Rail transit operation risk assessment management method and system based on knowledge graph

Abstract

The present application provides a rail transit operation risk assessment management method and system based on a knowledge graph, which relates to the field of risk assessment management technology, including: retrieving and calling rail transit event records and multi-source spatiotemporal data; performing clustering and confidence verification respectively, executing knowledge system reconstruction and self-attention configuration, and generating a risk knowledge graph; supervising the training of risk assessment models; supervising the return of rail transit data, combining risk assessment models, conducting risk event determination and deduction prediction analysis, and determining risk orders; visualizing the terminal display interface, matching and determining risk plans, and performing multi-concurrent event timing configuration to determine risk management strategies; and conducting rail transit risk management. This application can solve the technical problems of insufficient timeliness and accuracy of risk identification in the prior art, achieve the technical goal of improving the timeliness and accuracy of risk identification in the rail transit system, and achieve the technical effect of ensuring the safe operation of the rail transit system.

Description

Technical Field

[0001] The present application relates to the technical field of risk assessment management, and in particular to a rail transit operation risk assessment management method and system based on a knowledge graph. Background Art

[0002] As an important part of modern urban transportation, rail transit plays a key role in improving traffic efficiency, reducing urban congestion and protecting the environment. However, with the increasing complexity of rail transit systems and the increase in passenger flow, risk management and safety assurance have become important issues that need to be addressed.

[0003] At present, in the existing technology, the commonly used risk assessment methods mainly include historical data analysis, expert experience judgment and simple mathematical models. These methods can identify and predict some common risk events to a certain extent, but they have obvious limitations when faced with complex, multi-source spatiotemporal data. First, traditional methods cannot make full use of massive real-time monitoring data, resulting in insufficient timeliness and accuracy in risk identification. Secondly, the existing technology lacks the ability to deeply mine and analyze the complex relationships between data, making it difficult to fully reflect the evolution trend of risk events. Finally, when multiple risk events occur concurrently, the existing response strategies often lack systematicity and foresight, making it difficult to achieve the optimal risk management effect.

[0004] In summary, there are technical problems in the existing technology due to low data utilization, insufficient analysis capabilities and imperfect response strategies, which lead to insufficient timeliness and accuracy in risk identification, further affecting the safe operation and management of the rail transit system. Summary of the invention

[0005] The purpose of this application is to provide a rail transit operation risk assessment management method and system based on knowledge graph, so as to solve the technical problems in the prior art that the timeliness and accuracy of risk identification are insufficient due to low data utilization, insufficient analysis capabilities and imperfect response strategies, which further affect the safe operation and management effect of the rail transit system.

[0006] In view of the above problems, this application provides a rail transit operation risk assessment management method and system based on knowledge graph.

[0007] In the first aspect, the present application provides a rail transit operation risk assessment management method based on a knowledge graph, which is implemented by a rail transit operation risk assessment management system based on a knowledge graph, including: based on a data interface, retrieving and calling rail transit event records and multi-source spatiotemporal data, wherein the data interface is connected to multiple data platforms; clustering and confidence checking the rail transit event records and the multi-source spatiotemporal data respectively, performing knowledge system reconstruction and self-attention configuration, and generating a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, and the graph layer is configured by limiting the data granularity, and multi-perspective features are fitted as reconstruction elements; with spatiotemporal convolution and grid evolution as structural functions, combined with the risk knowledge graph, supervised training of a risk assessment model; supervised feedback of rail transit data, combined with the risk assessment model, risk event judgment and deduction prediction analysis, and determination of a risk list, wherein the risk list is marked with a risk level and spatiotemporal location; terminal display interface visualization of the risk list, matching and determining the risk plan and performing multi-concurrent event timing configuration, and determining the risk management strategy; based on the risk management strategy, rail transit risk management is performed.

