Equipment diagnosis method and device and computer equipment

Abstract

The invention relates to an equipment diagnosis method and device and computer equipment. Comprising the steps that a target entity in a user query request is determined through a large model, diagnosis problems concerned by a user can be accurately recognized, multi-hop reasoning is carried out in a knowledge graph based on the target entity, associated knowledge and information can be deeply mined, and multiple initial reasoning paths are obtained. Probability analysis is carried out on the multiple initial reasoning paths according to the equipment operation data, the optimal target reasoning path is screened out, the target reasoning path can be accurately processed through a large model, and a diagnosis result is obtained. By adopting the method, more comprehensive and accurate equipment fault diagnosis can be carried out.

Description

Technical Field

[0001] The present application relates to the field of industrial control technology, and in particular to an equipment diagnosis method, device and computer equipment. Background Art

[0002] The stable operation of equipment plays a vital role in industrial production. For example, various industrial equipment in thermal power plants will inevitably experience various faults during long-term operation due to mechanical wear, improper operation, environmental factors, etc.

[0003] The current fault detection and diagnosis technology has made significant progress, such as rule-based expert systems, which perform fault diagnosis through preset if-then rules and continuously maintain the rules. However, the existing rule base is difficult to cover all industrial fault scenarios, and requires a lot of manual maintenance of the library, which often has limitations in actual application scenarios. Summary of the invention

[0004] Based on this, the purpose of this application is to provide a more comprehensive and accurate equipment diagnosis method, device and computer equipment to solve the technical problems mentioned in the above background technology.

[0005] In a first aspect, the present application provides a device diagnosis method, including:

[0006] Obtaining a user query request and device operation data, and determining a target entity in the user query request through a large model;

[0007] Based on the target entity, multi-hop reasoning is performed in the knowledge graph to obtain multiple initial reasoning paths; the knowledge graph includes nodes representing entities and edges representing relationships between entities;

[0008] Performing probability analysis on the multiple initial reasoning paths according to the device operation data to screen out a target reasoning path;

[0009] The target reasoning path is processed by a large model to obtain a diagnosis result.

[0010] In one embodiment, the process of constructing a knowledge graph includes: acquiring multi-source data and a preset first prompt template; processing the multi-source data and the first prompt template through a large model to obtain an extraction result; the extraction result includes multiple original entities and original relationships identified from the multi-source data; and constructing a graph of the extraction result to obtain a knowledge graph.

[0011] In one embodiment, a graph is constructed for the extraction results to obtain a knowledge graph, including: standardizing the extraction results to obtain triple data constituting an initial knowledge graph; constructing a second prompt template based on the triple data; the second prompt template includes the known information of the device and the relevant connections between each known information of the device; and updating the initial knowledge graph through the large model and the second prompt template to obtain a knowledge graph.

[0012] In one embodiment, the extraction results are standardized to obtain triple data constituting an initial knowledge graph, including: preprocessing the original entities and original relations in the extraction results; the original entities include entity subjects and entity objects; the preprocessing includes at least establishing a term dictionary, unifying synonyms and polysemous terms; and converting the preprocessed entity subjects, original relations and entity objects into triple data according to the triple structure.

[0013] In one embodiment, multi-hop reasoning is performed in a knowledge graph based on a target entity to obtain multiple initial reasoning paths, including: traversing from the node where the target entity is located in the knowledge graph as a starting point to determine multiple node information; the node information includes at least relationship strength and path length; when traversing to a special node, identifying the abnormal pattern of the special node through an anomaly detection algorithm; performing path scoring based on multiple node information and abnormal patterns to obtain multiple initial reasoning paths.

[0014] In one embodiment, probability analysis is performed on the multiple initial reasoning paths based on the equipment operation data to screen out a target reasoning path, including: integrating the initial reasoning paths and the equipment operation data through a Bayesian network to obtain multiple fault causes; constructing a decision tree associated with the equipment operation data; performing posterior probability processing through the decision tree and the Bayesian network, and taking the initial reasoning path corresponding to the fault cause whose probability exceeds a preset value as the target reasoning path.

[0015] In one embodiment, the above method also includes: obtaining optimization source information; the optimization source information at least includes user feedback, actual processing results of the device, diagnosis frequency of entities and relationships, information timeliness or expert scores; based on the graph attention network and the optimization source information, optimizing the weights between entities and relationships in the knowledge graph.

