Track emergency early warning method and device, storage medium and computer equipment

Abstract

The invention discloses a rail emergency early warning method and device, a storage medium and computer equipment. The method comprises the following steps: constructing a dynamic knowledge graph based on historical operation and maintenance data of a track; identifying a real-time operation event of the track to obtain event entities, and event semantic relationships and event time attributes among the event entities; according to the event entity, the event semantic relationship and the event time attribute, identifying a causal path of the real-time operation event in the dynamic knowledge graph; calculating the probability of occurrence of a causal path; if the occurrence probability is greater than a preset threshold value, determining the real-time operation event as an emergency; and according to the operation reasoning result in the causal path, early warning information of an emergency is determined, and the early warning information is displayed in the whole-station line building information model of the track. According to the invention, through the dynamic knowledge graph and the building information model, real-time judgment and spatial visual early warning of the track operation risk are realized, and the reliability of track operation and maintenance is improved.

Description

Technical Field

[0001] This application relates to the field of urban rail transit operation and maintenance technology, and in particular to a method, device, storage medium and computer equipment for early warning of rail emergencies. Background Technology

[0002] With the rapid development of urban rail transit, and the increasingly variable and complex operating conditions, rail accidents are frequent and unpredictable. Current technologies for predicting rail emergencies rely solely on experience or single indicators, lacking scientific quantification. Furthermore, maintenance personnel often struggle to intuitively perceive the scope of impact of rail emergencies, resulting in response measures that lack specificity and effectiveness. Summary of the Invention

[0003] In view of this, this application provides a method, device, storage medium and computer equipment for early warning of track emergencies. Through dynamic knowledge graph and building information model, it realizes real-time judgment and spatial visualization early warning of track operation risks, thereby improving the reliability of track operation and maintenance.

[0004] According to one aspect of this application, a method for early warning of orbital emergencies is provided, comprising:

[0005] A dynamic knowledge graph is constructed based on historical operation and maintenance data of the track.

[0006] Identify real-time running events of the track, obtain event entities, event semantic relationships between event entities, and event time attributes, wherein the real-time running events are determined based on real-time monitoring data of the track;

[0007] Based on the event entity, the event semantic relationship, and the event time attribute, identify the causal path of the real-time running event in the dynamic knowledge graph;

[0008] Calculate the probability of the occurrence of the causal path;

[0009] If the probability of occurrence is greater than a preset threshold, the real-time running event will be identified as a sudden event.

[0010] The early warning information of the emergency is determined based on the reasoning results in the causal path, and the early warning information is displayed in the full-station building information model of the track.

[0011] In one embodiment, constructing a dynamic knowledge graph based on historical track maintenance data specifically includes: identifying the historical maintenance data to obtain different types of entities and semantic relationships between them; matching the entities with the semantic relationships to generate triples; instantiating the triples into nodes and edges to generate an initial knowledge graph; adding time attributes to the triples based on time data in the historical maintenance data to generate quadruples; and updating the nodes and edges in the initial knowledge graph based on the quadruples to generate the dynamic knowledge graph.

[0012] In one embodiment, identifying the causal path of the real-time running event in the dynamic knowledge graph based on the event entity, the event semantic relationship, and the event time attribute specifically includes: identifying the relevant nodes of the event entity in the dynamic knowledge graph; based on the relevant nodes, instantiating the event entity, the event semantic relationship, and the event time attribute as nodes and edges in the dynamic knowledge graph to update the dynamic knowledge graph; converting the updated dynamic knowledge graph into a causal graph; and generating the causal path in the updated dynamic knowledge graph by traversing the causal graph starting from the event entity, according to the event time attribute.

[0013] In one embodiment, identifying the relevant nodes of the event entity in the dynamic knowledge graph specifically includes: encoding nodes in the dynamic knowledge graph into node feature vectors; clustering the node feature vectors to generate clusters; encoding the event entity into an event feature vector based on the event semantic relationship and the event temporal attribute; calculating the similarity between the event feature vector and the node feature vector corresponding to the cluster center in the cluster; if there is a similarity within a preset range, determining the relevant node based on the cluster corresponding to the similarity within the preset range; if there is no similarity within the preset range, designating the event entity as the relevant node.

