Data-driven rail transit risk prevention and control method and system

Abstract

The application provides a kind of data-driven rail transit risk prevention and control method and system, belongs to rail transit operation management technical field, based on "application scene-knowledge structure-concept, attribute and relationship mode" mapping rule, constructs rail transit risk prevention and control knowledge graph mode layer;Combined with the data-driven rail transit risk prevention and control knowledge extraction method, constructs rail transit risk prevention and control knowledge graph data layer;Adopt the graph database standardization storage triple structure and instance label, constructs rail transit risk prevention and control knowledge graph;Based on rail transit risk prevention and control knowledge graph semantic link relationship and graph database, four rail transit risk prevention and control function implementation rules of risk point risk portrait, implicit relationship mining, accident cause reasoning and risk prevention and control auxiliary decision are proposed.The application starts from rail transit risk prevention and control application scene, provides decision basis for realizing multi-dimensional, multi-level precision risk prevention and control and fine risk management.

Description

Technical Field

[0001] This invention relates to the field of rail transit operation management technology, specifically to a data-driven method for constructing and applying a knowledge graph for rail transit risk prevention and control. Background Technology

[0002] In recent years, with the deepening of "smart transportation" and "intelligent transportation," information systems for ensuring the safety of rail transit systems have been successively built and put into operation. However, these systems tend to focus more on platform applications of business processes and paper documents, lacking in-depth exploration and application of risk knowledge in rail transit systems. This results in key scientific problems in rail transit risk analysis and safety management, such as unclear perception objects, inaccurate risk evolution paths, and unscientific prevention and control measures and emergency decisions, making it difficult to support the effective implementation of proactive prevention and control models. At the same time, the field of rail transit risk prevention and control is a vast knowledge system, including risk point and risk status representation, risk evolution paths, prevention and control decisions, and the complex relationships between them, which is beyond the capabilities of the traditional relational databases commonly used in these systems. Summary of the Invention

[0003] The purpose of this invention is to provide a data-driven method and system for preventing and controlling risks in rail transit, so as to solve at least one of the technical problems existing in the background art.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] On the one hand, this invention provides a data-driven method for risk prevention and control in rail transit, comprising:

[0006] Based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern", a knowledge graph pattern layer for rail transit risk prevention and control is constructed.

[0007] Based on the characteristics of unstructured data, and combined with a data-driven knowledge extraction method for rail transit risk prevention and control, a knowledge graph data layer for rail transit risk prevention and control is constructed.

[0008] A knowledge graph for rail transit risk prevention and control is constructed by using a graph database to standardize the storage of triple structures and instance tags.

[0009] Based on the semantic linking relationships of the knowledge graph for rail transit risk prevention and control and the graph database, four rules for realizing rail transit risk prevention and control functions are proposed: risk profiling of risk points, mining of implicit relationships, reasoning about the causes of accidents, and auxiliary decision-making for risk prevention and control.

[0010] Preferably, a knowledge graph model layer for rail transit risk prevention and control is constructed, including:

[0011] Construct a knowledge graph application scenario U for rail transit risk prevention and control, and break down the application scenario into multiple application sub-scenarios based on its application field and scope of use;

[0012] Establish the mapping relationship and mapping function between the application scenarios and knowledge structure of the knowledge graph for rail transit risk prevention and control, and construct the knowledge structure of the knowledge graph for rail transit risk prevention and control.

[0013] The knowledge structure of the knowledge graph for rail transit risk prevention and control is analyzed layer by layer, defining concepts, concept attributes, and relationship patterns between concepts and between concepts and attributes, and constructing a relationship pattern deconstruction diagram;

[0014] ER diagrams are used to graphically represent concepts, concept attributes, and the relationship patterns between concepts and between concepts and attributes, thus constructing a knowledge graph ontology model for rail transit risk prevention and control.

[0015] Evaluation indicators for the knowledge graph ontology model of rail transit risk prevention and control are set, and the ontology model of rail transit risk prevention and control is evaluated, verified and optimized, and a pattern layer of rail transit risk prevention and control knowledge graph is constructed.

[0016] Preferably, the application scenario U of the knowledge graph for risk prevention and control in rail transit is broken down into three sub-scenarios: 1) exploring and characterizing all possible states of risk points from multiple dimensions to assist operation safety management personnel in formulating risk point control plans; 2) fully grasping the evolution patterns and development models of rail transit operation accidents to assist operation safety management personnel in cutting off risk evolution paths; and 3) clarifying on-site operation procedures, implementing job responsibilities, and forming a rail transit safety operation guidance plan.

[0017] Preferably, a mapping relationship is established between the application scenarios and knowledge structures of the knowledge graph for rail transit risk prevention and control. and mapping function Constructing a knowledge graph knowledge structure O for rail transit risk prevention and control, namely,

[0018]

[0019]

[0020] U=|u1,u2,u3,…,un|(n=1,2,3…)

[0021] O=|o1,o2,o3,…,on|(n=1,2,3…)

[0022]

[0023]

[0024] In the formula, un and on represent the nth constituent element in U and O, respectively. This indicates a mapping from un to on;

[0025] By analyzing the application scenarios of the knowledge graph for rail transit risk prevention and control, the research objects in the application scenarios are used as the basis for mapping the relationship between the knowledge structure and the research objects.

[0026]

[0027] In the formula, a ox This represents the mapping relationship between the application scenario (ox) and the knowledge structure.

[0028] We obtain the mapping relationships under application scenarios u1, u2, and u3. And the mapping results o1, o2, o3, that is,

[0029]

[0030]

[0031]

[0032] Preferably, the knowledge structure is analyzed layer by layer, defining concepts, concept attributes, and relationship patterns between concepts and between concepts and attributes, and constructing a relation pattern deconstruction diagram; wherein, concepts include risk points, risk events, measures, prevention and control positions, and prevention and control standards, and concept attributes are risk identification attributes; including:

[0033] Analyze the knowledge structure of the risk point layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts;

[0034] Analyze the knowledge structure of the risk chain layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts;

[0035] The analysis measures and responsibility layer knowledge structure define the concepts, attributes, and relational schemas within this structure, and construct a deconstruction diagram of the relational schemas between concepts.

[0036] Preferably, the knowledge graph ontology structure evaluation model for rail transit risk prevention and control consists of two parts: a correlation coefficient matrix G and an evaluation index weight vector W, i.e.

[0037]

[0038] B=[b1,b2...bi](i=1,2,...,m)

[0039] W=[w1,w2...wj](j=1,2,...,n)

[0040]

[0041] In the formula, B represents the evaluation result matrix of m ontology model design schemes, bi represents the evaluation result of the i-th ontology model; W represents the weight vector of each evaluation index, wj represents the weight of the j-th index; G represents the correlation coefficient matrix, and gij represents the correlation coefficient between the j-th index and the j-th optimal index in design scheme i.

[0042] Preferably, the knowledge graph for risk prevention and control in rail transit consists of a three-layer knowledge structure, with the risk evolution link as the intermediate layer and risk points and prevention and control as the upper and lower related layers.

[0043] The middle layer uses "risk events" and "accident scenarios" as nodes and coupling relationships as edges to express how risk events develop and evolve to ultimately lead to accident scenarios in a "chain" or "network" transmission process, and uses risk identification attributes as the node attributes of "risk events".

[0044] The upper layer uses "risk points" as nodes and is connected to the middle layer through a predicate relationship to describe the state manifestation of risk points evolving and escalating into risk events. The upper and lower levels within the layer are used as edges to express the hierarchical relationship between components within the system.

