Deep groove ball bearing fault diagnosis and classification using wavelet analysis and artificial neural network

Abstract

Now a days deep groove ball bearings are widely used to support the load of the shaft and to reduce friction in industrial machinery and domestic appliances. The major issue that arises in deep groove ball bearings is catastrophic failure which arises due to fatigue loading, electrical erosion, corrosion or spalls on various bearing components. Thus to ensure steadiness and continuous running of the machine, condition monitoring and defect detection of deep groove ball bearings are very essential. This research paper emphasizes on fault detection of deep groove ball bearings having specific defects present on various bearing elements using Debauchies Wavelet (DB-02) up to fourth level of decomposition. The vibration signals were recorded from a customized ball bearing test rig. The accelerometer and FFT analyzer is used to collect time and frequency domain vibration data and signature. Finally Artificial Neural Network (ANN) based Pattern recognition classifier is used for automatic bearing fault detection. The training of the network is done based on the collected data and the testing is done based on random data set. The highest classification rate was found to be 94%. This paper represents the implementation of Artificial Neural Network as a functional artificial intelligence tool for automatic bearing fault detection and classification without any human involvement.