Power plant abnormal data processing method based on knowledge graph

Abstract

The invention relates to the technical field of power plant data processing, and discloses a power plant abnormal data processing method based on a knowledge graph, which comprises the following steps: identifying entities and relationships from structured and unstructured data by utilizing a knowledge graph structured knowledge representation method of the knowledge graph; a complex data association relationship in power plant production is comprehensively captured and expressed, meanwhile, the occurrence condition of abnormal data is matched by utilizing knowledge fusion of a knowledge graph, and a corresponding reference processing method is provided; according to the method, more accurate operation data analysis is realized, and the processing strain capacity of abnormal conditions of the basic-level power plant is improved.

Description

Technical Field

[0001] The present application relates to the technical field of power plant data processing, and specifically to a method for processing abnormal data of a power plant based on a knowledge graph. Background Art

[0002] During generator set operation, power plant systems and personnel must constantly identify and process data that may appear within a short period of time and deviate from normal operating patterns. This abnormal data may be caused by equipment failure, operational errors, external interference, or other unexpected events, potentially impacting the stability and reliability of generator set operations. Therefore, timely and accurate processing of this transient abnormal data is crucial to ensuring the normal operation of power generation.

[0003] Currently, grassroots power generation units rely primarily on real-time monitoring and early warning systems to detect transient anomalies in data. These systems primarily monitor the operating status of the entire system, such as busbar voltage, main steam temperature, and drum water level, through sensors installed on key equipment. When monitored data exceeds normal ranges, the system immediately issues audible and visual warning signals, prompting operators to respond promptly. The real-time monitoring and early warning system primarily preprocesses the monitored data through simple data cleaning and aggregation methods. Simple data cleaning, based on a voting mechanism used in on-off control, uses pre-set rules to determine the final value for data collected from multiple similar or identical data sources, thereby verifying and eliminating redundancy. Data aggregation methods, based on common data aggregation techniques such as averaging, assume that most data points are accurate and reliable, and that outliers or errors are rare. This results in a more stable and representative value, reducing data variability and noise.

[0004] This data processing method makes the existing real-time monitoring and early warning system have the advantages of rapid response and comprehensive anomaly capture, but it still has some defects and shortcomings:

[0005] 1. Misjudgment: If abnormal data appears continuously, the majority voting mechanism may make the system unable to effectively identify the true abnormal trend; or if abnormal data appears in isolation, the system cannot determine whether the device has a real problem, thereby triggering unnecessary protection actions.

[0006] 2. Insensitivity to abnormal trends: The system will only trigger an audible and visual alarm when the operating data reaches the alarm value. If the equipment operating data is approaching the alarm value but has not yet reached the alarm value, the system will not issue an early warning, thus depriving the operation and maintenance personnel of the opportunity to prevent accidents.

[0007] 3. Insufficient dynamic updates: The existing DCS system control logic is continuously updated based on equipment updates and personnel design. After the unit logic is modified, the knowledge base update is delayed. During the personnel design process, logical loopholes are easily caused, which requires a lot of time and cost for repeated analysis and deduction. It is still difficult to ensure that the control logic covers various situations comprehensively.

[0008] 4. Unreliability of accident handling: The existing system can only warn operation and maintenance personnel of abnormal situations through sound and light alarms. However, the accident handling process is too dependent on the status and skill level of the personnel. If the personnel's response is not quick enough and the judgment is not accurate enough, the scope of the accident will be further expanded, greatly increasing the unreliability of accident handling. Summary of the Invention

[0009] The purpose of this application is to provide a power plant abnormal data processing method based on knowledge graph, which enables power plant operation and maintenance personnel to respond to abnormal data more quickly, judge abnormal situations more accurately, handle abnormal accidents more precisely, and reduce the risk of accidents caused by abnormal data being amplified.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] The present invention provides a method for processing abnormal data of a power plant based on a knowledge graph, comprising the following steps:

[0012] Step 1: Obtain the unit historical operation structured data set, including the unit historical normal operation data set and the unit historical abnormal operation data set of the power plant, clean the unit historical operation structured data set, and define tags;

[0013] Step 2: Obtain the unstructured data set of unit management, including the procedure and specification data set. Based on the tag definition of the unit historical operation structured data set, perform knowledge extraction on the procedure and specification data set to obtain a structured procedure and specification data set.