[0014] In one embodiment, converting the updated dynamic knowledge graph into a causal graph specifically includes: determining causal candidate edges in the updated dynamic knowledge graph based on a preset causal type and the time attributes of nodes in the updated dynamic knowledge graph; performing relational reasoning on the causal candidate edges using a graph neural network to filter out causal edges from the causal candidate edges; and generating a causal graph based on the causal edges.

[0015] In one embodiment, calculating the occurrence probability of the causal path specifically includes: determining the conditional probability of the edges in the causal path based on the historical operation and maintenance data; determining the initial occurrence probability of the causal path based on the conditional probability; and correcting the initial occurrence probability using a preset time decay factor based on the time attributes of the nodes in the causal path to obtain the occurrence probability of the causal path.

[0016] In one embodiment, the track emergency warning method further includes: acquiring design data and geographical location data of the track stations and track sections, and constructing a static model of the track; displaying the real-time monitoring data of the stations and track sections on the static model to obtain the full-track building information model, wherein the real-time monitoring data includes sensor data and image data.

[0017] According to another aspect of this application, a track emergency early warning device is provided, comprising:

[0018] The module is used to build a dynamic knowledge graph based on historical operation and maintenance data of the track;

[0019] The identification module is used to identify real-time running events of the track, obtain event entities, event semantic relationships between the event entities, and event time attributes, wherein the real-time running events are determined based on real-time monitoring data of the track; and, based on the event entities, the event semantic relationships, and the event time attributes, identify the causal path of the real-time running events in the dynamic knowledge graph; and calculate the probability of occurrence of the causal path.

[0020] The early warning module is used to determine the real-time running event as a sudden event if the probability of occurrence is greater than a preset threshold; and to determine the early warning information of the sudden event based on the running inference results in the causal path, and to display the early warning information in the full-station building information model of the track.

[0021] According to another aspect of this application, a readable storage medium is provided that stores a program or instructions thereon, which, when executed by a processor, implement the steps of the above-described orbital emergency warning method.

[0022] According to another aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the steps of the above-described orbital emergency warning method.

[0023] By employing the above technical solutions, this application provides a method, device, storage medium, and computer equipment for early warning of track emergencies. Based on historical track operation and maintenance data, a dynamic knowledge graph including a time dimension is constructed. This dynamic knowledge graph is used to infer the future operational outcomes of real-time track events and quantify their probability of occurrence. Emergencies are then selected from real-time operational events based on their probability of occurrence. Simultaneously, a full-station building information model of the track is used to achieve three-dimensional visualization of emergencies, thereby effectively improving the rail transit system's ability to respond to emergencies and ensuring operational safety and efficiency.

[0024] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0026] Figure 1 A flowchart illustrating the early warning method for track emergencies provided in an embodiment of this application is shown;

[0027] Figure 2 A structural block diagram of the track emergency early warning device provided in an embodiment of this application is shown. Detailed Implementation

[0028] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0029] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0030] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “attached” to another element, it can be directly connected or attached to the other element, or there may be intermediate elements. Furthermore, “connected” or “attached” as used herein can include wireless connections or wireless interconnections. The term “and / or” as used herein includes all or any unit and all combinations of one or more associated listed items.

[0031] Exemplary embodiments according to this application will now be described in more detail with reference to the accompanying drawings. However, these exemplary embodiments may be implemented in many different forms and should not be construed as being limited to the embodiments set forth herein. It should be understood that these embodiments are provided so that the disclosure of this application is thorough and complete, and that the concept of these exemplary embodiments is fully conveyed to those skilled in the art.

[0032] This embodiment provides a method for early warning of track emergencies, such as... Figure 1 As shown, the method includes:

[0033] Step 101: Construct a dynamic knowledge graph based on the historical operation and maintenance data of the track.

[0034] It should be noted that urban rail transit is a vehicle transportation system that uses track structures for load-bearing and guidance. Based on the requirements of the city's overall transportation plan, it involves setting up fully or partially enclosed dedicated track lines to transport a considerable volume of passengers in the form of trains or single vehicles. Examples include subways and trams.

