167 results about "Data density" patented technology

The invention relates to an optimization processing method for a streaming data dynamic window, and the method comprises the steps: taking streaming data as a dynamic system, and calculating the data disturbance energy at each moment on a time axis; constructing a window trigger point according to the potential energy curve; when the trigger potential locally reaches a peak value or the trend changes, adaptively dividing the window; the starting boundary and the ending boundary of each window are defined as elastic intervals; before the window slides or is closed, window boundary self-adjustment and complementation are carried out according to compensatory analysis of subsequent data streams; taking a large window as a container, and dividing sub-segments according to data density change; partial states are shared among the sub-windows, and fine-grained analysis including anomaly detection and instantaneous peak value recognition is executed; introducing a life cycle termination relationship driven by an entropy threshold: when the window activeness is lower than the entropy threshold, triggering state compression or early release; the accuracy and dynamics of window division are realized, the response capability of a system to data fluctuation and burst is enhanced, and the elasticity and fault-tolerant capability of window boundaries are improved.

The embodiment of the invention provides a recommendation model training method, an object recommendation method, a recommendation system and computing equipment. The recommendation model training method comprises the following steps: acquiring user data of a sample user for a sample object; performing confrontation pre-training on the initial recommendation model through a domain classifier of a user type based on coding features of user data output by the initial recommendation model to obtain a pre-trained recommendation model; dividing the user data into meta-learning task sets of a plurality of recommendation scenes, and training the pre-training recommendation model based on the meta-learning task sets to obtain a meta-parameter recommendation model; and based on the meta learning task set of the target recommendation scene, constructing an adaptive task set of the target recommendation scene, and based on the adaptive task set, training the meta parameter recommendation model to obtain an adaptive recommendation model of the target recommendation scene. Recommendation precision is guaranteed under low data density, balance between sparse data and high precision is achieved, and training efficiency is improved.

The invention discloses a multi-dimensional time series data dynamic evolution analysis method and device, a medium and electronic equipment. The method comprises the steps that time series data are obtained and subjected to structured preprocessing; performing multi-dimensional quantization and attribute endowing on each piece of time series data, and constructing a three-dimensional data node containing a space coordinate, an initial weight and a timestamp; based on the set of the three-dimensional data nodes in a specific time window, constructing a three-dimensional data density field capable of representing data hotspot distribution and intensity in the time window; in the three-dimensional data density field, identifying a density peak region as a core data cluster, calculating the cluster mass according to the density and the coverage range of the core data cluster, and calculating the association strength among different core data clusters; and serializing a plurality of continuous time window data field state snapshots to form a dynamic evolution view capable of showing data hot spot generation and extinguishing and data cluster fusion and splitting processes. According to the invention, abstract multi-dimensional data is converted into a visual three-dimensional density field.

The invention discloses a sensor data enhancement method and system, relates to the technical field of sensor data processing, and solves the problem that an existing sensor data enhancement scheme is based on a static, single and fixed processing mode. In order to solve the problems that data distribution characteristics of different scenes cannot be adapted, environmental interference or sensor inherent errors are not considered, and the influence of abnormal values is easily caused, the technical scheme is characterized in that sensor data enhancement is performed based on a dynamic adaptive grid and a multi-compensation mechanism, and dynamic grid division, sensor error compensation and abnormal value influence processing are realized; static interpolation is upgraded into a dynamic self-adaptive data interpolation engine, data values of unknown points in a whole research area are calculated through limited sampling point numerical values, and by means of the data processing and enhancing algorithm, the data density is improved, and the problems of data sparseness and data missing are solved.

The invention discloses an NUMA (Non Uniform Memory Access) architecture perceived high-scalability learning index construction method and system and an operation method. According to the construction method, a mixed node tree structure is adopted to organize data, the structure comprises internal nodes accurately searched by a linear model and leaf nodes for storing data, the leaf nodes are composed of ordered gap nodes and segment nodes, and the segment nodes support dynamic evolution from an ordered stage to a semi-ordered stage based on data density. Conflict data is managed using hierarchical benchmark buckets. According to the system, a self-adaptive node evolution mechanism based on an operation cost model is realized, index performance is optimized through node reconstruction triggered by foreground write operation and background hot and cold node compression, a multi-thread concurrency control strategy is adopted, and a memory management and thread scheduling framework perceived by NUMA is adopted. The operation method provides an efficient point query and point insertion operation process. The data layout can be dynamically optimized, the operation cost is reduced, and the method is suitable for high-performance database and memory management scenes.