[0045] The lower-level related layer takes "measures" as the central node and "standard sources and prevention and control positions" as related nodes, and connects with the middle layer through predicate relationships to explain the four stages of "prevention-treatment-control-rescue", thereby implementing the standards and norms for risk prevention and control and proactive safety assurance of the rail transit operation system.

[0046] Secondly, the present invention provides a data-driven rail transit risk prevention and control system, comprising:

[0047] The first construction module is used to build the knowledge graph pattern layer of rail transit risk prevention and control based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern".

[0048] The second construction module is used to construct a knowledge graph data layer for rail transit risk prevention and control based on the characteristics of unstructured data and combined with a data-driven knowledge extraction method for rail transit risk prevention and control.

[0049] The third construction module is used to construct a knowledge graph for rail transit risk prevention and control by using graph database to standardize the storage of triplet structures and instance tags.

[0050] The module determines the implementation rules for four rail transit risk prevention and control functions: risk profiling, implicit relationship mining, accident cause reasoning, and risk prevention and control auxiliary decision-making, based on the semantic link relationship of the rail transit risk prevention and control knowledge graph and graph database.

[0051] Thirdly, the present invention provides a non-transitory computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the data-driven rail transit risk prevention and control method described above.

[0052] Fourthly, the present invention provides a computer program product, including a computer program that, when run on one or more processors, is used to implement the data-driven rail transit risk prevention and control method described above.

[0053] Fifthly, the present invention provides an electronic device, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the data-driven rail transit risk prevention and control method as described above.

[0054] The beneficial effects of this invention are as follows: Starting from the needs of risk prevention and proactive safety assurance in rail transit systems, this invention proposes a data-driven method for constructing and applying a knowledge graph for rail transit risk prevention and control, providing intelligent means for the effective implementation of proactive safety assurance strategies in rail transit. Simultaneously, based on the application scenario of rail transit risk prevention and control—"managing risk points, cutting off risk chains, and implementing prevention and control responsibilities"—this invention proposes four rules for realizing rail transit risk prevention and control functions: risk profiling of risk points, mining of implicit relationships, accident causal reasoning, and risk prevention and control auxiliary decision-making, based on the semantic linking relationships of the rail transit risk prevention and control knowledge graph and the Cypher statements provided by the Neo4j graph database. This provides a decision-making basis for achieving multi-dimensional and multi-level precise risk prevention and control and refined risk management.

[0055] The advantages of additional aspects of the invention will be set forth more clearly in the following description or will be learned by practice of the invention. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating the data-driven knowledge graph construction and application method for rail transit risk prevention and control, as described in an embodiment of the present invention.

[0058] Figure 2This is a flowchart illustrating the construction process of the knowledge graph pattern layer for rail transit risk prevention and control, as described in an embodiment of the present invention.

[0059] Figure 3 This is an example diagram illustrating the coupling relationship of the data-driven knowledge graph construction and application method for rail transit risk prevention and control described in this invention.

[0060] Figure 4 This is the knowledge graph ontology model for rail transit risk prevention and control, as described in the data-driven method for constructing and applying rail transit risk prevention and control knowledge graphs in an embodiment of the present invention.

[0061] Figure 5 This is a flowchart illustrating the knowledge extraction process for rail transit risk prevention and control, as described in the data-driven knowledge graph construction and application method for rail transit risk prevention and control, as illustrated in an embodiment of the present invention.

[0062] Figure 6 This is a risk point risk profiling rule graph model for the data-driven knowledge graph construction and application method for rail transit risk prevention and control, as described in an embodiment of the present invention.

[0063] Figure 7 This is a hidden relationship mining rule graph model for the data-driven knowledge graph construction and application method for rail transit risk prevention and control described in the implementation example of this invention.

[0064] Figure 8 This is an accident causation reasoning rule graph model for the data-driven knowledge graph construction and application method for rail transit risk prevention and control described in the implementation example of this invention.

[0065] Figure 9 This is a risk prevention and control auxiliary decision-making rule graph model for the data-driven knowledge graph construction and application method for rail transit risk prevention and control described in the implementation example of the present invention. Detailed Implementation

[0066] Embodiments of the present invention are described in detail below, examples of which 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 the present invention, and should not be construed as limiting the present invention.

[0067] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0068] It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as here.

[0069] 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 specification 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, and / or groups thereof.

[0070] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0071] To facilitate understanding of the present invention, the present invention will be further explained and described below with reference to the accompanying drawings and specific embodiments. However, the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0072] Those skilled in the art should understand that the accompanying drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily essential for implementing the present invention.

[0073] Example 1

[0074] In this embodiment 1, a data-driven rail transit risk prevention and control system is first provided, including:

[0075] The first construction module is used to build the knowledge graph pattern layer of rail transit risk prevention and control based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern".

[0076] The second construction module is used to construct a knowledge graph data layer for rail transit risk prevention and control based on the characteristics of unstructured data and combined with a data-driven knowledge extraction method for rail transit risk prevention and control.

[0077] The third construction module is used to construct a knowledge graph for rail transit risk prevention and control by using graph database to standardize the storage of triplet structures and instance tags.

[0078] The module determines the implementation rules for four rail transit risk prevention and control functions: risk profiling, implicit relationship mining, accident cause reasoning, and risk prevention and control auxiliary decision-making, based on the semantic link relationship of the rail transit risk prevention and control knowledge graph and graph database.

[0079] In this embodiment 1, the above-described system is used to implement a data-driven rail transit risk prevention and control method, including:

[0080] Based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern", a knowledge graph pattern layer for rail transit risk prevention and control is constructed.

[0081] Based on the characteristics of unstructured data, and combined with a data-driven knowledge extraction method for rail transit risk prevention and control, a knowledge graph data layer for rail transit risk prevention and control is constructed.

[0082] A knowledge graph for rail transit risk prevention and control is constructed by using a graph database to standardize the storage of triple structures and instance tags.

[0083] Based on the semantic linking relationships of the knowledge graph for rail transit risk prevention and control and the graph database, four rules for realizing rail transit risk prevention and control functions are proposed: risk profiling of risk points, mining of implicit relationships, reasoning about the causes of accidents, and auxiliary decision-making for risk prevention and control.

[0084] Constructing a knowledge graph model layer for rail transit risk prevention and control, including:

[0085] Construct a knowledge graph application scenario U for rail transit risk prevention and control, and break down the application scenario into multiple application sub-scenarios based on its application field and scope of use;

[0086] Establish the mapping relationship and mapping function between the application scenarios and knowledge structure of the knowledge graph for rail transit risk prevention and control, and construct the knowledge structure of the knowledge graph for rail transit risk prevention and control.

[0087] The knowledge structure of the knowledge graph for rail transit risk prevention and control is analyzed layer by layer, defining concepts, concept attributes, and relationship patterns between concepts and between concepts and attributes, and constructing a relationship pattern deconstruction diagram;

[0088] ER diagrams are used to graphically represent concepts, concept attributes, and the relationship patterns between concepts and between concepts and attributes, thus constructing a knowledge graph ontology model for rail transit risk prevention and control.

[0089] Evaluation indicators for the knowledge graph ontology model of rail transit risk prevention and control are set, and the ontology model of rail transit risk prevention and control is evaluated, verified and optimized, and a pattern layer of rail transit risk prevention and control knowledge graph is constructed.

[0090] The application scenario U of the knowledge graph for risk prevention and control in rail transit is broken down into three sub-scenarios: 1) It is necessary to explore and characterize all possible states of risk points from multiple dimensions to assist operation safety management personnel in formulating risk point control plans; 2) It is necessary to fully grasp the evolution law and development mode of rail transit operation accidents to assist operation safety management personnel in cutting off the risk evolution path; 3) It is necessary to clarify on-site operation procedures, implement job responsibilities, and form a rail transit safety operation guidance plan.

