52 results about "Temporal clustering" patented technology

The invention discloses a traffic large model construction and decision-making method and device based on multi-modal two-way map reasoning, and the method comprises the steps: constructing a multi-modal data set of a text, an image and a track, generating fusion features through spatial-temporal clustering and cross-modal Transform coding, carrying out the two-way map reasoning in combination with a traffic knowledge map, and carrying out the decision-making of the traffic large model. The method comprises the following steps: generating an embedded representation through a forward graph neural network, reversely mapping a decision scheme generated by a language model to a graph to verify consistency, outputting knowledge to enhance embedding, fusing multi-modal features and knowledge embedding by adopting an LoRA multi-task joint fine tuning technology, adapting to traffic field tasks, deploying a real-time inference engine, and carrying out real-time inference on the traffic field. And processing the dynamic data flow through an aging perception attention mechanism, and outputting traffic event identification, path planning and scene question and answer results in parallel. Compared with the prior art, the method has the advantages that the problems of insufficient multi-source heterogeneous data fusion, low knowledge utilization efficiency and poor real-time decision consistency can be solved, and the semantic understanding and decision accuracy of the traffic large model is effectively improved.

The invention discloses a multi-stage spatial-temporal clustering method and system based on a fused mahalanobis distance. The method comprises the following steps: acquiring a spatio-temporal data set, and determining a spatio-temporal neighbor relation of samples in the data set; calculating a space communication distance and a time decay distance of the sample; fusing the space communication distance and the time decay distance by using a mahalanobis distance to obtain a relative distance of the sample; selecting a class cluster center from the data set according to the local density of the sample and the relative distance; adopting a multi-stage distribution strategy to distribute non-class-cluster center samples to corresponding class clusters; wherein the multi-stage allocation strategy comprises an inevitable allocation stage based on space-time shared neighbor and a similarity allocation stage based on a weighted similarity matrix. According to the method, the key problems that an existing space-time clustering algorithm is insufficient in space-time attribute coupling processing and sensitive to distribution errors are solved, and the clustering accuracy, robustness and practicability in the fields of intelligent traffic analysis, seismic sequence recognition and the like are remarkably improved.

The invention discloses a remote sensing image change detection method based on semantic interaction and differential representation, and the method comprises the following steps: S1, inputting a to-be-detected dual-temporal remote sensing image into an improved ResNet34 network, and obtaining the multi-scale features of the to-be-detected dual-temporal remote sensing image; s2, inputting the multi-scale features of the to-be-detected dual-temporal remote sensing image into a semantic interaction module to obtain multi-scale dual-temporal semantic features; s3, inputting the multi-scale dual-temporal semantic features into a difference representation module to obtain multi-scale difference features; s4, inputting the multi-scale difference features into a multi-scale feature aggregation module to obtain difference aggregation features; inputting the multi-scale features of the to-be-detected dual-time-phase remote sensing image into a multi-scale feature aggregation module to obtain dual-time-phase aggregation features; s5, inputting the difference aggregation features after up-sampling convolution into a binary change detection module to obtain a binary change graph; and S6, performing mask multiplication on the dual-temporal clustering features after up-sampling convolution and the binary change graph to obtain a dual-temporal semantic change graph. According to the invention, the accuracy of detecting the change position of the dual-temporal remote sensing image and identifying the change element type can be improved.

The invention relates to the technical field of data analysis and visualization, in particular to a sales data sparsity compensation and self-adaptive visualization method and a sales data sparsity compensation and self-adaptive visualization system, a knowledge graph containing a sales node time sequence relation is constructed through spatio-temporal clustering and attribute division, a bidirectional time sequence processing mechanism is adopted, historical evolution and future trend are considered at the same time, and the sales data sparsity compensation and self-adaptive visualization method is established. Historical and future relation strength among nodes is obtained, sales association probability among the nodes is predicted, comprehensive mastering of node relations is realized, prediction accuracy is improved, a missing value regression model of a multi-dimensional relation is constructed on the basis of a graph attention mechanism, attributes and states needing to be filled are identified, and the probability of sales association between the nodes is predicted. Potential relation reasoning is carried out through Monte Carlo tree search, an adaptive node attribute filling result is obtained, a visual layout strategy is designed according to node weights and attribute values, a sales flow change trend is displayed through a thermodynamic model, space-time correlation characteristics of sales data are effectively captured, and a solid foundation is provided for sparsity compensation.

