Machine learning and wavelet analysis for diagnosis & classification of faults in belt drive

The fault diagnosis and classification in rotating machine elements is a part of the condition monitoring. Application of Machine learning concepts in condition monitoring can enhance the quality of work. This condition monitoring can be achieved by making use of vibration data, that is collected using Data Acquisition system. The belt fault simulator is used for collecting healthy and faulty data in time domain, by using healthy and induced faulty condition. The vibration signals are enhanced by using filter to remove noise. The fault detection in belt drive is achieved by processing the filtered signal using wavelet enveloped power spectrum. The classification and diagnosis of fault in Belt drive using Machine Learning technique is presented in this paper. The SVM classifier is trained by making use of the time features, extracted from wavelet transform. A reliable method which can separate the fault condition of the Belt drive is demonstrated using the results.