The invention provides a polymer flooding reservoir model prediction method and device, equipment and a storage medium. Relates to the field of oil-gas field development engineering. The method comprises the steps that an injection-production system matrix is converted into a multi-channel field graph containing time codes, and a depth operator network is constructed in combination with three-dimensional physical field data output through reservoir numerical simulation; a branch network adopts a residual attention network to convert the injection-production system field graph into a parameter-dependent feature graph; the backbone network converts tensors of time coding into spatial modulation weights, and initial operators of a pressure field, an oil saturation field and a polymer concentration field are generated through dot product operation; dynamically enhancing the data density of the high-error region by using an adaptive sampling mechanism based on a Gaussian mixture model; and realizing transient multi-physics field synchronization high-precision prediction by combining geometric similarity matching multiplexing physical tags. The problems that a traditional numerical simulator is low in calculation efficiency and a conventional agent model is poor in generalization ability are solved.

The invention provides an operation and maintenance performance assessment data management system based on dynamic weight optimization, and relates to the technical field of operation and maintenance management, and the system comprises a quantification module which is used for obtaining an original operation and maintenance data sequence through a data collection interface, and mapping the operation and maintenance data sequence to a multi-dimensional feature space; in the multi-dimensional feature space, quantitative analysis is carried out on spatial distribution of data points, and data density distribution characteristics of all areas are determined; the calculation module is used for executing unsupervised clustering operation based on the density distribution characteristics and the proximity relation of the data points to obtain a group of data sets with internal consistency; core statistical features are calculated for each generated data set, and a specific calibration coefficient is derived for the data set based on the core statistical features. According to the invention, precision, scenario and high efficiency of operation and maintenance performance assessment are realized.

A hybrid recommendation system for generating personalized film recommendations, the system comprising the following: a processing unit for receiving metadata, configured to receive descriptive data related to films, the descriptive data including plot summaries, genre identifiers, cast lists and director names; a text preprocessing module that is communicatively coupled with the metadata ingestion processing unit and is configured to perform tokenization, stopword removal, stemming, and lemmatization on the descriptive data to obtain a processed corpus; a TF-IDF vectorization unit configured to encode the processed corpus into high-dimensional semantic feature vectors by calculating TF-IDF inverse document frequency values ​​over the entire film dataset, with the semantic feature vectors representing the contextual meaning of terms for individual films; a collaborative filter engine comprising a user-element interaction matrix, wherein the engine is configured to compute latent preference signals using one or more techniques selected from the group consisting of cosine similarity, k-nearest neighbor similarity, and matrix factorization; a score fusion controller coupled to both the TF-IDF vectorization unit and the collaborative filter engine, wherein the module is configured to normalize the semantic feature vectors and the collaborative preference scores and dynamically combine them according to an adaptive weighting coefficient, the coefficient being determined as a function of data density, interaction sparsity, and user history length; and a recommendation output module configured to evaluate candidate films for each user based on the combined score and generate a top-N recommendation list.

The invention relates to the technical field of supply chain traceability, in particular to a mother and baby product production traceability management method and system based on a block chain, and the method comprises the following steps: obtaining raw material and packaging chain segment data, extracting batch information and detection numbers, recognizing nodes and recording cross chains, counting the cross frequency of finished product segments, and constructing a path index map. And screening an optimal main path, generating a main path identifier, judging a filling data density grade, marking a chain segment needing to be adjusted, and generating a scheduling strategy parameter table. According to the method, field splicing is performed on the batch number, the raw material batch number and the detection report number to compare the segment identifier, so that node connection identification and cross structure establishment are realized, chain segment relevance is enhanced, cross node frequency is counted, high-frequency node pairs are extracted to construct the path, and the path identification accuracy is improved; and screening a path with the fewest skipping times and complete node coverage as a main path, enhancing the traceability stability, sensing the frequency based on the time interval and the chain block number, and realizing the dynamic scheduling of the chain segment state.

