951 results about "Domain adaptation" patented technology

An intelligent analysis method based on medical document structure perception and multi-modal fusion comprises the following steps: carrying out structure topology modeling on a medical document, extracting visual layout, text meta-information, space coordinates and semantic keyword features, constructing a semantic topological graph and dynamically shielding irrelevant contents; selecting an extraction path according to a document type, performing deep semantic analysis and entity recognition on a text-type document, and performing visual enhancement OCR recognition on a scanning-type document; the features are injected into a medical knowledge graph, and feature fusion, semantic verification, relation reasoning and information completion are achieved through a graph neural network; a three-stage strategy optimization model of basic pre-training, domain adaptation and online reinforcement learning is adopted; and large-scale processing is realized through a dynamically aggregated distributed architecture. The method is used for intelligent analysis and structured conversion of documents of hospitals, medical insurance and medical scientific research. The problems that heterogeneous medical document analysis adaptability is poor, multi-modal fusion is difficult, medical knowledge utilization is insufficient, and large-scale processing efficiency is low are solved.

The invention relates to the technical field of power transmission and distribution production, and discloses a data integration and multi-modal diagnosis method based on a power transmission and distribution scene, and the method comprises the following steps: S1, enhanced integration of multi-source heterogeneous data: collecting time sequence monitoring data, text procedures and image data of power transmission and distribution equipment, constructing an equipment topological correlation graph through a graph attention neural network, and carrying out the enhanced integration of the multi-source heterogeneous data; node feature embedding is optimized through self-supervised comparative learning, an adversarial variational auto-encoder is designed for edge data, and an enhanced sample is generated in combination with physical constraints of equipment. According to the data integration and multi-modal diagnosis method based on the power transmission and distribution scene, the field adaptability and reliability of a diagnosis result are improved while the model fine tuning cost is reduced, and the knowledge migration problem of a general model in the power transmission and distribution scene is solved; the introduction of a dynamic knowledge graph and a multi-dimensional evaluation system realizes the real-time integration of new regulation knowledge and the comprehensive evaluation of model performance, and ensures the sustainable evolution ability and decision transparency of the diagnosis model.

Provided is a method for perceiving and eliminating an abnormal state of active distribution network based on data enhancement, including: acquiring, by synchrophasor measurement device, data of each node of active distribution network in target domain in real-time and transmitting to processor; inputting the acquired data into a classification model, and outputting abnormal detection and classification results in real time; and analyzing the abnormal detection and classification results, and transmitting an abnormal state eliminating instruction to a distribution terminal to eliminate the abnormal state. Wherein, hidden distribution features in node data of active distribution network are mined through dynamic clustering, a large amount of unlabeled data are clustered, a data label is generated through self-coding and label correction rule, training samples with balanced category distribution is generated through data enhancement and is used to train the classification model based on dynamic graph attention network by domain adaption method.

The invention discloses an artificial intelligence speech recognition system, and the system comprises a multi-modal feature extraction module which employs an improved Conformer architecture to synchronously extract the time-frequency features and text embedding vectors of speech signals; the joint training module is used for performing joint optimization on ASR and NMT loss functions through an adversarial training strategy, learning voice recognition and machine translation tasks at the same time through joint training, and completing direct mapping from voice features to a target language; the context perception translation engine is used for integrating an attention mechanism of a pre-training language model, carrying out deep coding on the extracted speech features and generating cross-language semantic representation; the self-adaptive post-processing module is used for dynamically optimizing an output result by adopting a reinforcement learning framework, dynamically adjusting the output result according to a reward function, and optimizing translation quality and a speech synthesis effect; the dynamic language recognition module is a real-time language classifier based on a Wave2Vec 2.0 framework and is used for recognizing the language of the input voice in real time; and the incremental field adaptation module is used for quickly updating a field term library by using a LoRA fine tuning technology.

