Hydropower station intelligent operation and maintenance method and assistant system

Abstract

The invention belongs to the technical field of hydropower operation and maintenance, and relates to a hydropower station intelligent operation and maintenance method and an assistant system. The method comprises the following steps: constructing an expert knowledge base subsystem; constructing an equipment facility knowledge base subsystem; constructing a large model based on the knowledge graph; calling the large model to generate alarm information, and performing enhanced retrieval; the large model analyzes the retrieval result, pushes the fault disposal suggestions to the operation and maintenance personnel for disposal or performs searching through networking of the large model, and after integration, generates fault disposal suggestions and pushes the fault disposal suggestions to the operation and maintenance personnel for disposal; and when no retrieval result exists, a fault phenomenon and disposal method first draft is generated, adjusted by operation and maintenance personnel and input into the expert knowledge base subsystem. According to the method, hydropower station equipment facilities are taken as an entity starting point, and the knowledge graph is constructed by combining a large model for reasoning, so that the purposes of quickly identifying potential faults and providing accurate disposal suggestions are achieved; intelligent alarm prediction and fault disposal information pushing are realized, and the correctness and efficiency of fault disposal of hydropower station operation and maintenance personnel are effectively improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of hydropower operation and maintenance, and in particular relates to a hydropower station intelligent operation and maintenance method and an assistant system. BACKGROUND

[0002] The hydropower industry is undergoing a paradigm shift from traditional operation and maintenance to digital and intelligent operation and maintenance. The traditional operation and maintenance mode is difficult to cope with the challenges of increasingly complex power grid peak shaving and accident disposal. The data of various equipment of the hydropower station are rich in sources, but due to insufficient multi-source data fusion, the following problems may occur:

[0003] (1) Cross-system data (such as unit data and auxiliary equipment data) are fragmented and cannot be correlated and analyzed, the data value is not fully tapped, and it is difficult to support global decision-making;

[0004] (2) The equipment failure trend cannot be obtained in time. At the present stage, only relying on the monitoring, inspection and curve analysis of the running personnel to find faults, the judgment and decision-making ability of the running personnel is required to be high, and there are problems of insufficient prediction ability and slow response speed;

[0005] (3) The cross-departmental collaborative decision-making efficiency is low. The archived files of various departments (such as operation, maintenance and hydraulic departments) of the hydropower station are separately distributed, lack of sharing mechanism, and it is difficult for cross-departmental personnel to make collaborative decisions, the operation and maintenance efficiency is low, and the safety and economy of the operation of the power station are affected;

[0006] (4) Unstructured data (such as documents and audio) and structured data (such as sensor data) are difficult to combine to form callable information, and valuable disposal experience cannot be systematically inherited, resulting in loss of tacit knowledge. SUMMARY

[0007] In order to solve the above technical problems, the application provides a hydropower station intelligent operation and maintenance method and an assistant system.

[0008] In a first aspect, the application provides a hydropower station intelligent operation and maintenance method, comprising:

[0009] collecting typical cases of fault disposal to construct an expert knowledge base subsystem;

[0010] constructing a multi-dimensional knowledge graph of equipment and facilities of the hydropower station to obtain an equipment and facility knowledge base subsystem;

[0011] The intelligent decision-making subsystem constructs a large model based on the knowledge graph;

[0012] The intelligent monitoring and alarm system calls the large model, generates alarm information according to the hydropower station operation and maintenance data, and pushes the alarm information to the expert knowledge base subsystem and the equipment and facility knowledge base subsystem for enhanced retrieval;

[0013] The intelligent monitoring alarm system inputs the retrieval result, the knowledge graph query result and the alarm information into the large model for analysis;

[0014] The large model analyzes the retrieval result, when the retrieval result contains completely corresponding fault phenomena and disposal methods, adopts the original text output mode to push the fault disposal suggestion to the operation and maintenance personnel for disposal, when the retrieval result does not contain completely corresponding fault phenomena and disposal methods, similar contents are recognized and extracted, and the large model is integrated to generate the fault disposal suggestion and push the fault disposal suggestion to the operation and maintenance personnel for disposal, when the expert knowledge base and the equipment and facility knowledge base subsystem both do not have similar contents, the large model is connected to search, the search result is integrated to generate the fault disposal suggestion and push the fault disposal suggestion to the operation and maintenance personnel for disposal, and if the intelligent monitoring alarm system has no retrieval result, the large model inputs the fault information and the disposal scheme into the system to generate a draft of fault phenomena and disposal methods, which is adjusted by the operation and maintenance personnel and then input into the expert knowledge base subsystem.