[0008] In the second aspect, the present application also provides a rail transit operation risk assessment and management system based on a knowledge graph, which is used to execute the rail transit operation risk assessment and management method based on the knowledge graph as described in the first aspect, which includes: a multi-source spatiotemporal data retrieval module, which is used to retrieve and call rail transit event records and multi-source spatiotemporal data based on a data interface, wherein the data interface is connected to multiple data platforms; a risk knowledge graph generation module, which is used to cluster and confidence check the rail transit event records and the multi-source spatiotemporal data respectively, perform knowledge system reconstruction and self-attention configuration, and generate a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, and the graph layer is configured by limiting the data granularity, and the multi-perspective features are realized. Features are fitted as reconstruction elements; a risk assessment model training module, the risk assessment model training module is used to supervise the training of the risk assessment model with spatiotemporal convolution and grid evolution as structural functions, combined with the risk knowledge graph; a risk list determination module, the risk list determination module is used to supervise the return of rail transit data, combined with the risk assessment model, to determine risk events and deduce predictive analysis, and determine the risk list, the risk list is marked with risk level and spatiotemporal location; a risk management strategy determination module, the risk management strategy determination module is used to visualize the risk list on the terminal display interface, match and determine the risk plan and perform multi-concurrent event timing configuration, and determine the risk management strategy; a risk management module, the risk management module is used to perform rail transit risk management based on the risk management strategy.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] Based on the data interface, rail transit event records and multi-source spatiotemporal data are retrieved and called, wherein the data interface is connected to multiple data platforms; the rail transit event records and the multi-source spatiotemporal data are clustered and confidence-checked respectively, and the knowledge system reconstruction and self-attention configuration are performed to generate a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, and the graph layer is configured by limiting the data granularity, and multi-view features are fitted as reconstruction elements; with spatiotemporal convolution and grid evolution as structural functions, the risk knowledge graph is combined to supervise the training of the risk assessment model; the rail transit data is supervised and transmitted back to form a risk assessment model; The risk assessment model is combined to determine risk events and conduct deduction and prediction analysis to determine a risk list, which is marked with risk level and spatiotemporal location; the risk list is visualized on a terminal display interface, the risk plan is matched and determined, and the timing of multiple concurrent events is configured to determine the risk management strategy; based on the risk management strategy, rail transit risk management is carried out, and ultimately the technical goal of efficient use of complex spatiotemporal data in the rail transit system and intelligent risk assessment is achieved, so as to achieve the technical effect of improving the timeliness and accuracy of risk identification and ensuring the safe operation of the rail transit system by optimizing the risk management strategy.

[0011] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0013] Figure 1 This is a flow chart of the rail transit operation risk assessment management method based on knowledge graph in this application;

[0014] Figure 2 This is a structural diagram of the rail transit operation risk assessment management system based on the knowledge graph in this application.

[0015] Description of reference numerals:

[0016] Multi-source spatiotemporal data retrieval module 11, risk knowledge graph generation module 12, risk assessment model training module 13, risk list determination module 14, risk management strategy determination module 15, risk management module 16. DETAILED DESCRIPTION

[0017] This application provides a rail transit operation risk assessment management method and system based on knowledge graphs, which solves the technical problems in the prior art that the timeliness and accuracy of risk identification are insufficient due to low data utilization, insufficient analysis capabilities and imperfect response strategies, further affecting the safe operation and management of the rail transit system. The technical goal of achieving efficient utilization of complex spatiotemporal data in the rail transit system and intelligent risk assessment is achieved, achieving the technical effect of improving the timeliness and accuracy of risk identification and ensuring the safe operation of the rail transit system by optimizing risk management strategies.

[0018] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0019] For example, please refer to the attached Figure 1 , this application provides a rail transit operation risk assessment management method based on knowledge graph, which specifically includes the following steps:

[0020] Step 1: Based on a data interface, retrieve and call rail transit event records and multi-source spatiotemporal data, wherein the data interface is connected to multiple data platforms.

[0021] Specifically, the role of the data interface is to connect and obtain data resources from multiple data platforms, including traffic management systems, surveillance cameras, GPS and other equipment, to ensure the breadth and comprehensiveness of the data. By using the data interface, rail transit event records and multi-source spatiotemporal data can be retrieved and called, ensuring that information obtained from different data sources can be uniformly processed and analyzed, providing rich basic data for subsequent risk assessment.

[0022] Step 2: Cluster and confidence check the rail transit event records and the multi-source spatiotemporal data respectively, perform knowledge system reconstruction and self-attention configuration, and generate a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, configures the graph layer by limiting the data granularity, and fits multi-perspective features as reconstruction elements.

[0023] Specifically, clustering and confidence checking are performed on rail transit event records and multi-source spatiotemporal data, that is, similar events and data are grouped for easy analysis and identification. Cluster analysis helps identify common features in the data, while confidence checking evaluates the reliability and accuracy of the data. By ensuring that the data used has high accuracy and credibility, it lays the foundation for building a reliable knowledge system. When performing knowledge system reconstruction and self-attention configuration, the data is converted into a knowledge graph by analyzing and reorganizing it. The knowledge graph is a structured knowledge representation method that presents the relationship between data in the form of nodes and edges. Self-attention configuration uses artificial intelligence technology to automatically identify and highlight important information and patterns. Knowledge system reconstruction and self-attention configuration not only optimize the organization of data, but also improve the ability to understand and use data, and generate more valuable risk assessment information. The generated risk knowledge graph has a unified data structure, which ensures that all data can be compared and analyzed within the same framework. The graph layer is configured by limiting the data granularity. Limiting the data granularity means processing the data at a specific level of detail to ensure the accuracy and relevance of the graph. Multi-view features are fitted as reconstruction elements. Multi-perspective feature fitting means analyzing and reconstructing data from multiple perspectives to ensure that the risk factors reflected in the map are more comprehensive and realistic.

[0024] Step 3: Using spatiotemporal convolution and grid evolution as structural functions, combined with the risk knowledge graph, supervise the training of the risk assessment model.

