Power plant personalized fault case recommendation method, system, equipment and medium

Abstract

The invention discloses a power plant personalized fault case recommendation method, system and device and a medium, and relates to the technical field of power plant intelligent fault management and recommendation, and the method comprises the steps: carrying out the multi-source fault data fusion and entity relation modeling through a knowledge graph; based on the operation data and the user behaviors, constructing a personalized recommendation model and generating a recommendation case; and through semantic matching and dynamic analysis, generating and pushing priority-sorted recommendation cases. According to the method, through data cleaning and knowledge graph construction, data from different sources are systematically and effectively integrated, and a multi-dimensional incidence relation network is constructed; compared with the prior art, the method has the advantages that complex relations among equipment, faults and solutions can be expressed more comprehensively, analysis deviation caused by data isolation is avoided, user input is deeply analyzed by means of a semantic matching technology in combination with user behaviors and equipment operation data, and a recommendation scheme is dynamically adjusted in combination with real-time data; and a stronger decision support capability is provided for power plant fault management.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent fault management and recommendation of power plants, and in particular to a method, system, device and medium for recommending personalized fault cases of power plants. Background Art

[0002] With the development of industrial automation and information technology, the management and operation of modern power plants are gradually moving towards digitalization and intelligence. In terms of fault management, traditional manual experience records and paper documents are gradually replaced by digital fault records and analysis systems. At the same time, big data and artificial intelligence technologies are introduced into power plant operation and maintenance to improve fault diagnosis and processing efficiency. Knowledge graphs, as an emerging intelligent technology, provide power plant management systems with more efficient data expression and analysis capabilities by integrating multi-source heterogeneous data into structured knowledge networks. Personalized recommendation technology can achieve more accurate content recommendations by analyzing user behavior data and equipment operation data, providing decision support for power plant fault handling. However, although these technologies have been applied in some fields, the personalized recommendation and knowledge level of power plant fault management systems are still in the initial development stage, and the integration and practicality of the technologies need to be further improved.

[0003] The existing power plant fault management system mainly relies on simple matching algorithms based on static data, and lacks the ability to dynamically integrate and comprehensively analyze multi-source data. In data processing, the existing technology usually only performs simple cleaning on a single data source, and cannot fully solve the conflict problem of multi-source heterogeneous data, resulting in low data consistency and accuracy. The existing fault recommendation system lacks the ability to comprehensively analyze user behavior and equipment operating status, and the recommended content often cannot accurately reflect user concerns and the current status of the equipment, which easily leads to poor relevance of the recommendation results. The application of semantic parsing and matching technology is insufficient, and the existing system cannot effectively handle the pronoun reference and semantic ambiguity problems in complex fault case texts, resulting in limited context understanding ability of case recommendations. In addition, it is difficult for the existing technology to combine the multi-dimensional correlation characteristics of the knowledge graph to achieve dynamic priority sorting, and the pushed content cannot fully reflect the immediate response and pertinence to user needs. Therefore, the existing system has significant deficiencies in multi-source data fusion, recommendation accuracy and intelligent push, and it is difficult to meet the efficient and intelligent fault management needs of modern power plants. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: the existing power plant fault recommendation method has insufficient multi-source data fusion capabilities, poor recommendation accuracy, and cannot dynamically push priority content, as well as the problem of how to achieve fault case recommendation and push based on knowledge graphs.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a personalized fault case recommendation method for a power plant, comprising using a knowledge graph to perform multi-source fault data fusion and entity relationship modeling; building a personalized recommendation model based on operating data and user behavior and generating recommended cases; generating and pushing prioritized recommended cases through semantic matching and dynamic analysis.

[0007] As a preferred solution of the method for recommending personalized fault cases of power plants described in the present invention, the multi-source fault data fusion includes: cleaning processing through entity disambiguation, identifying faults with the same name and merging them into unique entities; using reference resolution, linking the same reference objects in different fault cases through semantic analysis, and generating a consistent data representation.