[0016] In one embodiment, the above method is applied in an intelligent agent cluster, which includes multiple intelligent agent systems that share information; the above method also includes: determining the target type of the target device corresponding to the device operation data, and determining the target intelligent agent system from multiple intelligent agent systems based on the target type; through the target intelligent agent system, the process of determining the target entity in the user query request through a large model is implemented until a diagnosis result is obtained.

[0017] In a second aspect, the present application also provides a device diagnostic apparatus, including:

[0018] A data acquisition module, used to acquire user query requests and device operation data, and determine the target entity in the user query request through a large model;

[0019] A path reasoning module, used to perform multi-hop reasoning in a knowledge graph based on the target entity to obtain multiple initial reasoning paths; the knowledge graph includes nodes representing entities and edges representing relationships connecting entities;

[0020] The result determination module is used to perform probability analysis on the multiple initial reasoning paths according to the equipment operation data, and screen out the target reasoning path; and process the target reasoning path through the large model to obtain the diagnosis result.

[0021] In a third aspect, the present application further provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned device diagnosis method when executing the computer program.

[0022] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned device diagnosis method are implemented.

[0023] The above-mentioned equipment diagnosis method, device, computer equipment and readable storage medium can accurately identify the diagnostic problems that users are concerned about by determining the target entity in the user query request through the large model, and can deeply mine the related knowledge and information based on the target entity and obtain multiple initial reasoning paths by performing multi-hop reasoning in the knowledge graph. Then, the multiple initial reasoning paths are probability analyzed according to the equipment operation data to screen out the optimal target reasoning path. In this way, the target reasoning path can be accurately processed through the large model to obtain the diagnosis result. Therefore, this solution integrates multi-source data such as user query requests and equipment operation data. Through the semantic understanding ability of the large model and the reasoning ability of the knowledge graph, it can accurately screen out the optimal target reasoning path, and realize more comprehensive, more accurate and traceable fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 An application environment diagram of a device diagnosis method in an embodiment;

[0025] Figure 2 A schematic diagram of a flow chart of a device diagnosis method in one embodiment;

[0026] Figure 3 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0028] The device diagnosis method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, and through the interaction between the terminal 102 and the server 104. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The terminal 102 can obtain user query requests and device operation data. The server 104 is used to determine the target entity in the user query request through a large model, and perform multi-hop reasoning in the knowledge graph based on the target entity to obtain multiple initial reasoning paths. The server 104 is also used to perform probability analysis on multiple initial reasoning paths according to the device operation data, screen out the target reasoning path, and process the target reasoning path through the large model to obtain a diagnosis result, so that the terminal 102 displays the diagnosis result. Among them, the terminal 102 can be, but is not limited to, various personal computers, smart phones, tablet computers, drone equipment, smart vehicle-mounted devices, portable wearable devices, etc. Server 104 is implemented by an independent server or a server cluster composed of multiple servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0029] In one embodiment, Figure 2 As shown, a device diagnosis method is provided, which is applied to Figure 1 The computer equipment including the server and the terminal is used as an example to illustrate, including the following steps:

[0030] Step 202, obtaining the user query request and the device operation data, and determining the target entity in the user query request through the big model.

[0031] Among them, the big model is the abbreviation of Large Language Model, which is a large-scale neural network language model. Through pre-training, it has acquired powerful language understanding and generation capabilities on massive text data. The big model can generate realistic text, understand complex language, answer questions, etc.

[0032] Specifically, the computer device obtains the device operation data of the target device in real time, and implements the voice recognition function through the interactive interface of the natural language interactive interface to convert the input voice of the user's query request into text, and then uses the large model to perform intent recognition and entity extraction on the user input. Intent recognition is mainly used to understand the user's purpose or request type, such as determining whether the user wants to query the device status, report a fault, or request maintenance guidance; the extracted target entity includes the device name, component name, parameter index, and fault type that the user is helped to identify in the interactive interface based on the request type.

[0033] When the target device is a steam turbine, taking the abnormal vibration fault of the steam turbine as an example, the operation and maintenance personnel will use the observed abnormal phenomenon as a user query request for voice input: "No. 1 steam turbine vibration abnormality". The computer equipment converts the voice into text and determines the intent recognition as "report fault" through the big model. The extracted target entities are "No. 1 steam turbine" and "abnormal vibration".

[0034] Step 204, perform multi-hop reasoning in the knowledge graph based on the target entity to obtain multiple initial reasoning paths.

