Knowledge graph-based switching operation intelligent scheduling method and system

Abstract

The invention discloses a switching operation intelligent scheduling method and system based on a knowledge graph, and the method comprises the steps: building a transformer substation power equipment knowledge graph and an operation logic knowledge graph, and fusing the knowledge graph to obtain a switching operation knowledge graph; constructing a reinforcement learning model for solving the switching operation sequence, and obtaining a to-be-checked switching operation task sequence; constructing an anti-error regulation knowledge graph, fusing the anti-error regulation knowledge graph with a power equipment knowledge graph, carrying out deep reinforcement learning through an anti-error algorithm to obtain an anti-error checking knowledge graph, and carrying out anti-error checking to obtain a visual anti-error checking result and a switching operation task sequence; according to the method, the reinforcement learning model is constructed, the switching operation sequence is generated based on the knowledge graph, and meanwhile, anti-error checking is automatically carried out, so that the operation sequence generation efficiency and accuracy are improved, and the anti-error level of the transformer substation is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent dispatching for switching operations, and particularly relates to an intelligent dispatching method and system for switching operations based on a knowledge graph. Background Art

[0002] Electrical equipment is divided into three states: running, standby (cold standby and hot standby), and maintenance. Switching operation refers to the operation of changing the state of equipment from one state to another. By operating disconnectors, circuit breakers, and hanging and removing grounding wires, the electrical equipment is converted from one state to another or the operation mode of the system is changed. Switching operations must implement the operation ticket system and the work supervision system.

[0003] Currently, the switching operations in substations mainly use a graphical topology visualization interface for assistance and an expert knowledge base system for intelligent ticket generation, and then are realized through five-prevention drills. However, the graphical interface assistance method largely depends on the knowledge and cognitive level of on-site operators, is subjective, has a low degree of automation and intelligence, and may result in misoperations due to insufficient knowledge reserves and operating experience, causing power grid safety accidents; while the expert knowledge base system requires timely updating of the expert knowledge base, which is time-consuming and laborious to maintain.

[0004] Therefore, to solve the above problems, it is necessary to develop an intelligent dispatching method and system for switching operations based on a knowledge graph. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and provide an intelligent dispatching method and system for switching operations based on a knowledge graph. By constructing a reinforcement learning model, generating a switching operation sequence based on the knowledge graph, and automatically performing anti-error verification, the efficiency and accuracy of generating the operation sequence are improved, and the anti-error level of the substation is enhanced.

[0006] The purpose of the present invention is achieved as follows: An intelligent dispatching method for switching operations based on a knowledge graph includes: S1. Based on the substation topology data, taking the power equipment in the substation as entities and the associations between the power equipment as relationships, establish a knowledge graph of the power equipment in the substation; S2. Based on the historical switching operation task data, extract the switching operation logic and the switching equipment state update rules, and construct an operation logic knowledge graph; S3. Based on the graph fusion algorithm, fuse the power equipment knowledge graph and the operation logic knowledge graph to obtain a switching operation knowledge graph; S4. Construct a reinforcement learning model for solving the switching operation sequence; S5. Based on the switching operation knowledge graph, combined with the current dispatching task operation scope and the task device status update rules, obtain the switching operation task sequence to be verified through the reinforcement learning model; S6. Construct an anti-error regulation knowledge graph based on the data of switching operation anti-error regulations; S7. Based on the graph fusion algorithm, fuse the power equipment knowledge graph and the anti-error regulation knowledge graph, and perform deep reinforcement learning through the anti-error algorithm to obtain the anti-error verification knowledge graph; S8. Based on the anti-error verification knowledge graph, extract and analyze the obtained switching operation task sequence to be verified, and perform anti-error verification to obtain the visual anti-error verification result; S9. If the verification is passed, obtain the switching operation task sequence and form a switching operation ticket; if the verification fails, issue an alarm.

[0007] Further, in step S1, establishing the substation power equipment knowledge graph includes completing the extraction of entity information through a statistics-based information extraction method, and then combining the obtained entity results with a rule-based information extraction method to extract relationship information for entities and relationship triples, and storing them through the Neo4j graph database.

[0008] Further, the switching operation logic in step S2 includes the task device, the set of operable devices that have an electrical connection relationship with the task device and can switch the task device status, and the relevant path formed between the task device and the operable devices based on the relevant relationship; the switching device status update rule includes the association relationship between the change of the operating state of the task device and the operating states of all relevant operable devices in the path.

