Railway train equipment fault diagnosis method and equipment based on domain knowledge graph

Abstract

The invention relates to a railway train equipment fault diagnosis method and system based on a domain knowledge graph. The method comprises the following steps: acquiring and screening railway train equipment fault data; extracting a structured entity in the fault data of the railway train equipment, linking the structured entity and the defined relationship, and generating a triple; calculating the similarity between the structured entities, and performing short alignment updating on the triple according to the similarity between the structured entities; the updated triple is stored in a graph database, and construction of the knowledge graph is completed; and performing fault diagnosis by combining retrieval of a graph database and reasoning of a knowledge graph, and determining a fault reason. According to the technical scheme, railway train equipment fault diagnosis based on the domain knowledge graph is achieved, the problem that historical text data cannot be fully utilized in traditional fault diagnosis can be solved, and the problems that fault analysis association is weak and efficiency is low are solved by combining fuzzy query and knowledge graph reasoning.

Description

Technical Field

[0001] The present application relates to the technical field of railway fault diagnosis, and in particular to a railway operating equipment fault diagnosis method and equipment based on a domain knowledge graph. Background Art

[0002] With the rapid development of my country's railway system, the scale of its network has continued to expand, which has not only greatly promoted the rapid growth of the economy, but also improved people's travel efficiency. However, at the same time, the complexity of the railway system and the requirements for safe operation are also increasing. As a key link in the safe and stable operation of the railway system, the state of the driving equipment is directly related to the safety and efficiency of the train operation. Once a driving equipment failure occurs, even if it does not constitute a railway traffic accident, it may cause train delays, endanger driving safety or affect the normal operation of the train. Therefore, the fault diagnosis research of driving equipment is particularly important.

[0003] Traditional fault diagnosis methods often rely on the experience and judgment of experts. Although this method can solve some problems to a certain extent, its efficiency and accuracy are often difficult to guarantee when faced with massive historical fault text data. In addition, traditional fault diagnosis methods cannot effectively represent the relationship between various fault factors, making fault analysis complicated and difficult. Summary of the invention

[0004] In order to at least overcome the complex and difficult problems of railway operating equipment fault diagnosis in related technologies in a certain extent, the present application provides a railway operating equipment fault diagnosis method and equipment based on a domain knowledge graph.

[0005] The scheme of this application is as follows:

[0006] According to a first aspect of an embodiment of the present application, a method for diagnosing railway vehicle equipment faults based on a domain knowledge graph is provided, comprising:

[0007] Obtain and filter railway running equipment failure data;

[0008] Extract structured entities from railway equipment fault data, link structured entities and defined relationships, and generate triples;

[0009] Calculate the similarity between structured entities and perform short alignment updates on triples based on the similarity between structured entities;

[0010] Store the updated triples in the graph database to complete the construction of the knowledge graph;

[0011] Combine the retrieval of graph database and the reasoning of knowledge graph to perform fault diagnosis and determine the cause of the fault.

[0012] Preferably, obtaining and screening railway vehicle equipment fault data includes:

[0013] According to the pre-defined fault classification, filter the railway operating equipment fault data that meets the classification.

[0014] Preferably, the structured entities in the railway vehicle equipment fault data include: fault phenomenon, fault cause, fault category and fault location;

[0015] The relationships defined include: occurs in, causes, and belongs to.

[0016] Preferably, the method further comprises:

[0017] Remove redundant information from railway equipment failure data based on the constructed rule base;

[0018] The redundant information at least includes: processing flow, failure impact and working principle.

[0019] Preferably, structured entities in railway vehicle equipment fault data are extracted, and structured entities and defined relationships are linked to generate triples, including:

[0020] Use the BIO annotation method to annotate each piece of data and obtain three types of labels;

[0021] The BERT pre-trained model is used as the embedding layer. The forward and backward sequence features of the labeled data are extracted through the bidirectional LSTM. The score of each word vector corresponding to the label is calculated and sent to the CRF layer. The score of each word vector corresponding to the label and the transition probability between labels are calculated through the CRF layer, and the label sequence with the highest score is output as the final extraction result.