[0035] In this embodiment, the historical operation and maintenance data of the track includes historical incidents and historical maintenance records. The historical incident records document past incidents, including the track's operating status at the time of the incident, the type of incident, the time of occurrence, and the resulting consequences. The historical maintenance records correspond to the historical incident records and include the incident code, the faulty equipment that caused the incident, the installation location of the faulty equipment, the emergency measures taken, and the repair time.

[0036] For specific examples, historical incident records could be records of signal delays causing train delays on a certain date, month, and year. Correspondingly, historical maintenance records could be records of repairs performed on a faulty signal at a certain location on a certain date, month, and year, such as replacing the mainboard, so that the signal could be restored to normal.

[0037] Then, the historical operation and maintenance data is preprocessed, including cleaning outliers, missing values, duplicate values, and unstructured data from the data source, in order to update the historical operation and maintenance data.

[0038] Based on historical operation and maintenance data, a time dimension is introduced to construct a dynamic knowledge graph for track operation and maintenance. This allows for the integration and standardization of scattered and unstructured track operation and maintenance data, reflecting the correlation between historical emergencies and improving the efficiency and safety of track operation and maintenance.

[0039] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, a dynamic knowledge graph is constructed based on the historical operation and maintenance data of the track. Specifically, this includes: identifying the historical operation and maintenance data to obtain different types of entities and semantic relationships between entities; matching entities with semantic relationships to generate triples; instantiating triples into nodes and edges to generate an initial knowledge graph; adding time attributes to triples based on time data in the historical operation and maintenance data to generate quadruples; and updating the nodes and edges in the initial knowledge graph based on the quadruples to generate a dynamic knowledge graph.

[0040] In this embodiment, Natural Language Processing (NLP) technology or machine learning algorithms are used to identify historical operation and maintenance data and extract different types of entities and semantic relationships between them. For example, entity types can include equipment type, fault type, and emergency measure type. For instance, the entity of equipment type can be signal equipment, power supply equipment, or vehicle equipment. Fault types can be signal interruption, power supply abnormality, mechanical failure, etc. Emergency measure types can be equipment restart, spare parts replacement, manual intervention, etc. Semantic relationships can include "equipment-occurrence-fault", "fault-leading-sudden result", "fault-triggering-emergency measure", and "equipment-associated-maintenance record", etc. Entities and semantic relationships can then be matched to generate triples. For example, for the operation and maintenance record "Signal equipment A experienced a signal interruption, and emergency plan B was executed", "signal equipment A" is extracted as the "equipment" entity, "signal interruption" as the "fault" entity, and "emergency plan B" as the "emergency measure" entity. Triples are then generated according to semantic relationships: (signal equipment A, experienced a fault, signal interruption) and (signal interruption, corresponding measure, emergency plan B). Then, the entities in the triples are instantiated as nodes, and the semantic relations are instantiated as edges to generate an initial knowledge graph, so as to discover more implicit knowledge through reasoning ability.

[0041] In this embodiment, a time dimension is introduced. The time information corresponding to the triples in historical operation and maintenance data is added to the triples as a time attribute in the form of timestamps. This transforms the static triple representation in the initial knowledge graph into a quadruple structure, thereby forming a time-series graph model, i.e., a dynamic knowledge graph, based on the quadruples. For example, for the operation and maintenance record "Signal device A experienced a signal interruption on [date], and emergency plan B was executed", a time attribute is added to the triple corresponding to this operation and maintenance record based on the time information "[date]".

[0042] It is worth mentioning that dynamic knowledge graphs can divide the graph into multiple subgraphs according to time intervals through time slicing technology, thereby quickly locating nodes of state change. It can not only support historical backtracking, but also infer future trends through time series prediction technology.

[0043] Step 102: Identify the real-time running events of the track and obtain the event entities, the event semantic relationships between event entities, and the event time attributes.

[0044] Among them, real-time operation events are determined based on real-time monitoring data of the track.

[0045] Step 103: Identify the causal path of real-time running events in the dynamic knowledge graph based on event entities, event semantic relationships, and event time attributes.

[0046] In this embodiment, sensors positioned within the track's operating range collect real-time monitoring data of the track, such as temperature sensor data, vibration sensor data, and displacement sensor data. This real-time monitoring data is further analyzed to generate real-time track operation events. For example, if the signal acquisition sensor data for a signal falls below a specified value, a real-time operation event indicating a signal interruption is generated.

