Bearing Fault Diagnosis Using Neural Network and Genetic Algorithms with the Trace Criterion

Abstract

Condition monitoring of rolling element bearings through the use of vibration analysis is an established technique for detecting early stages of component degradation. This study is presented to compare the performance of bearing fault detection using artificial neural networks (ANNs) and genetic algorithms (GA). The time domain vibration signals of a rotating machine with normal and defective bearings are processed for feature extraction. The extracted features from preprocessed signals are used as inputs to ANN classifier for five-class recognition (one normal and four with different levels of fault). The system of features selection is based on genetic algorithms as optimization method and the trace criterion from the linear discriminant analysis (LDA) as evaluation function. The ANNs are trained with a subset of the experimental data for known machine conditions, and tested using the remaining set of data. The procedure is illustrated with and without features selection. The results show the efficiency of the proposed methodology.