[0009] Further, in step S4, constructing the reinforcement learning model for solving the switching operation sequence specifically includes regarding the switching operation knowledge graph as the reinforcement learning training environment, training in an exploratory trial-and-error manner, using the obtained rewards or punishments to guide the switching decision, and adjusting and optimizing the strategy according to the action behavior effect to adapt to the environment and obtain the optimal solution.

[0010] Further, in steps S3 and S7, the graph fusion algorithm uses a pre-trained model for feature extraction and matching to establish the mapping relationship in the multi-source knowledge base.

[0011] A switching operation intelligent dispatching system based on a knowledge graph includes a data acquisition module, a data processing module, a knowledge graph construction module, a knowledge graph fusion module, a reinforcement learning module, an operation sequence generation module, an anti-error verification module, a result output module, a visualization module, and a human-computer interaction module.

[0012] Further, the data acquisition module and the data processing module are used to collect and process various types of data of relevant devices involved in the whole process, which are used as the corpus for constructing the knowledge graph, including structured data and unstructured data.

[0013] Further, the data processing module performs data cleaning, integration, and quality improvement on the structured data, and extracts key information from the unstructured data through information extraction technology.

[0014] Due to the adoption of the above technical solution, the beneficial effects of the present invention are as follows: By collecting various data, a substation electrical equipment knowledge graph, a substation electrical equipment knowledge graph, and an anti-misoperation regulation knowledge graph are respectively established, and through the graph fusion algorithm, they are fused to obtain a switching operation knowledge graph and an anti-misoperation verification knowledge graph. Reinforcement learning models are respectively constructed, and continuously iterated to generate the optimal switching operation sequence, and automatically perform anti-misoperation verification. The efficiency of generating the switching operation sequence is better, the accuracy is higher, and at the same time, the anti-misoperation level of the substation is effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flowchart of the present invention.

[0016] Figure 2 is a system structure block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The technical solution of the present invention will be further specifically described below through embodiments and in combination with the drawings.

[0018] As Figure 1 、 Figure 2 shown, a switching operation intelligent scheduling method based on a knowledge graph includes: S1. Based on the substation topology data, taking the electrical equipment in the substation as entities and the associations between the electrical equipment as relationships, a substation electrical equipment knowledge graph is established; including extracting entity information through a statistics-based information extraction method, and then combining the obtained entity results with a rule-based information extraction method to extract entity and relationship triples to obtain relationship information, and storing it through the Neo4j graph database.

[0019] S2. Based on the historical switching operation task data, extract the switching operation logic and the switching equipment state update rule, and construct an operation logic knowledge graph; wherein, the switching operation logic includes task equipment, an operable equipment set that has an electrical connection relationship with the task equipment and can switch the state of the task equipment, and the relevant paths formed between the task equipment and the operable equipment based on the relevant relationship; the switching equipment state update rule includes the association relationship between the change in the operating state of the task equipment and the operating states of all relevant operable equipment in the path.

[0020] S3. Based on the graph fusion algorithm, fuse the power equipment knowledge graph and the operation logic knowledge graph to obtain the switching operation knowledge graph; the graph fusion algorithm uses a pre-trained model for feature extraction and matching to establish the mapping relationship in the multi-source knowledge base.

[0021] S4. Construct a reinforcement learning model for solving the switching operation sequence; specifically, constructing a reinforcement learning model for solving the switching operation sequence includes regarding the switching operation knowledge graph as the reinforcement learning training environment, training in an exploratory trial-and-error manner, using the obtained rewards or punishments to guide the switching decision-making, adjusting and optimizing the strategy according to the action behavior effects, and adapting to the environment to obtain the optimal solution; after a given task operation, the reinforcement learning model calls the switching operation knowledge graph, obtains the current state of the task equipment, makes a switching decision and executes the action. As the environmental state changes, the reinforcement learning model continuously tries errors until the maximum reward value is obtained, and the optimal action strategy is obtained.

[0022] S5. Based on the switching operation knowledge graph, and combined with the current dispatching task operation scope and the task equipment status update rule, obtain the to-be-verified switching operation task sequence through the reinforcement learning model.

[0023] S6. Construct an anti-error regulation knowledge graph based on the anti-error regulation data of the switching operation.

[0024] S7. Based on the graph fusion algorithm, fuse the power equipment knowledge graph and the anti-error regulation knowledge graph, and perform deep reinforcement learning through the anti-error algorithm to obtain the anti-error verification knowledge graph; the graph fusion algorithm uses a pre-trained model for feature extraction and matching to establish the mapping relationship in the multi-source knowledge base.

[0025] S8. Extract and analyze the obtained to-be-verified switching operation task sequence based on the anti-error verification knowledge graph, and perform anti-error verification to obtain the visualized anti-error verification result.