[0035] Among them, the knowledge graph includes nodes that represent entities and edges that represent the relationships between entities. The knowledge graph is also a structured knowledge base that can provide structured information for search, recommendation, question and answer, etc. Multi-hop reasoning represents the process of starting from an entity and synthesizing information through multiple steps or relationship jumps to obtain iterative and multi-level conclusions.

[0036] Specifically, multi-hop reasoning starts with the symptoms described by the user, such as "abnormal vibration". The computer device locates the node where the target entity is located in the knowledge graph, and uses a graph traversal algorithm such as depth-first search or breadth-first search to explore possible causes of the abnormal vibration, forming multiple initial reasoning paths, each of which has an initial confidence level that decreases as the path lengthens.

[0037] Step 206, perform probability analysis on multiple initial reasoning paths according to the equipment operation data, screen out the target reasoning path, process the target reasoning path through the large model, and obtain the diagnosis result.

[0038] Among them, equipment operation data such as vibration spectrum, bearing temperature curve, etc.

[0039] Specifically, the computer equipment uses a Bayesian network to integrate multiple reasoning paths and equipment operation data to perform a posteriori probability analysis to obtain the most likely cause of the fault. The computer equipment then uses the generation capability of the large model to process the target reasoning path of the fault cause and output the diagnosis results and treatment suggestions in natural language.

[0040] For example, through knowledge graph reasoning and data analysis, the possible cause of the fault is determined to be "rotor imbalance". Multiple initial reasoning paths of processing suggestions related to "rotor imbalance" are queried in the knowledge graph, and combined with the current equipment operation data, the large model generates natural language diagnosis results and processing suggestions, such as "The diagnosis results show that there may be a rotor imbalance problem in turbine No. 1. It is recommended to perform the following operations immediately: 1. Check whether the blades are damaged or fouled; 2. Perform on-site dynamic balancing tests; 3. If conditions permit, consider reducing the load operation of the unit. At the same time, please pay close attention to the vibration trend. If the vibration continues to worsen, you may need to consider shutting down for maintenance."

[0041] In the above-mentioned equipment diagnosis method, the target entity in the user query request is determined by the large model, which can accurately identify the diagnostic problem that the user is concerned about. Based on the target entity, multi-hop reasoning in the knowledge graph can deeply mine the related knowledge and information to obtain multiple initial reasoning paths. Then, the multiple initial reasoning paths are probability analyzed according to the equipment operation data to screen out the optimal target reasoning path. In this way, the target reasoning path can be accurately processed through the large model to obtain the diagnosis result. Therefore, this solution integrates multi-source data such as user query requests and equipment operation data. Through the semantic understanding ability of the large model and the reasoning ability of the knowledge graph, it can accurately screen out the optimal target reasoning path, and realize more comprehensive, more accurate and traceable fault diagnosis.

[0042] In addition, the natural language interactive interface greatly improves the ease of use of this solution and reduces the requirements for user expertise. Operation and maintenance personnel can interact through voice or text to obtain diagnostic results and processing suggestions, allowing inexperienced users to quickly and accurately diagnose and handle faults.

[0043] In one embodiment, the process of constructing a knowledge graph includes: acquiring multi-source data and a preset first prompt template; processing the multi-source data and the first prompt template through a large model to obtain extraction results; and constructing a graph of the extraction results to obtain a knowledge graph.

[0044] The extraction results include multiple original entities and original relationships identified from multi-source data.

[0045] Specifically, the computer equipment collects and preprocesses multi-source data, such as collecting multi-source data such as thermal power plant equipment operation data, equipment manuals, maintenance manuals, fault reports, etc. Then, the structured multi-source data is cleaned and standardized, and the unstructured text is converted into a processable format using, for example, optical character recognition technology. In addition, the first prompt template for a specific field is pre-designed in the computer equipment, such as "Please identify the entities related to thermal power plant field equipment in the following text, including equipment name, component name, parameter index, fault type, etc., and indicate the relationship between them. Text: [Enter text]".

[0046] The computer device also includes a pre-trained large model, which can be used to perform tasks such as named entity recognition and relationship extraction. By inputting the constructed first prompt template and the pre-processed multi-source data into the large model, the large model extracts information based on the pre-trained knowledge and the requirements in the first prompt template, and preliminarily outputs the recognized entities and relationships. Then, the output of the large model is parsed using regular expressions, and the entity and relationship information is extracted to form a standardized output format to obtain the original entities and original relationships, so that a knowledge graph can be constructed based on the original entities and original relationships.