[0025] Specifically, a risk assessment model structure combining spatiotemporal convolution and raster evolution is constructed. Spatiotemporal convolution is used to capture the temporal and spatial characteristics of traffic data, while raster evolution is used to predict future trends. During the model training process, the rich information in the risk knowledge graph is combined to provide detailed background and associated data. Through supervised learning methods, model parameters are continuously adjusted to improve the accuracy of risk event recognition and prediction, ensuring that risk assessment can be carried out effectively.

[0026] Step 4: monitor the returned rail transit data, combine it with the risk assessment model, conduct risk event determination and prediction analysis, and determine the risk list, which contains the risk level and time and space location.

[0027] Specifically, after real-time feedback of rail transit data, the trained risk assessment model is input to analyze the data, determine whether there are risk events, and make deductions and predictions, not only identifying current risk events, but also predicting their future development trends. The results are output in the form of a risk list, which contains the risk level of each event and its corresponding time and space location, providing a detailed description of the risk status.

[0028] Step 5: Visualize the risk list on the terminal display interface, match and determine the risk plan, configure the timing of multiple concurrent events, and determine the risk management strategy.

[0029] Specifically, the generated risk list is visualized on the terminal display interface, so that each risk event and its spatial and temporal location and risk level can be intuitively seen. At the same time, the corresponding risk plan is automatically matched, and the timing configuration of multiple concurrent events is performed according to the characteristics and severity of different risk events, ensuring that when facing multiple simultaneous risk events, the response measures can be reasonably arranged and the best risk management strategy can be determined.

[0030] Step 6: Conduct rail transit risk management based on the risk management strategy.

[0031] Specifically, according to the determined risk management strategy, the risk management measures for rail transit are implemented, including taking preventive and response measures to reduce the possibility of risk events. Or when risk events occur, quick actions are taken to mitigate their impact. Through the systematic management process, the safety and reliability of rail transit are significantly improved, ensuring the safety of passengers and staff.

[0032] The knowledge graph-based rail transit operation risk assessment management method is applied to the knowledge graph-based rail transit operation risk assessment management system, which can achieve the technical goals of efficient utilization of complex spatiotemporal data in the rail transit system and intelligent risk assessment, and achieve the technical effect of improving the timeliness and accuracy of risk identification and ensuring the safe operation of the rail transit system by optimizing risk management strategies.

[0033] Furthermore, the present application also includes:

[0034] Based on the rail transit event records, the knowledge system is reconstructed to generate a one-dimensional knowledge graph; based on the multi-source spatiotemporal data and the spatiotemporal data trends, the knowledge system is reconstructed to generate a two-dimensional knowledge graph; based on the risk influencing factors, the spectral association between the one-dimensional knowledge graph and the two-dimensional knowledge graph is established to generate the risk knowledge graph.

[0035] Specifically, by systematically analyzing and reorganizing all rail transit event records in rail transit, the event data is converted into structured knowledge nodes and edges to generate a one-dimensional knowledge graph. The one-dimensional knowledge graph mainly records the characteristics of the event itself, such as event type, time, location, cause and consequence. Through the analysis and connection of the one-dimensional knowledge graph, a graph that can intuitively show the relationship between rail transit events is formed, which can reveal the potential relationship between events and provide basic data for further risk assessment.

[0036] Then, we receive spatiotemporal data from different sources, such as geographic location information, timestamps, and environmental data, and build a two-dimensional knowledge graph by analyzing the temporal and spatial trends of spatiotemporal data from different sources. The two-dimensional knowledge graph not only shows the distribution and change trend of data in time and space, but also helps identify potential risk areas and time periods. The two-dimensional display method makes the spatiotemporal characteristics of data intuitively presented, making it easier to identify and understand complex spatiotemporal patterns and trends.

[0037] Next, by identifying and analyzing various risk influencing factors, risk influencing factors refer to the impact of multi-source spatiotemporal data on traffic risks. For example, the impact of seasonal fog on visibility. Using the relationship and correlation between graphs, the relevant nodes and edges in the one-dimensional and two-dimensional knowledge graphs are associated to form a comprehensive risk knowledge graph. The comprehensive graph can provide a comprehensive perspective, revealing the interactions between complex risk factors, and help decision makers conduct risk assessment and management more accurately.

[0038] By reconstructing rail transit event records and multi-source spatiotemporal data, one-dimensional and two-dimensional knowledge graphs are generated, and through the association of risk influencing factors, a comprehensive risk knowledge graph is finally formed, which helps to improve the risk assessment and management efficiency of rail transit operations.

[0039] Furthermore, the present application also includes:

[0040] Identify the rail transit event records, and determine multiple clusters based on a preset similarity of cluster features; traverse the multiple clusters, set a confidence interval for data credibility screening, and extract the event feature group of the first cluster; determine the main data structure, perform data isomorphism conversion on the event feature group, and determine the first valid data; based on the first valid data, construct a first sub-graph; complete the construction of the Nth sub-graph, integrate and associate the first sub-graph until the Nth sub-graph, and generate the one-dimensional knowledge graph.

