Online joint replacement-order optimization driven by a nonlinear ensemble remaining useful life prediction method

Remaining useful life (RUL) prediction and maintenance optimization are two critical and sequentially connected modules in the prognostics and health management of machines. Due to the advantages of obtaining more accurate RUL prediction results and the effectiveness of addressing replacement scheduling and spare parts provision dynamically, ensemble RUL prediction and online joint replacement-order optimization are paid specific attention to. Despite substantial works on those two aspects, there are still two limitations that compromise their performances in practical applications: 1) Existing ensemble RUL prediction methods neglected the nonlinear relationships among individual prediction models. 2) No online joint optimization model that utilizes ensemble RUL information is available. Faced with these two limitations, this paper first proposes a nonlinear ensemble RUL prediction method, which takes nonlinear relationships among models into consideration. Furthermore, an online joint replacement-order model is formulated using the ensemble RUL prediction results, and an iterated local search based optimization algorithm is utilized for dynamically finding the near-optimal joint policies. Through the experimental study of milling cutter life tests, the proposed nonlinear ensemble RUL prediction method is verified with higher accuracy, and the joint optimization model utilizing the ensemble RUL results is shown to provide more effective joint policies.

Automated summary

This paper proposes a method for ensemble RUL prediction that takes into account the nonlinear relationships among individual prediction models and formulates an online joint replacement-order model using the ensemble RUL prediction results. Experimental results show that the proposed method has higher accuracy and provides more effective joint policies.