[0047] For example, the multi-source data in the input text is: "Unit 2 steam turbine generator set was found to be abnormal during routine inspection. The operator reported hearing abnormal noise and observed that the rotor vibration increased significantly. Further inspection found that the turbine bearing temperature had exceeded the normal operating range and reached 95°C. Preliminary judgment may be due to rotor imbalance or bearing wear. It is recommended to conduct a detailed inspection immediately and consider shutting down for maintenance if necessary to prevent the fault from expanding." Therefore, the extraction results obtained by information extraction are: (Unit 2 steam turbine generator set, found, abnormal situation), (abnormal noise, yes, symptom), (rotor vibration, significantly increased, no), (turbine bearing temperature, exceeded, normal operating range).

[0048] In this embodiment, by automatically integrating multi-source data and processing it using a large model and the first prompt template, the comprehensiveness and accuracy of the knowledge graph construction are improved. The structured representation of the knowledge graph makes knowledge easier to discover and retrieve, and supports complex query and reasoning operations. The application of large models improves the intelligence level of knowledge graph construction, allowing the graph to understand and process complex data relationships more deeply, thereby providing comprehensive, efficient and intelligent data support for various application scenarios. Therefore, combining multi-source data fusion, natural language processing with large model reasoning and knowledge graphs significantly improves the accuracy, coverage and interpretability of subsequent fault diagnosis.

[0049] In one embodiment, a computer device uses a graph database to store and manage a knowledge graph.

[0050] In one embodiment, a graph is constructed for the extraction results to obtain a knowledge graph, including: standardizing the extraction results to obtain triple data constituting an initial knowledge graph; constructing a second prompt template based on the triple data; and updating the initial knowledge graph through the large model and the second prompt template to obtain a knowledge graph.

[0051] The second prompt template includes the known information of the device and the relevant relationship between each known information of the device.

[0052] Specifically, the computer device converts the format of the extracted original entities and original relationships according to the preset triple structure to obtain triple data that constitutes the initial knowledge graph. Then it is necessary to determine whether there are missing relationships in the initial knowledge graph and perform prediction, completion and update, such as determining whether there are missing relationships through statistical analysis, rule checking or expert review.

[0053] After determining the missing, the computer device determines the known information of the device based on the triple data, and infers the possible related connections between the known information of the device. For example, the known information of the device is: the steam turbine is the core equipment of the generator set, the rotor is the key component of the steam turbine, the bearing supports the operation of the rotor, and the increased vibration is a common symptom of equipment failure. The possible related connections inferred are: What is the possible relationship between the steam turbine and the increased vibration? What is the possible relationship between the bearing and the increased vibration? What is the possible relationship between the rotor and the increased vibration?

[0054] Then, the computer device inputs the constructed second prompt template into the large model, obtains the output results and analyzes the output of the large model to extract possible new relationships. For example, the large model outputs: 1. Steam turbine and increased vibration: Steam turbine may experience increased vibration, and increased vibration may be a symptom of turbine failure; 2. Bearing and increased vibration: Bearing failure may cause increased vibration, and increased vibration may indicate a bearing problem; 3. Rotor and increased vibration: Rotor imbalance may cause increased vibration, and increased vibration may reflect abnormal rotor status.

[0055] Finally, the computer equipment standardizes the new relationships output by the large model, such as: (steam turbine, possible fault symptoms, increased vibration), (bearing fault, may cause, increased vibration), (rotor imbalance, may cause, increased vibration). The computer equipment will use the rule engine or expert knowledge to verify the rationality of these new relationships, add the verified new relationships to the initial knowledge graph, and mark their source and confidence to obtain the knowledge graph.

[0056] In this embodiment, by standardizing the extraction results into triple data and using this to construct the second prompt template, and then using the large model to dynamically update the initial knowledge graph, the accuracy and consistency of the knowledge graph can be significantly improved, enabling it to quickly respond to new information and maintain the timeliness and relevance of knowledge.

[0057] In one embodiment, an incremental update mechanism is also designed in the computer device to continuously learn from new data and update the knowledge graph. The incremental update mechanism of the knowledge graph is a dynamic and continuous process that aims to maintain the timeliness and accuracy of the graph. This mechanism receives new knowledge provided by experts through a manual input interface, and then processes this information using modules such as change detection, knowledge extraction, consistency checking, and knowledge fusion. In practical applications, for example, if a new maintenance report is manually uploaded, the mechanism converts it into structured knowledge and integrates it with the existing knowledge graph. Through version control and rollback mechanisms, the system can also manage update history to ensure the traceability and reliability of knowledge.