A system and method for cross-domain knowledge transfer in federated compression networks. The system enables efficient lossless data compression across diverse data types by intelligently sharing compression strategies between domains. A cross-domain knowledge transfer system identifies relationships between different data domains, adapts compression parameters accordingly, and optimizes learning processes to maximize knowledge reuse. The architecture may include a knowledge repository for storing domain features and compression patterns, domain mapping components that identify similarities, and transfer learning optimization that enables efficient adaptation with minimal examples. This approach significantly accelerates model training for new domains while improving compression performance. Applications include satellite telemetry systems where efficient compression is critical for transmitting large information sets between distant locations. The system may employ probability prediction driven arithmetic coding paired with long short-term memory networks, enhanced by cross-domain knowledge sharing that adapts successful compression strategies from one domain to another while preserving domain-specific optimization.

The invention discloses a lithium battery health state estimation method based on transfer learning, and the method comprises the steps: firstly carrying out the normalization and time sequence reconstruction of a battery charging voltage-capacity curve, and constructing a unified input sequence; extracting long-time-sequence degradation characteristics based on a Mama network, and completing SOH regression prediction through a two-stage full connection layer; in the cross-domain adaptation stage, in combination with an alignment strategy of dynamic time warping and weighted maximum mean difference, time sequence matching and feature distribution alignment in the degradation stage are realized; meanwhile, on the basis of a sample weighting mechanism of a Wasserstein distance, the effectiveness of migrating source domain knowledge to a target domain is improved; through a two-stage strategy of source domain pre-training and source-target joint training, a relatively low prediction error and a relatively high fitting degree can be kept under the condition that a target domain is not labeled; the method shows good generalization ability and robustness under different battery types and different working conditions, and can provide reference for health management of the electric vehicle.

The invention relates to the technical field of large model algorithms, in particular to a knowledge distillation-based field large model lightweight adaptive method and system, and the method comprises the steps: obtaining knowledge distillation parameters and student model performance parameters, and building a nonlinear mapping relation between the knowledge distillation parameters and the student model performance parameters; the optimal parameter combination is optimized and solved based on the mapping relation, and target knowledge distillation parameters are generated; issuing the target parameters to a training engine, monitoring performance deviation in real time and triggering re-optimization; in the reasoning process, performance fluctuation is monitored, and model characteristics are managed and controlled; target domain data characteristics are collected, a mapping relation is corrected in combination with big data analysis, and the domain adaptation capacity is improved; a knowledge base and a case base of historical distillation data are constructed, a standardized adjustment scheme is formed, and self-adaptive matching is achieved. According to the scheme, through precise modeling, dynamic optimization, real-time monitoring and knowledge reuse, the knowledge distillation efficiency, model robustness and field adaptability are remarkably improved, and systematic technical support is provided for large model lightweight.

The invention discloses an information reasoning method, system and equipment based on knowledge enhancement and a medium. The method comprises the following steps: constructing a target knowledge graph and a target knowledge graph index according to a target entity, a target attribute and a target constraint condition in to-be-reasoned information; determining an atlas sub-graph of the target entity through the first-level entity hash index; screening an entity attribute set matched with the target attribute from the atlas sub-graph based on the secondary attribute classification index; filtering the entity attribute set matched with the target attribute according to the three-level space-time dimension index to obtain a target dynamic attribute; recalling a target retrieval fact from the target knowledge graph according to the graph sub-graph, the entity attribute set and the target dynamic attribute; the to-be-reasoned information and the target retrieval facts are input into the preset large language model to obtain the target reasoning result, knowledge enhancement reasoning can be carried out by constructing the multi-level index target knowledge graph and fusing the dynamic attribute constraints in combination with the large language model, and then the accuracy, the real-time performance and the field adaptability of the reasoning result are improved.

