Review on Various Signal Processing Techniques for Predictive Maintenance

Abstract

Electrical equipment is the workhorse of industry, and its failure will result in a plant’s complete shutdown. Diagnostic improvements can boost reliability or even avoid an unanticipated catastrophe. The purpose of condition monitoring (CM) is to recognize a defect or a degrading process which has reached an indicative level and to appoint a warning of the anomaly in advance of the breakdown. Faults in electrical machines can be detected with the help of condition monitoring techniques. Various parameters can be used for condition monitoring like vibration, current, sound etc. Comparison has been carried out between various signal processing techniques applicable under the parameters. Technologies such as fuzzy-logic-based systems, genetic algorithms, neural networks, wavelet method, and so on, have largely replaced human-based defect identification. Dyadic Wavelet Transformation, FFT, and Artificial Intelligence were among the techniques explored in this paper. As a result, while acknowledging the need for further study, this review study provides a bird’s-eye view on the different techniques for fault identification.