The present application relates to the technical field of storage, and discloses a data attribute-aware storage system and data management method for a key-value database. The method comprises two parts, i.e., data hotness attribute-based writing and data density attribute-based merging. The data hotness attribute-based writing comprises classifying key-value pairs into cold key-value pairs and hot key-value pairs, to realize separate storage of cold data and hot data by means of a data hotness attribute-based writing strategy. The data density attribute-based merging comprises: when key-value pairs in an SSTable are extracted during a merging operation and form a sorted key-value pair sequence, extracting the key-value pairs one by one, and calculating a binary difference between every two adjacent key-value pairs; and if the difference is greater than a set threshold, stopping filling the current SSTable, and creating a new SSTable. In the present invention, by analyzing data hotness and data density, data having different attributes are classified to different data partitions, thereby reducing I / O amplification caused by repeated reading and writing in a compression process.

The invention relates to the technical field of supply chain data storage, and discloses a supply chain performance data block chain storage method based on multi-party security computing. The method comprises the following steps: monitoring performance data states and relative interaction relationships of all participants of a supply chain in real time, and judging whether the data density in a target supply chain region exceeds a safety threshold value or not; when a threshold value is exceeded, analyzing the interaction influence of data of each participant through a multi-party security computing protocol, and evaluating the hidden danger of privacy disclosure caused by data sharing; analyzing and processing the spatial and temporal change of the historical performance data by adopting distributed feature learning and combining with a time sequence, and evaluating the potential interference degree of historical data fluctuation on the integrity of the current performance data; and based on the privacy leakage hidden danger and the potential interference degree of historical data fluctuation, whether overall block chain evidence storage optimization of the target supply chain region is carried out is judged. According to the method, the problems of supply chain data density and privacy protection can be solved in a targeted manner, and evidence storage optimization is realized by combining historical data spatial-temporal characteristics.

The invention relates to a space-time big data dynamic index construction method and device and a storage medium. The method comprises the following steps: inputting a spatio-temporal data stream into a distributed system to carry out fragmentation processing and index tree structure construction to obtain an initial index structure, and carrying out parallel feature extraction on the spatio-temporal data stream to obtain a spatio-temporal distribution feature vector; calculating a space-time overlapping degree, a data density similarity and a query frequency correlation degree of adjacent nodes based on the space-time distribution feature vectors to obtain a combined evaluation score of the node pair; performing multi-objective optimization based on the merging evaluation score to generate an optimal merging threshold value; and according to the optimal merging threshold, performing incremental merging operation on the node pairs, and performing data recombination and range boundary updating to obtain an optimized index tree structure. According to the method, smooth adjustment of the index structure is realized, the influence of structure adjustment on online query service is minimized, and the index structure can be continuously optimized and kept in an efficient state.

The invention discloses a geological mineral data intelligent modeling method and system based on multilevel space-time intrinsic coding, and the method comprises the steps: generating a 9-level adaptive grid through employing an improved HEALPix subdivision algorithm for the geological complexity and target data density of a to-be-detected region; based on a 9-level adaptive grid, encoding space time-varying grid units of existing geological mineral resource data to obtain a geological feature space matrix; constructing a multi-modal data register based on Transform, inputting the 9-level adaptive grid and the geologic feature space matrix into the multi-modal data register, and outputting an associated feature matrix; using a hybrid interpolation model to perform hybrid interpolation on the correlation feature matrix to obtain a full-grid correlation correlation interpolation matrix; and a CNN error correction module is used to optimize the full-grid correlation degree correlation interpolation matrix to obtain a geological mineral resource space-time intrinsic code, and geological mineral resource data intelligent modeling is realized based on the geological mineral resource space-time intrinsic code.

The present invention discloses a method for identifying a graphics rendering mechanism based on rendering pipeline alignment. The method pre-executes a benchmark test program in CPU mode and GPU mode respectively, completes pre-alignment of the key stage processing interval and the stable working interval in the two modes to form a first sampling interval to achieve alignment of the rendering pipeline. The second sampling interval is formed based on the pre-alignment of frame complexity. The target sampling interval is obtained by finding the intersection of the first sampling interval and the second sampling interval. During actual testing, CPU load data of each of the two modes is obtained, the CPU load data is sampled according to the target sampling interval, and the correlation between the two sampling data is calculated to determine the graphics rendering mechanism of the tested system. The unified rendering action ensures that the sampling data covers the complete rendering cycle. The reasonable distribution of data density under different load scenarios is ensured by analyzing the frame complexity. The combination of the two improves the representativeness and statistical effectiveness of the data, and effectively improves the accuracy of identifying the graphics rendering mechanism.