[0047] Then, based on a dynamic knowledge graph, natural language processing techniques or machine learning algorithms are used to identify real-time running events, extracting event entities, semantic relationships between event entities, and event temporal attributes. By reasoning about event entities, semantic relationships, and temporal attributes within the dynamic knowledge graph, causal paths are obtained, including the future operational inference results of the real-time running events. This allows for the quantification of the probability of real-time running events leading to abnormal results through causal paths.

[0048] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, based on the event entity, the semantic relationship of the event, and the event time attribute, the causal path of the real-time running event is identified in the dynamic knowledge graph. Specifically, this includes: identifying the relevant nodes of the event entity in the dynamic knowledge graph; based on the relevant nodes, instantiating the event entity, the semantic relationship of the event, and the event time attribute into nodes and edges in the dynamic knowledge graph to update the dynamic knowledge graph; converting the updated dynamic knowledge graph into a causal graph; and, based on the event time attribute, traversing the causal graph from the event entity in the updated dynamic knowledge graph to generate a causal path.

[0049] In this embodiment, real-time running events are treated as new data streams in the dynamic knowledge graph. An incremental update algorithm is used to only partially update the relevant nodes affected in the dynamic knowledge graph, rather than reconstructing them entirely, thereby improving the efficiency of identifying sudden events.

[0050] For example, the relevant nodes of the event entity in the dynamic knowledge graph are identified, the event entity, the event semantic relationship, and the event time attribute are instantiated as nodes and edges in the dynamic knowledge graph and added to the dynamic knowledge graph, and the semantic relationship between the event entity and the relevant nodes is established so that the edges of the event entity are connected to the relevant nodes, thereby realizing the incremental update of the dynamic knowledge graph.

[0051] It should be noted that if there are many related nodes for an event entity, the semantic similarity between the event entity and related nodes can be extracted by relation extraction or calculation to obtain directly related nodes among the related nodes, so as to connect the event entity to the most relevant directly related nodes in the dynamic knowledge graph.

[0052] Furthermore, by distinguishing between relevance and causality in the updated dynamic knowledge graph to avoid spurious associations, a causal graph is obtained. This causal graph is a directed acyclic graph.

[0053] Therefore, in the causal graph, starting from the event entity, the causal graph is traversed using the Breadth-First-Search (BFS) algorithm. Paths with reasonable time sequence are selected based on the event's temporal attributes to generate causal paths. In each causal path, the first node is the event entity, and the last node is the inference result of the real-time event in future execution. For example, a causal path could be "equipment failure - signal interruption - train delay".

[0054] It should be noted that during the traversal of the causal graph, one can start from an event entity and traverse forward (in the result direction) along the directed edges of the causal graph to find possible consequence chains. Alternatively, one can start from the current event node and traverse backward (in the cause direction) along the directed edges of the causal graph to find the root cause chain that led to the event. Furthermore, the time of each node that is a result in the causal path must be later than the time of the node that is a cause.

[0055] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, identifying the relevant nodes of event entities in the dynamic knowledge graph specifically includes: encoding nodes in the dynamic knowledge graph into node feature vectors; clustering the node feature vectors to generate clusters; encoding event entities into event feature vectors based on event semantic relationships and event time attributes; calculating the similarity between the event feature vectors and the node feature vectors corresponding to the cluster centers in the clusters; if there is a similarity within a preset range, determining the relevant nodes based on the clusters corresponding to the similarity within the preset range; if there is no similarity within the preset range, the event entity is regarded as a relevant node.

[0056] In this embodiment, a streaming clustering algorithm is used to associate event entities with existing nodes in a dynamic knowledge graph to facilitate the generation of subsequent causal paths.

[0057] Specifically, the attributes and relationships of each node in the dynamic knowledge graph are encoded into a feature vector. This feature vector can include categorical features, numerical features, temporal features, and spatial features. Similarly, event entities are encoded into event feature vectors in the same format. Then, the feature vectors of the nodes in the dynamic knowledge graph are clustered to generate a predetermined number of clusters. Next, the similarity between the event feature vectors and the feature vectors of the nodes corresponding to the cluster centers is calculated. Here, the similarity can be either the cosine similarity between the event feature vectors and the feature vectors of the nodes corresponding to each cluster center, or the Euclidean distance between the event feature vectors and the feature vectors of the nodes corresponding to each cluster center.

