A maintenance decision-making method based on stochastic processes and evidential variables considering small sample conditions

Maintenance is crucial for increasing the reliability of a system during its usage. In this paper, a maintenance decision-making optimisation method, based on stochastic processes and evidential variables, is proposed. This method reduces the high maintenance costs caused by the inability to accurately evaluate the remaining useful life (RUL) of the underlying equipment under small sample sizes when based on random variables. A degradation model is first developed based on stochastic processes supported by evidential variables, the RUL interval of the equipment and its corresponding Basic Probability Assignments (BPAs) are accurately predicted. A maintenance decision-making model is then developed based on evidential variables, where the expected maintenance cost per unit time is considered as the objective function while the ordering and maintenance times of spare parts are considered as decision variables. Finally, the objective function is optimised to determine the optimal ordering and maintenance times. Considering the piston pump as an example, it is demonstrated that the proposed method has high effectiveness in predicting the RUL and making maintenance decisions. It reduces the expected maintenance cost per unit time by 0.1054- 2.5171%.