Optimizing Predictive Maintenance With Machine Learning for Reliability Improvement

Reliability is the probability that a system performs a required function without failure under the stated conditions for a specified period of time [1]. For industrial equipment, one reliability objective is to identify, evaluate and control the failures and take proper measures to eliminate failures that can lead to unplanned and costly downtime [1,2]. Currently, unexpected equipment failures can cause significant loss of throughput, production damage, reduced reliability, availability and service revenue, decreased tool lifetime, unplanned maintenance and repair, extra cost and project delay [2–4]. A survey in U.S. indicates that nearly 46% of major equipment repairs are caused by unexpected failures [3]. Moreover, recent studies reveal that losses and costs from unplanned equipment failures are estimated to be over $50 million per year for industrial manufacturers [5].