本研究旨在探討某軍方基修工廠中,各項武器裝備在維修排程問題方面之求解模式與決策方法,研究內容分成兩階段:(1)四專案維修排程問題,每一專案視為一種武器裝備之維修,並進一步探討多目標決策模式;(2)以多專案維修排程問題之成果作為基礎,對該基修工廠之年度最佳維修專案排程進行探討,以作為相關決策單位之參考。 針對第一階段:四專案同時進廠維修排程問題,本研究將建立數學規劃模式,運用LINGO軟體求解,模式中所考量目標計有五項:(1)最大完工時間(Makespan,Cmax)最小化,(2) 總流程時間(Total Flow Time)最小化,(3) 最大延遲時間(Maximum Tardiness)最小化,(4) 總延遲時間(Total Tardiness)最小化,以及(5)延遲專案個數(Number of Tardy Jobs)最小化;並將上述目標依權重兩兩組合成單一目標後,分別求解各組目標於此問題之非支配解集合,再運用決策分析選擇最佳方案。 第二個階段:基修工廠年度最佳維修產能問題。年度最佳維修專案排程須滿足兩項限制:(1)所有預定數量之系統裝備必須能夠在年度內完成;(2)若提升任何系統裝備一台,即不可能在年度內完成。研究方法將以模擬退火演算法求得近似最佳解集合,之後再依相關決策因素,決定年度最佳維修能量與其規劃之年度專案排程計劃。
The thesis explores a decision problem on selecting the best alternative among various maintenance schedules for four types of military equipments that are currently in use. The problem under study involves two multi-criteria decision making issues: Selecting the best schedule (1) that maintains and repairs simultaneously four different types of equipments and (2) that maximizes the productivity of the maintenance working centers within a whole year. The research results will provide a valuable reference for the relevant management to make good decisions. The problem solving approach to the first issue consists of fives phases: (1) Determine the objectives that the management is concerned with; (2) Construct multi-objective linear integer models for the problem; (3) Collect relevant data from the working centers; (4) Find a set of Pareto optimal solutions by employing LINGO optimizer to the models; (5) Provide the best alternatives to the management based on their concerns. A simulated annealing algorithm is developed to solve the second issue; i.e., to find the maximum maintenance capability of the working centers within one year. The alternatives under consideration have a makespan between 360 and 365 days. Similar to the first issue, a limited number of alternatives, as well as the analysis of benefits and losses, are provided to help the management make a good decision. Additionally, a sensitivity analysis is performed on a particular and important working center in the situation where one additional unit of such center is constructed. A detailed discussion and decision making based on the sensitivity numerical results are included in the thesis.