[0015] In the second aspect, the present application provides an intelligent operation and maintenance assistant system for a hydropower station, which comprises an intelligent monitoring alarm system, an expert knowledge base subsystem, an equipment and facility knowledge base subsystem, an intelligent decision subsystem and a fault phenomenon analysis and input subsystem;

[0016] The intelligent monitoring alarm system is used for calling a large model, generating alarm information according to hydropower station operation and maintenance data, pushing the alarm information to the expert knowledge base subsystem and the equipment and facility knowledge base subsystem for enhanced retrieval, and inputting the retrieval result, the knowledge graph query result and the alarm information into the large model for analysis;

[0017] The expert knowledge base subsystem is used for storing typical fault disposal cases;

[0018] The equipment and facility knowledge base subsystem is used for storing a multi-dimensional knowledge graph of equipment and facilities in the hydropower station;

[0019] The intelligent decision subsystem is used for storing a large model based on a knowledge graph; the large model analyzes the retrieval result, when the retrieval result contains completely corresponding fault phenomena and disposal methods, adopts the original text output mode to push the fault disposal suggestion to the operation and maintenance personnel for disposal, when the retrieval result does not contain completely corresponding fault phenomena and disposal methods, similar contents are recognized and extracted, and the large model is integrated to generate the fault disposal suggestion and push the fault disposal suggestion to the operation and maintenance personnel for disposal, when the expert knowledge base and the equipment and facility knowledge base subsystem both do not have similar contents, the large model is connected to search, the search result is integrated to generate the fault disposal suggestion and push the fault disposal suggestion to the operation and maintenance personnel for disposal;

[0020] The fault phenomenon analysis and input subsystem is used for, when the intelligent monitoring alarm system has no retrieval result, inputting the fault information and the disposal scheme into the system by the large model to generate a draft of fault phenomena and disposal methods, which is adjusted by the operation and maintenance personnel and then input into the expert knowledge base subsystem.

[0021] Based on the technical scheme, the application can be further improved as follows.

[0022] Further, the expert knowledge base subsystem includes a fault handling data collection system and a preventive handling data collection system; the fault handling data collection system and the preventive handling data collection system extract the event summary, the event cause, the exposed problem and the handling measure of the typical fault handling case respectively, and perform manual revision and arrangement.

[0023] Further, a multi-dimensional knowledge graph of the equipment and facilities of the hydropower station is constructed, including: taking the water turbine generator system, the transformer system, the high-voltage power supply system, the speed regulator system, the excitation system and the hydraulic structure system as the entity starting point, radiating the subsystem downward, defining the physical connection relationship between the entities, constructing the affiliation relationship network between the entities through the functional dependency relationship and the hierarchical relationship, and fusing the multi-source operation parameters to construct the multi-dimensional knowledge graph of the equipment and facilities of the hydropower station.

[0024] Further, the functional dependency relationship includes a data transmission path; the hierarchical relationship includes equipment, components and parts; and the multi-source operation parameters include voltage, current, frequency and power.

[0025] Further, the equipment and facility knowledge base subsystem includes equipment and facility manuals, equipment and facility working principle diagrams, operation and maintenance procedures, equipment setting value sheets, equipment operation modes, equipment operation data, equipment operation manuals and written technical summaries.

[0026] Further, the intelligent decision-making subsystem constructs a large model based on the knowledge graph, including: adopting a matching relationship method based on a rule engine, combining the functional dependency and the hierarchical relationship to construct the entity relationship extraction of the equipment and facilities of the hydropower station, simultaneously optimizing the attribute mapping through data cleaning and alignment, and storing the equipment topology relationship by using a Neo4j graph database; and enhancing the retrieval from the expert knowledge base subsystem and the equipment and facility knowledge base subsystem through a prompt engineering.

[0027] Further, the intelligent monitoring and alarm system calls the large model to generate alarm information according to the hydropower station operation and maintenance data, including: the large model combines the real-time monitoring data and the fault cases stored in the knowledge base to perform multi-modal feature data matching analysis on potential faults, and generates alarm information by dynamically evaluating the hydropower station operation and maintenance data through a rule engine.