[0008] As a preferred solution of the personalized fault case recommendation method for power plants described in the present invention, the entity relationship modeling includes: entity labeling of the cleaned fault data through a knowledge graph, semantic classification and labeling of equipment attributes, fault modes and maintenance suggestions; building a relationship network based on the correlation between entities, using a graph structure to define entity nodes and relationship edges, and dynamically calculating the weights of relationship edges through historical data analysis; optimizing the relationship network through a semantic rule engine, merging redundant relationships and removing weakly related entity edges; generating a final knowledge graph model, which includes multi-dimensional associations between equipment types, fault characteristics and corresponding solutions.

[0009] As a preferred solution of the method for recommending personalized fault cases of power plants described in the present invention, the construction of a personalized recommendation model includes collecting the user's operating behavior in the system, including search records, click history, access frequency and user role information; analyzing the real-time monitoring data of the running equipment, including equipment status parameters, fault alarm information and historical trend data; fusing the user behavior data with the equipment operation data, and using a weighted factor model to quantify the user's interest preferences; establishing a personalized recommendation model through a machine learning method, prioritizing the fault cases that the user is concerned about, and generating preliminary recommended cases according to the recommendation logic.

[0010] As a preferred solution of the method for recommending personalized fault cases for power plants described in the present invention, the recommendation logic includes: when user behavior data indicates that the user is concerned about a specific case type that is frequently accessed, the same type of cases in the case collection are prioritized according to the access frequency, and the sorting rule is based on a weighted calculation of the user's historical access frequency and the success rate of solving related cases; when the equipment operation status is displayed as abnormal parameter fluctuations and no alarm is triggered, the recommendation logic prioritizes matching the case closest to the current operating parameter fluctuation range, and weights the case according to the historical fault frequency and processing time of the equipment associated with the case; when the equipment generates an alarm and is associated with multiple fault modes, the recommendation logic prioritizes recommending cases associated with high-level alarms according to the alarm level, and generates optimized recommended cases based on the complexity of the case's processing suggestions and the user's role permissions.

[0011] As a preferred solution of the personalized fault case recommendation method for power plants described in the present invention, the semantic matching includes: constructing a semantic representation of pronouns for pronouns and referential phrases in the fault case text, and establishing a candidate set of referential entities based on pronoun context information and syntactic dependencies; screening the optimal referential match by calculating the semantic similarity between pronouns and candidate entities; storing the referential matching results in the knowledge graph, and updating the semantic description of related entities.

[0012] As a preferred solution of the personalized fault case recommendation method for power plants described in the present invention, the generation and push of priority-ranked recommended cases includes: performing natural language processing on keywords or phrases input by users to extract key entities and relationships; retrieving case data related to the extracted key entities in the knowledge graph, including similar fault cases, equipment-related information and solutions; combining real-time operation data analysis to filter case data that best matches the current equipment status; sorting the filtering results according to user portraits and case relevance, and generating a priority recommendation list; and pushing the highest priority case information to the user terminal through a dynamic push function, including detailed content of the case, associated equipment and recommended processing suggestions.

[0013] Another object of the present invention is to provide a personalized fault case recommendation system for power plants, which can construct the relationship between equipment, faults and solutions through knowledge graph technology, thereby solving the problem that the current power plant fault recommendation technology has insufficient multi-source data fusion capabilities.