[0014] Step 3: Perform knowledge fusion on the labeled and defined historical normal operation dataset of the unit and the structured procedure and specification dataset to obtain a power plant data logical relationship map that fully describes the physical operation standards of the power plant;

[0015] Step 4: Obtain the unstructured data set of unit management including the exception processing data set, extract knowledge from the exception processing data set according to the logical relationship diagram of the power plant data, and obtain a structured exception processing data set;

[0016] Step 5: Perform knowledge fusion on the labeled and defined historical abnormal operation dataset of the unit and the structured abnormal processing dataset to obtain an abnormal case map that fully describes the abnormal operation of the power plant entity;

[0017] Step 6: Obtain the real-time operation data of the power plant's units, and determine the unit's operating status based on the power plant data's logical relationship diagram;

[0018] Step 7: If normal, convert the real-time operation data of the unit into the historical normal operation data of the unit, enter it into the power plant data logical relationship map, and update the power plant data logical relationship map;

[0019] Step 8: If an abnormality is detected, further abnormality analysis and processing will be performed on the abnormal situation and trend according to the power plant data logic map. If it is a known historical abnormality, a solution will be output to the operation and maintenance personnel based on the abnormal case map. If it is an unknown abnormality, the maintenance personnel will be notified for further inspection to obtain abnormality report data.

[0020] Step 9: Integrate the abnormal operation data of the unit, enter the abnormal case map, update the abnormal case map, and output a standardized accident report.

[0021] Furthermore, the method for data cleaning the unit historical operation structured data set includes: using the Python Pandas library to define cleaning rules to eliminate outliers exceeding ±3σ and equipment maintenance period data, and clearing data in the time period of equipment maintenance, feedback loss, and exceeding the upper and lower limit thresholds to obtain valid unit historical operation structured data.

[0022] Furthermore, the method for defining and labeling the cleaned unit historical operation structured data set includes: labeling entities in the unit historical operation structured data set, and defining attributes and relationships of corresponding entities.

[0023] Furthermore, the method for extracting knowledge from the crew management unstructured data set includes: using the BERT-Base model to identify entities in the crew management unstructured data set, and constructing an entity-relationship graph through Neo4j.

[0024] Furthermore, a method for performing knowledge fusion on the labeled and defined historical normal operation dataset of the unit and the structured procedure and specification dataset to obtain a power plant data logical relationship map for completely describing the physical operation standards of the power plant includes:

[0025] Align the entities, attributes, and relationships of the historical normal operation dataset of the unit with the entities, attributes, and relationships of the structured procedure specification dataset to obtain a logical relationship map of the power plant data;

[0026] Furthermore, a method for fusing the labeled historical abnormal operation dataset of the unit with the structured abnormal processing dataset to obtain an abnormal case map for fully describing the abnormal operation of the power plant entity includes:

[0027] The entities, attributes, and relationships of the group's historical abnormal operation dataset are aligned with the entities, attributes, and relationships of the structured exception handling dataset to obtain an abnormal case graph. The exception handling dataset includes emergency plans, technical measures, and accident report data.

[0028] Furthermore, if there is an abnormality, the method for further abnormal analysis and processing of the abnormal situation and trend based on the power plant data logic map includes:

[0029] According to the power plant data logic map, if the real-time operation data of the unit is normal but has an abnormal trend or is abnormal but has not triggered the main protection, the maintenance personnel will be notified to further investigate the cause of the abnormality, output the abnormal operation report, and enter the abnormal operation report into the abnormal case map;

[0030] According to the power plant data logic map, if the real-time operation data of the unit is abnormal and the main protection is triggered, the matching degree is calculated according to the power plant data logic map, and the case with the highest matching degree is aligned and output;

[0031] The abnormal case map is used to perform path retrieval, and the processing path is fed back to the operation and maintenance personnel for reference and selection. A processing plan is generated based on the processing path selected by the operation and maintenance personnel, and the processing plan is fed back to the power plant's main control logic system.