[0026] S9. If the verification passes, obtain the switching operation task sequence and form a switching operation ticket; if the verification fails, issue an alarm.

[0027] A switching operation intelligent dispatching system based on a knowledge graph, including a data collection module, a data processing module, a knowledge graph construction module, a knowledge graph fusion module, a reinforcement learning module, an operation sequence generation module, an anti-error verification module, a result output module, a visualization module, and a human-computer interaction module.

[0028] Among them, the data collection module and the data processing module are used to collect and process various data of the relevant equipment involved in the whole process respectively, and are used as the corpus for constructing the knowledge graph, including structured data and unstructured data.

[0029] Among them, the data processing module performs data cleaning, integration, and quality improvement on structured data, and extracts key information from unstructured data through information extraction technology.

[0030] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that it is still possible to modify the specific implementation manners of the present invention or make equivalent replacements. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the scope of the claims of the present invention.

Claims

1. An intelligent dispatching method for switching operations based on a knowledge graph, characterized in that: Including: S1. Based on the substation topology data, taking the power equipment in the substation as entities and the associations between power equipment as relationships, establish a knowledge graph of substation power equipment; S2. Based on the historical switching operation task data, extract the switching operation logic and the update rules of the switching equipment status, and construct an operation logic knowledge graph; S3. Based on the graph fusion algorithm, fuse the power equipment knowledge graph and the operation logic knowledge graph to obtain a switching operation knowledge graph; S4. Construct a reinforcement learning model for solving the switching operation sequence; S5. Based on the switching operation knowledge graph, combined with the current dispatching task operation scope and the task equipment status update rules, obtain the switching operation task sequence to be verified through the reinforcement learning model; S6. Construct an anti-error regulation knowledge graph based on the anti-error regulation data of switching operations; S7. Based on the graph fusion algorithm, fuse the power equipment knowledge graph and the anti-error regulation knowledge graph, and perform deep reinforcement learning through the anti-error algorithm to obtain an anti-error verification knowledge graph; S8. Based on the anti-error verification knowledge graph, extract and analyze the obtained switching operation task sequence to be verified, and perform anti-error verification to obtain a visual anti-error verification result; S9. If the verification is passed, obtain the switching operation task sequence and form a switching operation ticket; if the verification fails, issue an alarm.

2. The intelligent scheduling method for switching operation based on a knowledge graph according to claim 1, wherein: In the step S1 of establishing the knowledge graph of substation power equipment, it includes completing the extraction of entity information through a statistics-based information extraction method, and then combining the obtained entity results with a rule-based information extraction method to extract relationship information for entity and relationship triples, and storing them in the Neo4j graph database.

3. The intelligent dispatching method for switching operation based on knowledge graph according to claim 1, characterized in that: In the step S2, the switching operation logic includes the task equipment, the set of operable equipment that has an electrical connection relationship with the task equipment and can switch the status of the task equipment, and the relevant paths formed between the task equipment and the operable equipment based on the relevant relationship; The update rule of the switching equipment status includes the association relationship between the change of the running status of the task equipment and the running status of all relevant operable equipment in the path.

4. The intelligent dispatching method for switching operation based on knowledge graph according to claim 1, characterized in that: In the step S4, constructing a reinforcement learning model for solving the switching operation sequence specifically includes regarding the switching operation knowledge graph as a reinforcement learning training environment, training in an exploratory trial-and-error manner, guiding the switching decision with the obtained rewards or punishments, adjusting and optimizing the strategy according to the action behavior effect, and adapting to the environment to obtain the optimal solution.

5. The intelligent dispatching method for switching operation based on a knowledge graph according to claim 1, characterized in that: In the steps S3 and S7, the graph fusion algorithm uses a pre-trained model for feature extraction and matching to establish the mapping relationship in the multi-source knowledge base.

6. An intelligent dispatching system for switching operations based on a knowledge graph, characterized in that: Including a data acquisition module, a data processing module, a knowledge graph construction module, a knowledge graph fusion module, a reinforcement learning module, an operation sequence generation module, an anti-error verification module, a result output module, a visualization module, and a human-computer interaction module.

7. An intelligent dispatching system for switching operations based on a knowledge graph according to claim 6, characterized in that: The data acquisition module and the data processing module are used to collect and process various data of relevant equipment involved in the whole process respectively, which are used as the corpus for constructing the knowledge graph, including structured data and unstructured data.

8. The intelligent dispatching system for switching operation based on a knowledge graph according to claim 7, wherein: The data processing module performs data cleaning, integration, and quality improvement on structured data, and extracts key information from unstructured data through information extraction technology.

Images