The invention discloses a virtual space-time environment automatic completion method and system based on data density dynamic perception and a medium, and relates to the field of computer data processing. The method comprises the following steps: monitoring a data activeness index (such as spatio-temporal information entropy) of a spatio-temporal grid in real time in a distributed index database of a virtual spatio-temporal environment; when the index meets a preset data sparseness condition, generating a completion request; analyzing the space-time attribute tag of the target grid and acquiring associated multi-modal reference data from an external heterogeneous data source; carrying out feature recombination on the reference data by utilizing a generative neural network model, and generating discrete spatio-temporal data slices containing spatio-temporal anchor points and scene content attributes (such as environment illumination and material mapping); and finally, injecting the slices into the database to fill data holes. According to the method, a passive trigger mechanism and a structured reconstruction technology of unstructured data are introduced, so that the computing power cost of a large-scale virtual environment is effectively reduced, the problems of data sparseness and cold start of a long-tail region are solved, and the historical authenticity and time-space consistency of a virtual scene are ensured.

The invention discloses a genome variation cold and hot spot region prediction method and device, and the method comprises the steps: S1, carrying out the slicing of a target genome region according to a preset sliding window length, and constructing a multi-modal input tensor corresponding to a window; s2, inputting the multi-modal input tensor into a pre-trained deep learning prediction model, and outputting a cold and hot spot prediction score of each site in the window through a full connection layer; and S3, according to the cold and hot spot prediction scores, identifying a variation cold spot region and a variation hot spot region in the target genome region. According to the technical scheme provided by the invention, the dependence on the existing variation data density is eliminated, and non-blind area coverage in the whole exon group range is realized; meanwhile, the structured output based on the preset transcript coordinates can directly support clinical variation interpretation, a quantitative basis is provided for PM1 and cold spot evidence, and the proportion of unclear significance variation is effectively reduced.

The invention relates to the technical field of data storage, in particular to a distributed parallel data management method and system based on hierarchical storage, and the method comprises the steps: obtaining a distributed data set, numbering each data object in the distributed data set to obtain a first data object to an Nth data object, acquiring data type sequences, data size sequences and access frequency sequences of the first data object to the Nth data object, performing data density analysis to obtain first data density to Nth data density, and performing reference-dependency conjoint analysis on the first data object to the Nth data object to obtain a first weight matrix, the first correction density to the Nth correction density are obtained through density correction according to the first weight matrix, and hierarchical storage is performed according to the first correction density to the Nth correction density to obtain a data storage hierarchy allocation result, so that more accurate and efficient storage management of the distributed parallel data is realized.

The invention discloses a method for identifying a public charging infrastructure boundary in combination with Internet of Vehicles data and an artificial intelligence algorithm. The method comprises the steps of S1, data acquisition; s2, constructing a buffer area; s3, carrying out buffer region fusion; s4, decomposing and splitting a plurality of regions; s5, area calculation and small region elimination; s6, convex hull construction and softening treatment; and S7, determining the boundary, and determining the boundary of the charging infrastructure according to the softened boundary after the step S6. The technical scheme of the invention has the following advantages: (1) automatic identification is realized, and boundary information is extracted from actual behaviors without depending on manual delimitation or operator declaration; (2) the adaptability is high, the method can be suitable for different cities and different time periods, and different data densities and distributions can be flexibly adapted by adjusting parameters; (3) high-precision positioning is carried out, and real range identification can be realized by combining desensitized latitude and longitude data (4-bit precision) with a buffer zone strategy; and (5) the expandability is good, and boundary types and grades can be refined by combining other characteristics.

The invention relates to the technical field of data processing, in particular to a portable terminal with an electric power payment service intelligent management function, the system comprises a processor and a memory, and the processor executes a computer program of the memory to realize the following steps: obtaining an index combination for grouping users; clustering the users according to any index combination to obtain a cluster; obtaining an effect evaluation factor of any index combination according to the correlation of the historical data of the user behavior index of each user in each cluster, the data density in each cluster and the distance between every two clusters, obtaining an effective index combination according to the effect evaluation factor of each index combination, and obtaining the effective index combination according to the effective index combination. The users are grouped according to each effective index combination, and the user portraits are generated for providing personalized payment reminding services for each user, so that the accuracy of pushing the personalized payment reminding services for the users is improved.