[0041] Specifically, all rail transit event records are analyzed using a clustering algorithm and grouped according to the similarity of event characteristics. The preset proximity is a key parameter used to determine the degree of similarity between events and help divide the data to generate multiple clusters. Clusters represent groups of similar events that occur under specific conditions. By determining multiple clusters, the characteristics and distribution of different types of events can be more clearly understood.

[0042] Then, each cluster is checked one by one, and the set confidence interval is applied to filter out reliable data. The confidence interval is used to measure the reliability of the data and ensure that only data that is considered statistically reliable is retained. Then, the event feature group is extracted from a random cluster. The feature group contains the main features describing the events in the cluster, such as time, location, type, and impact.

[0043] Next, a unified master data structure is defined to standardize and integrate data from different sources. Then, the extracted event feature groups are converted into isomorphisms to make them conform to the requirements of the master data structure. Through standardization, all data are converted into a consistent format and structure to form the first valid data, ensuring the uniformity and comparability of the data and laying the foundation for subsequent graph construction.

[0044] Next, the first valid data is used to construct the first sub-graph. The first sub-graph contains all event features extracted and converted from the first cluster, forming a small-scale knowledge graph. Each first sub-graph represents a specific event group and its internal relationships. By constructing in stages, complex data relationships can be gradually accumulated and organized.

[0045] Then, after all sub-graphs are constructed, the first sub-graph is integrated and associated. By analyzing and connecting the nodes and edges in each sub-graph, a complete one-dimensional knowledge graph is finally formed. The one-dimensional knowledge graph integrates the information of all sub-graphs and provides a holistic perspective, showing all events and their interrelationships.

[0046] By identifying, clustering, filtering and transforming data, sub-graphs are gradually constructed and integrated, and finally a comprehensive one-dimensional knowledge graph is formed, which provides systematic data support for rail transit event analysis and risk assessment.

[0047] Furthermore, the present application also includes:

[0048] Determine multi-level data granularity from the bottom up, and configure the graph layer based on the multi-level data granularity; based on the multi-level data granularity, perform hierarchical decomposition on the first valid data to determine the decomposed data system; traverse the decomposed data system, perform multi-perspective fitting and coreference disambiguation processing on homologous data, combine the graph layer to perform data distribution and hierarchical association, and determine the first sub-graph.

[0049] Specifically, the data is refined through a bottom-up approach to determine multiple levels of granularity. Starting from the most basic data, gradually rise to higher levels of data granularity to ensure that there is a fine division and definition at each level. After determining the multi-level data granularity, the atlas layer is configured according to the multi-level data granularity. The configuration of the atlas layer will be based on the data hierarchy of different granularities to ensure that the data of each layer can be accurately mapped to the corresponding atlas layer, thereby ensuring that the organization and management of the data is more systematic and orderly.

[0050] Then, using multi-level data granularity, the first valid data is hierarchically decomposed and divided into different levels. Each level corresponds to a specific granularity, which helps to analyze and process the data in detail, so that the data has a clear structure and relationship at each level. Through hierarchical decomposition, a complete decomposed data system can be formed, which can provide a clear framework and guidance for subsequent data processing and analysis.

[0051] Next, the entire decomposed data system is accessed in turn to perform multi-perspective fitting and coreference disambiguation on homologous data. Multi-perspective fitting refers to analyzing and matching the same data from multiple perspectives, while coreference disambiguation is to solve the problem of unclear or ambiguous references that may exist in the data. Combined with the graph layer, the data is distributed and hierarchically associated to ensure that the data at each layer can be correctly associated and mapped to the graph. Finally, the first sub-graph is determined and constructed, which contains the hierarchically processed and associated data with high accuracy and relevance.

[0052] By determining multi-level data granularity, hierarchically decomposing data, performing multi-perspective fitting and coreference disambiguation processing, and combining the graph layer for data distribution and hierarchical association, we ultimately constructed a detailed and accurate first sub-graph, laying a solid foundation for the construction of the overall knowledge graph.

[0053] Furthermore, the present application also includes:

[0054] Determine self-attention configuration rules, wherein attention is weighted once based on distinctiveness and twice based on relevance; traverse the risk knowledge graph, identify the commonalities and individualities of graph nodes based on classified events, determine a first weight distribution, analyze the graph nodes based on the relevance of classified events, and determine a second weight distribution; fit the first weight distribution and the second weight distribution to determine a self-attention level; and mark the graph nodes of the risk knowledge graph based on the self-attention level.

[0055] Specifically, the self-attention configuration rules are formulated to optimize the node analysis of the risk knowledge graph. First, the nodes are weighted once based on distinctiveness, and initial weights are given to them according to their unique characteristics and attributes. Then, secondary attention weighting is performed based on the relevance between nodes, and their weights are adjusted by analyzing the mutual relationship and influence between nodes. Through two weightings, it can be ensured that the attention weight of each node reflects both its own uniqueness and its correlation with other nodes, thereby more accurately reflecting the actual situation of the graph.

[0056] Then, the entire risk knowledge graph is traversed and visited, and each node is analyzed in detail. First, by identifying the commonalities and individualities in classified events, the importance and particularity of nodes in specific categories can be determined, which helps to determine the first weight distribution of nodes, that is, the weight distribution based on event classification, which reflects the relative importance of nodes in different classified events, thereby providing basic data for subsequent weight adjustments.