[0091] Establish a mapping relationship between application scenarios and knowledge structures of the knowledge graph for rail transit risk prevention and control. and mapping function Constructing a knowledge graph knowledge structure O for rail transit risk prevention and control, namely,

[0092]

[0093]

[0094] U=|u1,u2,u3,…,un|(n=1,2,3…)

[0095] O=|o1,o2,o3,…,on|(n=1,2,3…)

[0096]

[0097]

[0098] In the formula, un and on represent the nth constituent element in U and O, respectively. This indicates a mapping from un to on;

[0099] By analyzing the application scenarios of the knowledge graph for rail transit risk prevention and control, the research objects in the application scenarios are used as the basis for mapping the relationship between the knowledge structure and the research objects.

[0100]

[0101] In the formula, a ox This represents the mapping relationship between the application scenario (ox) and the knowledge structure.

[0102] We obtain the mapping relationships under application scenarios u1, u2, and u3. And the mapping results o1, o2, o3, that is,

[0103]

[0104]

[0105]

[0106] The knowledge structure is analyzed layer by layer, defining concepts, concept attributes, and relationship patterns between concepts and between concepts and attributes, and constructing a relational pattern deconstruction diagram; among them, concepts include risk points, risk events, measures, prevention and control positions, and prevention and control standards, and concept attributes are risk identification attributes; including:

[0107] Analyze the knowledge structure of the risk point layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts;

[0108] Analyze the knowledge structure of the risk chain layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts;

[0109] The analysis measures and responsibility layer knowledge structure define the concepts, attributes, and relational schemas within this structure, and construct a deconstruction diagram of the relational schemas between concepts.

[0110] The knowledge graph ontology structure evaluation model for rail transit risk prevention and control consists of two parts: the correlation coefficient matrix G and the evaluation index weight vector W.

[0111]

[0112] B=[b1,b2...bi](i=1,2,...,m)

[0113] W=[w1,w2...wj](j=1,2,...,n)

[0114]

[0115] In the formula, B represents the evaluation result matrix of m ontology model design schemes, bi represents the evaluation result of the i-th ontology model; W represents the weight vector of each evaluation index, wj represents the weight of the j-th index; G represents the correlation coefficient matrix, and gij represents the correlation coefficient between the j-th index and the j-th optimal index in design scheme i.

[0116] The knowledge graph for risk prevention and control in rail transit consists of a three-layer knowledge structure, with the risk evolution link as the intermediate layer and risk points and prevention and control as the upper and lower related layers.

[0117] The middle layer uses "risk events" and "accident scenarios" as nodes and coupling relationships as edges to express how risk events develop and evolve to ultimately lead to accident scenarios in a "chain" or "network" transmission process, and uses risk identification attributes as the node attributes of "risk events".

[0118] The upper layer uses "risk points" as nodes and is connected to the middle layer through a predicate relationship to describe the state manifestation of risk points evolving and escalating into risk events. The upper and lower levels within the layer are used as edges to express the hierarchical relationship between components within the system.

[0119] The lower-level related layer takes "measures" as the central node and "standard sources and prevention and control positions" as related nodes, and connects with the middle layer through predicate relationships to explain the four stages of "prevention-treatment-control-rescue", thereby implementing the standards and norms for risk prevention and control and proactive safety assurance of the rail transit operation system.

[0120] Example 2

[0121] Reference Figure 1 In this embodiment 2, a flowchart of a data-driven method for constructing and applying a knowledge graph for rail transit risk prevention and control is shown. The specific method includes:

[0122] Step 101: Constructing a data-driven knowledge graph model layer for rail transit risk prevention and control;

[0123] This paper constructs application scenarios for a knowledge graph of rail transit risk prevention and control, and decomposes these scenarios into three sub-scenarios based on their application fields and scope of use: "u1 Controlling risk points, u2 Cutting off risk chains, and u3 Implementing prevention and control responsibilities." It establishes mapping relationships and mapping functions between the application scenarios and knowledge structures of the rail transit risk prevention and control knowledge graph, constructing the knowledge structure O. The knowledge structure is then analyzed layer by layer, defining concepts (risk points, risk events, measures, prevention and control positions, prevention and control standards), concept attributes (risk identification attributes), and relationship patterns between concepts and between concepts and attributes. Relationship pattern deconstruction graphs GP, GC, and GE are constructed. ER diagrams are used to graphically represent concepts, concept attributes, and relationship patterns, constructing the ontology model of the rail transit risk prevention and control knowledge graph. Evaluation indicators for the pattern layer of the rail transit risk prevention and control knowledge graph are set, and the ontology model is evaluated, verified, and optimized to construct the pattern layer of the rail transit risk prevention and control knowledge graph.

[0124] Step 103: Construction of the data layer of the knowledge graph for rail transit risk prevention and control based on data-driven approach;

[0125] S21: Construct a professional dictionary {word_id} for the rail transit field to standardize vocabulary in text data; train word vectors for the domain vocabulary based on the word2vec model to construct a word vector table {word_index}; construct event pair extraction templates based on Chinese logical rules to extract risk links; construct event structured algorithm rules based on dependency parsing to extract the main structure of risk events; construct knowledge fusion rules based on text similarity to perform knowledge fusion on risk events and accident scenarios, and construct relevant triples for the risk chain layer; construct risk point extraction rules based on constituent parsing to extract risk points and construct relevant triples for the risk point layer; extract measures, prevention and control positions, and sources of standards based on industry norms to construct relevant triples for the prevention and control responsibility layer.

[0126] Step 105: Construction of a data-driven knowledge graph for rail transit risk prevention and control;

[0127] The extracted instances are labeled based on concepts, concept attributes, and relational schemas; based on the knowledge graph schema layer and graph database storage specifications for rail transit risk prevention and control, the triple structure and labels are imported into the Neo4j graph database to construct a knowledge graph for rail transit risk prevention and control.

[0128] Step 107: Application analysis of data-driven knowledge graph for rail transit risk prevention and control;

[0129] For the u1 application scenario of controlling risk points, construct risk characterization rules for risk points; for the u2 application scenario of cutting off risk chains, construct rules for mining implicit relationships in risk chains; for the u2 application scenario of cutting off risk chains, construct rules for inferring the causes of accidents; for the u3 application scenario of implementing prevention and control responsibilities, construct rules for assisting decision-making in risk prevention and control.

[0130] Reference Figure 2 This paper presents a flowchart illustrating the construction process of a knowledge graph model layer for rail transit risk prevention and control, based on a data-driven approach. The specific steps are as follows:

[0131] Step 1: Construct a knowledge graph application scenario U for rail transit risk prevention and control, and decompose the application scenario into n sub-scenarios |u1,u2,u3,…,un| according to its application field and scope of use, that is,

[0132] U=|u1,u2,u3,…,un|(n=1,2,3…)

[0133] By analyzing the internal characteristics of the rail transit operation system and the need for proactive risk prevention and control, this invention decomposes the application scenario U of the rail transit risk prevention and control knowledge graph into three sub-scenarios, namely:

[0134] U = |u1,u2,u3|

[0135]

[0136] u3 = Implementing Prevention and Control Responsibilities

[0137] In the formula, u1 represents the need to explore and characterize all possible states of risk points from multiple dimensions to assist operation and safety management personnel in formulating risk point control plans in a comprehensive and objective manner;

[0138] u2 indicates the need to fully understand the evolution patterns and development models of rail transit operation accidents, and to assist operation safety management personnel in quickly and effectively cutting off the risk evolution path;

[0139] u3 indicates the need to clarify on-site operating procedures, implement job responsibilities, and form a standardized and unified rail transit safety operation guidance plan that links the entire system, so as to provide strong decision support for the safety assurance of the rail transit system.

