CN-121981240-A - Intelligent management method and device for fault data of wind turbine generator

CN121981240ACN 121981240 ACN121981240 ACN 121981240ACN-121981240-A

Abstract

The invention relates to the technical field of wind turbine generator fault data management, and discloses an intelligent management method and device for wind turbine generator fault data, wherein the method can completely record the whole process data from normal operation to performance degradation to fault of equipment by correlating time sequence relations of four types of entities and constructing a time sequence knowledge graph; and finally, the structured time sequence knowledge graph is combined with an intelligent agent, so that data and conclusions generated by each fault analysis can be precipitated into reusable fault data assets, and the data can be continuously updated and optimized along with data accumulation, thereby providing stable data support for long-term fault prediction and health management of the wind turbine generator.

Inventors

GUO LIANG
KONG DETONG
ZHOU YUHAO
WU YAWEN
PAN QIAOBO
HUANG YUHAO
CUI GUANG
MA YUEMING
ZHAO ZIXUAN
ZHANG DAOQUAN
ZHANG LEPING

Assignees

华电电力科学研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20251216

Claims (10)

1. An intelligent management method for fault data of a wind turbine generator is characterized by comprising the following steps: Acquiring operation log data of a wind turbine generator, and inputting the operation log data into a preset target recognition model; Identifying a wind turbine generator entity, a fault component entity, a fault type entity and a fault case entity in the operation log data through the target identification model; associating time sequence relations respectively corresponding to the wind turbine generator system entity, the fault component entity, the fault type entity and the fault case entity; Constructing a target time sequence knowledge graph based on the wind turbine generator entity, the fault component entity, the fault type entity, the fault case entity and the time sequence relations respectively corresponding to the fault type entity and the fault case entity; and constructing an intelligent body according to the target time sequence knowledge graph of the wind turbine and the multiple databases.
2. The method according to claim 1, wherein the obtaining the operation log data of the wind turbine generator and inputting the operation log data into a preset target recognition model includes: Acquiring operation log data of a wind turbine generator; performing data cleaning treatment on the operation log data; and inputting the operation log data after the data cleaning treatment into a preset target recognition model.
3. The method according to claim 2, wherein the inputting the log data after the data cleansing process into the preset target recognition model includes: Acquiring initial historical operation log data of the wind turbine generator; Performing data cleaning processing on the initial historical operation log data to generate updated historical operation log data; Dividing the updated historical running log data into a training set and a testing set according to a preset proportion; Inputting the updated historical running log data of the training set into a preset initial recognition model for training, and generating an updated recognition model; inputting the updated historical operation log data of the test set into the updated identification model for testing, and generating a target identification model; and inputting the operation log data after the data cleaning processing into the target recognition model.
4. The method according to claim 1, wherein said associating the respective corresponding timing relationships of the wind turbine entity, the faulty component entity, the fault type entity and the fault case entity comprises: Inputting the operation log data, the wind turbine generator system entity, the fault component entity, the fault type entity and the fault case entity into a preset time sequence relation classification model; respectively extracting time sequence information matched with the wind turbine generator set entity, the fault component entity, the fault type entity and the fault case entity through the time sequence relation classification model; And establishing a time sequence relation corresponding to the wind turbine generator entity, the fault component entity, the fault type entity and the fault case entity respectively based on the wind turbine generator entity, the fault component entity, the fault type entity and the fault case entity and the time sequence information corresponding to each other.
5. The method according to claim 1, wherein the constructing a target timing knowledge graph based on the wind turbine entity, the fault component entity, the fault type entity, and the fault case entity, and the respective corresponding timing relationships, comprises: constructing an initial time sequence knowledge graph by adopting the wind turbine generator entity, the fault component entity, the fault type entity, the fault case entity and the time sequence relations respectively corresponding to the fault type entity and the fault case entity; Acquiring real-time operation log data of the wind turbine generator, and extracting fault information of the real-time operation log data; Respectively extracting fault data matched with the wind turbine generator set entity, the fault component entity, the fault type entity and the fault case entity from the fault information; Constructing structural fault case data by adopting the wind turbine generator entity, the fault component entity, the fault type entity, the fault case entity and the fault data respectively corresponding to the fault case entity; And storing the structured fault case data into the initial time sequence knowledge graph, updating the initial time sequence knowledge graph, and generating a target time sequence knowledge graph.
6. The method as recited in claim 5, further comprising: when the intelligent agent receives a fault data retrieval request, acquiring a database query condition corresponding to the fault data retrieval request; Retrieving, from each of the databases, device operation time sequence data, state monitoring data, structured fault case data of the device operation time sequence data, and structured fault case data of the state monitoring data, which are matched with the database query conditions; and packaging the equipment operation time sequence data, the state monitoring data and the respectively corresponding structured fault case data, generating a fault data packet and sending the fault data packet.
7. An intelligent management device for fault data of a wind turbine generator, which is characterized by comprising: The acquisition module is used for acquiring the operation log data of the wind turbine generator and inputting the operation log data into a preset target identification model; The identification module is used for identifying a wind turbine generator entity, a fault component entity, a fault type entity and a fault case entity in the running log data through the target identification model; The association module is used for associating the time sequence relations respectively corresponding to the wind turbine generator system entity, the fault component entity, the fault type entity and the fault case entity; The first construction module is used for constructing a target time sequence knowledge graph based on the wind turbine generator entity, the fault component entity, the fault type entity, the fault case entity and the time sequence relations respectively corresponding to the fault type entity and the fault case entity; And the second construction module is used for constructing the intelligent body according to the target time sequence knowledge graph of the wind turbine and the plurality of databases.
8. An electronic device, comprising: the intelligent management system comprises a memory and a processor, wherein the memory and the processor are in communication connection, the memory stores computer instructions, and the processor executes the computer instructions, so that the intelligent management method for the fault data of the wind turbine is executed.
9. A computer-readable storage medium, wherein computer instructions for causing a computer to execute the wind turbine fault data intelligent management method according to any one of claims 1 to 6 are stored on the computer-readable storage medium.
10. A computer program product comprising computer instructions for causing a computer to perform the wind turbine fault data intelligent management method of any one of claims 1 to 6.

