捷運系統場站設施維修排程最佳化之研究(3/3)

The well-scheduled maintenance plan and adjustment can effectively assignmaintenance staffs under limited maintenance resources to improve maintenanceperformance, so that a stable operation of the MRT system can be ensured. Thecurrent maintenance schedule plan and adjustment of MRT station facilities ismainly performed based on the planner’s experience with a trial and error methodwithout efficiency, neglecting the overall maintenance manpower cost. Therefore,it is important to develop planning tools for efficient utilization of overallmaintenance manpower. This study intends to conduct research on threeproblems in an MRT Corporation. The first is a weekly maintenance scheduleplanning problem. A weekly maintenance schedule useful for each dailyoperation needs to be carried out considering the existing and out-sourcedvendors’ manpower and skilled licenses. The second is a daily maintenanceschedule planning problem. In a daily operation, there are correctivemaintenance items besides the planned periodic maintenance items.Maintenance tasks are practically performed by the operator owned staffs or fromoutsourcing vendors’ manpower to handle the next daily maintenance andcorrective maintenance items. The third is a real-time maintenance scheduleadjustment problem. In real-time operations, random equipment failures mayoccur and emergency repairs have to be performed. Considering therequirements for the remaining periodic and corrective maintenances tasks,coupled with the emergency repairs, the planner must adjust the remainingmaintenance schedule of the day by rearranging the maintenance staffs with thesupport of other units of the same plant.Since the content of this study is large, the study is divided into a three-yearproject. In the first year, given the regular maintenance and estimated correctivemaintenance items, we will develop a weekly maintenance schedule planningmodel, aiming to minimize the total maintenance cost, by considering the existingmanpower and skillful certificates for maintaining all equipment. We will employLagrangian Relaxation with sub-gradient method, coupled with meta-heuristics, todevelop a heuristic algorithm for efficiently solving the model. In the second year,in response to the demands of equipment troubleshooting, we will develop a dailymaintenance schedule planning model based on the weekly maintenanceschedule, further considering the corrective maintenance items and the supportof outsourced vendors’ manpower, with the objective of minimizing the totalmaintenance cost to plan the next day maintenance schedule. We will modify thealgorithm developed in the first year to efficiently solve the model. In the thirdyear, in response to the major equipment failures, we will develop a real-timemaintenance schedule adjustment model based on the daily maintenanceschedule, further considering the emergency maintenance items and the supportof maintenance manpower from other units in the same plant, with the objectiveof minimizing the total maintenance cost to plan the maintenance schedule of therest operation day. We will modify the algorithm developed in the second year toefficiently solve the model. The proposed models and solution methods shall notonly be useful references for academics, but also will provide MRT Operatorseffective planning tools to improve their maintenance efficiencies and reduce theircosts.