[0140] Step 2: Establish a mapping relationship between application scenarios and knowledge structures of the rail transit risk prevention and control knowledge graph. and mapping function Constructing a knowledge graph knowledge structure O for rail transit risk prevention and control, namely,

[0141]

[0142]

[0143] U=|u1,u2,u3,…,un|(n=1,2,3…)

[0144] O=|o1,o2,o3,…,on|(n=1,2,3…)

[0145]

[0146]

[0147] In the formula, un and on represent the nth constituent element in U and O, respectively. This indicates a mapping from un to on;

[0148] By analyzing the application scenarios of the knowledge graph for rail transit risk prevention and control, this invention uses the research objects in the application scenarios as the basis for mapping the relationship between them and the knowledge structure.

[0149]

[0150] In the formula, a ox This represents the mapping relationship between the application scenario (ox) and the knowledge structure.

[0151] We obtain the mapping relationships under application scenarios u1, u2, and u3. And the mapping results o1, o2, o3, that is,

[0152]

[0153]

[0154]

[0155] Step 3: Analyze the knowledge structure from Step 2 layer by layer, defining concepts (risk points, risk events, measures, prevention and control positions, prevention and control standards), concept attributes (risk identification attributes), and the relationship patterns between concepts and between concepts and attributes, and constructing a relational pattern deconstruction diagram G. P G C G E ;

[0156] 1) Analyze the knowledge structure of the risk point layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts.

[0157] From a systems theory perspective, risk points refer to the personnel, equipment, environment, and management aspects associated with risks in a rail transit operation system; that is,

[0158] RP = {H; EI; E; M}

[0159] In the formula, H, EI, E, and M represent the risk point sets of the rail transit system in terms of personnel, equipment, environment, and management, respectively. H, EI, E, and M can be further subdivided into subsets based on the composition of the rail transit system until the smallest constituent unit is reached.

[0160] Typically, a single risk point is accompanied by multiple risk states, that is,

[0161]

[0162]

[0163]

[0164]

[0165] In the formula, This represents the k-th type of risk for risk point i in personnel category. This represents the k-th risk level of risk point i in equipment category. To represent the k-th type of risk at environmental risk point i, This represents the k-th type of risk at risk point i in management category;

[0166] When a risk point, due to its own problems or the coupling effect of other risk points, causes its state value to change at a certain moment and exceed its own safety threshold, its state will evolve from a risk to a risk event, that is,

[0167]

[0168]

[0169]

[0170]

[0171] In the formula, taking personnel-related risk points as an example, express The state value at a certain moment;

[0172] ρ h * Indicates risk point h i The threshold for one's own safety status is related to the specific prevention and control targets and risk perception methods.

[0173] when The time indicates the risk point h i If its own safety threshold is exceeded, the risk point status will change from risk to risk. Evolutionary upgrades lead to risk events

[0174] Therefore, risk points include two states: risks and risk events. The difference lies in that "risk" generally describes a potential danger that has not yet occurred, while a risk event indicates that the potential danger has already turned into an actual loss. There is a transformation relationship T between the two, which is related to the safety state threshold of the risk point. Taking personnel-related risk points as an example, that is...

[0175]

[0176]

[0177]

[0178] In the formula, Representing risk points h respectively i From risk status To risk events The transformation relationships and transformation functions;

[0179] Based on the above-mentioned risk point state evolution process, a relational schema deconstruction graph G is constructed with the risk points as the starting nodes. P The relation schema expression is as follows:

[0180] G P ={<Pi,Rx,Pj> ,<Pj,Rx,Pk> ,<Pi,Rj,Ei>}

[0181] In the formula, Pi, Pj, and Pk represent risk point entities at different levels, Ei represents associated risk event entities, Rx represents hierarchical relationships, and Rj represents predicate relationships - risk events;

[0182] Among them, the hierarchical relationship represents the internal hierarchical relationship of risk points in the rail transit system. Logically, it belongs to the genus-species relationship and can be interpreted as the relationship of inclusion and being included between the superior risk point and the subordinate risk point. According to the hierarchical concept expansion method, the risk point set is divided from the perspective of "human-machine-environment-management" until the smallest constituent unit is reached; the predicate relationship represents the description of the risk event state of the risk point; the triple structure between the risk point and other concepts is shown in Table 1.

[0183] Table 1

[0184]

[0185] 2) Analyze the knowledge structure of the risk chain layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts.

[0186] The risk chain reflects the development pattern and evolution process of an accident, and consists of three elements: risk point, risk event, and the coupling relationship between risk events.

[0187] When a risk point exceeds its own safety threshold, resulting in unsafe human behavior, unsafe equipment conditions, abnormal environmental changes, or management deficiencies, if it is not effectively managed and controlled, one or more overlapping risk events will dynamically correlate and couple with other risk events in a certain time and space through coupling effects. This ultimately evolves and escalates into a "chain-like" or "network-like" propagation path leading to an accident scenario.

[0188]

[0189] In the formula, These represent unsafe acts by humans, unsafe conditions of equipment, abnormal changes in the environment, and management deficiencies, respectively; A represents the risk chain, which is a specific sequence of accident causes formed by risk events linked together through a coupling mechanism.

[0190] Based on the aforementioned process of risk events developing, evolving, and ultimately forming accidents, a relational schema deconstruction graph G is constructed with risk events as the starting nodes. E The relation schema expression is as follows:

[0191] G E ={<Ei,Ro,Ej> ,<Ei,Ro,Ai> ,<Ei,Rf,Pi>}

[0192] In the formula, Ei and Ej represent two different risk event entities, Ai represents the associated accident scenario entity, Pi represents the risk identification attribute value, Ro represents the coupling relationship, and Rf represents the risk identification attribute.

[0193] Among them, coupling relationship represents the interaction between risk events or between risk events and accident scenarios; risk identification attribute represents the custom attribute that determines whether a risk event is triggered, which is related to the specific prevention and control object and risk perception method; the triple structure between risk events and other concepts is shown in Table 2.

[0194] Table 2

[0195]

[0196] By analyzing the data on rail transit operation accidents, the coupling relationships are expanded to four categories: "causality, condition, sequence, and contrast."

[0197] Ro = {Ry; Rt; Rs; Rz}

[0198] <Ei,Ro,Ej> ={<Ei,Ry,Ej> ,<Ei,Rt,Ej> ,<Ei,Rs,Ej> ,<Ei,Rz,Ej>}

[0199] <Ei,Ro,Ai> ={<Ei,Ry,Ai> ,<Ei,Rt,Ai> ,<Ei,Rs,Ai> ,<Ei,Rz,Ai>}

[0200] In the formula, Ry, Rt, Rs, and Rz represent causal, conditional, sequential, and adversative relationships, respectively.

[0201] Reference Figure 3 This paper presents an example of the coupling relationship between a data-driven knowledge graph construction and application method for rail transit risk prevention and control. The details are as follows:

[0202] Causality is used to describe a cause-and-effect relationship, where a previous risk event leads to the occurrence of the next risk event (accident scenario). A specific example is <"short circuit due to carbon fiber detachment in the tunnel section", causal relationship: "power supply contact network tripping">.

[0203] Conditional relationships are used to describe the relationship between the occurrence of a certain risk event and the occurrence of the next risk event (accident scenario). Specific examples are <“Power supply contact network tripping”, conditional relationship _, “Signal system mis-transmitted speed signal”> and <“Temporary abnormal route reversal reversal”, conditional relationship _, “Signal system mis-transmitted speed signal”>.

