737 results about "Characteristic space" patented technology

The invention belongs to the field of artificial intelligence, particularly relates to a large model illusion suppression method, system and equipment based on a dynamic knowledge base and multi-modal consistency constraint, and aims at solving the problem that factual illusion is likely to occur when an existing large language model generates content. The method comprises the steps that a knowledge base of multi-source heterogeneous data is constructed and dynamically maintained, and a dynamic credibility weight fusing data source authority, knowledge timeliness and multi-modal consistency is calculated for each piece of knowledge in the knowledge base; when the content is generated by the model, high-credibility related knowledge is retrieved from the knowledge base according to the current context; in the decoding stage of the model, a constraint loss item is designed, and the generation probability is adjusted in real time by calculating the similarity between the currently generated content and the retrieval knowledge in the feature space. According to the method, the multi-modal knowledge base for dynamic credibility evaluation is introduced, and the real-time consistency constraint is applied in the generation and decoding link, so that the accuracy and the reliability of the generated content are remarkably improved.

The invention relates to the technical field of intelligent manufacturing equipment fault diagnosis, and particularly discloses a Meta-Transform driven multi-working-condition equipment bearing fault diagnosis method and system. The method aims at bearing fatigue damage risks caused by dynamic adjustment of technological parameters of a numerical control machine tool in the aerospace manufacturing process and challenges such as feature distribution offset and fault sample scarcity caused by variable working conditions. The diagnosis system is constructed through three core modules. The method comprises the following steps: firstly, reconstructing an original bearing signal into a multi-scale time-frequency feature space by adopting continuous wavelet transform; then designing a causal Transform architecture with a strict lower triangle attention mask, and realizing feature extraction and classification according to a physical causal law of fault propagation; and finally, integrating the mechanisms into a model-independent element learning framework, and realizing cross-working-condition rapid self-adaption through a self-adaption gradient pruning strategy. The bearing fault diagnosis accuracy under the condition of few samples is improved, the interpretability and generalization ability of the model are enhanced, and the industrial application practicability of bearing fault diagnosis is improved.

The invention relates to a PCB production line process parameter intelligent matching method based on feature space mapping, and the method comprises the steps: collecting and fusing the material attributes, structure parameters, sizes and historical process records of a plurality of batches of PCB products, carrying out the normalization preprocessing, removing abnormal data, and constructing a high-quality feature matrix; after multi-dimensional feature expression is realized by utilizing a multi-scale embedded network, a mapping relation between features and process parameters is dynamically learned on the basis of an adaptive space mapping network in combination with a soft constraint multi-objective loss function, and gradient cutting, step length adjustment and a disturbance elasticity pool mechanism are introduced in a parameter recommendation process to guarantee convergence and stability. After the parameters are implemented, feedback data are collected in real time, periodic iteration distillation optimization and loss function recalibration are carried out, and finally a fine-tuning parameter recommendation scheme marked with conflict indexes, confidence intervals and weight suggestions is output for an engineer. According to the scheme, the intelligence, the traceability and the field adaptation capability of parameter recommendation are improved.

The invention relates to the technical field of mode recognition and data recognition, and discloses an intelligent visual detection method for non-standard precision part surface microdefects, which comprises the following steps: acquiring surface gray level image data of a to-be-detected part, physically abandoning low-frequency components through discrete wavelet transform, and reserving high-frequency detail components to construct a frequency domain input tensor; constructing a double-flow reconstruction model containing a space domain coding network and a frequency domain coding network, and minimizing the distribution difference of the same feature between double-domain characterization through potential feature space consistency constraint joint optimization; the method comprises the following steps of: calculating a spatial domain residual image and a frequency domain residual image, combining a texture topological residual image extracted by structural tensor characteristic decomposition, and generating a comprehensive abnormal response image through weighted fusion to judge the defect, and effectively inhibiting macroscopic geometric contour interference through frequency domain decoupling and a topological check mechanism on the premise of not needing a standard geometric template. And sensitive perception and accurate identification of weak texture defects on the surface of the non-standard part are realized.