Description

Intelligent management method and device for fault data of wind turbine generator Technical Field The invention relates to the technical field of wind turbine generator fault data management, in particular to an intelligent wind turbine generator fault data management method and device. Background In the long-term running process of the wind turbine generator, massive operation and maintenance data can be continuously generated, and SCADA data for recording the running state of equipment, CMS vibration data for monitoring the vibration condition of components, operation logs manually recorded by operation and maintenance personnel and the like are covered. The data contains rich fault related information, and is a core basis for developing unit fault diagnosis and maintenance decision. However, the current wind turbine generator system fault data presents significant complexity characteristics, the data types are various, the formats are not uniform, and the fault information is scattered in different data sources, so that great challenges are brought to effective management and utilization of the fault data, and if the fault data cannot be properly processed, the running reliability and the running maintenance efficiency of the wind turbine generator system are seriously affected. Therefore, the method generally adopts a knowledge graph mode to manage fault data of the wind turbine generator, and builds entity association through the graph to integrate the data, but the method is static graph, only can realize basic association function, does not introduce time attribute, can not model dynamic data such as equipment health time sequence, fault state switching time stamp and the like, is difficult to embody the time sequence evolution process of equipment from normal to degradation to fault, needs repeated retrieval and arrangement during subsequent fault analysis and model training, has high data multiplexing cost, and is difficult to form recyclable fault data assets. Disclosure of Invention The invention provides an intelligent management method and device for fault data of a wind turbine generator, which are used for solving the problems that the prior art is difficult to embody the time sequence evolution process of equipment from normal to degradation to fault, repeated retrieval and arrangement are needed during subsequent fault analysis and model training, the data multiplexing cost is high, and recyclable fault data assets are difficult to form. In a first aspect, the invention provides an intelligent management method for fault data of a wind turbine, which comprises the following steps: Acquiring operation log data of a wind turbine generator, and inputting the operation log data into a preset target recognition model; Identifying a wind turbine generator entity, a fault component entity, a fault type entity and a fault case entity in the operation log data through the target identification model; associating time sequence relations respectively corresponding to the wind turbine generator system entity, the fault component entity, the fault type entity and the fault case entity; Constructing a target time sequence knowledge graph based on the wind turbine generator entity, the fault component entity, the fault type entity, the fault case entity and the time sequence relations respectively corresponding to the fault type entity and the fault case entity; and constructing an intelligent body according to the target time sequence knowledge graph of the wind turbine and the multiple databases. The method can completely record the whole process data from normal operation to performance degradation to failure of the equipment through correlating the time sequence relations of the four types of entities and constructing the time sequence knowledge graph, clearly presents the time sequence evolution track of normal, degradation and failure, solves the problem that the process cannot be embodied in the prior art, constructs and correlates the scattered failure related data, does not need to repeatedly search and sort data from massive logs when failure analysis or model training is carried out subsequently, directly invokes structural information in the knowledge graph, greatly reduces the data multiplexing cost, combines the structured time sequence knowledge graph with an intelligent agent, can precipitate the data and conclusions generated by each failure analysis into reusable failure data assets, and can be updated and optimized continuously along with the data accumulation, thereby thoroughly solving the dilemma that the prior art is difficult to form recyclable failure data assets and providing stable data support for long-term failure prediction and health management of wind power units. In an optional implementation manner, the acquiring the operation log data of the wind turbine generator and inputting the operation log data into a preset target recognition model includes: Acquiring operation lo