[0014] As a preferred solution of the personalized fault case recommendation system for power plants described in the present invention, it includes: a data fusion module, a case generation module, and a case push module; the data fusion module includes a multi-source data cleaning module and an entity relationship modeling module, the multi-source data cleaning module is used to solve the redundancy and conflict problems in different data sources through entity disambiguation and reference resolution technology, and the entity relationship modeling module is used to build the association relationship between equipment, faults and solutions through knowledge graph technology; the case generation module includes a user behavior analysis module, an operation data processing module, and a personalized recommendation model module, the user behavior analysis module is used to collect and analyze the user's operation behavior and extract the user's focus, the operation data processing module is used to extract key features from the real-time operation data of the equipment, and the personalized recommendation model module is used to generate recommendation solutions using machine learning methods; the case push module includes a semantic parsing module, a case screening module, and a dynamic push module, the semantic parsing module is used to parse the keywords input by the user through natural language processing technology, the case screening module is used to screen case data matching the equipment status and user needs from the knowledge graph, and the dynamic push module is used to push the sorted cases to the user terminal to assist in decision-making.

[0015] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a method for recommending personalized fault cases of a power plant.

[0016] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for recommending personalized fault cases for a power plant.

[0017] Beneficial effects of the present invention: The personalized fault case recommendation method for power plants provided by the present invention systematically and effectively integrates data from different sources and constructs a multi-dimensional association relationship network through data cleaning and knowledge graph construction; compared with the prior art, it can more comprehensively express the complex relationship between equipment, faults and solutions, avoid analysis bias caused by data isolation, combine user behavior and equipment operation data, and quantify user interests and equipment requirements through a weighted factor model; this method not only improves the relevance of recommendations, but also realizes the upgrade of recommended content from a single dimension (such as historical visits) to a multi-dimensional comprehensive analysis, making recommendations more accurate and efficient, using semantic matching technology to deeply analyze user input, and dynamically adjusting recommendation schemes in combination with real-time data; compared with traditional static recommendations, the immediacy, relevance and high personalization of recommended content are achieved, providing stronger decision-making support capabilities for power plant fault management, and the present invention achieves better results in terms of accuracy, efficiency and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 This is an overall flow chart of the method for recommending personalized fault cases for power plants provided in the first embodiment of the present invention.

[0020] Figure 2 This is an overall flow chart of a personalized fault case recommendation system for a power plant provided in the third embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0022] Example 1, reference Figure 1 , which is an embodiment of the present invention, provides a method for recommending personalized fault cases of power plants, including:

[0023] S1: Use knowledge graphs for multi-source fault data fusion and entity relationship modeling.

[0024] Furthermore, multi-source fault data fusion includes cleaning processing through entity disambiguation to identify faults with the same name and merge them into unique entities; using reference resolution to link the same referent objects in different fault cases through semantic analysis to generate consistent data representation.

[0025] It should be noted that entity-relationship modeling includes, through the knowledge graph, entity labeling of the cleaned fault data, semantic classification and labeling of equipment attributes, fault modes and maintenance suggestions; building a relationship network based on the association between entities, using a graph structure to define entity nodes and relationship edges, and dynamically calculating the weights of relationship edges through historical data analysis; optimizing the relationship network through a semantic rule engine, merging redundant relationships and removing weakly related entity edges; generating a final knowledge graph model, which includes multi-dimensional associations between equipment types, fault characteristics and corresponding solutions.

[0026] It should also be noted that in the process of knowledge fusion, different text data sources may have different names but point to the same entity. For example, when equipment fails, "contour axis" and "load-bearing axis" both represent an entity. Therefore, in entity disambiguation, it is necessary to identify and disambiguate the entities appearing in the text through algorithms and other means, and then determine whether the entity has appeared in other data sources and link it to the correct entity to ensure the accuracy and consistency of knowledge. Similar to entity disambiguation, reference resolution is also to eliminate the situation where different texts refer to the same entity but have different pronouns. For example, a fault case describes the failure of a gear, while another case uses the pronoun "it" to replace the gear. In reference resolution, natural language processing technology is needed to determine which entity pronouns point to the same or similar entities and link them, thereby improving the integrity and depth of knowledge.