The application discloses a feature editing method for large model content security, which compares and analyzes the sparse coding features of a chat assistant constructed based on a large language model under positive user input and negative user input, extracts the internal response differences of the model to different semantic directions, and the mechanism can automatically and accurately identify the key feature dimensions highly related to the semantic direction of the target attribute. The model activation is mapped to a sparse feature space by using a sparse autoencoder, and each dimension of the feature has independent and interpretable semantic meaning. By injecting a feature guide vector in the space, the interference of the control process on the text grammar, fluency and information density is significantly reduced. The sparse representation mechanism is introduced to structure the intermediate activation features in the reasoning process of the large language model and to intervene in a targeted manner, so that the reply of the chat assistant to the user input conforms to the preset safety specification, and the safety and controllability of the chat assistant in the interaction with the user are improved.

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 discloses a self-adaptive visual admittance control method fusing fluid characteristics and multi-modal perception, which comprises the following steps: designing a self-adaptive Bingham-shear thickening fluid virtual damping coefficient through nonlinear mapping based on sigmoid, and combining a threshold triggering behavior of a Bingham fluid and a sudden stiffening characteristic under the impact of the shear thickening fluid; the flexibility is enhanced under the action of small force, and the anti-interference capability is improved under impact. Besides, a force auxiliary function based on force amplitude is introduced, an anisotropic compliance strategy is combined, rigidity and damping are dynamically adjusted by identifying the main force direction, and the mechanism can reduce sensitivity to noise of a micro sensor and ensure stability and accuracy in the task execution process. Meanwhile, an environment attraction domain model is established in a feature space, Lyapunov analysis shows that the system has consistent final boundaries, stable convergence is ensured, and secondary correction is supported.

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 provides an industrial anomaly detection and root localization method and system based on data driving, and the method comprises the steps: carrying out the cleaning, feature extraction and normalization processing of original data collected in an industrial production process, and constructing a feature space; based on a local anomaly factor LOF and a mahalanobis distance MD method, jointly detecting local anomaly and global anomaly, and identifying an abnormal working condition; extracting space and time correlation characteristics of the abnormal variables through Pearson correlation weighting and Granger causal test to form a space-time correlation matrix; constructing an abnormal causal network based on the matrix, and tracing an abnormal root and a propagation path through depth-first search and abnormal propagation intensity evaluation; and finally, dynamic optimization of the anomaly detection and diagnosis method is realized based on parameter self-adaption and model incremental learning. According to the method, the anomaly detection accuracy and the anomaly traceability interpretation capability can be effectively improved, and the intelligent level and the self-adaptive capability of data processing are enhanced.

The invention discloses a dynamic planning method and system for an intelligent patrol point location of power transformation equipment, and belongs to the technical field of intelligent patrol of power systems, the dynamic planning method for the intelligent patrol point location of the power transformation equipment comprises the following steps: mapping dynamic features and static attributes to the same feature space and carrying out cross-modal association; constructing a defect severity model, and introducing defect severity in risk quantification calculation to generate a point location priority list; a transformer substation three-dimensional point cloud model is built, path nodes are initialized according to a point location priority list, and a greedy algorithm is utilized. A closed-loop decision-making system of multi-modal data fusion is constructed; according to the method, automatic planning driven by a risk quantification model based on defect history, generation of a three-dimensional space non-blind area coverage path, dynamic adjustment triggered by two factors of equipment change and inspection effect, deep collaborative analysis of machine account-defect-real-time data and continuous inspection of a complex scene are guaranteed by a semantic compensation mechanism.

The embodiment of the invention discloses an air conditioner maintenance data classification method and system based on machine learning, and the method comprises the steps: integrating multi-source heterogeneous maintenance records generated in the maintenance process of air conditioner equipment, building a correlation index through a common identification field, and fusing dispersed data into a maintenance data set in a uniform format; performing hierarchical semantic analysis on unstructured texts in the set to generate structured semantic features, and performing time sequence feature extraction on structured data; then constructing a hybrid classification model training framework fusing semantic and time sequence features, and generating a maintenance data classification model through feature space alignment, dynamic weight distribution, hyper-parameter optimization and an early stop strategy; and finally, classifying newly-added maintenance records by applying the model, checking by combining an expert knowledge base, manually rechecking conflict results, and returning corrected data as an incremental sample back to the model to realize continuous optimization.