[0204] The sequential relationship is used to describe the continuous relationship between one risk event and the next. A specific example is <"Sudden power supply contact network tripping", sequential relationship _, "Temporary abnormal route reversal">.

[0205] The transition relationship is used to describe the opposition between two risk events. A specific example is <“Maintenance personnel inspect the pantograph of the vehicle”, transition relationship _, “Maintenance personnel did not find any abnormality in the pantograph”>. The triple structure of the coupling relationship is shown in Table 3.

[0206] Table 3

[0207]

[0208] 3) Analyze the knowledge structure of the responsibility layer, define the concepts, attributes and relational schemas within the structure, and construct a deconstruction diagram of the relational schemas between concepts.

[0209] In the daily operation and maintenance of rail transit systems, relevant management personnel continuously analyze accident reports and fault logs to summarize the causes and patterns of accidents. Based on the controllability, severity, and scope of impact of accidents and incidents, they formulate on-site handling plans to guide personnel in clarifying on-site operating procedures and handling processes, so as to effectively respond to similar incidents.

[0210] Based on safety assurance and control regulations, production and business technical procedures, and other information, a deconstruction diagram G is constructed, starting with measures as the initial node, to establish a relational schema. C The relation schema expression is as follows:

[0211] G C ={<Ci,Rg,Fi> ,<Ci,Rl,Gi> ,<Ei,Rc,Ci> ,<Ai,Rc,Ci>}

[0212] In the formula, Ci represents the measure entity, Fi represents the associated prevention and control post entity, Gi represents the associated normative source entity, Ei represents the associated risk event entity, and Ai represents the associated accident scenario entity; Rg, Rl, and Rc all belong to the predicate relationship; the triple structure between the measure and other concepts is shown in Table 4.

[0213] Table 4

[0214]

[0215] To further optimize and adjust the safe operation and management of the rail transit system, based on the evolution logic of risk links and the connotation of safety assurance work for rail transit operation systems, the concept of measures is further expanded to form four subordinate concepts: "preventive measures, governance measures, control measures, and rescue measures."

[0216] Ci = {Cf; Cz; Ck; Cj}

[0217] In the formula, Cf, Cz, Ck, and Cj represent the four stages of "prevention, treatment, control, and rescue," respectively.

[0218] The "prevention" phase involves taking all proactive and effective preventative measures to prevent risk events from occurring and to maximize operational safety, based on a comprehensive identification of all risk points affecting the safe operation of the rail transit system and their manifestations.

[0219] The "treatment" stage refers to taking immediate treatment measures to eliminate risks when the prevention stage fails and the risk points become abnormal, so as to ensure that the rail transit system can operate safely, effectively and continuously in a coordinated manner.

[0220] The "control" phase refers to the immediate implementation of control strategies when the governance phase fails and the problem cannot be eliminated, in order to mitigate the consequences as much as possible and prevent a "small incident" from escalating into a "major accident".

[0221] The "rescue" phase refers to the immediate activation of an emergency response when the control phase fails, ultimately leading to an accident. This involves organizing emergency repairs and rescue efforts to restore normal operations in the shortest possible time and minimize accident losses.

[0222] Reference Figure 4 This paper presents a knowledge graph ontology model for rail transit risk prevention and control, based on a data-driven approach to constructing and applying such a knowledge graph.

[0223] Step 4: Use ER diagrams to graphically represent the concepts, concept attributes, and relational schemas from Step 3, and construct an ontology model of the knowledge graph for rail transit risk prevention and control.

[0224] Define a graphical construction standard S, and determine the set of concepts, attributes, and relationships D.

[0225] G = {S, D, C}

[0226] 1) Define the graphical construction standard S

[0227] Rectangular boxes represent concepts

[0228] The elliptical box represents the attribute of the concept set.

[0229] The diamond shape represents concepts and attributes, and the relationships between concepts.

[0230] 2) Determine the concepts, attributes, and relation set D

[0231] The knowledge graph for risk prevention and control in rail transit includes six types of concept nodes, one type of attribute, and three types of relationships;

[0232] 3) Determine the knowledge structure C of the knowledge graph

[0233] The knowledge graph for risk prevention and control in rail transit consists of a three-layer knowledge structure, with the risk evolution chain as the middle layer and risk points and prevention and control as the upper and lower related layers.

[0234] The middle layer uses "risk events" and "accident scenarios" as nodes, and coupling relationships (sequential, causal, conditional, and transitional relationships) as edges to express how risk events develop and evolve to ultimately form a "chain" or "network" transmission process that leads to accident scenarios, and uses risk identification attributes as the node attributes of "risk events".

[0235] The upper layer uses "risk points" as nodes and is connected to the middle layer through a predicate relationship to describe the state manifestation of risk points evolving and escalating into risk events. The upper and lower levels within the layer are used as edges to express the hierarchical relationship between components within the system.

[0236] The lower-level related layer takes "measures" as the central node and "standard sources and prevention and control positions" as related nodes, and connects with the middle layer through predicate relationships to explain the four stages of "prevention-treatment-control-rescue", thereby implementing the standards and norms for risk prevention and control and proactive safety assurance of the rail transit operation system.

[0237] Step 5: Set evaluation indicators for the ontology model of the knowledge graph of rail transit risk prevention and control, evaluate, verify and optimize the ontology model in Step 4, and construct the mode layer of the knowledge graph of rail transit risk prevention and control.

[0238] The knowledge graph ontology structure evaluation model for rail transit risk prevention and control consists of two parts: the correlation coefficient matrix G and the evaluation index weight vector W.

[0239] B = W * G

[0240] B=[b1,b2...bi](i=1,2,...,m)

[0241] W=[w1,w2...wj](j=1,2,...,n)

[0242]

[0243] In the formula, B represents the evaluation result matrix of m ontology model design schemes, and bi represents the evaluation result of the i-th ontology model.

[0244] W represents the weight vector of each evaluation index, and wj represents the weight of the j-th index.

[0245] G represents the correlation coefficient matrix, and gij represents the correlation coefficient between the j-th index and the j-th optimal index in design scheme i.

[0246] The evaluation criteria for the weight vector W and correlation coefficient matrix G, as well as the evaluation criteria for the evaluation result matrix B, are as follows:

[0247] 1) Evaluation index weight vector W

[0248] This invention uses clarity, consistency, completeness, and scalability as four evaluation indicators for the knowledge graph ontology model of rail transit risk prevention and control, and the sum of the weights of each indicator is 1, that is,

[0249] w1+w2+w3+w4=1

[0250] 2) Correlation coefficient matrix G

[0251] The correlation between the indicator vectors of the ontology model design scheme and the relatively optimal scheme can be used as a criterion for judging whether the ontology model scheme meets the design criteria. The evaluation matrix R0 = [r...] of the relatively optimal scheme is set. 01 ,r 02 ,…,r 0n Ontology model design scheme index r ij The scores are given by domain experts, and the scores need to be normalized to obtain the evaluation matrix R of the ontology model design scheme.

[0252]

[0253] Furthermore, the index r of the ontology model design scheme is calculated. ij Compared with the relative optimal solution index r 0j The correlation coefficient between them, that is,

[0254]

[0255] In the formula, ρ represents the resolution coefficient, ρ∈(0,1), and usually ρ=0.5;

[0256] We obtain the correlation coefficient matrix G.

[0257]

[0258] 3) Evaluation Result Matrix B

[0259] The weighted correlation degree bi between the ontology model design scheme and the relatively optimal scheme is calculated to obtain the evaluation result matrix B, where α represents the threshold of the ontology model design scheme, i.e.,

[0260]

[0261] It reflects the degree of correlation between various ontology model design schemes and the optimal scheme. Since the optimal scheme is set to 1, the closer it is to 1, the higher the degree of fit between the scheme and the optimal scheme.