[0022] A rule-based method is used to extract relations from fault information, and the extracted structured entities and defined relations are concatenated to generate triples.

[0023] Preferably, a short alignment update is performed on the triples according to the similarity between the structured entities, including:

[0024] When the similarity between two structured entities is higher than a preset threshold, the structured entity with the longer text is replaced with the structured entity with the shorter text.

[0025] Preferably, the fault diagnosis is performed by combining the retrieval of the graph database and the reasoning of the knowledge graph to determine the cause of the fault, including:

[0026] The fault frequency and impact degree are stored as attributes in the knowledge graph, and fuzzy query is performed on the knowledge graph to obtain the first analysis result of the fault cause;

[0027] Based on cosine similarity matching and knowledge graph reasoning, the second analysis result of the fault cause is obtained.

[0028] According to a second aspect of an embodiment of the present application, a railway operating equipment fault diagnosis system based on a domain knowledge graph is provided, comprising:

[0029] Processor and memory;

[0030] The processor and the memory are connected via a communication bus:

[0031] Wherein, the processor is used to call and execute the program stored in the memory;

[0032] The memory is used to store a program, and the program is at least used to execute a railway operating equipment fault diagnosis method based on a domain knowledge graph as described in any one of the above items.

[0033] The technical solution provided by this application may have the following beneficial effects:

[0034] The railway operating equipment fault diagnosis method based on domain knowledge graph in the present application includes: obtaining and screening railway operating equipment fault data; extracting structured entities in the railway operating equipment fault data, linking structured entities and defined relationships, and generating triples; calculating the similarity between structured entities, and performing short alignment update on triples according to the similarity between structured entities; storing the updated triples in a graph database to complete the construction of the knowledge graph; combining the retrieval of the graph database and the reasoning of the knowledge graph to perform fault diagnosis and determine the cause of the fault.

[0035] In this technical solution, the coverage and usage of knowledge are focused on the specific field of railway operating equipment fault diagnosis through the domain knowledge graph, which has higher professionalism and accuracy. The structure of the knowledge graph is used to construct and display the association between knowledge through nodes (representing entities) and edges (representing the relationship between entities). It not only has powerful expression ability and modeling flexibility, but also can process massive amounts of data and discover patterns and relationships hidden in the data. Therefore, the use of knowledge graph technology to assist railway workers in fault analysis and processing is more accurate and efficient.

[0036] This technical solution realizes railway operating equipment fault diagnosis based on domain knowledge graph, which can solve the problem that traditional fault diagnosis cannot fully utilize historical text data. It combines fuzzy query and knowledge graph reasoning to solve the problem of weak correlation and low efficiency in fault analysis.

[0037] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0039] Figure 1 It is a flowchart of a method for diagnosing railway vehicle equipment faults based on a domain knowledge graph provided by an embodiment of the present application;

[0040] Figure 2 This is a schematic diagram of original data of railway vehicle equipment failure data provided by an embodiment of the present application;

[0041] Figure 3 This is a schematic diagram of a rule base constructed according to an embodiment of the present application;

[0042] Figure 4 A schematic diagram of a process for extracting structured entities from railway vehicle equipment fault data, linking structured entities and defined relationships, and generating triples is provided by an embodiment of the present application;

[0043] Figure 5 This is a schematic diagram of an entity alignment process provided by an embodiment of the present application;

[0044] Figure 6 It is a schematic diagram of a fault diagnosis and analysis process provided by an embodiment of the present application;

[0045] Figure 7 It is a flow chart of a railway operating equipment fault diagnosis system based on a domain knowledge graph provided by an embodiment of the present application.