The invention relates to the technical field of deep counterfeit image detection, and discloses a deep counterfeit detection method and system, a storage medium and computer equipment. The method comprises the following steps: firstly, constructing a reference data set containing a forged image and an original real image; secondly, through an integrated model, generating antagonistic samples for the reference data set, and integrating the successfully attacked antagonistic samples into an antagonistic sample set; and finally, merging the reference data set and the adversarial sample set, and constructing a robustness enhanced data set containing four types of samples. In the model training stage, multi-classification cross entropy loss and comparative learning loss are combined, and expression of the model in a feature space is optimized through comparative learning constraint, so that the model learns discriminative features with more compact intra-class features and more dispersed inter-class features. The model trained by the method not only can effectively defend against attack and improve robustness, but also surpasses original detection performance on clean samples, and has remarkable technical advantages and application value.

The invention relates to the technical field of computer vision, in particular to an image target detection system and method based on deep learning, and the system comprises a dynamic feature alignment unit, a motion blur compensation unit and a feature fusion control unit. A dynamic feature alignment unit generates spatial deformation parameters through a deformable convolutional layer and an offset prediction sub-network, resamples a shallow high-resolution feature map, and realizes deep and shallow feature space alignment, and a motion blur compensation unit generates a motion vector based on brightness gradient field difference, constructs a mask and weights a suppression blur region, so as to realize deep and shallow feature space alignment. The feature fusion control unit analyzes local entropy and target size distribution, dynamically distributes feature weights and feeds back and optimizes offset parameters, a closed-loop learning loop is formed by the method, and the problems of inaccurate feature alignment, fuzzy interference and poor scene adaptation are solved.

The invention relates to an animal wound multi-modal intelligent identification method based on artificial intelligence, and the method comprises the steps: carrying out the feature extraction and semantic constraint through multi-modal sample collection and metadata extraction, employing an image preprocessing and text natural language processing technology, and combining a mixed visual model of a convolutional neural network and a visual Transform, and a large language model. A cross-modal attention mechanism and a semantic trigger are utilized to realize feature reweighting, clinical standard soft boundary constraints are introduced, and fuzzy semantic rules are converted into learnable constraints in a feature space, so that the accuracy and consistency of model judgment are improved. The method has adaptive optimization and incremental learning capabilities, and is beneficial to improving generalization and clinical applicability of exposure level intelligent judgment under different animals and complex wound types.

The invention discloses an industrial defect detection method based on self-supervised pre-training and feature space generation. Firstly, features are extracted and mapped to a unified potential space; secondly, introducing a SimMIM framework, which is one of mainstream technologies in the current industrial vision pre-training field, to carry out self-supervised pre-training to improve the representation capability, and using an OfficientForme network to improve the reasoning speed and reduce the memory demand; secondly, performing abnormal synthesis in a feature space by adopting a generative adversarial network and enhanced Perlin noise, outputting a feature increment and a soft mask by a generator, and performing linkage updating with a pixel-level mask and an image-level label; thirdly, a segmentation-classification double-head framework is adopted, a segmentation head outputs a pixel-level anomaly graph, and a classification head outputs an image-level anomaly score; and finally, training optimization is carried out through strategies such as abnormal graph up-sampling smoothing, a grouping learning rate, multi-stage scheduling and the like. The method realizes accurate and controllable synthesis of the abnormal region under the scene without, with or with mixed supervision, and significantly improves the robustness and real-time performance.

The invention provides a multi-modal sentiment analysis method and system based on main modal two-stage guidance, and relates to the technical field of sentiment analysis. Inputting the multi-modal data into a multi-modal sentiment analysis model, and extracting language, visual and acoustic features from the multi-modal data through a feature extraction module; semantically decoupling the multi-modal features into modal invariant features and modal unique features through a feature space distribution alignment module, and realizing feature distribution alignment dominated by language modals through alignment reconstruction constraints; performing self-attention modeling on the modal invariant feature through an attention enhancement module to obtain a first enhanced feature, and adaptively enhancing the visual and acoustic unique features through a cross-modal attention mechanism by taking the language unique feature as a dominant feature to obtain a second enhanced feature; the first enhanced feature and the second enhanced feature are fused through the emotion prediction module, an emotion intensity prediction result is obtained through regression prediction, and the accuracy and robustness of emotion analysis in a complex scene are improved.