The invention relates to the technical field of tourism information data analysis, and discloses a tourism information data analysis method based on multiple data sources. According to the method, multi-source heterogeneous tourism data such as tourist position tracks, social media comments, scenic spot real-time flow and meteorological environments are obtained. Performing spatial-temporal clustering analysis on the tourist position trajectory data to generate a behavior pattern map; extracting an emotional tendency index from the social media comment data; and associating and matching the scenic spot real-time flow with the meteorological environment data to generate an environment bearing pressure coefficient. And inputting the results into a multi-modal fusion model to obtain an analysis decision vector, constructing a personalized recommendation strategy graph through a collaborative filtering algorithm, and outputting a tourism service optimization scheme. According to the invention, deep analysis of multi-source data can be integrated, scenic spots can be assisted to know behaviors and demands of tourists, operation management is optimized, personalized services are provided, and tourism service quality and scenic spot competitiveness are improved.

The invention particularly relates to a spatial analysis method based on large model agent driving. According to the spatial analysis method based on large model agent driving, syntactic analysis and intention recognition are carried out on a spatial problem described by a natural language through a large model, and spatial-temporal clustering, network analysis and a field simulation algorithm module are dynamically matched; packaging a heterogeneous algorithm into a standardized service component based on a model context protocol (MCP), and automatically constructing an analysis process chain; and integrating a GIS engine and a visual template library, generating a dynamic thematic map according to the processed spatio-temporal data through the visual engine, and superposing multi-dimensional analysis annotations. According to the spatial analysis method based on large model agent driving, the full-chain intelligent capability from data processing to decision support is constructed, the efficiency and precision of spatio-temporal evolution modeling are remarkably improved, meanwhile, the system integration and maintenance cost is reduced, the method is suitable for high-complexity scenes needing real-time spatio-temporal data analysis, and the method is suitable for popularization and application. And the method has remarkable technical innovation and industrial application value.

The invention discloses an expressway traffic flow prediction method based on a spatial-temporal clustering generative adversarial network under an emergency, belongs to the field of expressway traffic flow prediction, and remarkably improves the precision of expressway traffic flow prediction under the emergency. The method comprises the following steps: step 1, acquiring traffic flow data from a traffic detection system, and constructing a traffic network graph structure; 2, integrating geographical location information, traffic function similarity and delay propagation characteristics to construct a comprehensive clustering algorithm, and realizing the comprehensive clustering algorithm; 3, designing a generator network based on a space-time attention encoder and a discriminator network structure fusing a bidirectional long-short-term memory network and graph convolution based on the clustering result and the feature representation in the step 2, and constructing a complete prediction model; 4, performing prediction model training based on the composite loss function; and step 5, applying the trained prediction model to real-time data prediction to realize traffic flow prediction under emergencies.

The invention discloses a power distribution switch terminal communication state analysis method and system, and belongs to the technical field of power distribution network communication. The method comprises the following steps: acquiring dual-source parameters of a terminal and a communication link, and preprocessing to form a time sequence data sequence; an improved local outlier factor algorithm is combined with spatio-temporal clustering to identify abnormal communication events; extracting an analysis sequence containing event context based on the dynamic window; respectively performing communication state evaluation on a terminal side and a link side through a multi-task deep learning model, and outputting state classification and severity score; and when the evaluation results on the two sides conflict, intelligent decision making is performed by adopting a multi-level arbitration mechanism based on an evidence theory and integrating factors such as confidence difference, a historical trend and topological association, so that accurate responsibility determination of a fault source is realized. The method solves the problems that the traditional threshold alarm is high in false alarm rate and the fault source cannot be positioned, and remarkably improves the accuracy and operation and maintenance efficiency of communication state diagnosis.

The invention relates to the technical field of water resource monitoring, in particular to a water resource monitoring data analysis method based on data mining, which improves the data quality through wavelet threshold denoising and directional Kriging interpolation, enhances the prediction precision in combination with improved k-means + + spatial-temporal clustering and a Stacking integrated model, and improves the prediction accuracy. A reverse Darcy flow tracking algorithm is adopted to realize accurate pollution traceability, and an intelligent decision-making system is constructed and formed based on Bayesian hyper-parameter optimization and a knowledge graph, so that the water quality anomaly detection efficiency, the pollution source positioning accuracy and the system adaptive capacity are remarkably improved.