[0058] Furthermore, relevant nodes of the event entity are filtered based on similarity. For example, if there is a cluster with a Euclidean distance less than a first preset threshold or a cosine similarity greater than a second preset threshold (i.e., a preset range), the event entity is added to that cluster, and all nodes in that cluster are considered relevant nodes of the event entity. Otherwise, a new cluster is created, the event entity is placed in the new cluster, and the event entity is considered its own relevant node.

[0059] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, the updated dynamic knowledge graph is converted into a causal graph, specifically including: determining causal candidate edges in the updated dynamic knowledge graph based on the preset causal type and the time attributes of the nodes in the updated dynamic knowledge graph; using a graph neural network to perform relational reasoning on the causal candidate edges to filter out causal edges from the causal candidate edges; and generating a causal graph based on the causal edges.

[0060] In this embodiment, potentially causal relationship types can be pre-defined in the updated dynamic knowledge graph, such as semantic tags like "cause," "lead," and "trigger," to distinguish them from non-causal relationships like "connect" and "belong to." Furthermore, causal relationships must satisfy the constraint that the node acting as the cause occurs earlier than the node acting as the result, thus obtaining causal candidate edges in the dynamic knowledge graph. For example, in the dynamic knowledge graph, if the timestamp of the node "device failure" is earlier than the timestamp of the node "signal interruption," then "device failure - signal interruption" may be a causal candidate edge; otherwise, it is excluded.

[0061] Furthermore, a graph network is used to perform relational reasoning on causal candidate edges, filtering out causal edges from the candidate edges to ensure the reliability of the reasoning. Thus, a causal graph is generated based on the causal edges.

[0062] It should be noted that a cause-effect graph is a directed acyclic graph, where each edge consists of a node that is the cause and a node that is the result.

[0063] Step 104: Calculate the probability of occurrence of the causal path.

[0064] In this embodiment, by calculating the probability of the occurrence of a causal path, the probability of an anomaly occurring in the real-time operation of the orbit in the future is quantified, thereby improving the accuracy of risk prediction.

[0065] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully explain the specific implementation process of this embodiment, the calculation of the occurrence probability of the causal path specifically includes: determining the conditional probability of the edges in the causal path based on historical operation and maintenance data; determining the initial occurrence probability of the causal path based on the conditional probability; and correcting the initial occurrence probability using a preset time decay factor based on the time attributes of the nodes in the causal path to obtain the occurrence probability of the causal path.

[0066] In this embodiment, the probability of a causal path occurring is calculated using the Bayesian conditional probability formula.

[0067] Specifically, the conditional probability of each edge in the causal path is calculated from historical operation and maintenance data.

[0068] For example, the conditional probability table of edges in a causal path is represented as follows:

[0069]

[0070] Wherein, p(E) j |E i ) indicates that the two endpoints are E i With E j The edge, where E i Pointing to E j E i As the node that serves as the cause, E j For nodes that are the results, it is represented as

[0071] Here, n(E) j +E i E in historical operation and maintenance data j With E i The frequency of simultaneous occurrence. n(E) i E occurred in historical operation and maintenance data. i The frequency.

[0072] Furthermore, causal pathways The initial probability of occurrence can be expressed as:

[0073] p path =p(E i )p(E j |E i )p(E e |E j ),

[0074] Where, p path causal path The initial probability of occurrence, p(E) i ) refers to E in historical operation and maintenance data i The probability of occurrence is based on historical operation and maintenance data. i The ratio of the frequency to historical operation and maintenance data is used to determine p(E). e |E j ) represents the edge of a causal path. The conditional probability.

[0075] Furthermore, the probability of a causal path occurring is expressed as:

[0076]

[0077] Where, p path ′ represents the causal path The probability of occurrence, λ is the preset time decay factor, Δt ei For node E e With Ei The difference in timestamps.

[0078] It should be noted that the preset time decay factor is determined based on the accuracy required for calculating the probability. For example, the preset time decay factor can be set to 0.1 / minute.