The invention relates to the field of target detection, and particularly discloses an electric power work target detection method based on multi-modal large model knowledge distillation, which utilizes a vision-language multi-modal large model as a teacher model, and improves the target detection efficiency by expanding prompt word guidance. A high-quality pseudo label and a region-text pair are generated for an unlabeled electric power work image as a supervision signal, and on this basis, through joint optimization of detection loss, feature distillation loss, logic distillation loss and multi-modal contrast learning loss, a lightweight YOLO student model is guided to learn positioning and classification knowledge and to learn a multi-modal contrast learning loss. And deep alignment with the open vocabulary understanding ability of the teacher model is carried out on the feature space and semantic level, so that a semantic gap between closed category detection and open world perception is effectively bridged. Through the mode, the detection precision and generalization ability of the student model on common, rare and even unseen targets in the electric power work scene are remarkably improved.

The invention discloses a large model content security-oriented feature editing method, which comprises the following steps of: comparing and analyzing sparse coding features of a chat assistant constructed on the basis of a large language model under positive user input and negative user input, and extracting internal response differences of the model in different semantic directions; the mechanism can automatically and accurately identify key feature dimensions highly related to the semantic direction of the target attribute. A sparse auto-encoder is utilized to activate and map the model to a sparse feature space, and each dimension of feature has an independent and interpretable semantic meaning. By injecting a feature guide vector into the space, the interference of a control process on text grammar, fluency and information density is remarkably reduced. A sparse representation mechanism is introduced, structural modeling and targeted intervention are carried out on intermediate activation features in a big language model reasoning process, replies input by a chat assistant to a user are guided to conform to a preset safety specification, and the safety and controllability of the chat assistant in interaction with the user are improved.

The invention relates to the technical field of radar signal processing, and discloses a complex ground reflection environment-oriented dual-path target detection method, which comprises the following steps of: performing quantile truncation and discretization processing on original range profile data, mapping a processed signal to a D-channel discrete feature space and generating a D-dimensional discrete feature matrix; the D-dimensional discrete feature matrix is input into a dual-channel parallel processing architecture, a CFAR adaptive detection module is adopted in a first path to suppress strong clutters to generate a first response matrix, and a false attention mechanism is adopted in a second path to enhance weak targets to generate a second response matrix; performing spatial alignment and weighting processing on the first response matrix and the second response matrix, and performing cross-domain confidence fusion to output a comprehensive response matrix; and carrying out fixed threshold binarization processing on the comprehensive response matrix based on a global threshold to obtain a target detection result. The method has both local gain compensation and global adaptive suppression, and can realize more robust detection of a weak target under a complex background.

The invention discloses a complex scene target detection method based on a model dynamic joint mechanism, and the method comprises the steps: carrying out the preliminary detection through an RT-DETR model, retaining more potential targets through dynamic threshold adjustment, and projecting a generated detection frame to a feature space of an improved YOLOv12 model through dual-mode feature mapping; the improved YOLOv12 model integrates an SEAM attention module and a rejection loss function so as to enhance feature representation and positioning compactness of an occluded target. Then, a model joint mechanism is adopted to process preliminary results of the two models; through difficult case mining and online learning, missing detection targets are supplemented, and the RT-DETR model is optimized; and intelligently fusing the detection results of the two models through dynamic weight distribution based on scene complexity and hierarchical fusion of a decision tree. And finally, post-processing is carried out by using an improved non-maximum suppression algorithm, and mistaken deletion is reduced. According to the method, the problems of missing detection, false detection and inaccurate positioning of the target in a complex scene are effectively solved, and the recall rate and the accuracy rate of detection are remarkably improved.

The invention discloses an intelligent decision-making method and system based on deep learning, and relates to the field of data processing. The method comprises the steps of obtaining place data and corresponding attribute data of a to-be-decided project; establishing a fuzzy relation matrix between the places and the attributes; generating a network model reflecting trust relationship strength among users through trust propagation operation of the graph neural network; identifying a community structure containing a community overlapping degree through a community discovery algorithm; and according to the trust relationship strength between the users and the community overlapping degree, calculating an influence weight of a decision maker, forming a group consensus through a robust optimization method, and generating a decision result of the project to be decided. Aiming at low network relation modeling precision caused by multi-source heterogeneous data in bus station layout decision making in the prior art, the method and the device have the advantages that the network relation modeling precision is low through accurate modeling of an information propagation path in a complex trusted network, effective dimension reduction representation of a high-dimensional feature space and robust optimization solution in an uncertain environment; therefore, the calculation precision and robustness of the bus station layout intelligent decision-making system are improved.

