Predicting remaining useful life (RUL) becomes significant to implement prognostics and health management of industrial systems. The relevant studies have contributed to creating RUL prediction models and validating their acceptable performance; however, they are confined to drive reasonable preventive maintenance strategies derived from and connected with such predictive models. This paper proposes a data-driven preventive maintenance method that predicts RUL of industrial systems and determines the optimal replacement time intervals to lead to cost minimization in preventive maintenance. The proposed method comprises: (1) generating RUL prediction models through learning historical process data by using machine learning techniques including random forest and extreme gradient boosting, and (2) applying the system failure time derived from the RUL prediction models to the Weibull distribution-based minimum-repair block replacement model for finding the cost-optimal block replacement time. The paper includes a case study to demonstrate the feasibility of the proposed method using an open dataset, wherein sensor data are generated and recorded from turbofan engine systems.

Systems such as database and socal network systems have been broadly used, and their unexpected failure, with great losses and sometimes a social confusion, has received attention in recent years. Therefore, it is an important issue to find optimal maintenance plans for such kind of systems from the points of system reliability and maintaining cost. However, it is difficult to maintain a system during its working cycle, since stopping works might incur users some troubles. From the above viewpoint, this paper discusses minimal repair maintenance policy with periodic replacement, while considering the random working cycles. The random working cycle and periodic replacement policies with minimal repair has been discussed in traditional literatures by usually analyzing cases for the nonstopping works. However, maintenance can be more conveniently done at discrete time and even during the working cycle in real applications. So, we propose that periodic replacement is planned at discrete times while considering the random working cycle, and moreover provide a model in which system, with a minimal repair at failures between replacements, is replaced at the minimum of discrete times KT and random cycles Y. The average cost rate model is used to determine the optimal number of periodic replacement.

This paper deals with two forms of preventive replacement policy with minimal repair at failure. Those are, 1. the replacement policy I based on the cumulative operating time. 2. the replacement policy II based on the number of failures. The basic assumptions are; (1) the cost of minimal repair at failure is increasing with the number of failures since the last replacement, (2) the equipment fails stochastically with time.