A multimodal metadata extraction system may be provided with a scene detector having a video content input and an output representing scene boundaries. The metadata extractor may be responsive to the content of a scene to extract metadata corresponding to several, plural, or multiple extraction modes. A metadata embedding may be used for each of the modes. An embedding aggregator responsive to the embedding operates to formulate an aggregated embedding for each scene thereby indexing the content of the scene. The scene detector may include a frame analyzer for identifying consecutive frames having similar characteristics. A boundary detector may be provided to identify boundaries of consecutive frames having sufficiently similar characteristics that they likely belong to the same shot. An embedding system may be provided to formulate a composite distance matrix capturing the distance between shot embeddings. A temporal clustering system may be connected to the composite distance matrix. An output of the temporal clustering system identifies the scene boundaries of the content. An embedding database may be connected to the embedding aggregator for storing the aggregated embedding for use as a search index for scenes identified in the content.

The invention provides an intelligent adaptive calibration method and system, and relates to the technical field of calibration, and the method comprises the steps: carrying out the spatial-temporal clustering of a real-time asynchronous event stream based on real-time motion data, carrying out the time alignment with a current RGB image frame, and generating a purification event cluster; accumulating the purification event clusters into an event grey-scale map, inputting the event grey-scale map and the current RGB image frame into a dynamic sensing network, and outputting a static mask; based on the purification event cluster, mapping the event coordinates to a coordinate system of the current RGB image frame, and performing space projection filtering by combining a static mask to form a static feature point set; constructing an objective function based on the static feature point set and performing nonlinear optimization to obtain real-time calibration parameters and calibration residual errors; the motion constraint parameters are updated based on the real-time calibration parameters, meanwhile, the static mask segmentation threshold value of the dynamic sensing network is dynamically adjusted according to the calibration residual error, and the robustness and practicability of the multi-mode visual system in the complex dynamic environment are improved.

The present application provides a method and apparatus for screening close contact objects based on trajectories. A method for screening close contact objects based on trajectories, which is used to screen the close contact objects of a target moving object, includes: extracting trajectory points based on the target trajectory generated by the movement of the target moving object, and each of the extracted trajectory points is associated with time information and space information; performing spatio-temporal clustering on the trajectory points based on the time information and the space information to obtain one or more target trajectory point sets; for each target trajectory point set, determining the spatio-temporal range covered by the target trajectory point set, and screening out the trajectory data that matches the spatio-temporal range from the full amount of trajectory data, and determining the attribution object of the screened trajectory data as the close contact object of the target moving object. By using the above method, close contact objects can be screened out from the full amount of trajectory data, reducing the number of trajectories to be processed subsequently and improving the efficiency.

The invention belongs to the technical field of communication network operation and maintenance and artificial intelligence crossing, and particularly relates to a base station fault root cause positioning method and system based on a knowledge graph, and the method comprises the steps: constructing a communication network knowledge graph carrying causal direction attributes and propagation weights, and carrying out the adaptive spatial-temporal clustering convergence of alarm data, mapping the alarm event to a knowledge graph to activate a causal propagation sub-graph, calculating a root cause probability score through multi-hop message transmission perceived along a causal relationship edge execution direction by a graph attention network, outputting a node which is ranked in the front as a root cause, and continuously updating a graph weight through an expert feedback closed loop, according to the method, root cause positioning is reconstructed from a feature classification normal form to a probability propagation normal form of a causal relationship graph, the accuracy rate reaches 85%, and the positioning duration is shortened to 3 min.

The application discloses a traffic anomaly event-oriented spatio-temporal clustering method, comprising: obtaining traffic anomaly clues and highway basic data in a clustering period; importing the traffic anomaly clues and the highway basic data into the same geographic information system, constructing a unified topological space based on road network topological data, so that all data elements have unified topological relations; in the unified topological space, clustering the traffic anomaly clues based on topological proximity relations, generating initial anomaly clusters; performing effectiveness verification on the initial anomaly clusters, and eliminating invalid clusters; marking the anomaly clusters that pass the effectiveness verification as valid anomaly events, and searching all continuous road segments covered by the valid anomaly events in the unified topological space; and outputting a linear anomaly road segment set composed of the continuous road segments. The application can provide a clear range reference for traffic anomaly event disposal, and help to improve the pertinence and efficiency of the disposal.

This invention discloses a spatiotemporal clustering method for small target detection based on sparse event points. The method involves preprocessing raw data captured by an event camera to obtain preprocessed event data; traversing all event points in the event data, assigning a weight value to each event point, and sorting all event points in descending order according to their weight values; using the average weight of the bottom M% of event points as a clustering threshold, and performing nearest neighbor clustering on the event data based on this threshold to obtain preliminary detection results; and finally, performing point cloud filtering on the preliminary detection results to obtain the final detection results. This invention can directly extract small target features from event data for detection. Under static backgrounds, it can effectively filter out event point clutter triggered by thermal noise and interference event points generated by the movement of large objects, achieving high detection accuracy.

