Fault Diagnosis Of Electric Actuator In Gas Turbine Control System

With China’s efforts to promote gas turbine product development and industrial system,gas turbine is increasingly widely used in electric power industry production,gas turbine failures occur from time to time,among which the failure rate of electric actuator in gas turbine control system ranks the top.If the fault of gas turbine electric actuator cannot be found in time,it will cause its shutdown and production,seriously affecting the safety and economic benefits of gas turbine operation.Therefore,it is urgent to study the fault diagnosis method of gas turbine electric actuator.This paper focuses on fault feature extraction and fault diagnosis of gas turbine,takes gas turbine electric actuator as the research object,builds a semi-physical simulation platform to simulate and inject fault test according to the actual working conditions,and carries out in-depth research and simulation on fault feature extraction and diagnosis method of gas turbine electric actuator.In the simulation platform building and fault classificati on,this paper expounds the gas turbine control system of electric actuators are the structure characteristics and working scenes,the construction of the electric actuators experiment platform,introduces the electric actuator fault type,for gas turbine electric actuator fault feature extraction and diagnosis methods research lay the foundation.In terms of fault feature extraction,this paper expounds the time domain analysis,frequency domain analysis and time-frequency domain analysis of three commonly used fault feature extraction methods,from the time and frequency domain analysis method,analyzes the experience mode decomposition,the wavelet transform,the variational mode decomposition of three kind of fault characteristic signal decomposition method,the simulation to compare the three different performance.The results show that the variational mode decomposition is the best method to suppress the mode aliasing effect.In the aspect of fault diagnosis,the principle and process of T-SNE dimensionality reduction visualization method are described.An example is analyzed based on semi-physical simulation platform.Three fault characteristic signal decomposition methods,i.e.,empirical mode decomposition,empirical wavelet transform and variational mode decomposition,are used respectively to carry out fault diagnosis test combined with t-SNE method.Compared with the original signal sent directly to T-SNE,the classification effect of each diagnostic method was compared.The results show that the fault diagnosis method combined with variational mode decomposition and T-SNE dimension reduction visualization method has the best diagnosis effect without misclassification and aliasing,and is suitable for the fault diagnosis of electric actuator effectiv ely.The fault diagnosis method of VMD,T-SNE and K-means clustering is adopted to decompose,reduce dimension and classify original data in turn.Based on semi-physical simulation platform,10 common faults are studied and verified respectively in 50% and 80%working conditions.The results show that,The fault diagnosis method of electric actuator can effectively diagnose different common faults,and the diagnosis accuracy is greatly improved.