Modeling And Joint Optimization Of A Maintenance System Based On Closed Queuing Network Theory

With the development of large systems such as manufacture, logistics, railwaysetc, the complexity of the systems becomes increasingly more. An efficientmaintenance recourse allocation plan guarantees the system to work safely,effectively, and reliably. For such a system, generally we need to focus on how tominimize the system’s total cost, while keep the reliability or availability aboverequired values. The total cost is composed of inventory cost, maintenance cost, andoperation cost. The availability and the reliability depend on several factors likemaintenance frequency, resource allocation plan (materials, machinery, staff, etc),component inventory.A common approach to model the maintenance system is to divide the originalsystem into several subsystems and make decisions separately. However, there areobvious relations between subsystems. In this research, we develops a jointoptimization model for the maintenance system of railway locomotives based on aclosed queuing network, so as to optimize the maintenance resources allocation andimprove the system performance. With the railway locomotives’ maintenanceapplication, the approximate optimal quantities of maintenance channels, locomotivesspares, and operating locomotives are achieved upon high system availability atminimal costs. A modified genetic algorithm is applied to search for the solution. It isshown that there are major improvements for the system in both resources allocationand the system performance in our paper. The Taguchi design method is used toanalyze the independent and interaction effects of parameters to the systemperformance. The research focus on both the single component and muti-componentcases. The conclusions will be indicated in our research for the real application ofmaintenance systems.