The invention provides a layered space-time clustering method of a time-varying load model, and relates to the technical field of power system load modeling. The method comprises the steps of obtaining a load model data set of load nodes through periodic identification, wherein the load model data set comprises a dynamic load proportion, an active static load parameter, a reactive static load parameter and a dynamic load model parameter; clustering the identification result of each load node by using a time clustering algorithm, and extracting a node typical load model set; and uploading the typical load model of each load node to a control center for spatial clustering, and further extracting a typical load model set of the system, including a typical active static load model set, a reactive static load model set and a dynamic load model set of the system. By introducing a clustering algorithm based on local density and distance deviator, the load modeling precision and efficiency can be improved, the problem caused by response lag of a traditional method to load change is avoided, and the adaptability and reliability of a load model are improved.

This invention discloses a method and system for probable quantification of detrital zircon based on deep learning, relating to the field of biogeochemical cycle system technology. The method includes: step S1, constructing a detrital zircon U-Pb geochronology database; step S2, using a deep temporal clustering-based probable clustering model to delineate direct probable regions within the database constructed in step S1; step S3, using a temporal convolutional network-based probable quantification model to learn the data features of the direct probable regions based on the results of step S2, and performing probable quantification on the sink regions to obtain the quantification results. This invention effectively solves the problem that mainstream temporal clustering and classification models do not pay enough attention to the absolute location information of age distribution data, making them well-suited for detrital zircon U-Pb geochronology data.

The invention provides a leakage hot area spatio-temporal clustering method and system based on an inspection robot, and relates to the technical field of pipeline detection, and the method comprises the steps: obtaining time sequence multi-source collection data of a pipe network through the inspection robot carrying a multi-modal sensor, carrying out the extraction of a time sequence and spatial feature relation, and constructing a spatio-temporal data grid; performing data feature association analysis based on the spatio-temporal data grids, performing multi-granularity aggregation on time and space according to association features, determining spatio-temporal relationship features among the grids, performing leakage hot area identification, performing visual conversion in the spatio-temporal data grids, and generating identification visual information. The technical problem that the pipe network leakage risk identification accuracy and reliability are not high due to the lack of comprehensive judgment capability for the pipe network state in the prior art is solved. The technical effects of comprehensively discriminating and analyzing the state of the pipe network and improving the accuracy and reliability of identifying the leakage hot area of the pipe network by introducing the multi-mode sensing data of the inspection robot are achieved.

The application provides a kind of city creative area identification method and system based on ST-DBSCAN space-time clustering algorithm, comprising: step 1, collecting target enterprise data of target city and preprocessing;Step 2, generate city creative enterprise electronic map based on the enterprise address of the target enterprise;Step 3, on the city creative enterprise electronic map, the target enterprise is clustered and partitioned for the first time based on the DBSCAN algorithm of spatial distance, and the DBSCAN candidate area is preliminarily identified;Step 4, based on the target enterprise data, the target enterprise calculation index is screened, the weight of each calculation index of the target enterprise is determined by entropy method, weighted summation, and the enterprise creativity is calculated;Step 5, in the DBSCAN candidate area, the kernel density is calculated with the enterprise creativity as the weight, and the density partition is formed;different grades of creative areas are divided according to the different density levels.The application provides a kind of city creative area identification and method and system based on DBSCAN-KDE joint clustering algorithm, to realize the accurate identification of city creative area.

The embodiment of the invention provides a behavior portrait generation method, electronic equipment, a storage medium and a program product. The method comprises the following steps: performing spatio-temporal clustering processing based on a plurality of historical track data of a user to obtain key interest points; extracting a plurality of path fragments from the plurality of pieces of historical track data; determining preference path data according to the frequency characteristics of the plurality of path segments; determining abnormal path data according to the multi-dimensional dynamic behavior data of the plurality of path segments; and determining a behavior portrait of the user according to the key interest point, the preference path data and the abnormal path data. The method is used for improving the accuracy of the behavior portrait.

