Machine learning approach for studying the influencing factors affecting the operational reliability and remaining useful life

Abstract

PurposeReliability study plays a significant role in supporting the operation of any machinery working in a dynamic and harsh environment. This quality is inherently uncertain and a stochastic variable of any system. This study will be beneficial for designing an appropriate maintenance schedule, reducing unplanned production downtime and reducing maintenance cost of electrical motor operated particularly in dynamic and harsh environmental industries.Design/methodology/approachThis study focused on the effects of operating conditions (OCs) on the operational reliability and remaining useful life (RUL) of machinery. A probabilistic graphical method called Bayesian network (BN) was used for studying the effect of OCs on the system performance. The developed methodology has been demonstrated by analyzing the operational reliability and predicting the RUL of electrical motors operated in a heavy mining machinery.FindingsThe failure probabilities estimated from the historical data of the motor system are failure likelihood, and OCs are the evidence in the developed BN model. It has been observed that the performance and RUL of the motor are significantly influenced by OCs and maintenance. A threshold value of reliability at which the motor system requires maintenance or replacement has been proposed to guide management in decision making.Originality/valueThe Bayesian approach for studying the covariate of motor reliability and RUL estimation is a novel approach. This study will be beneficial for designing an appropriate maintenance schedule, reducing unplanned production downtime and reducing maintenance cost of electrical motor operated particularly in dynamic and harsh environmental industries.