Imperfect maintenance strategy of a three-stage failure process and a two-level inspection for train bogie system

Abstract With the gradual expansion of the scale of urban rail transit, its safety and cost issues have aroused widespread concern. Aiming at the common phenomena of "over-maintenance" or "under-maintenance" in urban rail trains, the imperfect preventive maintenance model with a two-level inspection strategy based on a three-stage failure process for train bogie system is proposed. Firstly, three parameters of Weibull distribution of components are obtained by analyzing historical fault data. According to the reliability curve and failure rate obtained by Weibull three parameters, the single components are divided into three stages: normal driving stage, initial deterioration stage and severe deterioration stage, and the three stages are identified by a two-level inspection. Among them, the normal driving stage and the severe deterioration stage are identified by the low-level inspection, and at the same time, it is possible to identify the initial deterioration stage, while all stages are identified accurately by the high-level inspection. Secondly, maintenance economy and train availability are taken as the optimization condition of the imperfect maintenance strategy, and the obtained optimal imperfect maintenance interval is taken as the high-level inspection time in the two-level inspection. Then an inspection threshold is proposed to determine the functional relationship between the high-level inspection and the low-level inspection. With the minimum expected cost as the goal, the expected cost of single components is calculated through different low-level inspection time, inspection threshold and preventive repair threshold. Finally, taking certain city urban rail train bogie system as an example, it is found that the minimum expected cost of axle box subassembly is CNY 43.28 a day, the maintenance period is 428 days. The minimum expected cost of gear box subassembly is CNY 27.26 a day, and the maintenance period is 651 days. Compared with the similar models, the maintenance period is prolonged and the expected cost is reduced, which verifies the effectiveness and superiority of the model and provides a theoretical basis for scientific maintenance.