Machine Fault Detection Using Neuro-Fuzzy Inference System and Genetic Algorithms

Abstract

A study is presented to show the performance of machine fault detection using adaptive neuro-fuzzy inference system (ANFIS) and genetic algorithms (GAs), termed here as GA-ANFIS. The time domain vibration signals of a rotating machine with normal and defective gears are processed for feature extraction. The extracted features from original and preprocessed signals are used as inputs to GA-ANFIS for two class (normal or fault) recognition. The number and the parameters of membership functions used in ANFIS along with the features are selected using GAs maximizing the classification success. The results of fault detection are compared with GA based artificial neural network (ANN), termed here as GA-ANN. In GA-ANN, the number of hidden nodes and the selection of input features are optimized using GAs. For each trial, the GA-ANFIS and GA-ANN are trained with a subset of the experimental data for known machine conditions. The trained GA-ANFIS and GA-ANN are tested using the remaining set of data. The procedure is illustrated using the experimental vibration data of a gearbox. The results compare the effectiveness of both types of classifiers (ANFIS and ANN) with GA based selection of features and classifier parameters.