[0032] Furthermore, the GNN algorithm is used to calculate the matching degree, and the cases where the similarity of the matching degree threshold parameter is greater than or equal to 70% are aligned and output.

[0033] A power plant abnormal data processing system based on knowledge graph, including:

[0034] A multi-source data integration module for obtaining the power plant's unit historical operation structured data set, unit management unstructured data set, and unit operation real-time data;

[0035] Knowledge graph building blocks for data processing;

[0036] Real-time decision-making module for exception handling and decision support.

[0037] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements any one of the methods for processing abnormal data of a power plant based on a knowledge graph.

[0038] Compared with the prior art, the present invention has the following advantages:

[0039] 1. Data Association: This invention introduces knowledge graph technology into the power plant abnormal data processing and decision support system. Through the knowledge graph's structured knowledge representation method, it identifies entities and relationships, as well as their attributes, from unstructured data. This provides a graph-based data structure that facilitates rapid semantic search and reasoning, and can comprehensively capture and express the complex data associations in power plant production.

[0040] 2. Data Integration: This invention can efficiently integrate data from different sources (such as sensor data, operation logs, historical fault records, technical measures, and procedures and specifications), not limited to structured, single data sources. It builds a unified data view and forms a comprehensive power system knowledge base to achieve better knowledge sharing and inheritance.

[0041] 3. Automated Intelligent Diagnosis: This system can store knowledge about power plant equipment operation and control. By analyzing abnormal data and conducting knowledge reasoning, combined with expert knowledge and machine learning algorithms, it can perform historical matching and in-depth analysis of abnormal situations to achieve intelligent diagnosis.

[0042] 4. Interactive Decision Support: This invention uses knowledge graphs to provide decision support features, helping operations personnel quickly respond to abnormal events; for example, providing graph alignment;

[0043] 5. Real-time: The knowledge graph framework can process real-time data and provide rapid responses to abnormal unit operations;

[0044] 6. Flexibility and scalability: The knowledge graph framework can easily add new data sources and knowledge fields to meet the ever-evolving needs of grassroots power plant operations and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the implementation methods of the present application, the following is a brief introduction to the drawings required for use in the implementation methods. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 It is a schematic diagram of the method flow of this application;

[0047] Figure 2 This is the system architecture diagram of this application;

[0048] Figure 3 This is a flowchart of the power plant system knowledge graph construction method of this application;

[0049] Figure 4This is a flow chart of the power plant operation and maintenance data processing and auxiliary decision-making system based on knowledge graph in this application. DETAILED DESCRIPTION

[0050] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0051] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention. Therefore, the drawings only show the layers related to the present invention and are not drawn according to the number, shape and size of the layers in actual implementation. In actual implementation, the type, quantity and proportion of each layer can be changed arbitrarily, and the layer layout type may also be more complicated.

[0052] In the following description, numerous details are set forth to provide a more thorough explanation of the embodiments of the present invention; however, it is apparent to one skilled in the art that the embodiments of the present invention may be practiced without these specific details.

[0053] See also Figures 1 to 4 A power plant abnormal data processing method based on knowledge graph includes: obtaining historical operation structured data of power plant units, including normal historical operation data sets and abnormal historical operation data sets of power plant units, and performing data cleaning on the structured data sets, extracting useful information from the original data, removing noise and redundancy, and obtaining effective normal historical operation data sets and abnormal historical operation data sets of units.

[0054] Data cleaning uses the Python Pandas library to define cleaning rules to eliminate outliers exceeding ±3σ and data during equipment maintenance periods. Data during equipment maintenance, feedback loss, and data exceeding upper and lower thresholds will be cleared.

[0055] The entities in the valid historical normal operation data set and the historical abnormal operation data set of the unit are marked, and the attributes and relationships of the marked entities are defined.