[0046] Reference numerals: processor-21; memory-22. DETAILED DESCRIPTION

[0047] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0048] Embodiment 1

[0049] Figure 1 is a flow chart of a method for diagnosing railway vehicle equipment faults based on a domain knowledge graph provided by an embodiment of the present application, with reference to Figure 1 , a railway driving equipment fault diagnosis method based on domain knowledge graph, comprising:

[0050] S11: Acquire and filter railway running equipment failure data;

[0051] S12: extracting structured entities from railway running equipment fault data, linking structured entities and defined relationships, and generating triples;

[0052] S13: Calculate the similarity between structured entities, and perform short alignment update on the triples according to the similarity between structured entities;

[0053] S14: Store the updated triples in the graph database to complete the construction of the knowledge graph;

[0054] S15: Combine the retrieval of the graph database and the reasoning of the knowledge graph to perform fault diagnosis and determine the cause of the fault.

[0055] It should be noted that obtaining and screening railway operating equipment failure data includes:

[0056] According to the pre-defined fault classification, filter the railway operating equipment fault data that meets the classification.

[0057] In specific practice, the pre-defined fault categories include: G1: locomotive failure; G2: vehicle failure; G3: EMU failure; G4: railway ferry equipment failure; G5: self-propelled special equipment failure; G6: line, bridge and tunnel equipment failure; G7: signal equipment failure; G8: communication equipment failure; G9: power supply equipment failure; G10: water supply equipment failure; G11: information system equipment failure; G12: tail device failure; G13: monitoring and surveillance equipment failure; G14: damage to line safety protection equipment and facilities; G15: water damage, landslide, falling rocks, fallen trees; G16 other equipment failures.

[0058] The structured entities in the railway running equipment fault data include: fault phenomenon, fault cause, fault category and fault location;

[0059] The relationships defined include: occurs in, causes, and belongs to.

[0060] Fault symptoms and fault causes need to be extracted from unstructured data overview and causes.

[0061] It should be noted that the method also includes:

[0062] Remove redundant information from railway equipment failure data based on the constructed rule base;

[0063] Redundant information at least includes: processing flow, failure impact and working principle.

[0064] The original data of railway running equipment failure data is as follows Figure 2As shown, in the historical data details, the fault type exists in the form of a classification label, which is used to identify the type of fault. The fault location describes the specific location or component where the fault occurs. Both are structured data. The fault overview is a detailed description of the fault situation, including the time, location, phenomenon, etc. of the fault. The fault cause is used to explain the cause of the fault. The contents of the fault overview and the fault cause are unstructured data, and both are long text data.

[0065] The structured entity recognition is performed on the railway running equipment fault data text. The overview and cause fields contain not only the required entity types, but also redundant information such as processing flow, fault impact, working principle, etc. Removing this redundant information in advance during the extraction process helps reduce the loss of the model when obtaining long text context information. The content after removing "notification" or "arrangement" is summarized, and the rule library constructed is as follows Figure 3 As shown, L l With L r Represents the left and right boundary words of the definition, such as...L l TL r The contents of … will be retained. Figure 2 For example, the original data is summarized as "At xx:xx on xx / xx / xx, xx line xx yard of xx bureau reported that a red light flashed on xx track in the station. The engineering, electrical and public security departments were notified to handle it. At xx:xx, the engineering and electrical departments completed the handling and normal traffic was restored. Traffic was not affected." The retained result is "At xx:xx on xx / xx / xx, xx line xx yard of xx bureau reported that a red light flashed on xx track in the station."

[0066] The original reason data is "The cause of the failure of the red light band is that the two 1.8m power lead wire plug nails at the receiving end broke at the same time at the waist of the rail." Then the retained result is "The two 1.8m power lead wire plug nails at the receiving end broke at the same time at the waist of the rail."

[0067] Extract structured entities from railway equipment fault data, link structured entities and defined relationships, generate triples, and refer to Figure 4 ,include:

[0068] S121: Use the BIO annotation method to annotate each piece of data and obtain three types of labels;

[0069] S122: Use the BERT pre-trained model as the embedding layer, extract the forward and backward sequence features of the labeled data through the bidirectional LSTM, calculate the score of each word vector corresponding to the label and send it to the CRF layer, calculate the score of each word vector corresponding to the label and the transition probability between labels through the CRF layer, and output the label sequence with the highest score as the final extraction result;

[0070] S123: Use a rule-based method to extract relations from fault information, concatenate the extracted structured entities and defined relations, and generate triples.