[0058] In one embodiment, the extraction results are standardized to obtain triple data constituting an initial knowledge graph, including: preprocessing the original entities and original relations in the extraction results; and converting the preprocessed entity subjects, original relations, and entity objects into triple data according to the triple structure.

[0059] The original entity includes entity subject and entity object; the preprocessing at least includes establishing a term dictionary, unifying synonyms and polysemous terms.

[0060] Specifically, the computer equipment will standardize and preprocess the extracted original entities and original relationships, such as unified naming, elimination of synonyms, etc., and then transform them according to the triple structure (subject, relationship, object). For the thermal power plant field, the computer equipment creates a standard terminology dictionary, including equipment name, component name, parameter index, fault type, etc. For example, equipment name: {steam turbine, boiler, generator...}, component name: {rotor, stator, bearing, blade...}, parameter index: {temperature, pressure, vibration, flow...}, fault type: {imbalance, axis deviation, crack, corrosion...}.

[0061] Then, use the large model to identify and unify synonyms or near-synonyms, such as "bearing heating" -> "bearing temperature rise", "vibration" -> "vibration". Then unify the naming format of the original entities and original relationships. For example, the equipment name uses the full name: "No. 1 turbine" instead of "1# unit", and the parameter indicators add units "temperature (℃)" and "pressure (MPa)". The relationship name uses verbs or noun phrases: "cause", "includes components", "is a symptom of the fault". And deal with terms with multiple meanings. For example, "shaft" may refer to "rotor shaft" or "drive shaft", and the specific reference to "pressure" may be "steam pressure" or "oil pressure" according to the context.

[0062] Finally, the computer device converts the preprocessed entity subject, original relationship and entity object into triple data, that is, into the form of (entity subject, original relationship, entity object). For example: (No. 2 steam turbine generator, fault symptoms, abnormal noise), (No. 2 steam turbine generator, fault symptoms, increased vibration), (No. 2 steam turbine generator, including components, rotor), (No. 2 steam turbine generator, including components, bearings), (bearings, abnormal parameters, increased temperature), (bearing temperature, value, 95°C).

[0063] In this embodiment, by preprocessing the original entities and original relationships and converting them into a standardized triple data structure, the standardization and readability of the data can be significantly improved, making the data easier to manage and operate, and facilitating subsequent information retrieval, data analysis, and knowledge graph construction.

[0064] In one embodiment, multi-hop reasoning is performed in the knowledge graph based on the target entity to obtain multiple initial reasoning paths, including: traversing from the node where the target entity is located in the knowledge graph as the starting point to determine multiple node information; when traversing to a special node, identifying the abnormal pattern of the special node through an anomaly detection algorithm; and scoring the path according to the multiple node information and abnormal patterns to obtain multiple initial reasoning paths.

[0065] The node information at least includes the relationship strength and the path length. The special nodes include nodes related to sensor data.

[0066] Specifically, in the process of traversing from the node where the target entity is located as the starting point, the computer device will read the node information such as the relationship strength and path length in the graph, and when exploring a special node such as "bearing temperature", it will obtain the sensor data related to the special node in real time. After the computer device pre-processes these sensor data, such as denoising and normalization, it uses an abnormality detection algorithm such as isolation forest or autoencoder to identify abnormal patterns of special nodes. The computer device considers the relationship strength, path length, abnormality of abnormal patterns and historical case similarity in the graph, and determines the corresponding different weight coefficients, so as to perform a comprehensive path score according to the scoring function, and then screen out multiple initial reasoning paths according to the score size.

[0067] For example, the scoring function is as follows: Score = w1*R+w2*(1 / L)+w3*A+w4*S, where R is the relationship strength, L is the path length, A is the abnormality of the abnormal pattern, S is the similarity of historical cases, and w1-w4 are the corresponding weight coefficients.

[0068] In this embodiment, by traversing the knowledge graph and using anomaly detection algorithms to identify abnormal patterns of special nodes, the accuracy and efficiency of anomaly detection can be improved, and the path scoring mechanism can score the reasoning path based on multiple node information and abnormal patterns, which provides a more detailed and quantitative basis for decision-making and improves the accuracy of the initial reasoning path determination.

