Fault Identification in Doubly Fed Induction Generator Using FFT and Neural Networks

Abstract

This paper presents a fault identification system for doubly fed induction generator (DFIG). It considers cases of single phase short-circuits and load switching. The system uses the fast fourier transform (FFT) to preprocessor data, which consist by the stator line currents. The principal component analysis (PCA) is employed to reduce the dimensionality of the output data of FFT and the fault identification is made by means of artificial neural network (ANN). Also, a post-processing (PP) is employed in order to increase the network reliability, which reduces the error of ANN. The system is simulated in the MATLAB simulation software using a database with different voltage, speed and load. The results show that the system with PCA, FFT and ANN has a good performance and accuracy with the PP to fault identification in the DFIG.