[0057] Then, after determining the first weight distribution, the correlation between nodes is further analyzed. The correlation analysis is performed by evaluating the mutual relationship and influence of nodes in the classified events. Through the correlation analysis, the second weight distribution of nodes can be determined, that is, the weight distribution based on the correlation of classified events, which reflects the degree of association and mutual influence between nodes in different events, and provides more detailed data support for the subsequent self-attention level determination.

[0058] Next, the self-attention level is determined by fitting the first weight distribution with the second weight distribution and combining the data of the two weight distributions. The self-attention level is a comprehensive indicator that reflects the relative importance and attention of each node in the entire knowledge graph. Through fitting, the attention level of the node can be more accurately evaluated and marked, thereby providing guidance for further processing of the graph node.

[0059] Then, after determining the self-attention level, the nodes of the risk knowledge graph are marked according to the self-attention level. Node marking is achieved by applying the self-attention level information to the graph nodes, so that the mark of each node reflects its importance and attention in the graph, ensuring that the structure and information of the graph are clearer and more orderly, which is helpful for subsequent risk assessment and decision-making.

[0060] By formulating self-attention configuration rules, analyzing the commonality, individuality and relevance of nodes, fitting weight distribution, and finally determining the self-attention level and marking nodes, a refined and highly accurate risk knowledge graph is constructed, providing a solid data support and analysis basis for risk assessment of rail transit operations.

[0061] Furthermore, the present application also includes:

[0062] Using a convolutional network with a preset spatiotemporal grid, construct spatiotemporal convolution areas and perform multi-area twinning, and integrate in parallel to construct a risk identification unit; using grid evolution based on event trend prediction as a function, construct an evolution prediction unit; post-connect the evolution prediction unit to the risk identification unit to determine the main structure of the model; combining the risk knowledge graph with the main structure of the model, perform data-driven modeling to generate the risk assessment model.

[0063] Specifically, the preset space-time grid refers to the preset event span and spatial region, and the event convolution feature extraction and risk event determination within the space-time are performed. Through the preset space-time grid, the convolution network can refine the data in the spatial and temporal dimensions. The space-time convolution area captures local features by dividing the data into multiple small blocks and applying convolution operations on each small block. Multi-zone twins refer to the simultaneous convolution processing in multiple different areas to ensure that information is extracted from different angles and scales. Among them, parallel processing of multiple data can be performed at the same time, or when the area to be analyzed is large, partitioned parallel processing can be performed to improve processing efficiency. After the results of the parallel processing of the convolution area are integrated, a unit that can efficiently identify risks is constructed, which can effectively analyze and identify risks of complex space-time data.

[0064] Then, by analyzing the temporal and spatial trends of events, a grid evolution prediction unit is constructed. Grid evolution predicts possible future events by observing and analyzing the changing trends of data in time and space. Through prediction, potential risk points and changing trends can be identified in advance, providing a forward-looking reference for risk assessment. The evolution prediction unit combines historical data and trend analysis algorithms to make the prediction results more accurate and reliable.

[0065] Next, the output of the evolution prediction unit is combined with the results of the risk identification unit to form the main structure of the model. The post-connection of the evolution prediction unit to the risk identification unit ensures that the prediction results can serve as a supplement and verification of the identification results, providing a more comprehensive risk assessment. The determination of the main structure of the model enables the entire system to operate in an orderly manner, which can not only identify current risks but also predict future trends, providing double protection for decision-making.

[0066] Next, the risk knowledge graph is combined with the main structure of the model for data-driven modeling. The risk knowledge graph provides rich background information and related data, and the main structure of the model provides a framework for identification and prediction. Through data-driven methods, the two can be organically combined to generate a comprehensive risk assessment model that can not only analyze the current risk status, but also identify potential risks in advance through prediction functions, thereby providing more efficient and comprehensive risk management solutions.

[0067] By constructing spatiotemporal convolution areas and evolutionary prediction units and combining the two with risk knowledge graphs for data-driven modeling, a comprehensive, accurate and reliable risk assessment model is generated, providing advanced technical support for risk management of rail transit operations.

[0068] Furthermore, the present application also includes:

[0069] Transmit traffic monitoring data back, activate the target spatiotemporal convolution area to extract traffic convolution features and determine risk events, and generate a risk decision result; if the risk decision result is no, output the result; if the risk decision result is yes, transfer the risk decision result to the evolution prediction unit to perform risk event deduction and prediction to determine the risk evolution trend; integrate the risk decision result and the risk evolution trend to generate the risk list.

[0070] Specifically, the latest traffic monitoring data is transmitted back, including real-time traffic flow, speed, accident records and other information. Then, the target spatiotemporal convolution area is activated to extract traffic features from the monitoring data. Through the convolution operation, key features in the data can be identified and risk events can be determined. Convolution feature extraction enables the system to efficiently process and analyze large amounts of spatiotemporal data, thereby generating preliminary risk decision results.