The invention discloses a local legislation compliance intelligent detection system and method based on deep semantic analysis and a multi-modal legal knowledge graph, and relates to the field of computer technology and law crossing technology, the method comprises the following steps: constructing a legal knowledge graph; designing a multi-dimensional rule sub-library with a dynamic weight adjustment mechanism; a large language model based on Deepseek is utilized to construct legal provisions to perform a deep semantic analysis model, so that the intelligence of generation of triads of laws and regulations in the national field is realized, and the field adaptability of triad generation is improved; executing multi-dimensional conflict detection based on the knowledge graph, a conflict detection rule base and a semantic analysis result; and a multi-dimensional law conflict report is automatically generated. According to the method, a legal knowledge graph is constructed, multi-source legal data is integrated, a multi-dimensional rule sub-library is combined, a Deepseek-based special model is used for carrying out deep semantic analysis on legal provisions, and explicit and implicit conflict judgment is carried out on contradictory point locations and contexts.

The invention belongs to the technical field of spacecraft pose estimation, and particularly relates to a cross-domain spacecraft pose estimation method based on mask self-distillation domain adaptation, and the method comprises the steps: 1, inputting a complete image, and employing a Faster R-CNN algorithm to position a spacecraft bounding box; the robustness of the model is improved by applying a track environment data enhancement strategy; and extracting a target ROI region as key point regression network input based on the detection frame. 2, dividing ROI (Region of Interest) data of a source domain and a target domain; and optimizing heat map supervision loss learning key point positioning knowledge. 3, inputting random mask enhanced target domain data into the student model; inputting original target domain data into the teacher model; a learnable shared prototype space is constructed, and self-distillation is guided through heat map consistency loss and semantic consistency loss. And 4, jointly optimizing the loss of the key point regression network 3, and realizing progressive migration of source domain annotation knowledge to a target domain. And 5, solving the 6D pose of the spacecraft relative to the camera through the EPnP. According to the invention, robust six-degree-of-freedom pose estimation of the target spacecraft is realized.

The invention relates to a multi-model-question-based group consensus large model illusion reduction method, which comprises the following steps of: screening a Top-N model from a candidate model set according to a multi-dimensional comprehensive scoring result, and executing full-combination bidirectional knowledge distillation on the Top-N model to obtain an initial model group, loading a plurality of domain knowledge bases for each model in the initial model group to carry out domain self-adaptive fine tuning, and constructing to obtain a group model set; giving a user question, triggering a plurality of fine-tuned field expert models in the group model set to perform parallel reasoning to generate an initial answer, performing iterative optimization by constructing a question set, generating a question instruction and updating the answer, and calculating the similarity of group answers by adopting a mixed kernel function in the iterative optimization process to obtain a group answer set; and when the similarity and the stability reach preset threshold values at the same time or reach the maximum number of iterations, stopping iteration and outputting a result. Compared with the prior art, the method has the advantages of high answering accuracy, high field adaptability and the like.

The invention discloses an energy storage battery health feature extraction and state evaluation method based on transfer learning. The method comprises the following steps: S1, constructing a source domain health feature library and pre-training a model; according to the method, dependence on complete cyclic data is broken through, high precision and robustness are still achieved under the conditions of data sparsity and working condition difference, and the method is suitable for intelligent operation and maintenance and predictive maintenance of an energy storage power station. And meanwhile, common incomplete and partial charge and discharge data fragments under actual working conditions can be directly utilized for feature extraction and state evaluation, dependence on complete charge and discharge cycles is avoided, and the application scene of the data driving method is greatly widened.

The invention discloses an open set domain adaptive image classification method based on a difference prompt learning technology of a pre-training vision-language model. According to the method, high-quality pseudo-open class images are generated, and de-noising text embedding and de-noising visual embedding are obtained by using a differential prompt learning technology, so that class characteristics of a source domain, a target domain and pseudo-open class samples are effectively extracted, and irrelevant noise is inhibited. According to the method, a vision-text comparison loss mechanism, a triple distance comparison loss mechanism and a negative sample penalty mechanism are further designed, a known category and an unknown category are effectively distinguished in a feature space, and the semantic alignment capability of cross-domain similar samples is enhanced. The method can significantly improve the classification accuracy and model robustness in an open set domain adaptation task, has the advantages of simple structure, high calculation efficiency, good generalization performance and the like, and is suitable for image classification, cross-domain transfer learning and other related application scenes.

