Remaining Useful Life Prediction Method for the Rolling Element of an Electrical Machine Using Linear Regression Analysis of the Vibration Signal of a Faulted Bearing

Abstract

The anticipation of potential failures and provision of early warning signals are enabled by predictive maintenance, playing a vital role in ensuring the optimal performance and reliability of electromechanical systems. In this context, the research presents an effective and efficient approach for predicting bearing faults, focusing on the analysis of vibration signals from rolling elements, particularly bearings. By applying linear regression analysis, the vibration signal from each bearing sample is transformed into the frequency domain, enabling the calculation of the area under the curve. To estimate the remaining useful life (RUL) of the bearing, the research utilizes linear regression analysis, where the slope of the regression line serves as a crucial indicator. A positive slope suggests accelerated wear or imminent failure, indicating a decrease in the RUL as the independent variable increases.By proactively detecting and resolving potential faults, industries can effectively minimize costs linked to unexpected downtime, urgent repairs, and component replacements. Notably, the study utilizes benchmark data sourced from the NASA prognostics data archive.