Fault Diagnosis Of Variable Frequency System Of Wind Turbine Based On Kernel Principal Component Analysis

Pitch control system is an important part of wind turbine control system,and its operating status is directly related to whether the wind turbine is safe and reliable operation.The wind turbine Supervisory Control and Data Acquisition(SCADA)system is provided with the function of data acquisition and condition monitoring,which can monitor the operating status of the pitch control system of wind turbine in real time.However,pitch control system is composed of complex electromechanical structure,which may result in a chain reaction or interaction between the faults of the pitch control system.So in many cases,maintenance personnel cannot timely and accurately identify fault source of the pitch control system through the wind turbines SCADA system.Therefore,it is of great academic significance and application value to research on the fault diagnosis of the pitch control system.Aiming at the problems of variable pitch control system,such as the large number of related parameters,nonlinearity and difficulty in accurate modeling.The Kernel Principal Component Analysis(KPCA)of data mining method made full use of the operating data monitored by the wind turbine SCADA system was presented in this thesis.Researching on fault detection and identification of pitch control system realized the KPCA-based wind turbine pitch control system fault diagnosis method.The simulation results verified the effectiveness of the proposed method.The main research contents of this thesis were as follows:(1)The faults mode,the faults cause and the relationship between the faults of the pitch control system were analyzed in this thesis.The results showed that the pitch control system is a nonlinear system with high failure rate,a large number of relevant operating parameters and mutual coupling and complex fault forms,which laid a theoretical foundation for the research of fault diagnosis of pitch control system.(2)In order to improve the rapidity and accuracy of the KPCA-based pitch control system fault diagnosis method,the operation parameters of the wind turbine SCADA system were analyzed in this thesis.The Relief algorithm was used to select the most representative and the best classification performance fault feature variables of the pitch control system,which was used to construct the observation vector of variable pitch system.(3)The optimization of kernel function parameters is very important for KPCA-based pitch control system fault diagnosis method.Thus,the kernel function parameter optimization method based on Particle Swarm Optimization(PSO)was applied in this thesis,which was used to obtain the optimal kernel function parameters;According to the observed data of the pitch control system monitored by the wind turbine SCADA system,a KPCA-based pitch control system fault diagnosis method was presented.Then,the fault detection and identification of pitch control system based on KPCA was carried out,which realized the fault diagnosis of the wind turbine pitch control system;The simulation research were carried out by using the fault information detected by the wind turbine SCADA system,and the validity of the KPCA-based pitch control system fault diagnosis method was verified.