The present invention relates to the field of data processing technology, and in particular to a data processing method for an electric power distribution device, comprising: constructing a data point set for reflecting the electric power load in the current area based on the collected electric power consumption data of users in the current area, clustering the data points in the data point set multiple times using the K-means clustering algorithm; obtaining multiple circular rings of each category in each clustering result, and determining the data density of each circular ring; sequentially determining the rationality of the intra-class density change, the rationality of the inter-class distribution, and the degree of preference of each clustering result; and determining the optimal clustering result of the K-means clustering algorithm based on the magnitude of the degree of preference, thereby realizing data processing of the electric power distribution device. The present invention deeply analyzes the distribution characteristics of the data points within the cluster category, taking into account the rationality of the intra-class density change and the rationality of the inter-class distribution, ensuring that the clustering results are more consistent with the spatial distribution and density differences of the actual electric power load, and more truly reflecting the electric power consumption behavior in different areas.

The invention discloses a power consumption data abnormity identification method and system, a medium and a product, and belongs to the field of power data abnormity identification, and the method comprises the steps: collecting the current original power consumption time sequence data of a plurality of loads in a target region, and obtaining the corresponding historical power consumption data and load flow calculation data; correcting the original power consumption time sequence data based on the historical power consumption data to obtain power consumption time sequence data; performing multi-scale time sequence feature extraction and prediction on the historical power consumption data by using a preset P-T imemi x algorithm to generate prediction data with higher data density; performing bidirectional alignment fusion on the power consumption time sequence data and the prediction data to obtain fused power consumption data; and combining the fused power consumption data with the power distribution network load flow calculation data, calculating a preset deviation index, judging whether each data point in the fused data is abnormal or not, and outputting an abnormality identification result. According to the invention, the problem that the abnormal power consumption data of low-cost acquisition equipment is difficult to identify accurately in the prior art can be solved.

The application discloses a cloud API service quality prediction method based on three-dimensional tensor high-order feature interaction, and belongs to the field of service quality prediction.The prediction method comprises the following steps: step 1, obtaining a cloud API service quality dataset; step 2, designing an abnormal point detection module to detect and screen out outliers of the cloud API service quality dataset; step 3, designing a spatial position information optimization module to fully extract information generated by latitude and longitude coordinates; step 4, constructing a self-adaptive three-dimensional tensor network model, inputting processed user features and cloud API features into the three-dimensional tensor network model to obtain new feature information generated after high-order feature interaction; and step 5, inputting the new feature information into a deep neural network layer for sufficient learning, and finally obtaining a cloud API service quality prediction value through a full connection layer.The method designed in the application has more accurate prediction effect and higher robustness in various data density scenes.

The invention discloses a joint compression method for building point cloud data based on geometric features, and the method comprises the steps: firstly carrying out the sampling and statistical filtering preprocessing of the building point cloud data under an octree, so as to remove noise and reduce the data density; secondly, accurately identifying and retaining key contour and structural features of the building through an edge point extraction algorithm based on curvature calculation, and then performing redundancy removal on non-edge points by adopting an adaptive multi-scale normal differential segmentation technology (DoN) which can adaptively adjust scale radius parameters according to geometric complexity of the surface of the building; and structure key points are effectively distinguished and reserved. And finally, merging the edge points and the points subjected to normal differential segmentation to form compressed point cloud data. Through verification of a precision evaluation index, the method can remarkably improve the quality of reconstructed point cloud while keeping a low compression ratio, and provides powerful support for efficient storage and transmission of building point cloud data.

The application discloses a database range filter design method and device based on a binary decision tree, and belongs to the technical field of database query. The method considers that data sets to be processed in a database are not uniform in most application scenarios, and an initial binary decision tree is constructed offline according to the distribution characteristics of the data sets. Each node of the initial binary decision tree is a range, and a root node represents a range interval of keys in the entire data set. Simulated online query is performed by using the initial binary decision tree, and pruning is performed according to the access frequencies of leaf nodes. Finally, compression coding is performed to obtain a database range filter. In the process of constructing the binary decision tree, the binary tree is grown according to data density and skew degree, so that the problem of high false positive rate in the case of uneven data distribution can be solved. The pruning operation of the binary tree according to the access frequency can reduce the space occupation. Finally, the database range filter can efficiently process data in the case of uneven data.