The invention relates to the technical field of fault early warning, in particular to an agricultural robot self-learning fault early warning method based on multi-source data fusion, and the method comprises the steps: obtaining the operation data and environment data of an agricultural robot, and generating a multi-source fusion tensor through time sequence synchronization and normalization processing; constructing a neural network comprising a self-supervised check branch and a minimum decoupling branch, and generating a reconstruction tensor; mapping operation and environment vectors by using an independent encoder, stripping linear and nonlinear environment features through an antagonism prediction path established by an orthogonal penalty term and a gradient inversion layer, and generating operation features; and constructing a feature topological graph, calculating local relative density, and generating a fault early warning signal. According to the method, deep association of multi-source data is mined through self-supervised verification, the problem of false alarm caused by environmental noise in an unstructured environment is effectively solved by forcibly eliminating environmental interference in a feature space by using orthogonal constraint and antagonism training, and the robustness and accuracy of fault early warning are remarkably improved.

The invention relates to a small sample target detection method and device based on a decoupling prototype and a medium, belongs to the technical field of crossing of remote sensing image processing and computer vision, and can effectively separate interference features such as category semantics, angles and backgrounds of remote sensing targets. A double-branch prototype decoupling structure of'category semantic branch + angle geometric branch 'is constructed, and a decoupling loss constraint based on cosine similarity is introduced, so that a category prototype and an angle prototype are approximately orthogonal in a feature space; and then the RPN and the detection head are cooperatively guided through double prototypes, and a two-stage training strategy is combined, so that efficient migration of basic class knowledge to new class small samples is realized, and the small sample target detection performance and robustness in a complex remote sensing scene are improved.

The invention discloses an intelligent customer intention recognition method and system based on multi-modal deep learning, and belongs to the technical field of artificial intelligence and natural language processing, and the method comprises the steps: obtaining customer service text data, customer service voice data and customer behavior data, and carrying out the preprocessing; constructing a topology-maintained heterogeneous feature mapping mechanism, mapping customer service text data features and customer service voice data features to a shared feature space, and maintaining respective topological structures; constructing a cross-modal attention calculation framework based on Riemannian manifold to obtain a text speech similarity score matrix; inputting the customer behavior data characteristics into a user behavior intention recognition model to obtain a user behavior intention recognition probability matrix; the two matrixes are input into an intelligent client intention recognition framework through a dynamic spectrum cluster fusion semantic matching degree calculation method, a client intention recognition result is output, and the client intention recognition accuracy is improved through multi-modal data fusion, topological structure keeping and multi-scale semantic matching.

The invention discloses an unmanned aerial vehicle aerial photography target detection method and device and a storage medium, and relates to the technical field of computer vision. The method comprises the steps of constructing a detection model based on a YOLO framework, replacing a convolution module of a feature pyramid in a backbone network of the detection model with a block local-global attention module, designing a triple feature fusion module in a neck network, and adopting a dynamic detection head fusing a triple attention mechanism at a detection head; the block local-global attention module is used for decomposing an input feature map into position-weighted sub-map blocks and fusing the sub-map blocks, enhancing each sub-map block in a compact feature space in parallel by utilizing unified perception context attention, meanwhile, carrying out implicit modeling on a global context, and finally, recovering an original structure through reverse mapping; and training and verifying the detection model by using the unmanned aerial vehicle aerial image data set. According to the method, high-speed reasoning is kept, and meanwhile, the detection precision of tiny and dense targets is remarkably improved.