[0069] In one embodiment, a probability analysis is performed on multiple initial reasoning paths based on the equipment operation data to screen out a target reasoning path, including: integrating each initial reasoning path and the equipment operation data through a Bayesian network to obtain multiple fault causes; constructing a decision tree associated with the equipment operation data; performing posterior probability processing through the decision tree and the Bayesian network, and taking the initial reasoning path corresponding to the fault cause whose probability exceeds a preset value as the target reasoning path.

[0070] Specifically, the computer device uses a Bayesian network to integrate multiple initial reasoning paths and real-time equipment operation data. Each reasoning path forms a conditional probability branch, and the equipment operation data is input into the network as observation evidence. The posterior probability is updated using variational inference or Markov chain Monte Carlo method to obtain the most likely cause of the fault. The computer device builds a decision tree to refine the diagnosis, selects the most discriminative equipment operation data, and uses historical cases for training and pruning. The probability distribution of the leaf nodes of the decision tree is then combined with the posterior probability of the Bayesian network to form the final ranking of multiple fault causes. Finally, the computer device extracts the initial reasoning path corresponding to the fault cause whose probability exceeds the preset value, and inputs it into the large model to generate a diagnostic report and processing suggestions in natural language.

[0071] In this embodiment, by constructing a decision tree associated with the equipment operation data, the system can make decisions based on real-time data, thereby improving the immediacy and response speed of decisions. By combining with the posterior probability processing of the Bayesian network, it is possible to evaluate the credibility of each fault cause and highlight high-probability fault causes, thereby providing more centralized and effective analysis.

[0072] In one embodiment, the above method also includes: obtaining optimization source information; optimizing the weights between entities and relationships in the knowledge graph based on the graph attention network and the optimization source information.

[0073] The optimization source information includes at least user feedback, actual processing results of the equipment, diagnosis frequency of entities and relationships, information timeliness or expert ratings. User feedback refers to the acceptance of the diagnosis results, actual application effects, etc.; actual processing of the equipment refers to the problem solving after the implementation of the diagnosis suggestions; the diagnosis frequency of entities and relationships refers to the frequency of occurrence of entities and relationships in successful diagnoses; information timeliness refers to the newness of the information; and expert ratings refer to the importance ratings of certain key relationships by domain experts.

[0074] Specifically, computer equipment also needs to conduct feedback learning and continuous optimization, such as recording user feedback and actual equipment processing results, and using reinforcement learning technology to continuously optimize the diagnostic model and knowledge graph. Based on the collected information from various optimization sources, the weights of entities and relationships in the knowledge graph are dynamically adjusted using the graph attention network (GAT). If the vibration amplitude is significantly reduced after the recommended operation is implemented, the computer device will use this positive result to strengthen the path in the knowledge graph. At the same time, through the comparative learning mechanism, the computer device will analyze the difference between this successful diagnosis and the failed diagnosis in similar situations before, thereby optimizing its decision-making process.

[0075] Among them, the graph attention network is a neural network model used to process graph structured data. It learns the weights between nodes through the attention mechanism to obtain the representation of the nodes.

[0076] In one embodiment, during the feedback learning and continuous optimization phase, the system not only records the direct feedback from the operation and maintenance personnel, but also analyzes the subsequent equipment operation data. For example, if the vibration amplitude is significantly reduced after the recommended operation is implemented, the system will use this positive result to strengthen the current diagnostic path. At the same time, through the comparative learning mechanism, the system will analyze the difference between this successful diagnosis and the failed diagnosis in similar situations before, thereby optimizing its decision-making process.

[0077] In one embodiment, the computer device can also evaluate the effect of the optimization of the knowledge graph through the following indicators. 1. Diagnostic accuracy: The optimized knowledge graph can improve the accuracy of fault diagnosis, which can be evaluated by comparing with manual diagnosis results, for example, the diagnostic accuracy is higher than 85%. 2. Response time: The optimization should reduce the reasoning steps required for diagnosis and shorten the response time, for example, the response time is less than 5s. 3. User satisfaction: Track the user's acceptance and feedback on the diagnosis results, for example, user satisfaction is more than 90%.

[0078] In the above embodiment, continuous autonomous learning and evolution from practice are achieved through multi-dimensional feedback collection, comparative learning, effect evaluation, etc. This all-round, closed-loop learning optimization mechanism can continuously improve its diagnostic capabilities, adapt to changing equipment conditions and newly emerging fault types, and truly realize intelligent equipment fault diagnosis.