[0028] Further, the multi-modal feature data of the hydropower station operation and maintenance data includes monitoring system alarm signals, electrical quantity data change amount feature data and trigger time feature data.

[0029] Further, the alarm information includes state abnormal alarm, trend abnormal alarm and process abnormal alarm.

[0030] The beneficial effects of the present application are: the present application takes the equipment and facilities of the hydropower station as the entity starting point, radiates to each subsystem downward, and realizes the construction of the knowledge graph in the manner of reasoning combined with the large model, so as to achieve the purpose of quickly identifying potential faults and providing accurate treatment suggestions; the intelligent monitoring data is combined with the large model, and the fault treatment analysis is carried out based on the knowledge base classic cases, so as to realize intelligent alarm; in addition, the present application realizes intelligent alarm prediction and fault treatment information pushing, and effectively improves the correctness and efficiency of the fault treatment of the hydropower station operation and maintenance personnel. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 A principle diagram of a hydropower station intelligent operation and maintenance method provided for the present application embodiment 1;

[0032] Figure 2 A specific flowchart of a hydropower station intelligent operation and maintenance method provided for the present application embodiment 1;

[0033] Figure 3 A principle block diagram of a hydropower station intelligent operation and maintenance assistant system provided for the present application embodiment 2. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical scheme and advantages of the present application embodiments clearer, the technical scheme of the present application embodiments will be described clearly and completely below in combination with the drawings in the present application embodiments. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the present application embodiments described and shown in the drawings here can be arranged and designed in various different configurations.

[0035] Embodiment 1

[0036] As an embodiment, as shown in the accompanying drawings, in order to solve the above technical problems, the present embodiment provides a hydropower station intelligent operation and maintenance method, which comprises: Figure 1

[0037] Collecting fault treatment typical cases to construct an expert knowledge base subsystem;

[0038] Constructing a hydropower station equipment and facility multi-dimensional knowledge graph to obtain an equipment and facility knowledge base subsystem;

[0039] An intelligent decision subsystem constructs a large model based on the knowledge graph;

[0040] An intelligent monitoring alarm system calls the large model, generates alarm information according to the hydropower station operation and maintenance data, and pushes the alarm information to the expert knowledge base subsystem and the equipment and facility knowledge base subsystem for enhanced retrieval;

[0041] The intelligent monitoring alarm system inputs the retrieval result, the knowledge graph query result and the alarm information into the large model for analysis;​

[0042] The large model analyzes the retrieval results. When the retrieval results contain completely corresponding fault phenomena and treatment methods, the fault treatment suggestion is pushed to the operation and maintenance personnel for treatment in the original output mode. When the retrieval results do not contain completely corresponding fault phenomena and treatment methods, similar contents are identified and extracted, and the fault treatment suggestion is generated after integration by the large model and pushed to the operation and maintenance personnel for treatment. When neither the expert knowledge base nor the equipment and facility knowledge base subsystem has similar contents, the large model is connected for searching, the search results are integrated, and the fault treatment suggestion is generated and pushed to the operation and maintenance personnel for treatment. If the intelligent monitoring alarm system has no retrieval results, the fault information and treatment scheme are input into the system by the large model, an initial draft of the fault phenomena and treatment methods is generated, and the operation and maintenance personnel adjust and input the initial draft into the expert knowledge base subsystem. The specific flowchart is shown in FIG. 8. Figure 2

[0043] When the intelligent monitoring system generates alarm information, the alarm information is pushed to the expert knowledge base subsystem and the equipment and facility knowledge base subsystem for enhanced retrieval, and the retrieval information, knowledge graph query results and alarm information are input into the large model. When the retrieved information has completely corresponding fault phenomena and treatment methods, the large model analyzes and pushes the fault treatment suggestion to the operation and maintenance personnel for treatment in the original output mode, ensuring the accuracy of the treatment scheme. When the large model analyzes and does not find the original text but has similar contents, the large model integrates and pushes the fault treatment suggestion to the operation and maintenance personnel for treatment. When neither the fault treatment typical case library nor the hydropower station equipment and facility multi-dimensional knowledge graph has corresponding contents, the large model is connected for searching, the search results are integrated, and the fault treatment suggestion is pushed to the operation and maintenance personnel for treatment, thereby improving the fault treatment efficiency of the operation and maintenance personnel. When no relevant contents are found in the knowledge base, the large model inputs the fault information and treatment scheme into the system, generates an initial draft of the fault phenomena and treatment methods, and the operation and maintenance personnel confirm and complete the input of the expert knowledge base database, thereby improving the knowledge base content reserve.