The invention provides a coarse and fine switching method and device in satellite top-crossing tracking and pointing, and the method comprises the steps: carrying out the state estimation and uncertainty quantification based on the instantaneous pointing error of a satellite top-crossing tracking and pointing system at a satellite top-crossing stage, generating and dynamically updating the two-dimensional probability density distribution describing the position of a light spot, and carrying out the dynamic updating of the two-dimensional probability density distribution; fusing with a light intensity distribution diagram influenced by atmospheric turbulence to obtain an effective receiving index, and determining a region of interest for fine aiming detection from probability distribution; performing spatio-temporal clustering analysis on multiple frames of detection data continuously collected by the fine aiming equipment, and calculating the coordinates of a fine aiming tracking point and the identification confidence of the fine aiming tracking point; and predicting the tracking interruption probability based on the atmospheric turbulence prediction data and the probability distribution, performing judgment in combination with the confidence coefficient, and if the judgment is passed, switching the control system to a fine aiming closed-loop tracking mode. According to the invention, rapid, reliable and anti-interference automatic switching from coarse aiming to fine aiming can be realized in a high dynamic scene in which a satellite rapidly passes through the top, so that rapid establishment and stable maintenance of a free space optical communication link are ensured.

The invention provides a centrifugal machine remote monitoring method and system, electronic equipment and a storage medium, and belongs to the technical field of industrial equipment monitoring, and the method comprises the steps: obtaining multi-dimensional operation state data collected in real time by an edge computing node deployed on a centrifugal machine body, constructing an operation state tensor with time and space correlation characteristics based on the multi-dimensional operation state data; performing multi-scale feature extraction on the running state tensor based on a depth time sequence clustering algorithm to obtain a first feature vector set, and determining a working mode of the centrifugal machine based on the first feature vector set and a feature-working mode mapping table; and constructing a monitoring strategy matrix based on a federated learning framework, activating the corresponding sensor group according to the working mode and the monitoring strategy matrix, and monitoring the centrifugal machine by using the sensor group. According to the centrifugal machine remote monitoring method and system, the electronic equipment and the storage medium, the complex operation state of the centrifugal machine can be accurately monitored and analyzed.

The invention discloses a radar target tracking method based on spatial-temporal clustering and a multi-physical model, and relates to the technical field of radar target tracking, and the method comprises the following steps: S1, inputting radar point target data; s2, removing noisy points through self-adaptive space-time clustering; s3, establishing a multi-physical model, and carrying out adaptive threshold weight matching point by point and track by track; s4, performing cyclic matching on the prediction points; s5, integrating the matching scores S1, S2,..., Sn in the matching score queue; and S6, taking a point corresponding to the optimal matching score as a result and outputting the result. According to the method, a multi-physical model matching judgment thought is provided, multiple target motion physical models are constructed, and the target historical motion state and the current trace point adaptability analysis are combined, so that the trace point attribution can be accurately identified even in complex scenes such as target motion mode change and multi-trace intersection, and the target motion mode can be accurately identified. The problems that a traditional method is insufficient in robustness and poor in adaptability are effectively solved, and the continuity and accuracy of target tracking in a complex scene are remarkably improved.

The invention discloses a spatial-temporal clustering method for traffic anomalies. The spatial-temporal clustering method comprises the following steps: acquiring traffic anomaly clues and highway basic data in a clustering period; the method comprises the following steps: importing traffic abnormal clues and highway basic data into the same geographic information system, and constructing a unified topological space based on road network topological data, so that all data elements have a unified topological relation; in the unified topological space, clustering processing is carried out on the traffic anomaly clues based on the topological proximity relation, and an initial anomaly cluster is generated; verifying the validity of the initial abnormal cluster, and eliminating invalid clusters; marking the abnormal clusters passing the validity verification as effective abnormal events, and searching all continuous road sections covered by the effective abnormal events in a unified topological space; and outputting a linear abnormal road section set formed by the continuous road sections. According to the invention, clear range reference can be provided for traffic abnormal event disposal, and the pertinence and efficiency of disposal can be improved.

The embodiment of the invention discloses a static meteorological satellite lightning positioning method and system based on a space grid index, and relates to the technical field of meteorological remote sensing, and the method comprises the steps: building a space index grid of a transient optical signal based on to-be-processed data with a first preset distance as a grid spacing; after determining a clustering core based on the spatial index grid of the transient optical signals, constructing a first sliding window, and clustering the transient optical signals meeting a first preset condition in the first sliding window to generate a transient optical signal group; taking the maximum latitude and longitude and the minimum latitude and longitude of the transient optical signal group as the boundary of a spatial index grid, and taking the second preset distance as the grid spacing to construct the spatial index grid of the transient optical signal group; and constructing a second sliding window by taking each grid point of the spatial index grid of the transient optical signal group as a center, performing spatial-temporal clustering analysis on the transient optical signal group meeting a second preset condition in the second sliding window, obtaining lightning corresponding to the to-be-processed data, and performing lightning positioning.