[0079] In one embodiment, the specific process of optimizing the weights in the knowledge graph includes: 1. Graph structure representation: convert the knowledge graph into a graph structure that can be processed by the graph attention network. 2. Feature vector construction: construct an initial feature vector for each node in the graph structure, including node type (such as device, symptom, cause, etc.), historical weight, usage frequency, and last update timestamp. 3. Attention mechanism: For each node, the graph attention network calculates its attention coefficient with adjacent nodes. 4. Information aggregation: The new representation of the node is calculated by the weighted sum of its neighbors: 5. Multi-head attention: Use multiple independent attention mechanisms and then connect or average the results. 6. Feedback integration: Convert user feedback and actual processing results of the device into reward signals for adjusting attention weights. 7. Weight update: Use the back-propagation algorithm to update the parameters of the graph attention network to adjust the weights of entities and relationships. Therefore, through weight optimization, common faults and effective solutions are more accessible in the knowledge graph, while rare or outdated information is weakened.

[0080] In one embodiment, the above method also includes: determining the target type of the target device corresponding to the device operation data, and determining the target intelligent agent system from multiple intelligent agent systems based on the target type; through the target intelligent agent system, realizing the process of determining the target entity in the user query request through a large model until a diagnosis result is obtained.

[0081] Among them, this solution is applied in an intelligent agent cluster, which includes multiple intelligent agent systems that share information.

[0082] Specifically, through collaborative learning among multiple intelligent agent systems, for example, the target intelligent agent system responsible for steam turbines may learn new vibration analysis techniques from intelligent agent systems responsible for other rotating equipment (such as fans, pumps, etc.), thereby obtaining the final diagnosis results and improving its own diagnostic capabilities. At the same time, the target intelligent agent system will also automatically generate new heuristic rules based on this diagnostic experience, such as "when the vibration of the steam turbine increases suddenly and the bearing temperature increases, give priority to the rotor imbalance problem."

[0083] Therefore, by introducing intelligent agent systems, each intelligent agent system is responsible for a specific type of equipment or fault, which enables the intelligent agent systems to share knowledge while maintaining their respective expertise.

[0084] In one embodiment, after obtaining the diagnosis results, the user can make further interactive inquiries with the natural language interactive interface based on the actual situation on site, such as "What if it is a bearing problem? How long does it take to complete the dynamic balance test?" etc. The computer equipment will give more detailed personalized suggestions based on the knowledge graph and historical cases, combined with the previous fault diagnosis results, to achieve intelligent interaction with the operation and maintenance personnel using the powerful language understanding ability of the big model, and provide explainable diagnosis results by combining the structured knowledge graph and the reasoning ability of the big model.

[0085] In one embodiment, when the computer device discovers during the diagnosis process that there is a high degree of uncertainty in the understanding of, for example, vibration characteristic parameters of a specific frequency band, it will proactively consult experts or suggest collecting more relevant data, thereby continuously improving its own knowledge system.

[0086] In summary, this solution can be applied to an intelligent operation and maintenance management system for industrial control, mainly including the following application scenarios: (1) Real-time fault diagnosis: Operation and maintenance personnel use mobile devices or control room terminals to describe the observed abnormal phenomena with voice, and quickly give fault diagnosis and treatment suggestions. (2) New employee training: Use knowledge graphs and interactive interfaces to provide new employees with an interactive equipment knowledge learning platform. Therefore, the operation and maintenance management system can automatically build and update the knowledge graph of the thermal power plant by integrating the equipment operation data of the thermal power plant, relevant literature on the Internet, and the descriptions of the operators. The operation and maintenance personnel can describe the fault phenomenon through voice input, and the operation and maintenance management system combines the knowledge graph for intelligent analysis, and finally gives accurate fault diagnosis and treatment suggestions, realizing the intelligent diagnosis of thermal power plant equipment failures and significantly improving the operation and maintenance efficiency and safety of thermal power plants.

[0087] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0088] Based on the same inventive concept, the embodiment of the present application also provides a device diagnostic apparatus for implementing the device diagnostic method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more device diagnostic apparatus embodiments provided below can refer to the limitations on the device diagnostic method above, and will not be repeated here.

[0089] In one embodiment, a device diagnosis apparatus is provided, comprising: a data acquisition module, a path reasoning module and a result determination module, wherein:

[0090] The data acquisition module is used to obtain user query requests and equipment operation data, and determine the target entity in the user query request through the large model.