[0027] It should also be noted that through entity disambiguation, entities with the same meaning are identified and merged to avoid data redundancy caused by different names; through reference resolution, the specific reference of pronouns in the case text is analyzed to ensure semantic consistency between data; data cleaning ensures the accuracy and consistency of fault information and avoids the interference of erroneous data on the analysis results; improves the system's processing capabilities for complex, multi-source data, and lays the foundation for the subsequent construction of knowledge graphs; through cleaning and resolution steps, the system can quickly integrate data from multiple sources, reduce the complexity of data processing and the need for manual intervention, provide more accurate input for the generation of knowledge graphs, and improve the efficiency and intelligence of power plant fault management.

[0028] S2: Build a personalized recommendation model based on operation data and user behavior and generate recommendation cases.

[0029] Furthermore, a personalized recommendation model is constructed, including collecting users' operating behaviors in the system, including search records, click histories, access frequencies, and user role information; analyzing real-time monitoring data of running equipment, including equipment status parameters, fault alarm information, and historical trend data; fusing user behavior data with equipment operation data, and using a weighted factor model to quantify user interest preferences; establishing a personalized recommendation model through machine learning methods, prioritizing fault cases that users are concerned about, and generating preliminary recommendation cases based on recommendation logic.

[0030] It should be noted that the recommendation logic includes, when user behavior data shows that users are concerned about a specific case type that is frequently accessed, priority is given to sorting cases of the same type in the case collection based on access frequency, and the sorting rules are based on a weighted calculation of the user's historical access frequency and the success rate of solving related cases; when the equipment operation status shows abnormal parameter fluctuations and no alarm is triggered, the recommendation logic prioritizes matching cases that are closest to the current operating parameter fluctuation range, and weights and sorts them according to the historical failure frequency and processing time of the equipment associated with the case; when the equipment generates an alarm and is associated with multiple failure modes, the recommendation logic prioritizes cases associated with high-level alarms based on the alarm level, and generates optimized recommended cases based on the complexity of the case's processing suggestions and user role permissions.

[0031] It should also be noted that a preferred solution for establishing a personalized recommendation model through machine learning methods includes the following specific operations: first, extracting key behavioral features based on user historical operation data, which include user search frequency, click preference, visit duration, and feedback behavior on cases, and associating these behavioral features with device operation status data; then, using feature selection methods to screen out behavioral patterns and device parameters that have the greatest impact on recommendation effects, and the screening process is grouped according to the synergy of behavioral frequency and parameter change trends; then, constructing an initial training data set based on the grouping characteristics, dividing the data set into positive samples and negative samples, where positive samples include cases in which recommended results in user historical behaviors are adopted, and negative samples include cases that are not adopted, and the ratio of positive and negative samples is balanced and adjusted Then, a hierarchical model training method is adopted to train the basic classification model through the initial training set. The basic model predicts the accuracy of the recommendation results according to the matching score between user behavior and case characteristics. The results of the basic classification model are input into the secondary optimization model. The secondary optimization model adjusts the weights of different features to enhance the personalized recommendation effect. The specific adjustment is based on the correlation between the user's historical click preference for the case and the device operation data. Finally, the final recommendation model is constructed and the optimized model is applied to the online recommendation scenario. The user's current behavior characteristics and real-time device status are dynamically input. The priority ranking of the recommendation results is generated through real-time calculation, and the recommended cases with the highest priority are pushed to the user. The online recommendation capability of the model is continuously updated and optimized through real-time feedback data.

[0032] It should also be noted that user operation behaviors (such as search records, click history, etc.) are collected to form user portraits; the operating status and alarm information of the equipment are analyzed to extract real-time features; user features and equipment features are weighted and fused to generate recommendation plans through machine learning algorithms; user portraits reflect the user's focus, which helps the system to more accurately meet personalized needs when making recommendations; the analysis of the equipment's operating status ensures the relevance of the recommended content to the actual equipment situation, avoiding the problem of being out of touch with the scenario; through comprehensive analysis of user and equipment characteristics, the system can generate recommendation plans that are more in line with user expectations and actual fault scenarios; this not only improves the accuracy of recommendations, but also provides users with more efficient and targeted fault resolution references.