The invention relates to the technical field of artificial intelligence, in particular to an automatic construction method and system for a dynamic mode knowledge graph. Comprising the steps of obtaining an unstructured text, and performing semantic dicing on the unstructured text to generate at least one text block; processing the text block to extract non-standardized related information, and automatically generating metadata used for describing a knowledge graph structure; performing standardization processing on the non-standardized entity information, the type information and the relation information to obtain standardized related information; and according to the metadata, storing the standardized entity information, type information and relation information into a knowledge graph database. According to the method, the dependency of the traditional technology on a static predefined mode is overcome, so that the structure of the knowledge graph can be dynamically evolved according to new data, the cost of manually designing and maintaining the graph mode is remarkably reduced, and the field adaptability and expandability of the system are enhanced.

An industrial big data-driven vertical federated transfer-based anomaly detection method and system. The method comprises: acquiring a source domain data set and a target domain data set from an industrial scenario, the source domain data set being constructed on the basis of industrial data having known anomaly labels, and the target domain data set being constructed on the basis of industrial data without anomaly labels; and on the basis of a preset vertical federated transfer model: performing vertical federated feature extraction: mapping the source domain data set and the target domain data set into a common feature space to obtain potential features; performing domain adaptation: extracting features having domain invariance and discriminability from among the potential features; and performing joint domain alignment: aligning the distance between domains, and mapping the features having domain invariance and discriminability to obtain the anomaly labels.

The invention discloses a bearing variable working condition fault diagnosis method fusing model migration and feature migration learning, and the method comprises the steps: processing bearing vibration signals of a source domain and a target domain through wavelet transform, and extracting a time-frequency diagram; expanding the two-dimensional time-frequency graph data set by using DCGAN, and balancing the number of the two-dimensional time-frequency graph data set; then, model parameter migration is adopted, AlexNet network parameters pre-trained in a source domain are migrated, a migrated AlexNet network is constructed, and depth features are extracted; then, a domain adaptation method based on improved migration joint matching is provided, multiple strategies are fused, and a low-dimensional feature space with small distribution difference and good discrimination performance is obtained; and finally, on the basis of a labeled source domain feature data training model after domain adaptation, realizing identification and classification of unlabeled target domain feature data. The method is ideal in diagnosis performance and high in accuracy under variable working conditions and data imbalance, domain data distribution difference can be reduced by improving the migration joint matching method, and feature discrimination performance and fault diagnosis accuracy are improved.

The invention relates to a database table field description intelligent generation and dynamic maintenance method and device based on RAG and a large model. The method comprises the following steps: constructing a domain-based dynamic knowledge base system through multi-source corpus collection, domain adaptation vectorization model processing and a dynamic knowledge base construction and updating process; receiving a description generation request of a table or a field, recalling related knowledge from the knowledge base through an RAG technology, assembling cue words, and then calling a large model for reasoning to generate a description result; the generated description result is confirmed through manual auditing, the knowledge base is updated according to confirmation feedback, and continuous optimization of the knowledge base is achieved. According to the method provided by the invention, the problems of low accuracy, high maintenance cost, poor interpretability and the like in a traditional method are effectively solved by constructing a field-enhanced RAG architecture and combining a three-layer knowledge system and a self-adaptive prompt framework which can be dynamically adjusted according to a business scene, and a new technical implementation path is provided for table and field description in intelligent data governance.

