國立臺灣大學電機工程學系暨研究所張時中2006-07-252018-07-062006-07-252018-07-062005-07-31http://ntur.lib.ntu.edu.tw//handle/246246/8022本報告總結第三年在本計畫的支持下的研究成 果,包括三個層次的排程佳化方法:(1)以平穩 (Stationary)馬可夫決策問題為載具,設計出一個 結合模擬排序佳化(Simulation-based Ordinal Optimization) 與目標函數值疊代(Simulation- Based Value Iteration, SBVI)的演算法，作為以快 速模擬排序佳化來選擇動態排程法則的理論基 礎。(2)針對反向拍賣競標(reverse auction)市場 機制,建立設計公司的訂單分配與合約製造商的 生產排程決策模型,以計量與模擬分析來探討合 約製造商間在競合關係下的生產排程策略。(3) 以衛星取像排程問題進行拉氏釋限(Lagrange relaxation)法與禁忌搜尋法於單機排程應用效能 的分析比較。The third year of research efforts developed scheduling methods in three aspects. (1) The design of ordinal optimization-based value iteration algorithm (OOBVI) combines ordinal optimization (OO) with simulation-based value iteration (SBVI) into an efficient method for solving complex stationary Markov decision problems (StMDPs). The method lays a foundation for dynamic composition of production scheduling policies via quick simulation. (2) A reverse auction-based model captures the gaming interactions between order assignments from a design house and the production planning/scheduling of individual contract manufacturers (fabs). The model enables quantitative analysis and simulation study of the production planning/scheduling policies in a co-opetition setting. (3) A study compares Lagrange relaxation-based and Tabu search-based single machine scheduling algorithm over a satellite imaging problem.application/pdf730020 bytesapplication/pdfzh-TW國立臺灣大學電機工程學系暨研究所排序佳化模擬目標函數值疊代演 算法設計反向拍賣競標生產排程Ordinal OptimizationSimulationValue IterationAlgorithm DesignReverse AuctionProduction Schedulingreporthttp://ntur.lib.ntu.edu.tw/bitstream/246246/8022/1/932212E002082.pdf