[0044] The method analyzes possible faults according to the operating state characteristics of the hydroelectric generator set equipment, and intelligently pushes fault treatment information to the operation and maintenance personnel. Specifically, the intelligent monitoring alarm system generates alarm information according to the fault alarm strategy, and pushes the information to the large model after enhanced retrieval in combination with the contents of the fault treatment typical case and the hydropower station equipment and facility multi-dimensional knowledge graph. The large model analyzes and judges the retrieval results, and pushes the decision information to the operation and maintenance personnel.

[0045] The large model-based intelligent operation and maintenance method of the hydropower station disclosed in the present application takes the equipment and facilities of the hydropower station as the starting point, radiates to each subsystem, and realizes the construction of the knowledge graph in combination with the reasoning of the large model, so as to quickly identify potential faults and provide accurate treatment suggestions.

[0046] ​The large model is combined with intelligent monitoring data and knowledge base cases to analyze fault disposal, state abnormal alarm, trend abnormal alarm and process abnormal alarm strategies are adopted, the coupling characteristics of analog quantity and switching quantity are quantified based on conditional mutual information, the alarm strategy is sorted and optimized, and intelligent alarm is realized. The patent realizes intelligent alarm prediction and fault disposal information pushing, and effectively improves the correctness and efficiency of the fault disposal of the hydroelectric station operation and maintenance personnel.

[0047] Optionally, the expert knowledge base subsystem includes a fault disposal data collection system and a preventive disposal data collection system; the fault disposal data collection system and the preventive disposal data collection system extract event summary, event reason, exposed problem and disposal measure of typical cases of fault disposal respectively, and perform manual revision and arrangement.

[0048] In actual application, fault disposal data is extracted from the fault disposal experience book, and preventive disposal data is extracted from the preventive disposal data experience book. In actual operation of the hydroelectric station, various types of faults have occurred, and rich typical cases of fault disposal have been formed. The existing cases are manually revised and arranged, each typical case is described according to four dimensions of event summary, event reason, exposed problem and disposal measure, to form a fault disposal experience book; the actual non-occurrence but the fault phenomenon can be inferred according to the equipment characteristics, the fault disposal plan is designed, and the accident experience of the station is arranged to form a preventive disposal experience book.

[0049] Optionally, a multi-dimensional knowledge graph of equipment and facilities of the hydroelectric station is constructed, including: taking the water turbine generator system, the transformer system, the high-voltage power supply system, the speed regulator system, the excitation system and the hydraulic structure system as the entity starting point, radiating the subsystem downward, defining the physical connection relationship between entities, constructing the affiliation relationship network between entities through functional dependency relationship and hierarchical relationship, and fusing multi-source operation parameters to construct the multi-dimensional knowledge graph of equipment and facilities of the hydroelectric station.

[0050] Optionally, the functional dependency relationship includes a data transmission path; the hierarchical relationship includes equipment, components and parts; and the multi-source operation parameters include voltage, current, frequency and power.

[0051] Optionally, the equipment and facility knowledge base subsystem includes equipment and facility manuals, equipment and facility working principle diagrams, operation procedures, equipment setting value sheets, equipment operation modes, equipment operation data, equipment operation manuals and written technical summaries.

[0052] Optionally, the intelligent decision-making subsystem constructs a large model based on a knowledge graph, including: adopting a matching relationship method based on a rule engine, combining functional dependencies and hierarchical relationships to construct a hydropower station equipment and facility entity relationship extraction, and simultaneously optimizing attribute mapping through data cleaning and alignment, and storing equipment topology relationships using a Neo4j graph database; and enhancing retrieval through a prompt engineering from an expert knowledge base subsystem and an equipment and facility knowledge base subsystem.

[0053] Specifically, an affiliation relationship network between entities is constructed through functional dependencies and hierarchical relationships, a matching relationship based on a rule engine is adopted, a Neo4j graph database is used, and a large model is combined for reasoning to achieve the construction of a knowledge graph, thereby achieving the purpose of quickly identifying potential faults and providing accurate disposal suggestions.