The invention discloses an information extraction data annotation and training system based on a large language model, and relates to the technical field of information extraction, and the system comprises a data set construction module which is used for constructing a pre-training data set and a fine tuning data set; the model continuous pre-training module is used for carrying out continuous pre-training on a preset general large language model based on the pre-training data set to generate a field adaptive pre-training model; the model fine tuning module is used for performing supervised fine tuning training on the domain adaptive pre-training model through a two-stage course learning strategy based on the fine tuning data set, and generating an information extraction model; the retrieval enhancement generation module is used for performing entity-semantic retrieval on an input text based on a preset knowledge base, outputting context information related to the input text, and outputting structured information of the input text based on the context information and an information extraction model, the problems of insufficient generalization ability, poor field adaptability and disastrous forgetting of a general large language model are solved, and the accuracy and robustness of information extraction are improved.

The invention discloses a multi-agent combat task cooperation method for a structure entropy guided graph neural network, and the method comprises the steps: S10, each combat agent interacts with an environment according to an action generated by a strategy network, the environment comprises environment information, task parameters and a preset task target, and the strategy of each combat agent is completely executed in a decentralized manner; collecting complete empirical trajectory data; s20, using the collected data for centralized training; performing value evaluation on the global state of each time step by using a value network; s30, calculating strategy loss and value loss by using a multi-agent near-end strategy optimization algorithm in combination with the output of the strategy network and the value estimation of the output of the value network; updating parameters of the strategy network and the value network by using a gradient descent method; and S40, performing loop iteration. The problems that in a traditional method, the battlefield game dynamic structure sensing ability is insufficient, the hierarchical strategy learning and generalization ability is limited, the adaptability of a model in a small sample area is poor, and the migration efficiency is low are solved.

The present invention relates to a method and system for predicting the remaining useful life of a rolling bearing. The method comprises: acquiring sample data to be predicted, and inputting same into a trained remaining useful life prediction neural network to obtain an output prediction result, wherein the remaining useful life prediction neural network comprises a feature encoder and a regression predictor, and the training process comprises using the feature encoder to preliminarily extract features from source domain sample data; using a temporal mixed contrastive domain adaptation training module to calculate contrastive loss, and using the contrastive loss to iteratively train the feature encoder, so as to further extract mutual information from target domain sample data features as a high-level feature; and using a fine-grained structural domain adaptation training module to calculate domain discrimination loss and a fine-grained matching degree between the source domain sample data and target domain sample data, and using the domain discrimination loss and the fine-grained matching degree to iteratively train the feature encoder, so as to further extract domain-invariant features between a source domain and a target domain. The system comprises an input interface, an output interface, a processor, a computer readable storage medium and stored program instructions, wherein the processor calls the program instruction to train the remaining useful life prediction neural network, calls the program instruction of the trained remaining useful life prediction neural network to instruct the feature extractor to perform feature extraction on rolling bearing vibration data to be detected, and inputs the extracted features into the regression predictor for prediction processing to obtain a prediction result. The present invention effectively improves the accuracy of the prediction result of the remaining useful life of rolling bearings.

The invention discloses a real-time domain adaptive defect detection method based on double alignment and uncertainty filtering. The method comprises the following steps: firstly, constructing a defect detection data set which comprises source domain data and target domain data; then, constructing a defect detection model which comprises a backbone network, an encoder, a decoder and a detection head; and finally, pre-training the defect detection model by using the source domain data to obtain a teacher model, and storing the multi-scale source domain features extracted by the teacher model backbone network into a feature database in groups according to scales. Initializing the defect detection model by using the teacher model parameters to obtain a student model; and carrying out collaborative optimization on the teacher model and the student model by using a pseudo-bounding box and feature distribution double-alignment strategy to realize real-time domain adaptive detection. According to the method, performance degradation caused by domain offset is effectively relieved through a double-alignment strategy, meanwhile, error tag accumulation is avoided through uncertainty perception and filtering of a pseudo-bounding box, and the stability during domain adaptation is improved.