[0262] If bi≥α, it means that the knowledge graph ontology model for rail transit risk prevention and control meets the requirements and can be used as the pattern layer for subsequent knowledge extraction and graph construction; if bi<α, it means that it does not meet the requirements and steps 2-5 need to be repeated to rebuild the knowledge graph ontology model for rail transit risk prevention and control.

[0263] Reference Figure 5 This paper presents a flowchart of a data-driven knowledge graph construction and application method for rail transit risk prevention and control, including the extraction of knowledge for rail transit risk prevention and control. The specific steps are as follows:

[0264] A. Construct a specialized dictionary {word_id} for the rail transit field. The specific steps are as follows:

[0265] Step 1: Collect and organize data in the rail transit field, including rail transit operation accident cases, rail transit design specifications, rail transit operation management specifications, and management methods for the classification and control of rail transit operation safety risks and the investigation and treatment of hidden dangers.

[0266] Step 2: Use the jieba word segmentation tool to extract words from the data in Step 1. At the same time, in order to ensure that the vocabulary in the rail transit field is objective, standardized, professional and authoritative, and can be reused and shared, the stop word list must be continuously updated through manual verification.

[0267] Step 3: Filter the words segmented in Step 2 by noun part-of-speech and use them as candidate words for the rail transit domain thesaurus;

[0268] Step 4: Conduct a comprehensive analysis of all levels and operational aspects of the rail transit system from four dimensions: personnel, equipment, environment, and management. Manually verify and supplement the candidate words in the rail transit field thesaurus from Step 3 to construct a professional dictionary for the rail transit field {word_id}.

[0269] Step 5: Standardize the vocabulary contained in the text data according to the professional dictionary {word_id} in the rail transit field;

[0270] B. Construct the word vector table {word_index}, the specific steps are as follows:

[0271] Step 6: Based on {word_id}, train word vectors for the rail transit domain using the word2vec model. Due to the limited domain data, this invention uses random numbers to map each word to a vector space, constructing a word vector table {word_index} for the rail transit domain, providing a basis for subsequent knowledge fusion and knowledge updates based on the similarity algorithm in this paper.

[0272] C. Construct event pair extraction templates based on Chinese logical rules. The specific steps are as follows:

[0273] Step 7: Set the Long_Sentence_keyword_list for segmenting long sentences in the accident report text data;

[0274] Sentence_keyword_list=["?",".",";","\n","!","..."]

[0275] Step 8: Use regular expressions to segment the domain data Report into long sentences, forming a sentence set Report_list consisting of several long event description sentences Sentence_list.

[0276] Report_list=[[Sentence_list1],[Sentence_list2],[Sentence_list3]...[Sentence_listn]]

[0277] Step 9: Based on Chinese logic, design matching sentence patterns for causal, sequential, conditional, and adversative relationships, and construct event pair extraction templates;

[0278] Step 10: Based on the event extraction template in Step 9, extract risk events and their coupling relationships from each Sentence_list in Step 8, forming several {E1, coupling relationship, E2} triples.

[0279] D. Construct event-structured algorithm rules based on dependency parsing. The specific steps are as follows:

[0280] Step 11: Analyze the dependency relationships between words in risk events based on the dependency parsing algorithm provided by LTP, extract the main structure of risk events, and improve the generalization ability of risk events while ensuring the semantic integrity of the risk event description and without changing its meaning. For example, "the track surface is relatively wet and slippery" becomes "the track surface is wet and slippery" after structuring.

[0281] Backbone structure = [SBV / POB] + [HED] + [VOB / FOB / POB]

[0282] E. Construct knowledge fusion rules based on text similarity, with the following specific steps:

[0283] Step 12: The word vector table {word_index} in the rail transit domain is used to sum the word vectors of each word in the risk event and take the average to obtain the sentence embedding.

[0284] Step 13: Calculate the cosine distance value of the sentence embedding, i.e., sim(s1,s2);

[0285] Step 14: If sim(s1,s2)>0.75 and passes both machine and manual verification, it means that the two are similar and can be merged into one node; otherwise, it means that they are not similar.

[0286] Step 15: The integration operation of new instances into the existing knowledge graph can be repeated from steps 12 to 14.

[0287] Step 16: Construct the risk chain layer knowledge structure related triplet structure, including <risk event, coupling relationship, risk event>, <risk event, coupling relationship, accident scenario>, <accident scenario, coupling relationship, risk event>;

[0288] F. Risk point extraction and related triplet construction extraction, the specific steps are as follows:

[0289] Step 17: Extract noun phrases related to risk events based on constituent syntactic analysis, and use them as candidate words for risk points;

[0290] Step 18: Determine whether the candidate words for risk points are in the professional dictionary {word_id} of the rail transit field based on the keyword recognition method; if so, extract the risk points; otherwise, manual verification is required, and {word_id} will be updated and improved.

[0291] Step 19: Construct the knowledge structure related triplet structure of the risk point layer, including <risk point, risk event, risk event>, <level 1 risk point, hierarchical relationship, level 2 risk>, <level 2 risk point, hierarchical relationship, level 3 risk>;

[0292] G. Based on industry standards, extract measures, prevention and control positions, and the source of the standards. The specific steps are as follows:

[0293] Step 20: Extract measures, prevention and control positions, and sources of regulations based on industry standards, and the relevant triplet for the prevention and control responsibility level, including the triplet of <measures, source of regulations, source of regulations> and <measures, prevention and control positions, prevention and control positions>.

[0294] Furthermore, a knowledge graph for risk prevention and control in rail transit will be constructed, with the following specific steps:

[0295] Step 1: Label the extracted instances based on concepts, concept attributes, and relational schemas using a tagging method;

[0296] Step 2: Use Python tools to connect to the Neo4j graph database, and based on the knowledge graph schema layer and graph database storage specifications for rail transit risk prevention and control, import the triplet structure and tags into the Neo4j graph database to construct a knowledge graph for rail transit risk prevention and control.

[0297] Reference Figure 6, showing a risk point risk portrait rule graph model of a data-driven method for constructing and applying a risk prevention and control knowledge graph for rail transit. The specific steps are as follows:

[0298] Rule1: For the application scenario of controlling risk points, construct a risk point risk characterization rule based on the association path between risk points and risk events to assist operation safety management personnel in formulating a risk point control plan comprehensively and objectively, and then conduct targeted daily risk inspections and hidden danger rectification. The specific steps are as follows:

[0299] Step 1: Input the statement "MATCH p=(a:'risk point')-[r:'risk event']->(b:'risk event') RETURN p", and the running result will return multiple visual association paths of "risk events" around the "risk point" node, obtaining a visual subgraph of risk inspection items related to the risk point;

[0300] Step 2: Modify the return command in Step 1 to "RETURN distinct a.Name AS risk point name b.Name AS risk inspection item", and the running result will return a list of risk inspection items for the risk point;

[0301] Step 3: Modify the return command in Step 1 to "RETURN COUNT(b.Name)", and the running result will directly return the number "x" of risk inspection items for the risk point, obtaining the number of risk inspection items for the risk point;

[0302] Step 4: Modify the return command in Step 1 to "RETURN b.name as risk event, count(*) as risk event quantity ORDER BY risk event quantity DESC", and the running result will sort in descending order according to the number of risk inspection items for the risk point, and then find the risk point that is most likely to undergo state transition.

