隨著交通需求日漸增加，大眾運輸不但要滿足民眾的需求外，對於民眾搭乘的安全體驗也須重視。捷運系統之營運須與維修保養同步進行與考量，設備藉由維修作業能使其維持良好狀態並能減緩故障之風險，其中捷運月台門作為人車安全之屏障，其維修更是不容忽視。營運管理單位藉由規劃排程以利維修作業之進行，目前實務之作法大多以人工經驗安排作業，此決策較缺乏系統性評估。因此，本研究以系統最佳化之觀點，發展捷運月台門預防檢修作業排程規劃模式，期能協助業者決策最佳之排程規劃。 本研究以時空網路流動技巧及數學規劃方式，定式維修工作隊於時空中流動之情形，在滿足實務相關限制條件下，以最小化總維修作業時間為目標，針對單一工作隊與多工作隊之排程，構建兩個捷運月台門預防檢修作業排程模式。在求解方法上，利用C++程式語言配合CPLEX數學規劃軟體進行模式求解。為評估模式之績效，本研究以國內某捷運公司之相關資料進行範例測試，並測試不同參數之敏感度分析和方案分析，結果顯示本研究之模式在實務上能有效運作，可提供捷運公司管理者作為月台門預防檢修作業規劃之參考。;With the increasing demand for transportation, mass transportation must not only satisfy people′s needs, but also attach importance to the safety experience. The operation of the MRT system must be synchronized with the maintenance and consideration. The equipment can maintain it in good condition and reduce the risk of failures through maintenance operations. The MRT platform screen doors serve as barriers to the safety of passengers and trains, and the maintenance cannot be ignored. The operation management facilitates the maintenance operation by scheduling. In practice, the scheduling is made mainly based on the personal experience of the decision maker, which lacks the perspective of system optimization. Therefore, this study takes the approach of system optimization and develops models for the maintenance scheduling of MRT platform screen doors, hoping to effectively assist business operators in deciding optimal scheduling. In this study, preventive maintenance scheduling models for MRT platform screen doors are developed, which are target one maintenance working team and multi maintenance working teams. And the time-space network flow technique is utilized to formulate the potential movements of working teams among all stations in the dimensions of time and space. The objective is to minimize the total working time. The C computer language, coupled with the CPLEX mathematical programming software, is employed to solve the problem. In order to evaluate the performance of the models, the numerical tests using real data from a domestic MRT company are performed. The results show that the proposed models and could be used in practice.