The invention discloses a cross-modal remote sensing image unsupervised adaptive method based on unreliable pseudo tag guidance. The method comprises the following steps: step 1, acquiring data of a source domain and a target domain and preprocessing the data; 2, constructing a teacher-student self-training framework, generating a pseudo tag for a target domain by a teacher network, and filtering noise through confidence evaluation; 3, dividing the image into reliable pixels and unreliable pixels according to the confidence coefficient, and generating a mask; 4, constructing positive samples, negative samples and anchor point features; 5, designing an unreliable sample guide pixel contrast loss function; 6, optimizing the loss function training model until the optimal performance is achieved; and 7, predicting test set data by adopting the trained model to obtain a semantic segmentation result. According to the method, the potential of pseudo labels of which modals are difficult to label is fully mined, the problem that unreliable pixels are insufficient in utilization during training is solved, and finally more effective cross-modal domain alignment and better cross-modal unsupervised domain adaptive semantic segmentation precision are realized.

The invention provides a code retrieval method and device for AI programming and related equipment, and the method comprises the steps: constructing a double-layer embedded vector model which comprises a basic semantic coding layer for parameter freezing and a plurality of independent lightweight adaptation layers which coexist; monitoring implicit feedback data generated by interaction of a user in the target code library to extract positive and negative sample pairs, and mining difficult negative samples based on a current model to form a training sample set; when the cumulative number of the samples reaches a preset trigger threshold value, optimizing and updating at least one lightweight adaptation layer parameter by utilizing a preset comparison loss function, and applying elastic weight constraint to key parameters of the lightweight adaptation layer parameter to obtain an updated double-layer embedded vector model; and performing multi-dimensional performance evaluation on the updated model, and when an evaluation result meets a preset publishing condition, deploying the model to a code retrieval service. The problems that in the prior art, cross-domain is difficult, the adaptation cost is high, and catastrophic forgetting is likely to happen are effectively solved, and zero-labeling self-evolution of the model is achieved.

The invention provides a pollution site multi-medium restoration intelligent decision-making system based on a large language model, which comprises a multi-source data input and standardization module, a site feature analysis module, a large language model knowledge inference engine, a dynamic matrix scoring system and a restoration scheme optimization and decision-making module, a set of full-process intelligent decision-making framework is constructed by integrating a large language model and professional knowledge in the environmental field, and intelligent matching, automatic scheme generation and optimization decision-making of a polluted site remediation technology are realized. The method is suitable for a plurality of complex pollution treatment scenes, and has good expandability, cross-domain adaptability and interpretability.

The invention discloses an unsupervised domain adaptive medical image segmentation method based on multi-view alignment and pseudo label optimization, and aims to solve the problems of insufficient segmentation precision and low pseudo label quality caused by domain offset. According to the technical scheme, firstly, image level alignment is executed through a frequency domain smooth fusion module, and a class target domain image is generated; pre-training a segmentation network by using the image and generating an initial pseudo tag; then, through a two-stage optimization process, the integrity and the structural rationality of the pseudo tag are improved through prototype-based potential foreground completion and SAM-based structural perception enhancement in the process; and finally, on the basis of the optimized high-quality pseudo tag, constructing a multi-view prototype contrast learning framework to carry out final feature level alignment training. According to the method, the segmentation precision of the model on the label-free target domain is improved, and an effective scheme is provided for solving the challenge of scarcity of annotation data in medical image segmentation.

The invention belongs to the technical field of industrial system integration, and provides a design method for rapid adaptive integration of a cross-domain system. Through the method, developers extract business logic rules of the industry, standardize characteristic mapping and select, develop and integrate functional modules, logic processes and scene components of the cross-domain system, so that the cross-domain system which conforms to industry characteristics and meets business circulation is formed, and meanwhile, the development efficiency of the cross-domain system is improved. The newly added industry logic component can be classified into each industry characteristic model based on the knowledge base, and a multi-dimensional characteristic vector is generated in the platform, so that cross-industry integrated component multiplexing is realized, and the cross-field applicability and flexibility of the platform are further improved. Modularized construction and configuration type development are supported, mapping and matching functions are carried out on various industrial manufacturing industry characteristics such as machining, electronic and electrical appliances and equipment sets, the requirements for subdivision production in the industrial field and rapid adaptation of business processes are met, the system integration cost is reduced, and the application development efficiency is greatly improved.