A method and system for evaluating multi-scale drought response under vegetation change. The method comprises: constructing a distributed time-varying gain hydrological model PML coupled with a remote sensing evapotranspiration module PML, fusing it with a traditional Palmer drought index PDSI, replacing the corresponding water flux of the traditional empirical water tank model with the hydrological simulation result of the DTVGM-PML, and generating an improved drought index PDSI-V considering vegetation dynamics; by setting different vegetation change scenarios, simulating hydrological processes under different vegetation conditions and calculating the corresponding PDSI-V time series; combining run theory and three-dimensional spatio-temporal clustering and other drought identification methods to extract multi-scale drought characteristic quantities; comparing and analyzing the differences in drought characteristics under different scenarios, quantitatively evaluating the influence of vegetation change on drought evolution, and revealing the drought response mechanism under vegetation change. The present application can provide more scientific technical support for regional drought monitoring, early warning and risk management.

The present invention describes a method for monitoring and analysing electroencephalography (EEG) signals to predict the probability of the development of post-operative delirium (POD) in patients undergoing surgery. The method involves receiving a patient's EEG signals, calculating the spectral power in predefined time windows using canonical frequency bands (delta, theta, alpha and beta) and comparing the calculated spectral powers with the centroids of at least two, preferably four, predefined brain state clusters. These clusters include low broadband power (LBP), high broadband power (HBP), alpha prevalent (AP) and delta prevalent (DP). On the basis of the comparison, the brain state is classified and the fractional occupancy of the LBP state, i.e. the proportion of time that the patient spends in this state, is calculated. The method then predicts the probability of developing POD according to the time spent in the LBP state. The invention also includes a system, a computer program product and a computer-readable storage means for implementing the method.

The invention relates to an air conditioner, a control method of the air conditioner and computer equipment, and the air conditioner comprises an indoor unit used for adjusting indoor air; the radar module is used for collecting human body detection data and object detection data of a target space according to a preset time interval; the controller is configured to perform spatial-temporal clustering on the human body detection data and the object detection data for each time interval to obtain a first clustering center and a second clustering center corresponding to each time interval; under the condition that the time sum of the time intervals reaches a preset time sum threshold value, clustering the first clustering center and the second clustering center corresponding to each time interval to generate a space-time weight matrix of the target space; acquiring a current position of a human body and time information corresponding to the current position through a radar module; and according to the current position, the time information and the space-time weight matrix, the operation mode of the air conditioner is adjusted, and therefore the intelligent level of the air conditioner is improved.

This application is applicable to the field of intelligent transportation technology and provides a vehicle collaborative positioning method based on federated transfer learning. The vehicle collaborative positioning method includes: the cloud clusters multiple regional states based on a multi-attribute spatio-temporal clustering algorithm to determine a regional state cluster corresponding to the target domain; the edge side aggregates the model weights corresponding to all regional states in the regional state cluster and uses a self-attention mechanism module to obtain the spatio-temporal characteristics of all regional states in the regional state cluster; the edge side filters the aggregated model weights using the spatio-temporal characteristics to obtain the final model weights and constructs a global model using the final model weights; the vehicle side receives the global model sent by the edge side and, based on federated learning, obtains a local model according to the global model; the local model outputs the GPS regional error; the vehicle side uses the GPS regional error to correct its own position information. The vehicle collaborative positioning method of this application can solve the problem of low vehicle positioning accuracy.

The present invention discloses a method and device for evaluating the power supply safety state of high-speed railways. The method includes: obtaining the original record data of faults and accidents on high-speed railways; extracting the time features and spatial features at the time of faults and accidents from the original record data of faults and accidents; using a spatio-temporal clustering analysis method to divide the levels of faults and accidents according to the time features to obtain the time coefficients at the time of accidents and faults; using a spatio-temporal clustering analysis method to divide the levels of faults and accidents according to the spatial features to obtain the spatial coefficients at the time of accidents and faults; determining the power supply safety index of high-speed railways according to the time and spatial coefficients at the time of accidents and faults; and evaluating the safety state of the high-speed railway power supply system according to the safety index. The present invention can quantitatively evaluate the safety state of the high-speed railway power supply system based on fault and accident data, and has scientific guiding significance for improving the safety performance of the high-speed railway power supply system.