Classification of bearing faults using extreme learning machine algorithm

Roller element bearing fault diagnosis is crucial in industry to maintain that the machine is in good condition so that there is no delay of work due to machine breakdown. This paper discusses the use of Extreme Learning Machine (ELM) algorithm to classify bearing faults. The performance of ELM is compared with Back Propagation (BP) algorithm. It was found that the results show that the ELM has smaller training error rate and testing error rate as compared to BP. ELM also required lesser time to train the neural network and at the same time, able to achieve higher accuracy than BP. Overall, the performance of ELM is encouraging.