[0079] In this embodiment, in calculating the probability of occurrence of a causal path, a preset time decay factor and the time attribute of the node are used to reduce the influence of earlier events and improve the accuracy of the calculation.

[0080] Step 105: If the probability of occurrence is greater than the preset threshold, the real-time running event is identified as a sudden event.

[0081] It is understandable that the causal path and the probability of its occurrence are based on historical emergencies, historical maintenance records, etc. Therefore, the reasoning result in the causal path is the risk result that the real-time running event may cause, and the probability of the causal path is the probability of the risk result caused by the real-time running event.

[0082] In this embodiment, if the probability of a causal path occurring is greater than a preset threshold, it indicates that the real-time running event poses a risk, and the real-time running event is identified as a sudden event.

[0083] It should be noted that urban rail emergencies include safety accidents, natural disasters, and man-made emergencies. Examples include equipment failures, earthquakes, and floods.

[0084] In one embodiment, the track emergency early warning method further includes: acquiring design data and geographical location data of track stations and track sections, and constructing a static model of the track; displaying real-time monitoring data of stations and track sections on the static model to obtain a full-track building information model.

[0085] The real-time monitoring data includes sensor data and image data.

[0086] In this embodiment, the geographical location data of the track stations and sections are obtained by oblique photography or GIS (Geographic Information System). The design data and geographical location data of the track stations and sections are converted into parametric models to perform fine modeling of the entire track line. Finally, lightweight processing is performed to form a static model of the entity for later display of dynamic sensing data.

[0087] Next, various sensor data from track stations and sections, such as temperature sensors, vibration sensors, and displacement sensors, are used to collect real-time physical parameters of the track facilities. Image and video information of the track operation area is acquired through a video surveillance system to monitor the real-time status of personnel and equipment. The collected data is then transmitted to a static model via a data interface to form a full-track BIM (Building Information Modeling) model, enabling dynamic perception of the track's operational status and dynamic display on the full-track BIM model. Simultaneously, visualization rendering is performed, and BIM equipment component data tags are added, allowing users to access real-time operational data when clicking on equipment in the full-track BIM model.

[0088] Step 106: Determine the early warning information of the emergency based on the reasoning results in the causal path, and display the early warning information in the track's full-station building information model.

[0089] In this embodiment, based on the operational reasoning results of emergencies in the causal path, a tiered early warning mode is set up. Risky equipment is highlighted in the full-station BIM model; high-risk equipment automatically triggers emergency procedures, while low-risk equipment is marked in yellow in the full-station BIM model, and maintenance suggestions and contingency plans are pushed to management for decision-making. Furthermore, the risk path corresponding to the causal path is highlighted in the full-station BIM model, enabling decision-makers to quickly locate problems and more intuitively perceive the impact and scope of emergencies through a visualized BIM model.

[0090] It should be noted that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0091] Furthermore, such as Figure 2 As shown, as a specific implementation of the above-mentioned track emergency warning method, this application embodiment provides a track emergency warning device 200, which includes: a construction module 201, an identification module 202, and a warning module 203.

[0092] Among them, the construction module 201 is used to build a dynamic knowledge graph based on the historical operation and maintenance data of the track;

[0093] The identification module 202 is used to identify real-time running events of the track, obtain event entities, event semantic relationships between event entities and event time attributes, and real-time running events are determined based on real-time monitoring data of the track; and, based on event entities, event semantic relationships and event time attributes, identify causal paths of real-time running events in a dynamic knowledge graph; and calculate the probability of occurrence of causal paths.

[0094] The early warning module 203 is used to determine the real-time running event as a sudden event if the probability of occurrence is greater than a preset threshold; and to determine the early warning information of the sudden event based on the running inference results in the causal path, and to display the early warning information in the track's full-station building information model.

[0095] In one embodiment, the construction module 201 is specifically used to identify historical operation and maintenance data to obtain different types of entities and semantic relationships between entities; match entities with semantic relationships to generate triples; instantiate triples into nodes and edges to generate an initial knowledge graph; add time attributes to triples based on time data in historical operation and maintenance data to generate quadruples; and update nodes and edges in the initial knowledge graph based on quadruples to generate a dynamic knowledge graph.

