enhanced, and overfitting is reduced. According to the method, the powerful graph structure learning ability of GNN is combined with an innovative dominant value noise adding mechanism, and the mixed reward function is elaborately designed, so that the MAPPO algorithm can more effectively learn a cooperation strategy between agents and optimize time delay and energy consumption in the aspect of edge computing task unloading, and the efficiency of the MAPPO algorithm is improved. And the exploration capability of the strategy and the avoidance capability of the communication risk can be obviously enhanced, so that a more robust and efficient intelligent task unloading scheme adapting to a dynamic environment can be obtained.

The invention discloses a wind turbine generator abnormal knowledge association method and device based on a knowledge graph and a medium, and the method comprises the steps: carrying out the collection of distributed multi-source heterogeneous data and knowledge based on the operation and maintenance demands of the wind power industry; establishing an OPC UA information model of the wind power equipment node; the method comprises the following steps: guiding an industrial multi-modal industrial large model to carry out standardized description and identification on node data through cue word engineering, completing entity extraction and attribute extraction, carrying out relation extraction and expression, and importing into a graph database to construct an abnormal knowledge graph; the semantic relation between the node entities is optimized and updated, knowledge merging and processing are completed, and the abnormal knowledge graph is dynamically updated; and for an abnormal knowledge application scene, performing knowledge reasoning based on abnormal knowledge graph mining entity association. According to the multi-modal knowledge graph construction method based on graph structure learning and fine tuning of the multi-modal industrial large model, relevance of different anomalies is explored, the knowledge base rich in abnormal semantics is constructed, and it is ensured that industrial abnormal faults are correctly processed.

The invention discloses a cross-page table recognition system and method based on reinforcement learning, and belongs to the technical field of computer vision and natural language processing. The system comprises a data construction module, a model training module and a structured output module; the data construction module is used for extracting a table sample from the PDF document and executing paging, structure labeling, rendering generation and image pairing; the model training module is used for carrying out feature extraction and structured learning on the image based on a vision-language multi-modal large model, introducing a generalized relative strategy to optimize a GRPO algorithm, and constructing a cross-page table recognition model; the structured output module is used for detecting a table area, identifying a cross-page relationship and outputting a complete structured file; the cross-page table recognition system and method based on reinforcement learning provided by the invention solve the problems of inaccurate cross-page table recognition, table structure breakage, merged cell loss and the like in the prior art.

The invention discloses a regional groundwater spatio-temporal evolution modeling and prediction method based on a graph neural network. The method is specifically implemented according to the following steps: step 1, collecting research region data and performing preprocessing; step 2, constructing a static boundary structure under hydrogeological constraints; step 3, extracting multi-scale time sequence features based on the Fourier neural network; 4, constructing a spatial dependency integration model based on the graph attention network; 5, introducing a dynamic graph structure to learn a modeling cross-regional association relationship; and step 6, predicting the dynamic change of the underground water level based on the graph Fourier network. The method solves the key technical problems of insufficient spatial-temporal feature coupling, limited adaptive capacity to hydrogeological heterogeneity and limited cross-regional generalization ability of an existing groundwater prediction model under complex hydrogeological conditions.

The embodiment of the invention discloses a graph neural network architecture search method and device based on a large language model and a medium. The search effect on a graph neural network architecture on graph data can be improved. According to the method, a graph neural network architecture and a graph structure are jointly optimized through an iterative updating method based on curriculum optimization so as to realize structure denoising; the large language model is integrated into a graph architecture search process to cope with semantic noise; therefore, graph structure learning and text modeling are incorporated into the denoising process in the search process, so that the search effect is improved, and a graph neural network architecture with better performance is obtained.

The invention relates to a category-independent attitude estimation method based on potential structure learning and electronic equipment, and the method comprises the steps: extracting features from a support image set and a query image set respectively, and obtaining support image set feature embedding and query image set features; generating two-dimensional coordinates of the key points based on feature embedding of the support image set; supporting image set features to be embedded and input into a variational auto-encoder based on iterative structure perception, and learning to output an adjacent matrix represented by a potential structure; querying a related fusion structure diagram by using a combined diagram structure transfer strategy based on the adjacent matrix sample and the global feature representation of the query image; and inputting the query related fusion structure diagram and the support image set feature embedding into the image convolutional network, updating the support image set feature embedding and the key point two-dimensional coordinates, and outputting the target key point coordinates of the query image after multi-layer iteration updating. Compared with the prior art, the method has the advantages of improving the accuracy and stability of category-independent attitude estimation and the like.