[0056] The unstructured data sets of the power plant are obtained, including procedures and specifications, emergency plans, technical measures, and accident report data sets. Natural language processing (NLP) technology is used to extract knowledge from the unstructured data sets including the procedures and specifications data sets based on the entities, attributes, and relationships defined in the valid historical normal operation data sets and historical abnormal operation data sets of the units, and the procedures and specifications data sets are converted into structured procedures and specifications data sets.

[0057] The knowledge fusion of the historical normal operation data set of the unit that has been marked and defined is carried out with the structured procedure and specification data set to obtain the power plant data logical relationship map that is used to fully describe the power plant entity operation standards.

[0058] According to the logical relationship diagram of power plant data, knowledge extraction is performed on the exception processing data set to obtain a structured exception processing data set. The exception processing data set includes emergency plans, technical measures, and accident report data.

[0059] The knowledge of the historical abnormal operation dataset of the unit defined by labels and the structured abnormal processing dataset are fused to obtain an abnormal case map that is used to fully describe the abnormal operation of the power plant entity.

[0060] Knowledge extraction uses the BERT-Base model to identify entities in the unstructured data set of unit management (such as "boiler" and "pressure threshold"), and constructs an entity-relationship graph through Neo4j.

[0061] The entities, attributes, and relationships of the unit's historical normal operation dataset were aligned with those of the structured procedures and specifications dataset, achieving entity alignment and conflict resolution. This resulted in a power plant data relationship graph that fully describes the plant's entity operating standards. Simultaneously, the entities, attributes, and relationships of the unit's historical abnormal operation dataset were aligned with those of the structured abnormality handling dataset, achieving entity alignment and conflict resolution. This resulted in an abnormality case graph that fully describes abnormal operation of the plant's entities. For example, the "boiler pressure" entity was aligned (similarity requirement reached 0.91) and merged into a unified graph (12,000 nodes and 35,000 relationships).

[0062] Knowledge fusion is based on the TransE graph embedding algorithm to calculate entity similarity (threshold ≥ 0.85), and conflicting data are verified through manual verification rules (e.g., the latest version is preferred).

[0063] Obtain the real-time operating data of the power plant's units, and determine whether the real-time operating data of the units is abnormal based on the power plant data relationship map.

[0064] The real-time operation data of the unit is calculated through the Flink window, and after an exception is triggered, the power plant data relationship graph is retrieved to match the case.

[0065] If it is abnormal, the real-time operation data of the unit will be judged based on the power plant data relationship map to determine whether it triggers the main protection; if it is normal, the real-time operation data of the unit will be judged to determine whether there is an abnormal trend.

[0066] If the real-time operation data of the unit is normal and has no abnormal trend, the real-time operation data of the unit is converted into historical normal operation data and entered into the power plant data logic map to update the power plant data logic map.

[0067] If the real-time operating data of the unit is abnormal and triggers the main protection, the GNN algorithm is used to calculate the matching degree, align the matching threshold parameter similarity greater than or equal to 70% and output it, use the abnormal case map to perform path retrieval, and feed back the processing path to the operation and maintenance personnel for reference and selection. The processing path selected by the operation and maintenance personnel is used to generate a processing plan, and the processing plan is fed back to the power plant's main control logic system.

[0068] A graph neural network (GNN) is used to calculate the similarity between real-time data and historical cases (cosine similarity ≥ 0.7 is considered a high match). If the match falls below a threshold (e.g., <70%), the results are pushed to the manual interface and the repair results are recorded in the anomaly case map. The repair results are automatically written back to the anomaly case map via the REST API, with an update cycle of ≤ 5 minutes.

[0069] If the real-time operating data of the unit is abnormal but does not trigger the main protection, or the real-time operating data of the unit is normal but has an abnormal trend, the maintenance personnel will be notified to further check the cause of the abnormality, output an abnormal operation report, and enter the abnormal operation report into the abnormal case map.

[0070] Integrate abnormal operation data of the unit, enter the abnormal case map, update the abnormal case map, and output standardized accident reports.

[0071] A power plant abnormal data processing system based on knowledge graph, including:

[0072] A multi-source data integration module for obtaining the power plant's unit historical operation structured data set, unit management unstructured data set, and unit operation real-time data;

[0073] Knowledge graph building blocks for data processing;

[0074] Real-time decision-making module for exception handling and decision support.

