近年來因為生產型態的變遷,再加上生產技術的提升,使得生產管理的工作日益複雜。尤其是面對工廠中大量的訂單、龐大的機台數、繁複的加工程序,造成許多不確定性的情形。而柔性計算的技術就以容許不確定性、強大的資料處理能力見長,所以在本論文中將柔性計算應用在生產管理工作中的交期指定與生產排程上。首先,在交期指定的問題上,分別利用倒傳遞網路與案例式推理法進行指定交期的任務,實驗結果發現案例式推理法在預測交期上有較佳的表現。接下來,本論文關心多目標流程型工廠的排程問題,並整合案例式推理法進行排程問題中工件交期之指定。針對此一問題,本論文提出了均勻分割空間搜尋法,以均勻地搜尋解空間中的柏拉圖最佳解。最後,並與變動權重法進行比較驗證的實驗,結果指出本研究所發展之均勻分割空間搜尋法在求解品質與效率的表現上都比變動權重法要好。綜合本論文之研究成果,可以瞭解將柔性計算的技術應用在生產管理問題是相當有潛力的。
With the changing of the production environment and the advancement of the production technology, the tasks of the production management are getting more and more complex. The plenty of orders, numerous machines and complicated process lead to many uncertain situations. Soft computing techniques are famous in the tolerance of uncertainty and efficiency in data process. Consequently, soft computing techniques were applied to the due-date assignment and production scheduling of the production management. At first, a backpropagation neural network and a case-based reasoning model were applied to the due-date assignment problem. The experimental result reported that the case-based reasoning model is superior in forecasting the due-date. Secondly, this dissertation concerned the multi-objective flowshop scheduling problem with combining the case-based reasoning due-date assignment model. Uniformly space-decomposed search approach was proposed to uniformly search the Pareto optimal solutions in the solution space. The comparisons with the variable weight approach have indicated that the uniformly space-decomposed search approach outperforms the variable weight approach effectively and efficiently. The entire result of this dissertation shows that it is potential to apply the soft computing techniques in production management problems.