[0071] Then, if the risk decision result is judged to be no, that is, no significant risk events are found in the current monitoring data, the result will be output directly, indicating that the current rail transit operation status is normal and there is no risk that requires further attention. This simplifies the processing flow, enables the system to respond quickly, and reduces unnecessary resource consumption.

[0072] Then, if the risk decision result is yes, the result will be passed to the evolution prediction unit. The evolution prediction unit will conduct in-depth deduction and prediction based on the current risk event, combined with historical data and trend analysis, aiming to determine the evolution trend of risk events, predict possible changes and impacts in the future, and thus provide more detailed risk assessment information.

[0073] Next, the preliminary risk decision results are integrated with the risk evolution trend generated by the evolution prediction unit. By combining the current risk determination and future risk prediction, a comprehensive risk data set is generated, which includes the current risk status and its future evolution trend, providing detailed data support and analysis basis for management and decision-making.

[0074] By transmitting back traffic monitoring data, activating the spatiotemporal convolution area for feature extraction and judgment, and combining the evolution prediction unit for deduction and prediction, a comprehensive risk list can be generated to provide detailed risk assessment and decision support.

[0075] In summary, the rail transit operation risk assessment management method based on knowledge graph provided in this application has the following technical effects:

[0076] Based on the data interface, rail transit event records and multi-source spatiotemporal data are retrieved and called, wherein the data interface is connected to multiple data platforms; the rail transit event records and the multi-source spatiotemporal data are clustered and confidence-checked respectively, and the knowledge system reconstruction and self-attention configuration are performed to generate a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, and the graph layer is configured by limiting the data granularity, and multi-view features are fitted as reconstruction elements; with spatiotemporal convolution and grid evolution as structural functions, the risk knowledge graph is combined to supervise the training of the risk assessment model; the rail transit data is supervised and transmitted back to form a risk assessment model; The risk assessment model is combined to determine risk events and conduct deduction and prediction analysis to determine a risk list, which is marked with risk level and spatiotemporal location; the risk list is visualized on a terminal display interface, the risk plan is matched and determined, and the timing of multiple concurrent events is configured to determine the risk management strategy; based on the risk management strategy, rail transit risk management is carried out, and ultimately the technical goal of efficient use of complex spatiotemporal data in the rail transit system and intelligent risk assessment is achieved, so as to achieve the technical effect of improving the timeliness and accuracy of risk identification and ensuring the safe operation of the rail transit system by optimizing the risk management strategy.

[0077] Embodiment 2, based on the rail transit operation risk assessment management method based on knowledge graph in the previous embodiment, the same inventive concept, this application also provides a rail transit operation risk assessment management system based on knowledge graph, please refer to the attached Figure 2 ,include:

[0078] A multi-source spatiotemporal data retrieval module 11, the multi-source spatiotemporal data retrieval module 11 is used to retrieve and call rail transit event records and multi-source spatiotemporal data based on a data interface, wherein the data interface is connected to multiple data platforms; a risk knowledge graph generation module 12, the risk knowledge graph generation module 12 is used to cluster and confidence-check the rail transit event records and the multi-source spatiotemporal data, respectively, perform knowledge system reconstruction and self-attention configuration, and generate a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, configures the graph layer by limiting the data granularity, and fits multi-perspective features as reconstruction elements; a risk assessment model training module 13, the risk assessment model training module 13 is used to Convolution and grid evolution into structural functions, combined with the risk knowledge graph, supervise the training of the risk assessment model; risk list determination module 14, the risk list determination module 14 is used to supervise the return of rail transit data, combined with the risk assessment model, to determine the risk event and deduce the prediction analysis, determine the risk list, the risk list is marked with the risk level and time and space location; risk management strategy determination module 15, the risk management strategy determination module 15 is used to visualize the risk list on the terminal display interface, match and determine the risk plan and perform multiple concurrent event timing configuration, and determine the risk management strategy; risk management module 16, the risk management module 16 is used to perform rail transit risk management based on the risk management strategy.

[0079] Furthermore, the risk knowledge graph generation module 12 in the rail transit operation risk assessment management system based on knowledge graph is also used for:

[0080] Based on the rail transit event records, the knowledge system is reconstructed to generate a one-dimensional knowledge graph; based on the multi-source spatiotemporal data and the spatiotemporal data trends, the knowledge system is reconstructed to generate a two-dimensional knowledge graph; based on the risk influencing factors, the spectral association between the one-dimensional knowledge graph and the two-dimensional knowledge graph is established to generate the risk knowledge graph.

[0081] Furthermore, the risk knowledge graph generation module 12 in the rail transit operation risk assessment management system based on knowledge graph is also used for:

[0082] Identify the rail transit event records, and determine multiple clusters based on a preset similarity of cluster features; traverse the multiple clusters, set a confidence interval for data credibility screening, and extract the event feature group of the first cluster; determine the main data structure, perform data isomorphism conversion on the event feature group, and determine the first valid data; based on the first valid data, construct a first sub-graph; complete the construction of the Nth sub-graph, integrate and associate the first sub-graph until the Nth sub-graph, and generate the one-dimensional knowledge graph.