[0303] Refer to Figure 7 , showing an implicit relationship mining graph model of a data-driven method for constructing and applying a risk prevention and control knowledge graph for rail transit. The specific steps are as follows:

[0304] Rule1: For the application scenario of cutting off the risk chain, propose an implicit relationship mining rule based on the complex chain network relationship inside the rail transit risk prevention and control knowledge graph to assist operation safety management personnel in formulating reasonable and effective risk chain control measures to achieve maximum operation safety. The specific steps are as follows:

[0305] Suppose Risk event _, Ai>, <B, risk event _, Bj>, <B, risk event _, Bn> (j ≠ n)

[0306] ∩<Bj, Coupling relationship _, Ai>;

[0307] → Initially infer that T = <Bn, Coupling relationship _, Ai>;

[0308] Create T = <Bn, Implicit relationship, Ai>;

[0309] → IF the implicit relationship ∈ {Causal relationship; Conditional relationship; Sequential relationship; Adversative relationship};

[0310] Create T = <Risk event Bn, Causal relationship / Conditional relationship / Sequential relationship / Adversative relationship, Risk event Ai>;

[0311] Refer to Figure 8 , which shows an accident cause inference rule graph model for the construction and application method of a rail transit risk prevention and control knowledge graph based on data-driven, realizing the inference of the accident evolution path. The specific rules are as follows:

[0312] Rule1: Based on the association path between the risk event and the accident scenario, propose accident cause inference rules to assist operation safety management personnel to infer the most likely accident causes according to the existing knowledge, and then formulate a rapid and effective deployment of emergency rescue resources to minimize the accident losses. The specific steps are as follows:

[0313] Step 1: Input the statement "MATCH(p1:`Level 3 risk point`),(p2:`Accident scenario`{name:′S accident scenario'}),p = ((p1)-[*..10]->(p2))where p1.name Contains'A'RETURN p,length(p)", and the returned result will show all the association paths and path lengths from the risk point A to the accident scenario S, obtaining a visualization subgraph of all possible accident causes and propagation paths under the known key information;

[0314] Step 2: Use the fault diagnosis method to check each inference result in Step 1 one by one, and continuously infer the accident causes through association rules until the root cause of the accident scenario S is found;

[0315] Refer to Figure 9 , which shows a risk prevention and control auxiliary decision-making rule graph model for the construction and application method of a rail transit risk prevention and control knowledge graph based on data-driven. The specific steps are as follows:

[0316] Rule1: For the application scenario of implementing prevention and control responsibilities, propose risk prevention and control auxiliary decision-making rules based on the association paths between measures and risk events, and between measures and accident scenarios, to guide daily risk proactive prevention and optimize the on-site emergency response process, providing new intelligent means for the implementation of rail transit risk prevention and control and proactive safety guarantee. The specific steps are as follows:

[0317] Step 1: Enter the statement "MATCH p=(a:`Risk Event`{name:′A′})-[r:`Measures`]->(b:`Prevention and Control Measures`)-[*..1]->()RETURN p LIMIT 25". The returned result will be displayed as a visual sub-graph of the corresponding prevention and control measures, prevention and control positions and standard sources for preventing or controlling the occurrence of risk event A.

[0318] Step 2: Enter the statement "MATCH p=(a:`Risk Event`{name:′A′})-[r:`Measures`]->(b:`Governance Measures`)-[*..1]->()RETURN p LIMIT 25". The returned result will be displayed as a visual sub-graph of the governance measures, prevention and control positions, and standard sources corresponding to the elimination of the risk event.

[0319] Step 3: Enter the statement "MATCH p=(a:`Risk Event`{name:′A′})-[r:`Measures`]->(b:`Control Measures`)-[*..1]->()RETURN p LIMIT 25". The returned result will display a visual sub-chart showing the corresponding control measures, prevention and control positions, and source of regulations for mitigating the adverse effects of risk events.

[0320] Step 4: Enter the statement “MATCHp=(a:`Accident Scene`{name:′B′})-[r:`Measures`]->(b:`Rescue Measures`)-[*..1]->()RETURN p LIMIT 25”. The returned result will display a visual sub-graph of the emergency rescue measures, prevention and control positions, and standard sources carried out to restore normal operations.

[0321] Example 3

[0322] This embodiment 3 provides a non-transitory computer-readable storage medium for storing computer instructions. When these computer instructions are executed by a processor, they implement the data-driven rail transit risk prevention and control method described above. The method includes:

[0323] Based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern", a knowledge graph pattern layer for rail transit risk prevention and control is constructed.

[0324] Based on the characteristics of unstructured data, and combined with a data-driven knowledge extraction method for rail transit risk prevention and control, a knowledge graph data layer for rail transit risk prevention and control is constructed.

[0325] A knowledge graph for rail transit risk prevention and control is constructed by using a graph database to standardize the storage of triple structures and instance tags.

[0326] Based on the semantic linking relationships of the knowledge graph for rail transit risk prevention and control and the graph database, four rules for realizing rail transit risk prevention and control functions are proposed: risk profiling of risk points, mining of implicit relationships, reasoning about the causes of accidents, and auxiliary decision-making for risk prevention and control.

[0327] Example 4

[0328] This embodiment 4 provides a computer program product, including a computer program that, when run on one or more processors, implements the data-driven rail transit risk prevention and control method described above. The method includes:

[0329] Based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern", a knowledge graph pattern layer for rail transit risk prevention and control is constructed.

[0330] Based on the characteristics of unstructured data, and combined with a data-driven knowledge extraction method for rail transit risk prevention and control, a knowledge graph data layer for rail transit risk prevention and control is constructed.

[0331] A knowledge graph for rail transit risk prevention and control is constructed by using a graph database to standardize the storage of triple structures and instance tags.

[0332] Based on the semantic linking relationships of the knowledge graph for rail transit risk prevention and control and the graph database, four rules for realizing rail transit risk prevention and control functions are proposed: risk profiling of risk points, mining of implicit relationships, reasoning about the causes of accidents, and auxiliary decision-making for risk prevention and control.

[0333] Example 5

[0334] This embodiment 5 provides an electronic device, including: a processor, a memory, and a computer program; wherein, the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes instructions to implement the data-driven rail transit risk prevention and control method described above, the method including:

[0335] Based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern", a knowledge graph pattern layer for rail transit risk prevention and control is constructed.

[0336] Based on the characteristics of unstructured data, and combined with a data-driven knowledge extraction method for rail transit risk prevention and control, a knowledge graph data layer for rail transit risk prevention and control is constructed.

[0337] A knowledge graph for rail transit risk prevention and control is constructed by using a graph database to standardize the storage of triple structures and instance tags.

[0338] Based on the semantic linking relationships of the knowledge graph for rail transit risk prevention and control and the graph database, four rules for realizing rail transit risk prevention and control functions are proposed: risk profiling of risk points, mining of implicit relationships, reasoning about the causes of accidents, and auxiliary decision-making for risk prevention and control.

[0339] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0340] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0341] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0342] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment, whereby a series of operational steps are performed to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0343] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that, based on the technical solutions disclosed in the present invention, various modifications or variations that can be made by those skilled in the art without creative effort should be included within the scope of protection of the present invention.