[0033] S3: Generate and push prioritized recommendation cases through semantic matching and dynamic analysis.

[0034] Furthermore, semantic matching includes constructing semantic representations of pronouns and referential phrases in the fault case text, establishing a candidate set of referential entities based on the pronoun context information and syntactic dependencies; screening the optimal referential match by calculating the semantic similarity between pronouns and candidate entities; storing the referential matching results in the knowledge graph and updating the semantic descriptions of related entities.

[0035] It should be noted that generating and pushing priority recommended cases includes: performing natural language processing on keywords or phrases input by users to extract key entities and relationships; retrieving case data related to the extracted key entities in the knowledge graph, including similar fault cases, equipment-related information, and solutions; combining real-time operation data analysis to filter case data that best matches the current equipment status; sorting the filtering results according to user portraits and case relevance, and generating a priority recommendation list; and pushing the highest priority case information to the user terminal through the dynamic push function, including detailed case content, related equipment, and recommended processing suggestions.

[0036] It should also be noted that natural language processing technology is used to parse user input and extract key entities and their relationships; relevant case data is retrieved and extracted from the knowledge graph, and dynamic analysis is performed in combination with device status; user portraits and case relevance are sorted to generate and push recommendation lists; semantic matching technology ensures the system's deep understanding of user needs and avoids the shortcomings of simple keyword matching; dynamic push combines user portraits and device status to achieve instant response and personalized customization of recommended content; through this process, the system can push cases that are highly matched with device status and user needs directly to the user terminal, reducing the time cost of user search and screening, and improving troubleshooting efficiency and user experience.

[0037] Example 2 is an embodiment of the present invention, which provides a method for recommending personalized fault cases for power plants. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0038] Firstly, a simulation experiment platform was built. The experimental data came from the historical fault cases of the power plant, real-time equipment operation data and user behavior records. The experiment used 1,000 randomly selected historical fault data and real-time monitoring data of 10 devices. The historical fault data of the power plant was collected from multi-source data. The entity disambiguation technology was used to identify entities with the same name but different meanings, and the semantically consistent entities were merged into a unique representation. The reference resolution technology was used to parse the specific reference of the pronouns in the case text. Subsequently, the cleaned data was annotated through the knowledge graph technology, and a multi-dimensional relationship network between equipment and faults was constructed. The relationship weights were dynamically adjusted in combination with historical data. Finally, a knowledge graph model containing equipment attributes, failure modes and maintenance suggestions was generated. User operation behaviors on the simulation platform were collected, including search frequency, click records and access preferences; at the same time, status parameters and alarm data are obtained from real-time equipment operation monitoring; user behavior is combined with equipment operation data, a weighted factor model is used to quantify user interest preferences, and a recommendation model is trained through machine learning methods; the model generates priority-ranked recommendation plans based on the adoption rate of user historical behavior, the relevance of recommended cases and real-time equipment status; users enter keywords in the system to trigger the recommendation process; the system parses the input content through natural language processing technology, extracts key entities and their relationships; combines knowledge graph retrieval with equipment operation data, screens the cases with the highest matching degree and sorts them; finally, the cases with the highest priority are pushed to the user terminal, and the push content includes case description, equipment information and recommendation reasons; refer to Table 1 to record and analyze the experimental data.