[0083] Furthermore, the risk knowledge graph generation module 12 in the rail transit operation risk assessment management system based on knowledge graph is also used for:

[0084] Determine multi-level data granularity from the bottom up, and configure the graph layer based on the multi-level data granularity; based on the multi-level data granularity, perform hierarchical decomposition on the first valid data to determine the decomposed data system; traverse the decomposed data system, perform multi-perspective fitting and coreference disambiguation processing on homologous data, combine the graph layer to perform data distribution and hierarchical association, and determine the first sub-graph.

[0085] Furthermore, the risk knowledge graph generation module 12 in the rail transit operation risk assessment management system based on knowledge graph is also used for:

[0086] Determine self-attention configuration rules, wherein attention is weighted once based on distinctiveness and twice based on relevance; traverse the risk knowledge graph, identify the commonalities and individualities of graph nodes based on classified events, determine a first weight distribution, analyze the graph nodes based on the relevance of classified events, and determine a second weight distribution; fit the first weight distribution and the second weight distribution to determine a self-attention level; and mark the graph nodes of the risk knowledge graph based on the self-attention level.

[0087] Furthermore, the risk assessment model training module 13 in the rail transit operation risk assessment management system based on knowledge graph is also used for:

[0088] Using a convolutional network with a preset spatiotemporal grid, construct spatiotemporal convolution areas and perform multi-area twinning, and integrate in parallel to construct a risk identification unit; using grid evolution based on event trend prediction as a function, construct an evolution prediction unit; post-connect the evolution prediction unit to the risk identification unit to determine the main structure of the model; combining the risk knowledge graph with the main structure of the model, perform data-driven modeling to generate the risk assessment model.

[0089] Furthermore, the risk list determination module 14 in the rail transit operation risk assessment management system based on knowledge graph is also used for:

[0090] Transmit traffic monitoring data back, activate the target spatiotemporal convolution area to extract traffic convolution features and determine risk events, and generate a risk decision result; if the risk decision result is no, output the result; if the risk decision result is yes, transfer the risk decision result to the evolution prediction unit to perform risk event deduction and prediction to determine the risk evolution trend; integrate the risk decision result and the risk evolution trend to generate the risk list.

[0091] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The rail transit operation risk assessment management method based on knowledge graph and the specific examples in the aforementioned embodiment 1 are also applicable to the rail transit operation risk assessment management system based on knowledge graph in this embodiment. Through the aforementioned detailed description of the rail transit operation risk assessment management method based on knowledge graph, those skilled in the art can clearly know the rail transit operation risk assessment management system based on knowledge graph in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0092] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0093] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A rail transit operation risk assessment management method based on knowledge graph, characterized in that: include: Retrieve and call rail transit event records and multi-source spatiotemporal data based on a data interface, wherein the data interface is connected to multiple data platforms; Clustering and confidence checking are performed on the rail transit event records and the multi-source spatiotemporal data respectively, and knowledge system reconstruction and self-attention configuration are performed to generate a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, and the graph layer is configured by limiting the data granularity, and multi-view feature fitting is used as a reconstruction element. The knowledge system reconstruction includes constructing a one-dimensional knowledge graph and a two-dimensional knowledge graph, and the self-attention configuration uses artificial intelligence technology to automatically identify and highlight important information and patterns. The multi-view feature fitting represents the analysis and reconstruction of data from multiple angles; Taking spatiotemporal convolution and grid evolution as structural functions, combined with the risk knowledge graph, supervise the training of the risk assessment model, wherein grid evolution is to predict possible future events by observing and analyzing the changing trend of data in time and space; Supervise the returned rail transit data, combine it with the risk assessment model, conduct risk event determination and deduction prediction analysis, and determine the risk list, which contains the risk level and time and space location; The risk list is displayed on the terminal, the risk plan is matched and multiple concurrent event sequence configuration is performed, and the risk management strategy is determined, wherein the multiple concurrent event sequence configuration refers to arranging response measures when facing multiple risk events occurring simultaneously; Based on the risk management strategy, conduct rail transit risk management; The construction of the one-dimensional knowledge graph and the two-dimensional knowledge graph includes: Based on the rail transit event records, a knowledge system is reconstructed to generate a one-dimensional knowledge graph, wherein all rail transit event records in rail transit are systematically analyzed and reorganized, and event data are converted into structured knowledge nodes and edges to generate a one-dimensional knowledge graph; Based on the multi-source spatiotemporal data and the spatiotemporal data trends, the knowledge system is reconstructed to generate a two-dimensional knowledge graph, wherein spatiotemporal data from different sources are received, including geographic location information, timestamps and environmental data, and the two-dimensional knowledge graph is generated by analyzing the time and space trends of the spatiotemporal data from different sources.