Claims

1. A data-driven method for risk prevention and control in rail transit, characterized in that, include: Based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern", a knowledge graph pattern layer for rail transit risk prevention and control is constructed. The construction of the knowledge graph model layer for rail transit risk prevention and control includes: constructing an application scenario U for the rail transit risk prevention and control knowledge graph, and decomposing the application scenario into multiple application sub-scenarios according to its application field and scope of use; establishing the mapping relationship and mapping function between the application scenario and the knowledge structure of the rail transit risk prevention and control knowledge graph, and constructing the knowledge structure of the rail transit risk prevention and control knowledge graph; parsing the constructed knowledge structure of the rail transit risk prevention and control knowledge graph layer by layer, defining concepts, concept attributes, and the relationship patterns between concepts and between concepts and attributes, and constructing a relation pattern deconstruction diagram; using ER diagrams to graphically represent concepts, concept attributes, and the relationship patterns between concepts and between concepts and attributes, and constructing the ontology model of the rail transit risk prevention and control knowledge graph; setting evaluation indicators for the ontology model of the rail transit risk prevention and control knowledge graph, evaluating, verifying, and optimizing the ontology model of the rail transit risk prevention and control knowledge graph, and constructing the knowledge graph model layer for rail transit risk prevention and control. Based on the characteristics of unstructured data, and combined with a data-driven knowledge extraction method for rail transit risk prevention and control, a knowledge graph data layer for rail transit risk prevention and control is constructed. A knowledge graph for rail transit risk prevention and control is constructed by using a graph database to standardize the storage of triple structures and instance tags. Based on the semantic linking relationship of the knowledge graph of rail transit risk prevention and control and the graph database, four rules for realizing the functions of rail transit risk prevention and control are proposed: risk point risk profiling, implicit relationship mining, accident cause reasoning, and risk prevention and control auxiliary decision-making. Among these, establishing a mapping relationship between application scenarios and knowledge structures for the knowledge graph of rail transit risk prevention and control is crucial. and mapping function Construct a knowledge graph knowledge structure O for rail transit risk prevention and control, that is, ； ； ； ； ； ； In the formula, , Let U and O represent the nth constituent element, respectively. Indicates from arrive The mapping relationship; By analyzing the application scenarios of the knowledge graph for rail transit risk prevention and control, the research objects in the application scenarios are used as the basis for mapping the relationship between the knowledge structure and the research objects. ； In the formula, This represents the mapping relationship between the application scenario (ox) and the knowledge structure. Obtain, application scenarios , , The following mapping relationship , , and mapping results , , ,Right now, ； ； 。 2. The data-driven rail transit risk prevention and control method according to claim 1, characterized in that, The application scenario U of the knowledge graph for risk prevention and control in rail transit is broken down into three sub-scenarios: 1) It is necessary to explore and characterize all possible states of risk points from multiple dimensions to assist operation safety management personnel in formulating risk point control plans; 2) It is necessary to fully grasp the evolution law and development mode of rail transit operation accidents to assist operation safety management personnel in cutting off the risk evolution path; 3) It is necessary to clarify on-site operation procedures, implement job responsibilities, and form a rail transit safety operation guidance plan.

3. The data-driven rail transit risk prevention and control method according to claim 1, characterized in that, The knowledge structure is analyzed layer by layer, defining concepts, concept attributes, and relationship patterns between concepts and between concepts and attributes, and constructing a relational pattern deconstruction diagram; among them, concepts include risk points, risk events, measures, prevention and control positions, and prevention and control standards, and concept attributes are risk identification attributes; including: Analyze the knowledge structure of the risk point layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts; Analyze the knowledge structure of the risk chain layer, define the concepts, attributes and relationship patterns within the structure, and construct a deconstruction diagram of the relationship patterns between concepts; The analysis measures and responsibility layer knowledge structure define the concepts, attributes, and relational schemas within this structure, and construct a deconstruction diagram of the relational schemas between concepts.

4. The data-driven rail transit risk prevention and control method according to claim 1, characterized in that, The knowledge graph ontology structure evaluation model for rail transit risk prevention and control consists of a correlation coefficient matrix. and evaluation index weight vector It consists of two parts, namely, In the formula, express Evaluation result matrix of various ontology model design schemes Indicates the first Evaluation results of various ontology models; This represents the weight vector of each evaluation indicator. Indicates the first The weight of each indicator; Represents the correlation coefficient matrix. Indicate design scheme The Middle The first indicator and the first The correlation coefficient of the optimal index.

5. The data-driven rail transit risk prevention and control method according to claim 1, characterized in that, The knowledge graph for risk prevention and control in rail transit consists of a three-layer knowledge structure, with the risk evolution link as the intermediate layer and risk points and prevention and control as the upper and lower related layers. The middle layer uses "risk events" and "accident scenarios" as nodes and coupling relationships as edges to express how risk events develop and evolve to ultimately lead to accident scenarios in a "chain" or "network" transmission process, and uses risk identification attributes as the node attributes of "risk events". The upper layer uses "risk points" as nodes and is connected to the middle layer through a predicate relationship to describe the state manifestation of risk points evolving and escalating into risk events. The upper and lower levels within the layer are used as edges to express the hierarchical relationship between the components within the system. The lower-level related layer takes "measures" as the central node and "standard sources and prevention and control positions" as related nodes, and connects with the middle layer through predicate relationships to explain the four stages of "prevention-treatment-control-rescue", thereby implementing the standards and norms for risk prevention and control and proactive safety assurance of the rail transit operation system.

6. A data-driven rail transit risk prevention and control system, characterized in that, include: The first construction module is used to build the knowledge graph pattern layer of rail transit risk prevention and control based on the mapping rules of "application scenario-knowledge structure-concept, attribute and relation pattern". The construction of the knowledge graph model layer for rail transit risk prevention and control includes: constructing an application scenario U for the rail transit risk prevention and control knowledge graph, and decomposing the application scenario into multiple application sub-scenarios according to its application field and scope of use; establishing the mapping relationship and mapping function between the application scenario and the knowledge structure of the rail transit risk prevention and control knowledge graph, and constructing the knowledge structure of the rail transit risk prevention and control knowledge graph; parsing the constructed knowledge structure of the rail transit risk prevention and control knowledge graph layer by layer, defining concepts, concept attributes, and the relationship patterns between concepts and between concepts and attributes, and constructing a relation pattern deconstruction diagram; using ER diagrams to graphically represent concepts, concept attributes, and the relationship patterns between concepts and between concepts and attributes, and constructing the ontology model of the rail transit risk prevention and control knowledge graph; setting evaluation indicators for the ontology model of the rail transit risk prevention and control knowledge graph, evaluating, verifying, and optimizing the ontology model of the rail transit risk prevention and control knowledge graph, and constructing the knowledge graph model layer for rail transit risk prevention and control. The second construction module is used to construct a knowledge graph data layer for rail transit risk prevention and control based on the characteristics of unstructured data and combined with a data-driven knowledge extraction method for rail transit risk prevention and control. The third construction module is used to construct a knowledge graph for rail transit risk prevention and control by using graph database to standardize the storage of triplet structures and instance tags. The module determines the implementation rules for four rail transit risk prevention and control functions: risk profiling, implicit relationship mining, accident cause reasoning, and risk prevention and control auxiliary decision-making, based on the semantic link relationship of the rail transit risk prevention and control knowledge graph and graph database. Among these, establishing a mapping relationship between application scenarios and knowledge structures for the knowledge graph of rail transit risk prevention and control is crucial. and mapping function Construct a knowledge graph knowledge structure O for rail transit risk prevention and control, that is, In the formula, , Let U and O represent the nth constituent element, respectively. Indicates from arrive The mapping relationship; By analyzing the application scenarios of the knowledge graph for rail transit risk prevention and control, the research objects in the application scenarios are used as the basis for mapping the relationship between the knowledge structure and the research objects. ； In the formula, This represents the mapping relationship between the application scenario (ox) and the knowledge structure. Obtain, application scenarios , , The following mapping relationship , , and mapping results , , ,Right now, ； ； 。 7. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the data-driven rail transit risk prevention and control method as described in any one of claims 1-5.

8. An electronic device comprising a processor, a memory, and a computer program; wherein, The processor is connected to a memory, and a computer program is stored in the memory. The electronic device is characterized in that, when it is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions that implement the data-driven rail transit risk prevention and control method as described in any one of claims 1-5.

Images