[0039] Table 1 Experimental data record table

[0040]

[0041] Experimental data show that after entity disambiguation and reference resolution, the total number of fault entities in each data set was reduced by about 20%-30%; this data shows that the knowledge graph construction process of the present invention effectively solves the problems of data redundancy and conflict, and improves the consistency and accuracy of data; the recommendation accuracy of each data set is maintained between 84% and 89%, which is about 15% higher than the traditional static recommendation method on average; this is because the system combines user behavior and equipment operating status, realizes multi-dimensional feature fusion through weighted factor model and machine learning method, and improves the relevance of recommended content; the time consumption of pushing recommended cases is within the range of 2.0-2.5 seconds, which meets the efficiency requirements of real-time decision-making of power plants; this effect is due to the efficient retrieval capability of the knowledge graph and the fast processing capability of semantic matching technology; the average user satisfaction score is 8.85, indicating that the recommended cases generated by the system can highly match user needs and provide effective decision-making support; the dynamic push module generates recommended content in real time based on user portraits and equipment status, ensuring the immediacy and pertinence of recommendations.

[0042] Example 3, reference Figure 2 , which is an embodiment of the present invention, provides a personalized fault case recommendation system for a power plant, including a data fusion module 100, a case generation module 200, and a case push module 300.

[0043] Among them, S4: the data fusion module 100 includes a multi-source data cleaning module 101 and an entity relationship modeling module 102. The multi-source data cleaning module 101 is used to solve the redundancy and conflict problems in different data sources through entity disambiguation and reference resolution technology, and the entity relationship modeling module 102 is used to build the association relationship between equipment, faults and solutions through knowledge graph technology.

[0044] It should also be noted that the knowledge graph constructed by the entity relationship modeling module 102 provides the user behavior analysis module 201 and the operation data processing module 202 with the association information between equipment and faults, so that the recommendation model 203 can generate accurate recommendations based on a high-quality knowledge base.

[0045] S5: The case generation module 200 includes a user behavior analysis module 201, an operation data processing module 202, and a personalized recommendation model module 203. The user behavior analysis module 201 is used to collect and analyze the user's operation behavior and extract the user's focus. The operation data processing module 202 is used to extract key features from the real-time operation data of the device. The personalized recommendation model module 203 is used to generate recommendation solutions using machine learning methods.

[0046] It should also be noted that the preliminary recommendation scheme generated by the personalized recommendation model module 203 is further optimized by the semantic analysis module 301, and the case screening module 302 screens the priority recommended content according to the knowledge graph and operation data.

[0047] S6: The case push module 300 includes a semantic analysis module 301, a case screening module 302, and a dynamic push module 303. The semantic analysis module 301 is used to analyze the keywords input by the user through natural language processing technology. The case screening module 302 is used to filter case data that matches the device status and user needs from the knowledge graph. The dynamic push module 303 is used to push the sorted cases to the user terminal to assist decision-making.

[0048] It should also be noted that the dynamic push module 303 collects user feedback information, transmits the newly added data back to the multi-source data cleaning module 101 and the entity relationship modeling module 102, updates the knowledge graph in real time, and forms the dynamic optimization capability of the system.

[0049] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0050] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0051] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0052] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0053] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A personalized fault case recommendation method for a power plant, characterized in that: include: Use knowledge graphs to perform multi-source fault data fusion and entity relationship modeling; Build personalized recommendation models based on operational data and user behavior and generate recommendation cases; Generate and push prioritized recommendation cases through semantic matching and dynamic analysis.

2. The method for recommending personalized fault cases for a power plant according to claim 1, characterized in that: The multi-source fault data fusion, include, Cleaning is performed through entity disambiguation to identify faults with the same name and merge them into unique entities; Co-reference resolution is adopted to link the same referents in different fault cases through semantic analysis to generate consistent data representation.

3. The method for recommending personalized fault cases for a power plant according to claim 1 or 2, characterized in that: The entity relationship modeling includes: The cleaned fault data is labeled with entities through the knowledge graph, and the equipment attributes, fault modes, and maintenance suggestions are semantically classified and labeled; Build a relationship network based on the association between entities, use a graph structure to define entity nodes and relationship edges, and dynamically calculate the weight of the relationship edge through historical data analysis; Optimize the relationship network through the semantic rule engine, merge redundant relationships and remove weakly related entity edges; Generate the final knowledge graph model, which includes the multi-dimensional relationship between equipment type, fault characteristics and corresponding solutions.