The invention relates to the technical field of financial transaction data visualization, in particular to a transaction data analysis method and system based on a regular polygon block thermodynamic diagram. By constructing a unified time-price two-dimensional data matrix, adaptive selection and rendering of various regular polygons such as rectangles, hexagons, parallelograms, triangles, linear views and the like are supported, and on this basis, information of untransacted buyers and sellers is superposed. According to the transaction data analysis method and system based on the regular polygon block thermodynamic diagram provided by the invention, multi-shape rendering is supported through the unified matrix, repeated calculation of data for different shapes is avoided, and the calculation complexity is reduced; the optimal shape is automatically selected according to the data density, the visual definition is improved, the problems of grid shape solidification, insufficient transaction granularity distinguishing, low multi-view linkage efficiency, missing of non-transaction data and the like in the prior art are solved, and the adaptability, precision and comprehensiveness of transaction data visualization are remarkably improved.

Embodiments of the present application provide a key point recognition device and method based on wireless radar signals. The method comprises: performing feature extraction on point cloud data obtained within a period of time to obtain first spatial feature data, Doppler velocity feature data, reflection energy feature data, density distribution feature data and time distribution feature data of the reflection point cloud; using a neural network-based fusion feature extraction model to detect the cascaded feature data to obtain fusion feature information; and using a neural network-based key point detection model to detect the fusion feature information to output key point data of the object. Thus, the key points of the object (such as a human body) can be detected according to the radar point cloud without limiting the action category, the required computing resources are less, and the detection accuracy is high; in addition, it is easy to implement, simple to operate, has strong noise resistance, and high privacy protection.

The invention provides a data transmission method, device and system, electronic equipment and a storage medium, and relates to the technical field of data communication, and the method comprises the steps: obtaining to-be-transmitted data, carrying out the packaging of the to-be-transmitted data according to a unified data transmission protocol, and obtaining a data package; wherein the unified data transmission protocol at least comprises transmission quality requirement parameters; dynamically determining a target transmission protocol according to the transmission quality requirement parameter and / or historical transmission data; according to the storage information and the data density of the equipment sending end and the cloud server, adjusting a data transmission parameter of the equipment sending end to obtain a first data transmission parameter; sending the data packet to a cloud server based on the target transmission protocol and the first data transmission parameter; the complexity of data processing is reduced through a unified data transmission protocol, the transmission layer protocol is dynamically determined in combination with the transmission quality requirement parameters and the historical transmission data, and the suitability and the data transmission quality during data transmission are improved.

One or more embodiments of the present application provide a full life cycle quality credible data collection method and related equipment. The method is applied to a distributed control system, comprising: determining an updated data density based on current full life cycle quality data, wherein the data density is used to represent the relative closeness between credible data; determining a credible data deployment node based on the updated data density; constructing a credible data path model based on the structure of the credible data deployment node, and screening target credible data in the credible data path model. Through the technical solution of the present application, the data collection stability can be effectively improved, the data collection delay can be reduced, and the influence of error elimination period on data transmission can be reduced.

The application provides a high-speed motor operation data intelligent analysis method and system, and relates to the technical field of data processing.The method comprises the following steps: performing a preprocessing operation on basic operation information and surrounding working condition factors to form initial integrated information; performing boundary contour calculation on a multidimensional data point set for the initial integrated information, and extracting an outer convex boundary point set of data distribution; determining a coverage range of data distribution based on the outer convex boundary point set, and establishing a minimum convex polygon verification interval containing all data points; performing adaptive hierarchical cutting on the verification interval according to a data density gradient and a working condition change amplitude to form a plurality of cutting sub-regions with different characteristic densities; extracting geometric features, distribution densities and boundary curvatures of each cutting sub-region as characteristic parameters, and mapping and converting the characteristic parameters into stability evaluation parameters and abnormality evaluation parameters of corresponding verification points.The application realizes accurate evaluation, real-time abnormality identification and dynamic adaptation of the running state of a high-speed motor.

The application discloses a bearing fault diagnosis method and system based on a CS-SHAP model, bearing operation signals are collected through acceleration, temperature and acoustic emission multi-sensor, multi-dimensional fault features are extracted from time domain, frequency domain and time-frequency domain after wavelet transform or EMD algorithm denoising to complete preprocessing, feature data is divided into similar clusters by K-means clustering, SHAP values are calculated cluster by cluster and weighted according to cluster data density or diagnosis influence degree to obtain feature importance score, a SVM classifier based on RBF kernel function is trained with screened key features, model parameters are optimized through 5-fold cross validation, new collected data is preprocessed and input into the trained model to obtain diagnosis results, feature SHAP values are calculated and visualized by using the CS-SHAP model, and key factors of faults are analyzed. The application realizes accurate diagnosis of bearing faults and interpretability of diagnosis results, and provides technical support for intelligent operation and maintenance of industrial equipment.