[0075] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements any one of the methods for processing abnormal data of a power plant based on a knowledge graph.

[0076] Data cleaning is performed on structured data sets such as the normal historical operation data of power plant units, and data from time periods such as equipment maintenance, feedback loss, and exceeding upper and lower thresholds are removed; at the same time, for non-(semi-)structured data sets such as power plant professional regulations and specifications, knowledge is extracted based on unified digital labels such as entity names, attributes, and relationships in the unit structured data sets, and through knowledge fusion, the unit system data is constructed into an interconnected power plant data logical map.

[0077] For non-(semi-)structured data sets such as emergency plans, technical measures and accident reports issued by the production or technical departments of power plants, as well as structured data sets such as historical abnormal operation data of units, knowledge extraction and fusion of data are performed according to the power plant data logic graph. Then, by introducing time nodes, human intervention nodes and other factors into the associated historical abnormal operation data, further knowledge processing is carried out to construct an abnormal case graph.

[0078] Real-time data is first determined to be abnormal through the existing real-time monitoring and early warning system. It is then input into the system, which uses the power plant's data logic map to determine whether the equipment associated with the data is abnormal. The system then outputs the most compatible scenario (up to three scenarios are output if the matching degree is equal). Based on the abnormal case map, the system provides personnel with a reference treatment method, which is then fed back to the power plant's main control logic system. When abnormal data occurs, the system processes and processes the abnormal historical operating data based on the abnormal case map. It then generates a standardized accident report and referenced standardized emergency response plans and technical measures, which are then entered into the abnormal case map.

[0079] The present invention utilizes the knowledge graph structured knowledge representation method of the knowledge graph to identify entities and relationships from structured and unstructured data, comprehensively capture and express the complex data association relationships in power plant production, and at the same time, utilizes the knowledge fusion of the knowledge graph to match the occurrence of abnormal data and provide corresponding reference processing methods, so that power plant operation and maintenance personnel can respond to abnormal data more quickly, judge abnormal situations more accurately, handle abnormal accidents more precisely, and reduce the risk of accidents caused by abnormal data being amplified.

[0080] The present invention can generate digital tags that are easy to associate and process from multi-source heterogeneous data such as power plant unit control logic, operating data, procedures and specifications, emergency plans, technical measures, accident reports, etc., to form a knowledge graph in the power plant field. Based on the knowledge graph, a system framework that can be used for abnormal data processing and auxiliary decision-making is constructed, which can be used to provide more accurate operating data analysis and improve the ability of grassroots power plants to handle abnormal situations.

[0081] In the above embodiments, although the present invention has been described in conjunction with specific embodiments of the present invention, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. The embodiments of the present invention are intended to cover all such replacements, modifications and variations that fall within the broad scope of the appended claims.

[0082] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. A method for processing abnormal data in a power plant based on a knowledge graph, characterized by: The following steps are involved: Step 1: Obtain the unit historical operation structured data set, including the unit historical normal operation data set and the unit historical abnormal operation data set of the power plant, clean the unit historical operation structured data set, and define tags; Step 2: Obtain the unstructured data set of unit management, including the procedure and specification data set. Based on the tag definition of the unit historical operation structured data set, perform knowledge extraction on the procedure and specification data set to obtain a structured procedure and specification data set. Step 3: Perform knowledge fusion on the labeled and defined historical normal operation dataset of the unit and the structured procedure and specification dataset to obtain a power plant data logical relationship map that fully describes the physical operation standards of the power plant; Step 4: Obtain the unstructured data set of unit management including the exception processing data set, extract knowledge from the exception processing data set according to the logical relationship diagram of the power plant data, and obtain a structured exception processing data set; Step 5: Perform knowledge fusion on the labeled and defined historical abnormal operation dataset of the unit and the structured abnormal processing dataset to obtain an abnormal case map that fully describes the abnormal operation of the power plant entity; Step 6: Obtain the real-time operation data of the power plant's units, and determine the unit's operating status based on the power plant data's logical relationship diagram; Step 7: If normal, convert the real-time operation data of the unit into the historical normal operation data of the unit, enter it into the power plant data logical relationship map, and update the power plant data logical relationship map; Step 8: If an abnormality is detected, further abnormality analysis and processing will be performed on the abnormal situation and trend according to the power plant data logic map. If it is a known historical abnormality, a solution will be output to the operation and maintenance personnel based on the abnormal case map. If it is an unknown abnormality, the maintenance personnel will be notified for further inspection to obtain abnormality report data. Step 9: Integrate the abnormal operation data of the unit, enter the abnormal case map, update the abnormal case map, and output a standardized accident report.