4. The method for recommending personalized fault cases for a power plant according to claim 1, characterized in that: The building of a personalized recommendation model includes: Collect user operation behaviors in the system, including search records, click history, access frequency, and user role information; Analyze the real-time monitoring data of running equipment, including equipment status parameters, fault alarm information and historical trend data; The user behavior data and the device operation data are integrated, and the weighted factor model is used to quantify the user's interest preferences. A personalized recommendation model is established through machine learning methods to prioritize the fault cases that users are concerned about, and generate preliminary recommendation cases based on the recommendation logic.

5. The method for recommending personalized fault cases for a power plant according to claim 4, characterized in that: The recommendation logic includes: When user behavior data indicates that users are interested in a specific case type that is frequently accessed, the case collection will be prioritized for sorting based on access frequency. The sorting rule is based on a weighted calculation of the user's historical access frequency and the success rate of solving related cases. When the equipment operation status shows abnormal parameter fluctuations and no alarm is triggered, the recommendation logic prioritizes matching the cases that are closest to the current operating parameter fluctuation range, and weights the cases according to the historical fault frequency and processing time of the equipment associated with the cases; When a device generates an alarm and is associated with multiple failure modes, the recommendation logic prioritizes cases associated with high-level alarms based on the alarm level, and generates optimized recommended cases based on the complexity of the case's processing suggestions and user role permissions.

6. The method for recommending personalized fault cases for a power plant according to claim 1, characterized in that: The semantic matching includes: For the pronouns and referential phrases in the fault case text, the semantic representation of the pronouns is constructed, and the referential entity candidate set is established based on the pronoun context information and syntactic dependencies; By calculating the semantic similarity between pronouns and candidate entities, the optimal referential match is selected; The reference matching results are stored in the knowledge graph and the semantic descriptions of related entities are updated.

7. The method for recommending personalized fault cases for a power plant according to claim 1, 2, 4 or 6, characterized in that: The generation and push of priority-ranked recommendation cases includes: Perform natural language processing on keywords or phrases input by users to extract key entities and relationships; Retrieve case data related to the extracted key entities in the knowledge graph, including similar fault cases, equipment association information, and solutions; Combined with real-time operation data analysis, filter the case data that best matches the current equipment status; Sort the screening results by user profile and case relevance, and generate a priority recommendation list; The dynamic push function pushes the highest priority case information to the user terminal, including the case details, associated devices, and recommended processing suggestions.

8. The personalized fault case recommendation system for power plants is characterized by: It includes a data fusion module (100), a case generation module (200), and a case push module (300); The data fusion module (100) comprises a multi-source data cleaning module (101) and an entity relationship modeling module (102), wherein the multi-source data cleaning module (101) is used to solve redundancy and conflict problems in different data sources through entity disambiguation and reference resolution technology, and the entity relationship modeling module (102) is used to construct an association relationship between equipment, faults and solutions through knowledge graph technology; The case generation module (200) comprises a user behavior analysis module (201), an operation data processing module (202), and a personalized recommendation model module (203), wherein the user behavior analysis module (201) is used to collect and analyze the user's operation behavior and extract the user's focus, the operation data processing module (202) is used to extract key features from the real-time operation data of the device, and the personalized recommendation model module (203) is used to generate a recommendation plan using a machine learning method; The case push module (300) comprises a semantic analysis module (301), a case screening module (302), and a dynamic push module (303). The semantic analysis module (301) is used to analyze keywords input by a user through natural language processing technology. The case screening module (302) is used to screen case data that matches the device status and user needs from the knowledge graph. The dynamic push module (303) is used to push the sorted cases to the user terminal to assist in decision-making.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for recommending personalized fault cases for a power plant according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for recommending personalized fault cases for a power plant according to any one of claims 1 to 7 are implemented.

Images