Fault Detection in Complex Mechanical Systems Using Wavelet Transforms and Autoregressive Coefficients

Abstract

Vibration monitoring techniques have played a major role in the detection of faults in rotating machinery. In the present work, individual (healthy and faulty shafts, outer race fault in bearings) and combined faults (outer race fault of bearings and misalignment of shaft) have been detected using discrete wavelet transform (DWT). An autoregressive (AR) model is then constructed from the detailed coefficients of DWT to highlight the severity of the combined faults as compared to the healthy and individual faults in the system. The result shows greater fluctuations in the AR coefficients as the complexity of the faults rises in the system. © 2020, Springer Nature Singapore Pte Ltd.