2. The method for processing abnormal data of a power plant based on a knowledge graph according to claim 1, characterized in that: The method for data cleaning of the unit's historical operation structured data set includes: using the Python Pandas library to define cleaning rules to eliminate outliers exceeding ±3σ and equipment maintenance period data, and clearing data during equipment maintenance, feedback loss, and exceeding upper and lower limit thresholds to obtain valid unit historical operation structured data.

3. The method for processing abnormal data of a power plant based on a knowledge graph according to claim 1, characterized in that: The method for defining and marking the cleaned unit historical operation structured data set includes: marking entities in the unit historical operation structured data set, and defining attributes and relationships of corresponding entities.

4. The method for processing abnormal data of a power plant based on a knowledge graph according to claim 3 is characterized in that: The method for extracting knowledge from the unstructured dataset of crew management includes: using the BERT-Base model to identify entities in the unstructured dataset of crew management and constructing an entity-relationship graph through Neo4j.

5. The method for processing abnormal data of a power plant based on a knowledge graph according to claim 4 is characterized in that: The method for fusing knowledge of a labeled and defined historical normal operation dataset of a unit with a structured procedure and specification dataset to obtain a power plant data logical relationship graph for completely describing the physical operation standards of the power plant includes: The entities, attributes and relationships of the unit's historical normal operation dataset are aligned with the entities, attributes and relationships of the structured procedure and specification dataset to obtain a logical relationship graph of the power plant data.

6. The method for processing abnormal data of a power plant based on a knowledge graph according to claim 4 is characterized in that: The method for fusing the labeled and defined historical abnormal operation dataset with the structured abnormal processing dataset to obtain an abnormal case map for fully describing the abnormal operation of the power plant entity includes: The entities, attributes, and relationships of the group's historical abnormal operation dataset are aligned with the entities, attributes, and relationships of the structured exception handling dataset to obtain an abnormal case graph. The exception handling dataset includes emergency plans, technical measures, and accident report data.

7. The method for processing abnormal data of a power plant based on a knowledge graph according to claim 1, characterized in that: If abnormal, further abnormal analysis and processing methods for abnormal conditions and trends are carried out according to the power plant data logic map, including: According to the power plant data logic map, if the real-time operation data of the unit is normal but has an abnormal trend or is abnormal but has not triggered the main protection, the maintenance personnel will be notified to further investigate the cause of the abnormality, output the abnormal operation report, and enter the abnormal operation report into the abnormal case map; According to the power plant data logic map, if the real-time operation data of the unit is abnormal and the main protection is triggered, the matching degree is calculated according to the power plant data logic map, and the case with the highest matching degree is aligned and output; The abnormal case map is used to perform path retrieval, and the processing path is fed back to the operation and maintenance personnel for reference and selection. A processing plan is generated based on the processing path selected by the operation and maintenance personnel, and the processing plan is fed back to the power plant's main control logic system.

8. The method for processing abnormal data of a power plant based on knowledge graph according to claim 7 is characterized in that: The GNN algorithm is used to calculate the matching degree, and the cases where the similarity of the matching degree threshold parameter is greater than or equal to 70% are aligned and output.

9. A power plant abnormal data processing system based on knowledge graph, characterized in that: include: A multi-source data integration module for obtaining the power plant's unit historical operation structured data set, unit management unstructured data set, and unit operation real-time data; Knowledge graph building blocks for data processing; Real-time decision-making module for exception handling and decision support.

10. 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 method according to any one of claims 1 to 8 is implemented.

Images