[0054] Optionally, the intelligent monitoring and alarm system calls a large model to generate alarm information according to hydropower station operation and maintenance data, including: the large model combines real-time monitoring data and stored fault cases in the knowledge base to perform multi-modal feature data matching analysis on potential faults, and generates alarm information through dynamic evaluation of the hydropower station operation and maintenance data by a rule engine.

[0055] The intelligent monitoring and alarm system sets alarm set values and alarm determination strategies for hydropower station operation and maintenance data, and generates and pushes alarm information when the alarm set values and alarm conditions are reached.

[0056] Optionally, the multi-modal feature data of the hydropower station operation and maintenance data includes monitoring system alarm signals, electrical quantity data change quantity feature data, and trigger time feature data.

[0057] The feature extraction method is, for example: a combined condition mining method is used to extract state timing and jump parameters for switch quantity signals; numerical segmentation statistics and trend fitting are performed on analog quantity signals according to numerical conditions; analog quantities and switch quantities are combined, and the coupling features of analog quantities and switch quantities are quantified based on conditional mutual information, and alarm strategies are sorted out.

[0058] Taking the slow melting fault disposal of the generator outlet PT as an example, when a fault occurs, the multi-modal collection quantity of fault information includes monitoring system alarm signals, electrical quantity data change quantity data, and trigger time data.

[0059] The monitoring system alarm signal includes generator outlet voltage, unit and main transformer fault recording and starting recording, etc.; the electrical quantity data change data includes: generator outlet voltage abnormal drop; the trigger time data includes the starting or occurrence time of the fault related signal or event. After multi-modal data is recognized by multi-modal recognition, fault feature extraction is performed, the text alarm keywords are extracted: "generator outlet no voltage", "recording wave starting"; the electrical quantity change trend features (such as voltage drop, current anomaly, etc.) are extracted; the time stamp is combined to align the multi-source signals. The above extracted multi-modal features are matched with the existing fault modes in the knowledge base, and the possible fault type is identified as: generator outlet PT fuse blown; combined with the knowledge base content and the large model analysis, the generator outlet PT fuse blown disposal method is pushed, and the pushing content includes on-site inspection, load adjustment and fuse replacement steps.

[0060] Optionally, the alarm information includes state abnormal alarm, trend abnormal alarm and process abnormal alarm.

[0061] Embodiment 2

[0062] Based on the same principle as the method shown in Embodiment 1 of the present application, as shown in the accompanying drawings, Figure 3 The present application also provides an intelligent operation and maintenance assistant system for a hydropower station in the embodiments, which includes an intelligent monitoring and alarm system, an expert knowledge base subsystem, a device and facility knowledge base subsystem, an intelligent decision-making subsystem, and a fault phenomenon analysis and input subsystem;

[0063] The intelligent monitoring and alarm system is used to call a large model, generate alarm information according to hydropower station operation and maintenance data, push the alarm information to the expert knowledge base subsystem and the device and facility knowledge base subsystem for enhanced retrieval, and input the retrieval results, knowledge graph query results and alarm information into the large model for analysis;

[0064] The expert knowledge base subsystem is used to store typical fault disposal cases;

[0065] The device and facility knowledge base subsystem is used to store a multi-dimensional knowledge graph of hydropower station devices and facilities;

[0066] The intelligent decision-making subsystem is used to store a large model based on a knowledge graph; the large model analyzes the retrieval results, when the retrieval results contain completely corresponding fault phenomena and disposal methods, the fault disposal suggestions are pushed to the operation and maintenance personnel for disposal in the form of original output; when the retrieval results do not contain completely corresponding fault phenomena and disposal methods, similar contents are identified and extracted, and the fault disposal suggestions are generated by the large model after integration and pushed to the operation and maintenance personnel for disposal; when there is no similar content in the expert knowledge base and the device and facility knowledge base subsystem, the large model is connected for search, the search results are integrated to generate fault disposal suggestions which are pushed to the operation and maintenance personnel for disposal;

[0067] The fault phenomenon analysis and input subsystem is used for inputting the fault information and treatment scheme into the system by the large model when there is no search result in the intelligent monitoring alarm system, to generate the initial draft of the fault phenomenon and treatment method, which is adjusted by the operation and maintenance personnel and then input into the expert knowledge base subsystem.

[0068] In actual application, the large model is deployed and called in the intranet through the intranet cloud platform of the hydropower station, and the client uses web pages, computer chat software, intelligent sound boxes and the like as human-computer interaction client carriers.