[0091] The path reasoning module is used to perform multi-hop reasoning in the knowledge graph based on the target entity to obtain multiple initial reasoning paths; the knowledge graph includes nodes representing entities and edges representing the relationships connecting each entity.

[0092] The result determination module is used to perform probability analysis on multiple initial reasoning paths according to the equipment operation data, screen out the target reasoning path, and process the target reasoning path through the large model to obtain the diagnosis result.

[0093] Each module in the above device diagnosis can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0094] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a device diagnosis method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0095] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0096] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0097] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0098] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program includes computer instructions, the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in the above-mentioned method embodiments.

[0099] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0100] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0101] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A device diagnosis method, characterized in that: The method comprises: Obtaining a user query request and device operation data, and determining a target entity in the user query request through a large model; Based on the target entity, multi-hop reasoning is performed in the knowledge graph to obtain multiple initial reasoning paths; the knowledge graph includes nodes representing entities and edges representing relationships between entities; A probability analysis is performed on multiple initial reasoning paths according to the equipment operation data, a target reasoning path is screened out, and the target reasoning path is processed through a large model to obtain a diagnosis result.

2. The method according to claim 1, characterized in that The construction process of the knowledge graph includes: Acquire multi-source data and preset first prompt templates; Processing the multi-source data and the first prompt template by means of a large model to obtain an extraction result; the extraction result includes a plurality of original entities and original relationships identified from the multi-source data; A graph is constructed for the extraction results to obtain a knowledge graph.

3. The method according to claim 2, characterized in that The graph construction of the extraction results to obtain a knowledge graph includes: Standardizing the extraction results to obtain triple data constituting an initial knowledge graph; Constructing a second prompt template according to the triple data; the second prompt template includes the known information of the device and the relevant relationship between each of the known information of the device; The initial knowledge graph is updated by the large model and the second prompt template to obtain a knowledge graph.

4. The method according to claim 3, characterized in that The extraction results are normalized to obtain triple data constituting an initial knowledge graph, including: Preprocessing the original entities and original relations in the extraction results; the original entities include entity subjects and entity objects; The preprocessing at least includes establishing a term dictionary, unifying synonyms and polysemous terms; According to the triple structure, the preprocessed entity subject, original relation and entity object are converted into triple data.

5. The method according to claim 1, characterized in that The multi-hop reasoning is performed in the knowledge graph based on the target entity to obtain multiple initial reasoning paths, including: Taking the node where the target entity in the knowledge graph is located as the starting point, traversing is performed to determine multiple node information; the node information at least includes relationship strength and path length; When traversing to a special node, an abnormal pattern of the special node is identified through an abnormality detection algorithm; Path scoring is performed based on multiple node information and abnormal patterns to obtain multiple initial reasoning paths.

6. The method according to claim 1, characterized in that The performing probability analysis on the plurality of initial reasoning paths according to the device operation data to screen out a target reasoning path includes: The initial reasoning path and equipment operation data are integrated through the Bayesian network to obtain multiple fault causes; constructing a decision tree associated with the device operation data; The decision tree and the Bayesian network are used to perform posterior probability processing, and the initial reasoning path corresponding to the fault cause whose probability exceeds a preset value is used as the target reasoning path.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: Obtaining optimization source information; the optimization source information at least includes user feedback, actual processing results of the device, diagnosis frequency of entities and relationships, information timeliness or expert ratings; Based on the graph attention network and the optimization source information, the weights between entities and relationships in the knowledge graph are optimized.

8. The method according to claim 1, characterized in that The method is applied in an agent cluster, wherein the agent cluster includes a plurality of agent systems that share information; the method further includes: Determining a target type of a target device corresponding to the device operation data, and determining a target agent system from a plurality of agent systems according to the target type; Through the target intelligent agent system, the process of determining the target entity in the user query request through the large model is implemented until a diagnosis result is obtained.

9. An equipment diagnostic device, characterized in that: The device comprises: A data acquisition module, used to acquire user query requests and device operation data, and determine the target entity in the user query request through a large model; A path reasoning module, used to perform multi-hop reasoning in a knowledge graph based on the target entity to obtain multiple initial reasoning paths; the knowledge graph includes nodes representing entities and edges representing relationships connecting entities; The result determination module is used to perform probability analysis on the multiple initial reasoning paths according to the equipment operation data, and screen out the target reasoning path; and process the target reasoning path through the large model to obtain the diagnosis result.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images