Fault Classification of Induction Motor Bearing Using Adaptive Neuro Fuzzy Inference System

Abstract

The current research of smart fault diagnosis is to mine different characteristics of a signal from vibration of a machine that can differentiate unusual fault categories. Generally the mechanical data which was observed from machines are having unique features. Based on the prior knowledge and previously obtained features, the inputs are given to artificial intelligent techniques for fault classification. In this paper, neuro fuzzy and neural network based fault classification techniques are proposed. This paper mainly comprises two main parts such as feature mining and fault categorization. To extract valid information or set of features from the vibration signal, various recent techniques included in the feature mining and lessening modules. Samples are collected from Case Western Reserve University Bearing Data Center and, more samples are obtained from matlab. Statistical features of a signal are evaluated using matlab. The samples obtained are given to neural network for training, testing and validation. The statistical features are given as input to Adaptive Neuro Fuzzy Inference System (ANFIS) for fault classification. An experimental result shows that training, testing and validation using neural network and fault classification using Adaptive Neuro fuzzy Inference System producing better results.