[0069] Optionally, the expert knowledge base subsystem includes a fault treatment data collection system and a preventive treatment data collection system; the fault treatment data collection system and the preventive treatment data collection system respectively extract the event summary, event cause, exposed problem and treatment measure of the typical case of fault treatment, and perform manual revision and arrangement.

[0070] Optionally, a multi-dimensional knowledge graph of the equipment and facilities of the hydropower station is constructed, including: taking the water turbine generator system, the transformer system, the high-voltage power supply system, the speed regulator system, the excitation system and the hydraulic structure system as the entity starting point, radiating the subsystem downward, defining the physical connection relationship between entities, constructing the affiliation relationship network between entities through the functional dependency relationship and the hierarchical relationship, and fusing the multi-source operation parameters to construct the multi-dimensional knowledge graph of the equipment and facilities of the hydropower station.

[0071] Optionally, the functional dependency relationship includes a data transmission path; the hierarchical relationship includes equipment, components and parts; and the multi-source operation parameters include voltage, current, frequency and power.

[0072] Optionally, the equipment and facility knowledge base subsystem includes equipment and facility manuals, equipment and facility working principle diagrams, operation and maintenance procedures, equipment setting value sheets, equipment operation modes, equipment operation data, equipment operation manuals and written technical summaries.

[0073] Optionally, the intelligent decision-making subsystem constructs a large model based on the knowledge graph, including: adopting a matching relationship method based on a rule engine, combining the functional dependency and the hierarchical relationship to construct the entity relationship extraction of the equipment and facilities of the hydropower station, simultaneously optimizing the attribute mapping through data cleaning and alignment, and storing the equipment topology relationship by using a Neo4j graph database; and enhancing the retrieval from the expert knowledge base subsystem and the equipment and facility knowledge base subsystem through a prompt engineering.

[0074] Optionally, the intelligent monitoring alarm system calls the large model to generate alarm information according to the operation and maintenance data of the hydropower station, including: the large model combines the real-time monitoring data and the fault cases stored in the knowledge base to perform multi-modal feature data matching analysis on potential faults, and generates alarm information by dynamically evaluating the operation and maintenance data of the hydropower station through a rule engine.

[0075] Optionally, the multi-modal feature data of the hydropower station operation and maintenance data comprises monitoring system alarm signals, electrical quantity data change amount feature data, and trigger time feature data.

[0076] Optionally, the alarm information comprises state abnormal alarm, trend abnormal alarm, and process abnormal alarm.

[0077] The above only is the preferred embodiment of the present application, and is not used to limit the present application, for the person skilled in the art, the present application can have various changes and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for intelligent operation and maintenance of a hydropower station, characterized in that: include: Collect typical fault handling cases and build an expert knowledge base subsystem; Construct a multi-dimensional knowledge graph of hydropower station equipment and facilities to obtain the equipment and facilities knowledge base subsystem; The intelligent decision-making subsystem builds a large model based on the knowledge graph; The intelligent monitoring and alarm system uses the large model to generate alarm information based on the hydropower station operation and maintenance data, and pushes the alarm information to the expert knowledge base subsystem and the equipment and facility knowledge base subsystem for enhanced retrieval; The intelligent monitoring and alarm system inputs the search results, knowledge graph query results and alarm information into the big model for analysis; The large model analyzes the search results. When the search results contain completely corresponding fault phenomena and solutions, the original text output is used to push the fault solution suggestions to the operation and maintenance personnel for handling. When the search results do not contain completely corresponding fault phenomena and solutions, similar content is identified and extracted, and the large model integrates them to generate fault solution suggestions and push them to the operation and maintenance personnel for handling. When there is no similar content in the expert knowledge base and the equipment and facility knowledge base subsystems, the large model conducts a network search, integrates the search results, generates fault handling suggestions, and pushes them to the operation and maintenance personnel for handling; If the intelligent monitoring and alarm system does not have any search results, the large model will enter the fault information and treatment plan into the system, generate a draft of the fault phenomenon and treatment method, and then enter it into the expert knowledge base subsystem after adjustment by the operation and maintenance personnel.

2. The intelligent operation and maintenance method of a hydropower station according to claim 1, characterized in that: The expert knowledge base subsystem includes a fault handling data collection system and a preventive handling data collection system; the fault handling data collection system and the preventive handling data collection system respectively extract the event overview, event causes, exposed problems and handling measures of typical fault handling cases, and manually revise and organize them.

