Fault Diagnosis Of Doubly-fed Wind Turbine Based On Data Driven Modeling

With the support of energy policies and the rapid development of science and technology,wind power technology at home and abroad has continued to improve on the road of innovation and development,and the type of wind power crew is constantly updating iteration.The current mainstream models are three types of transmission constant frequencies.The wind turbine is straight-drive,semi-straight-drive and dual-feed wind turbines.At the same time,the installed capacity of the wind turbine has continued to rise.As the wind turbine is mostly running in areas such as farmland,mountains,deserts,offshore,and even deep-sea,it is facing problems such as high failure rate and difficulty in maintaining maintenance.Monitoring,fault diagnosis and fault warning have become an important research direction in the industry todayAdvanced technology and technologies such as 5G communication networks,big data,artificial intelligence,digital twin and other advanced technology and technologies are continuously applied and used in the production,manufacturing,and monitoring of wind turbines.Essence In order to accurately and timely monitor the operating status of the wind turbine,drive the study of the important components of wind turbines with actual historical data or actual-time real-time data,and the research on important components of wind turbines has become the focus of domestic and foreign scholars’ attention.This article aims to use a data-driven method to use the double-feeding asynchronous wind turbine as the research goal,and conducts fault diagnosis and analysis through the mechanism model.First analyze the operation mechanism and energy transfer relationship between the dual-feeding asynchronous wind turbine units and the energy transmission relationship between the components,and build the data driving model of each component of the dual-feeding asynchronous wind turbine unit in Matlab/Simulinks.Machine models,transmission chain models,asynchronous generator models,side and network side-by-side flow models,and control strategies.Secondly,the idea of digital twins is based on the measured data to identify the important parameters in each component model in the data-driven model.In the drive model,the mapping relationship that can reflect the operating status of the wind turbine through the measured data,and it can be found.Through the historical data of a wind farm in Inner Mongolia,the mapping relationship of the data-driven model and the crew of the inner Mongolia,and continuously modified the model based on the test results.As a result,this method can well build a mapping relationship between the model and the unit.Finally,the data-driven model based on building a mapping relationship is simulated by using the measured historical data before the wind turbine failure,the method of comparing the results of the simulation with the measured data,and the diagnosis of the generator failure of the dual-feeding asynchronous wind turbine unit analyze.Analysis and verification of cases of failure in the measured historical data of a certain wind farm were analyzed.As a result,it proved that the method could find the abnormal state before the failure of the generator.