[0096] In one embodiment, the identification module 202 is specifically used to identify the relevant nodes of the event entity in the dynamic knowledge graph; based on the relevant nodes, the event entity, the event semantic relationship, and the event time attribute are instantiated as nodes and edges in the dynamic knowledge graph to update the dynamic knowledge graph; the updated dynamic knowledge graph is converted into a causal graph; according to the event time attribute, in the updated dynamic knowledge graph, starting from the event entity, the causal graph is traversed to generate a causal path.

[0097] In one embodiment, the identification module 202 is specifically used to encode nodes in the dynamic knowledge graph into node feature vectors; cluster the node feature vectors to generate clusters; encode event entities into event feature vectors according to the event semantic relationship and event time attribute; calculate the similarity between the event feature vector and the node feature vector corresponding to the cluster center in the cluster; if there is a similarity within a preset range, determine the relevant node according to the cluster corresponding to the similarity within the preset range; if there is no similarity within the preset range, the event entity is regarded as the relevant node.

[0098] In one embodiment, the identification module 202 is specifically used to determine causal candidate edges in the updated dynamic knowledge graph based on the preset causal type and the time attribute of the nodes in the updated dynamic knowledge graph; to use a graph neural network to perform relational reasoning on the causal candidate edges in order to filter out causal edges from the causal candidate edges; and to generate a causal graph based on the causal edges.

[0099] In one embodiment, the identification module 202 is specifically used to determine the conditional probability of the edges in the causal path based on historical operation and maintenance data; determine the initial occurrence probability of the causal path based on the conditional probability; and correct the initial occurrence probability using a preset time decay factor based on the time attributes of the nodes in the causal path to obtain the occurrence probability of the causal path.

[0100] In one embodiment, the construction module 201 is specifically used to acquire the design data and geographical location data of the stations and track sections of the track, and construct a static model of the track; display the real-time monitoring data of the stations and track sections on the static model to obtain a full-line building information model, wherein the real-time monitoring data includes sensor data and image data.

[0101] Specific limitations regarding the track emergency warning device can be found in the limitations of the track emergency warning method described above, and will not be repeated here. Each module in the aforementioned track emergency warning device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0102] Based on the above, Figure 1 Correspondingly, this application embodiment also provides a readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described track emergency warning method shown in the figure.

[0103] Based on this understanding, the technical solution of this application can be embodied in the form of a software product. This software product can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or portable hard drive), and includes several instructions to cause a computer device (such as a personal computer, server, or network device) to execute the methods described in the various implementation scenarios of this application.

[0104] Based on the above, Figure 1 The method shown, and Figure 2 To achieve the above objectives, the present application also provides a computer device, specifically a personal computer, server, network device, etc., as shown in the virtual device embodiment. This computer device includes a storage medium and a processor; the storage medium stores a computer program; the processor executes the computer program to achieve the above-described objectives. Figure 1 The method for early warning of track emergencies is shown.

[0105] In one embodiment, the computer device may further include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, an input unit such as a keyboard, etc., and optionally, a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a Wi-Fi interface), etc.

[0106] Those skilled in the art will understand that the computer device structure provided in this embodiment does not constitute a limitation on the computer device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0107] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and stores the hardware and software resources of a computer device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software within the physical device.

[0108] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platform, or the embodiments of this application can be implemented by hardware.

[0109] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application. Those skilled in the art will understand that the modules in the apparatus of the embodiment can be distributed within the apparatus of the embodiment as described, or can be modified to be located in one or more apparatuses different from this embodiment. The modules of the above-described embodiment can be combined into one module, or further divided into multiple sub-modules.

[0110] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of any particular implementation scenario. The above disclosures are merely a few specific implementation scenarios of this application; however, this application is not limited thereto, and any variations conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A method for early warning of track emergencies, characterized in that, The method includes: A dynamic knowledge graph is constructed based on historical operation and maintenance data of the track. Identify real-time running events of the track, obtain event entities, event semantic relationships between event entities, and event time attributes, wherein the real-time running events are determined based on real-time monitoring data of the track; Based on the event entity, the event semantic relationship, and the event time attribute, identify the causal path of the real-time running event in the dynamic knowledge graph; Calculate the probability of occurrence of the causal path; If the probability of occurrence is greater than a preset threshold, the real-time running event will be identified as a sudden event. The early warning information of the emergency is determined based on the reasoning results in the causal path, and the early warning information is displayed in the full-station building information model of the track.