3. The intelligent operation and maintenance method of a hydropower station according to claim 1, characterized in that: Construct a multi-dimensional knowledge graph of hydropower station equipment and facilities, including: taking the turbine generator system, transformer system, high-voltage power supply system, speed regulator system, excitation system and hydraulic structure system as the entity starting point, radiating subsystems downward, defining the physical connection relationship between entities, and constructing the subsidiary relationship network between entities through functional dependency and hierarchical relationships, integrating multi-source operating parameters, and constructing a multi-dimensional knowledge graph of hydropower station equipment and facilities.

4. The intelligent operation and maintenance method for a hydropower station according to claim 1, characterized in that: Functional dependencies include data transmission paths; hierarchical relationships include devices, components, and parts; and multi-source operating parameters include voltage, current, frequency, and power.

5. The intelligent operation and maintenance method of a hydropower station according to claim 1, characterized in that: The equipment and facility knowledge base subsystem includes equipment and facility manuals, equipment and facility working principle diagrams, operation and maintenance procedures, equipment value sheets, equipment operating modes, equipment operating data, equipment operation manuals and written technical summaries.

6. The intelligent operation and maintenance method of a hydropower station according to claim 1, characterized in that: The intelligent decision-making subsystem constructs a large model based on the knowledge graph, including: using a matching relationship method based on a rule engine, combining functional dependencies and hierarchical relationships to construct entity relationship extraction of hydropower station equipment and facilities, while optimizing attribute mapping through data cleaning and alignment, and using the Neo4j graph database to store equipment topology relationships; and enhancing retrieval from the expert knowledge base subsystem and the equipment and facility knowledge base subsystem through prompt engineering.

7. The intelligent operation and maintenance method of a hydropower station according to claim 1, characterized in that: The intelligent monitoring and alarm system calls the big model to generate alarm information based on the hydropower station operation and maintenance data, including: the big model combines real-time monitoring data with fault cases stored in the knowledge base, performs multimodal feature data matching analysis on potential faults, and dynamically evaluates the hydropower station operation and maintenance data through the rule engine to generate alarm information.

8. A hydropower station intelligent operation and maintenance method according to any one of claims 1 or 7, characterized in that: The multimodal characteristic data of hydropower station operation and maintenance data include monitoring system alarm signals, electrical quantity data change characteristic data and trigger time characteristic data.

9. A hydropower station intelligent operation and maintenance method according to any one of claims 1 or 7, characterized in that: Alarm information includes abnormal status alarm, abnormal trend alarm and abnormal process alarm.

10. An intelligent operation and maintenance assistant system for a hydropower station, characterized in that: It includes intelligent monitoring and alarm system, expert knowledge base subsystem, equipment and facility knowledge base subsystem, intelligent decision-making subsystem and fault phenomenon analysis and entry subsystem; The intelligent monitoring and alarm system is used to call the big model, generate alarm information based on the hydropower station operation and maintenance data, and push the alarm information to the expert knowledge base subsystem and the equipment and facility knowledge base subsystem for enhanced retrieval. The retrieval results, knowledge graph query results, and alarm information are input into the big model for analysis. Expert knowledge base subsystem, used to store typical fault handling cases; The equipment and facility knowledge base subsystem is used to store the multi-dimensional knowledge graph of hydropower station equipment and facilities; The intelligent decision-making subsystem is used to store a large model based on the knowledge graph. The large model analyzes the search results. When the search results contain a completely corresponding fault phenomenon and solution, the fault solution suggestions are pushed to the operation and maintenance personnel in the form of original text output. When the search results do not contain a completely corresponding fault phenomenon and solution, similar content is identified and extracted, and the large model integrates them to generate fault solution suggestions and push them to the operation and maintenance personnel for processing. When there is no similar content in the expert knowledge base and the equipment and facility knowledge base subsystems, the large model conducts a network search, integrates the search results, generates fault handling suggestions, and pushes them to the operation and maintenance personnel for handling; The fault phenomenon analysis and entry subsystem is used to enter the fault information and treatment plan into the system by the large model when there is no search result in the intelligent monitoring and alarm system, generate a preliminary draft of the fault phenomenon and treatment method, and then enter it into the expert knowledge base subsystem after adjustment by the operation and maintenance personnel.

Images