對航空公司的維修部門而言，如何使航機修護工作能有效率且確實，以確保航機在飛行中的安全，並能配合該公司營運飛航班表的準時起降，是一項重要且急迫的課題。傳統航空公司維修部門對於修護人員的排班上，係基於人工經驗，以簡單的排班及輪班規則進行修護人員的排班。此排班方式除費時外，往往無法與實際營運所需的修護需求相配合，易導致人力供需不均衡的現象，而使得修護部門的修護效率受到影響。本研究針對航空公司停機線修護人力供給之問題，利用數學規劃與電腦演算技巧，構建合適的模式與求解方法，以期幫助航空公司的修護部門有效率地規劃修護人員之排班及班次，提供下游修護人員指派之參考。 在求解的方法上，由於本研究所構建的混合整數規劃模式之規模甚大，難以求得最佳解，故本研究進一步發展一啟發式求解架構。此架構主要分三階段，各階段構建一模式。第一階段模式是以不分機型最佳化之觀點進行系統最佳排班班次規劃，第二階段主要是進行各單機型之排班人力供給規劃，第三階段則是針對單機型之人力供給規劃結果加以改善，進行混合機型之排班人力供給規劃。前二階段之模式，皆定式為整數規劃問題，第三階段則定式為混合整數規劃問題。在啟發式求解架構各階段的模式求解上，本研究皆自行以C程式語言撰寫執行輸入檔，並配合使用CPLEX套裝數學規劃軟體協助求解。最後，本研究以國內之一航空公司實際的修護資料，進行本研究模式之實例測試與分析，再依分析之結果提出結論及建議。 It is essential for airlines to efficiently perform aircraft maintenance in order to ensure aviation safety and punctuality. In tradition, the maintenance scheduling, based on staff’s experiences, was performed using a simple schedule rule. This scheduling method is not only time-consuming but also inefficient. In particular, the resulting manpower supply does not easily meet the minimum requirements, possibly causing imbalance between supply and demand. In this research, we use mathematical programs and computer algorithms to develop suitable models and solution methods, in order to help airlines efficiently and effectively plan their maintenance schedules and manpower supplies, which are then useful for downstream maintenance crew assignments. Because the problem size of the mixed integer programming model is expected to be huge, we developed a heuristic solution framework to solve the problem. The framework is divided into three stages. In the first two stages, two integer programs are formulated respectively. A mixed integer program is formulated in the third stage. The first model is used to determine the best shift plans. The second model is used to solve the maintenance manpower supply problem for each aircraft type. The third model helps simultaneously solve the maintenance manpower supply problem for mixed aircraft types, based on the fact that maintenance crew members are practically qualified for repairing different aircraft types in a work shift. We solve all stages by using the mathematical programming solver, CPLEX, and other self-developed computer programs. Finally, to evaluate the models and solution algorithms developed in the research, we perform a case study using the operating data from a major Taiwan airline. The results show that the models and the solution methods are useful.