2. The method for early warning of track emergencies according to claim 1, characterized in that, The dynamic knowledge graph constructed from the historical operation and maintenance data of the track specifically includes: Identify the historical operation and maintenance data to obtain the semantic relationships between different types of entities; The entities are matched with the semantic relationships to generate triples; The triples are instantiated as nodes and edges to generate an initial knowledge graph; Based on the time data in the historical operation and maintenance data, add a time attribute to the triplet to generate a quad. Based on the quadruple, update the nodes and edges in the initial knowledge graph to generate the dynamic knowledge graph.

3. The method for early warning of track emergencies according to claim 1, characterized in that, The step of identifying the causal path of the real-time running event in the dynamic knowledge graph based on the event entity, the event semantic relationship, and the event time attribute specifically includes: Identify the relevant nodes of the event entity in the dynamic knowledge graph; Based on the relevant nodes, the event entity, the event semantic relationship, and the event time attribute are instantiated as nodes and edges in the dynamic knowledge graph to update the dynamic knowledge graph; The updated dynamic knowledge graph is then converted into a causal graph. Based on the event time attribute, in the updated dynamic knowledge graph, starting from the event entity, the causal graph is traversed to generate the causal path.

4. The method for early warning of track emergencies according to claim 3, characterized in that, The identification of the relevant nodes of the event entity in the dynamic knowledge graph specifically includes: Encode the nodes in the dynamic knowledge graph into node feature vectors; Cluster the node feature vectors to generate clusters; Based on the semantic relationship of the event and the time attribute of the event, the event entity is encoded into an event feature vector; Calculate the similarity between the event feature vector and the node feature vector corresponding to the cluster center in the cluster; If a similarity score that falls within a preset range exists, the relevant node is determined based on the cluster corresponding to the similarity score that falls within the preset range. If no similarity exists within the preset range, the event entity is considered the relevant node.

5. The method for early warning of track emergencies according to claim 3, characterized in that, The step of converting the updated dynamic knowledge graph into a causal graph specifically includes: Based on the preset causal type and the time attribute of the nodes in the updated dynamic knowledge graph, determine the causal candidate edges in the updated dynamic knowledge graph; A graph neural network is used to perform relational reasoning on the causal candidate edges in order to filter out causal edges from the causal candidate edges; A causal graph is generated based on the causal edges.

6. The method for early warning of track emergencies according to claim 1, characterized in that, The calculation of the probability of occurrence of the causal path specifically includes: Based on the historical operation and maintenance data, determine the conditional probability of the edges in the causal path; Based on the conditional probability, determine the initial probability of occurrence of the causal path; Based on the time attributes of the nodes in the causal path, the initial occurrence probability is corrected using a preset time decay factor to obtain the occurrence probability of the causal path.

7. The method for early warning of track emergencies according to claim 1, characterized in that, The method further includes: Obtain the design data and geographical location data of the stations and track sections of the track, and construct a static model of the track; The real-time monitoring data of the station and the track section are displayed on the static model to obtain the full-station line building information model. The real-time monitoring data includes sensor data and image data.

8. A track emergency early warning device, characterized in that, The device includes: The module is used to build a dynamic knowledge graph based on historical operation and maintenance data of the track; The identification module is used to identify real-time running events of the track, obtain event entities, event semantic relationships between event entities, and event time attributes, wherein the real-time running events are determined based on real-time monitoring data of the track; and, Based on the event entities, the event semantic relationships, and the event temporal attributes, the causal path of the real-time running event is identified in the dynamic knowledge graph; and, Calculate the probability of occurrence of the causal path; The early warning module is used to determine the real-time running event as a sudden event if the probability of occurrence is greater than a preset threshold; and, The early warning information of the emergency is determined based on the reasoning results in the causal path, and the early warning information is displayed in the full-station building information model of the track.

9. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the steps of the orbital emergency early warning method as described in any one of claims 1 to 7.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the program, it implements the orbital emergency early warning method as described in any one of claims 1 to 7.

Images