Automatic Fault Classification for Journal Bearings Using ANN and DNN

Journal bearings are the most common type of in which a shaft freely rotates metallic sleeve. They find lot applications industry, especially where extremely high loads involved. Proper analysis various bearing faults and predicting modes failure beforehand essential to increase working life bearing. In current study, vibration data journal healthy condition five different fault conditions collected. A feature extraction method is employed classify conditions. Automatic classification performed using artificial neural networks (ANN). As probability correct prediction goes down for higher number ANN, made more robust by incorporating deep (DNN) with help autoencoders. Training was done scaled conjugate gradient algorithm performance calculated cross entropy method. Due increased hidden layers DNN, it possible achieve efficiency 100%