2. The rail transit operation risk assessment management method based on knowledge graph according to claim 1, characterized in that: The generating of the risk knowledge graph includes: Based on risk influencing factors, an inter-spectral association between the one-dimensional knowledge graph and the two-dimensional knowledge graph is established to generate the risk knowledge graph.

3. The rail transit operation risk assessment management method based on knowledge graph according to claim 2 is characterized in that: The generating of a one-dimensional knowledge graph comprises: Identifying the rail transit event records to determine a plurality of clusters based on a preset similarity of cluster features; Traversing the plurality of clusters, setting a confidence interval to perform data credibility screening, and extracting an event feature group of the first cluster; Determine a main data structure, perform data isomorphism conversion on the event feature group, and determine first valid data, wherein the main data structure is a standardized data structure constructed by the graph; Based on the first valid data, construct a first sub-graph; Complete the construction of the Nth sub-graph, associate the first sub-graph until the Nth sub-graph, and generate the one-dimensional knowledge graph.

4. The rail transit operation risk assessment management method based on knowledge graph according to claim 3 is characterized in that: Based on the first valid data, constructing a first sub-graph includes: Determine multi-level data granularity from bottom to top, and configure the atlas layer based on the multi-level data granularity; Based on the multi-level data granularity, the first valid data is hierarchically decomposed to determine a decomposed data system; The decomposed data system is traversed to perform multi-perspective fitting and coreference disambiguation processing on homologous data, and the data distribution and hierarchical association are performed in combination with the graph layer to determine the first sub-graph, wherein multi-perspective fitting refers to analyzing and matching the same data from multiple perspectives.

5. The rail transit operation risk assessment management method based on knowledge graph according to claim 1, characterized in that: Perform self-attention configuration, including: Determine the self-attention configuration rule, where the primary attention is weighted by distinctiveness and the secondary attention is weighted by relevance; Traversing the risk knowledge graph, identifying the commonality and individuality of graph nodes based on classified events, determining a first weight distribution, analyzing the graph nodes based on the relevance of classified events, and determining a second weight distribution; Fitting the first weight distribution and the second weight distribution to determine a self-attention level; Based on the self-attention level, the risk knowledge graph is marked with graph nodes.

6. The rail transit operation risk assessment management method based on knowledge graph according to claim 1, characterized in that: The supervised training risk assessment model comprises: A convolution network with a preset spatiotemporal grid is used to construct a spatiotemporal convolution area and perform multi-area twinning, and a risk identification unit is constructed by parallel integration. The preset spatiotemporal grid refers to a preset event span and spatial area, and event convolution feature extraction and risk event determination are performed within the spatiotemporal area. The spatiotemporal convolution area is to capture local features by dividing the data into multiple small blocks and applying convolution operations on each small block. The multi-area twinning refers to performing convolution processing in multiple different areas at the same time to ensure that information is extracted from different angles and scales, and performing parallel processing of multiple data or partitioned parallel processing at the same time. Taking the grid evolution based on event trend prediction as the function, an evolution prediction unit is constructed; The evolution prediction unit is post-connected to the risk identification unit to determine the main structure of the model; The risk knowledge graph is combined with the main structure of the model to perform data-driven modeling and generate the risk assessment model.

7. The rail transit operation risk assessment management method based on knowledge graph according to claim 6, characterized in that: The risk event determination and prediction analysis are carried out to determine the risk list, including: Transmit traffic monitoring data back, activate the target spatiotemporal convolution area to extract traffic convolution features and determine risk events, and generate risk decision results; If the risk decision result is no, output the result; If the risk decision result is yes, the risk decision result is transferred to the evolution prediction unit to perform risk event deduction and prediction to determine the risk evolution trend; The risk decision result and the risk evolution trend are integrated to generate the risk list.

8. The rail transit operation risk assessment and management system based on knowledge graph is characterized by: The steps for implementing the rail transit operation risk assessment management method based on knowledge graph as described in any one of claims 1 to 7 include: A multi-source spatiotemporal data retrieval module, which is used to retrieve and call rail transit event records and multi-source spatiotemporal data based on a data interface, wherein the data interface is connected to multiple data platforms; A risk knowledge graph generation module, which is used to perform clustering and confidence verification on the rail transit event records and the multi-source spatiotemporal data, respectively, perform knowledge system reconstruction and self-attention configuration, and generate a risk knowledge graph, wherein the risk knowledge graph has a unified data structure, configures the graph layer by limiting the data granularity, and fits multi-view features as reconstruction elements; A risk assessment model training module, wherein the risk assessment model training module is used to supervise the training of a risk assessment model by combining the risk knowledge graph with spatiotemporal convolution and grid evolution as structural functions; A risk list determination module, which is used to monitor the returned rail transit data, combine the risk assessment model, perform risk event determination and deduction prediction analysis, and determine the risk list, wherein the risk list is marked with risk level and time and space location; A risk management strategy determination module, which is used to visualize the risk list on a terminal display interface, match and determine the risk plan, perform multiple concurrent event timing configuration, and determine the risk management strategy; A risk management module, wherein the risk management module is used to perform rail transit risk management based on the risk management strategy.

Images