The invention discloses a sorting manipulator intelligent control system based on big data, and the system comprises a multi-source data collection module which is used for synchronously collecting the real-time operation parameters of a sorting manipulator, the multi-modal feature data of a to-be-sorted object, and the dynamic environment parameters; the cross-modal fusion module is used for performing space-time alignment on the multi-modal feature data through a dynamic weighting algorithm; the double-domain cooperative training module is used for constructing a mapping relation between a virtual simulation domain and a physical entity domain, migrating initial model parameters trained by the virtual domain to the entity domain based on transfer learning, minimizing double-domain data distribution difference through a domain adaptation loss function, and generating an intelligent decision model adaptive to an entity scene; and the adaptive control module is used for generating a real-time control sequence through a model prediction control algorithm according to the sorting path planning result output by the intelligent decision model. According to the invention, more intelligent sorting manipulator control is realized.

The invention belongs to the technical field of natural language processing and knowledge engineering, and discloses a method, a system, equipment and a medium for automatically constructing a multi-field scientific knowledge base, and the method comprises the following steps: based on an input identifier list or a field search word, retrieving a full text of a literature, analyzing and converting the full text into a structured text; combining a pre-configured large language model with a field cue word, extracting key information in the structured text, and outputting structured data; the automatic script filters irrelevant, repeated or incomplete structured data according to a pre-configured filtering rule; performing standardization processing on the filtered structured data so as to realize the consistency and comparability of the data; and inserting the standardized structured data into an interactive knowledge base or database, establishing data association, and verifying association logic. The method supports cross-field rapid adaptation, solves the problems of low efficiency and high error rate of a traditional method, and can be widely applied to the fields of biomedicine, material science, synthetic biology and the like.

The invention belongs to the field of target detection, and provides an unsupervised domain adaptive target detection method and system, and the method comprises the steps: carrying out the training of a student model through a source domain image, and copying the parameters of the student model to a teacher model; the teacher model generates an initial pseudo label for a target domain image, determines a pseudo label through a pseudo label screening method, and extracts a local class prototype of a teacher target domain; the student model is trained by using a source domain image and a target domain image with a pseudo label, and a source domain local class prototype and a student target domain local class prototype are extracted at the same time to calculate a global class prototype; determining a total loss function according to supervision loss, unsupervised loss, antagonistic loss, intra-domain class prototype comparison loss and inter-domain class prototype comparison loss; updating the student model by using the total loss function, and updating the teacher model based on the updated parameters of the student model to obtain a trained teacher model; and inputting the target domain image into the trained teacher model for target detection.

The invention discloses a zero-sequence current and vibration signal fusion optimal transmission-based transmission chain cross-domain diagnosis method. The method comprises the following steps of data acquisition and preprocessing; constructing a multi-sensor fusion feature optimal transmission domain adaptive diagnosis model: firstly, extracting multi-scale features in a source domain and a target domain by using multi-scale depth separable convolution, and enhancing feature representation through a space channel attention mechanism; then calculating domain adaptation loss, calculating local maximum mean value difference loss between source domain and target domain features, and aligning loss of source domain and target domain distribution by using an optimal transmission method; and carrying out model training and fault diagnosis. According to the method, the local maximum mean value difference loss and the optimal transmission distance loss are combined to construct a double-constraint optimization target, the local maximum mean value difference loss is responsible for capturing distribution differences more finely, the optimal transmission distance loss is responsible for finding an optimal transmission scheme to reduce the differences, and the generalization ability of the model is improved.