[0071] It should be noted that the label types in the data annotated using the BIO annotation method include B, I, and O. Among them, B represents the beginning of an entity, I represents the middle or end of an entity, and O represents a non-entity.

[0072] Perform short alignment updates on triples based on the similarity between structured entities, including:

[0073] When the similarity between two structured entities is higher than a preset threshold, the structured entity with the longer text is replaced with the structured entity with the shorter text.

[0074] The similarity scores between entities are calculated by combining cosine similarity, Jaccard similarity, and edit distance. Entities with shorter lengths greater than the set threshold are replaced to complete entity alignment and update triples. With the help of the Py2neo module, knowledge is stored in the Neo4j graph database to complete the construction of the knowledge graph.

[0075] It should be noted that the extracted railway accident fault descriptions have similar semantics but different expressions, such as "no indication of switch positioning", "no indication of switch reverse position", "no indication of switch from positioning to reverse position", "no indication of switch", etc. By pre-setting the similarity threshold, combining cosine similarity, Jaccard similarity, and edit distance to calculate the similarity score between entities, reconcile the disadvantages of various similarity calculations, if the score is higher than the similarity threshold, the shorter entity is retained, and the entity similar to it is replaced by the entity and processed in combination with manual inspection, such as Figure 5 As a graph database among non-relational databases, Neo4j displays and stores data and the relationships between data in the form of nodes and edges. This storage method enables Neo4j to efficiently handle complex network structures and associated queries, and is particularly suitable for application scenarios that require the expression and analysis of complex relationships between entities.

[0076] Combine the retrieval of the graph database and the reasoning of the knowledge graph to perform fault diagnosis and determine the cause of the fault, including:

[0077] The fault frequency and impact degree are stored as attributes in the knowledge graph, and fuzzy query is performed on the knowledge graph to obtain the first analysis result of the fault cause;

[0078] Based on cosine similarity matching and knowledge graph reasoning, the second analysis result of the fault cause is obtained.

[0079] In specific practice, historical occurrence data is collected and the fault frequency and impact degree are stored in the knowledge graph as attributes of the relationship cause, such as Figure 6 As shown in the figure, the possible fault causes can be intelligently sorted by using the knowledge graph for fuzzy retrieval. For entities that are not found, the cosine similarity is used to calculate the description and combine it with the entity relationship vector representation trained by TransH to obtain the reasoning result. In the TransH model, the representation of the relationship r is not limited to a translation vector r, but an additional hyperplane normal vector w is introduced. r For a triple (h, r, t), i.e. (head entity, relation, tail entity), the model first projects the head entity vector h and the tail entity vector t onto the hyperplane defined by the relation r to obtain h ⊥ and t ⊥ Then, on the projected hyperplane, minimize h ⊥ +r and t ⊥ In order to enhance sample diversity, improve model performance and generalization ability, negative sampling technology is added. For triples (h, r, t), let m be the average number of tail entities corresponding to each head entity, and n be the average number of head entities corresponding to each tail entity. Replace the head entity with a probability of m / (m+n) and replace the tail entity with a probability of n / (m+n). By continuously minimizing the distance difference of the correct triples on the relationship plane and minimizing the opposite number of the distance difference of the wrong triples, the optimal model is obtained. Figure 6 In the process shown, when using knowledge graph reasoning to solve the problem of asking the cause of a known fault phenomenon, candidate entities are regarded as invisible reference knowledge to help analysis because they have the same category or occur in the same location.

[0080] In this technical solution, the coverage and usage of knowledge are focused on the specific field of railway operating equipment fault diagnosis through the domain knowledge graph, which has higher professionalism and accuracy. The structure of the knowledge graph is used to construct and display the association between knowledge through nodes (representing entities) and edges (representing the relationship between entities). It not only has powerful expression ability and modeling flexibility, but also can process massive amounts of data and discover patterns and relationships hidden in the data. Therefore, the use of knowledge graph technology to assist railway workers in fault analysis and processing is more accurate and efficient.