The invention discloses a landslide displacement double-layer fusion prediction method and model, and the method comprises the steps: carrying out the decomposition of an original displacement time sequence through employing an ICEEMDAN algorithm, and obtaining a plurality of IMF components; performing feature engineering on each IMF component, representing a displacement trend by adopting a trend slope and a window mean value, representing mutation early warning by adopting kurtosis and a frequency spectrum entropy, representing a period rule by adopting a main frequency and a zero-crossing rate, representing system stability by adopting a sample entropy and a standard deviation, and constructing a three-dimensional feature space fusing a time domain and a frequency domain; data standardization is carried out on the extracted features, the interference effect of dimensions on the model is eliminated, and it is ensured that all feature dimensions are within a unified calculation scale range; a CNN-BiLSTM model is constructed for each IMF component; a CPO algorithm is used to optimize the CNN-BiLSTM model; according to the method, the data acquisition difficulty during model training and use can be reduced, and the usability of the model in actual deployment is enhanced; and the prediction precision and the accuracy of landslide displacement prediction are improved.

The invention belongs to the field of infusion monitoring, and particularly relates to a target container liquid level visual detection and early warning method and system.The method comprises the steps that an enhanced image and a three-dimensional point cloud sequence are obtained by collecting a target container deformation image, a three-dimensional point cloud and flexibility attribute data and conducting polarized light enhancement preprocessing; obtaining a deformation feature space and a deformation liquid level line feature sequence of the target container at different time points by means of a spatial-temporal feature extraction model, and constructing a three-dimensional dynamic deformation model in combination with three-dimensional modeling, finite element analysis and a NeoHookean model; based on the model, an undeformed standard model and a simulation algorithm, obtaining a correlation mapping function of the liquid level line deviation and a deformation characteristic space; the correlation mapping function is fed back to the spatial-temporal feature extraction model, the height and coordinates of the corrected deformation-free liquid level line are obtained in combination with the real-time target container deformation image, real-time monitoring and early warning in the infusion process are finally achieved in combination with a preset early warning rule, and the accuracy and reliability of infusion monitoring are improved.

The invention relates to the technical field of industrial process prediction, in particular to a tundish erosion prediction method based on a hierarchical hybrid expert framework. The method comprises the following steps: collecting time sequence physical field data of the tundish, and screening features to construct a unified feature space; shunting the feature space to obtain a time sequence feature and a statistical aggregation feature; a statistical aggregation feature training classifier is utilized to generate a calibration posterior probability, and a gating network is constructed; generating an initial mode subset based on a posterior probability and training a corresponding expert model; the confidence of data to be measured is obtained by the gating network, and a single expert model is selected for prediction or multiple expert models are fused through a self-adaptive strategy for weighted prediction according to whether the confidence exceeds a threshold value or not; and finally, reconstructing the predicted value into an erosion thickness absolute value through inverse transformation. According to the method, the accuracy and adaptability of tundish erosion prediction are effectively improved.

The invention discloses a motor bearing fault detection system and method based on robust deep learning, and belongs to the technical field of mechanical fault detection and intelligent perception. Feature extraction is carried out on an original vibration signal with a label based on a supervised learning branch network, and the original vibration signal is used as a reference sample; the samples with the same fault category and different fault categories as the reference samples are positive samples and negative samples, and inter-class separation and intra-class aggregation relations in a triple loss optimization embedding feature space are introduced to generate embedding representation; based on an unsupervised learning branch network, encoding the original vibration signal after time domain and frequency domain artificial feature extraction, and introducing triple loss to carry out unsupervised embedding learning to generate high-level feature embedding representation; and the embedded representations output by the two branch networks are fused, dual loss of triple loss and center loss is introduced for training, and a bearing fault detection model after training is completed is used for bearing fault detection.

The invention discloses a data asset visualization and collaborative governance method and system based on an AI agent, and the method comprises the steps: carrying out the mapping of attribute parameters, such as the type, scale, value and the like, of data assets, building a feature space, completing the nonlinear mapping from high dimension to low dimension through a t-SNE model, and achieving the visualization feature presentation of the data assets. Initializing a deep Q network agent, setting a reward function by taking visual features as a state space, governance operation as an action space and parameter change, selecting actions by the agent according to the reward function to execute collaborative governance, and storing empirical data into a buffer region to train the network; and training is repeated until network convergence, and a visual and cooperative treatment strategy is generated. The system is composed of a data asset feature mapping unit, a t-SNE conversion unit, an agent initialization unit and the like, all the units are mutually connected and cooperatively operate, data asset visualization and intellectualization and automation of cooperative governance are achieved, and the data asset management efficiency and the value mining capacity are effectively improved.