[0081] This technical solution realizes railway operating equipment fault diagnosis based on domain knowledge graph, which can solve the problem that traditional fault diagnosis cannot fully utilize historical text data. It combines fuzzy query and knowledge graph reasoning to solve the problem of weak correlation and low efficiency in fault analysis.

[0082] Embodiment 2

[0083] Figure 7This is a flow chart of a railway vehicle equipment fault diagnosis system based on a domain knowledge graph provided by an embodiment of the present application, referring to Figure 7 , a railway driving equipment fault diagnosis system based on domain knowledge graph, including:

[0084] Processor 21 and memory 22;

[0085] The processor 21 and the memory 22 are connected via a communication bus:

[0086] The processor 21 is used to call and execute the program stored in the memory 22;

[0087] The memory 22 is used to store programs, and the programs are used to execute at least one of the railway operating equipment fault diagnosis methods based on the domain knowledge graph in the above embodiments.

[0088] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0089] It should be noted that, in the description of this application, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of this application, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0090] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0091] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or a combination thereof: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0092] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0093] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0094] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0095] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0096] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A railway vehicle equipment fault diagnosis method based on domain knowledge graph, characterized in that: include: Acquire and filter railway running equipment failure data; Extract structured entities from railway equipment fault data, link structured entities and defined relationships, and generate triples; Calculate the similarity between structured entities and perform short alignment updates on triples based on the similarity between structured entities; Store the updated triples in the graph database to complete the construction of the knowledge graph; Combine the retrieval of graph database and the reasoning of knowledge graph to perform fault diagnosis and determine the cause of the fault.

2. The method according to claim 1, characterized in that: Obtain and filter railway equipment failure data, including: According to the pre-defined fault classification, filter the railway operating equipment fault data that meets the classification.

3. The method according to claim 1, characterized in that The structured entities in the railway running equipment fault data include: fault phenomenon, fault cause, fault category and fault location; The relationships defined include: occurs in, causes, and belongs to.

4. The method according to claim 1, characterized in that: The method further comprises: Remove redundant information from railway equipment failure data based on the constructed rule base; The redundant information includes at least: processing flow, failure impact and working principle.

5. The method according to claim 1, characterized in that Extract structured entities from railway equipment fault data, link structured entities and defined relationships, and generate triples, including: Use the BIO annotation method to annotate each piece of data and obtain three types of labels; The BERT pre-trained model is used as the embedding layer. The forward and backward sequence features of the labeled data are extracted through the bidirectional LSTM. The score of each word vector corresponding to the label is calculated and sent to the CRF layer. The score of each word vector corresponding to the label and the transition probability between labels are calculated through the CRF layer, and the label sequence with the highest score is output as the final extraction result. A rule-based method is used to extract relations from fault information, and the extracted structured entities and defined relations are concatenated to generate triples.

6. The method according to claim 1, characterized in that Perform short alignment updates on triples based on the similarity between structured entities, including: When the similarity between two structured entities is higher than a preset threshold, the structured entity with the longer text is replaced with the structured entity with the shorter text.

7. The method according to claim 1, characterized in that Combine the retrieval of the graph database and the reasoning of the knowledge graph to perform fault diagnosis and determine the cause of the fault, including: The fault frequency and impact degree are stored as attributes in the knowledge graph, and fuzzy query is performed on the knowledge graph to obtain the first analysis result of the fault cause; Based on cosine similarity matching and knowledge graph reasoning, the second analysis result of the fault cause is obtained.

8. A railway vehicle equipment fault diagnosis system based on domain knowledge graph, characterized in that: include: Processor and memory; The processor and the memory are connected via a communication bus: Wherein, the processor is used to call and execute the program stored in the memory; The memory is used to store a program, and the program is used at least to execute a railway operating equipment fault diagnosis method based on a domain knowledge graph as described in any one of claims 1-7.

Images