「資源撫平」,定義為在固定原有的工期之下,嘗試去減少因為資源消耗而產生之波動。高度的全球競爭下迫使許多公司在必須對昂貴的資源(如高成本之機台)去進行最佳與均勻地利用。因此,於資源限制專案排程下,如何減少資源使用所造成之波動所運用的方法,變得越來越重要。 在專案排程中,依照限制的不同,分為「資源限制下」與「時間限制下」之專案排程問題。而所謂「時間限制下」之專案排程問題,就是在某一固定的工期內,目的使其資源使用波動最小化,也就是「資源撫平」問題。曾有論文探討過單一模式下的資源撫平問題,但是加入多模式之考量後,會使得問題更加具有一般化。但是,必然也會增加其資源撫平的複雜性與困難度。 以往,曾有論文以遺傳演算法來進行此問題之求解。然而,對於遺傳演算法中的運運算元之參數設定,仍然沒有經過研究。本論文嘗試以田口方法中的「參數設計」以視基因演算法為一個系統的角度,並運用田口方法中的參數設計原理,進行實驗與分析驗證,找出在處理專案中資源撫平問題下,採用基因演算法求解所建議輸入的最佳參數組合,並以模擬實驗的結果,來實證經由參數設計找到之最佳組合確實能找到更佳之解。 “Resource Leveling” is defined that under the original fixed project duration it attempts to decrease the fluctuations incurred by resource consumption. Beneath highly global competition some companies must to utilize expensive resource, i.e. high cost machine optimally and uniformly. Therefore, the techniques toward how to diminish the fluctuations caused by resource demand are getting more important. Depends on different constraints, the project scheduling problem divides into two dimensions, which are “under time constraint” and “under resource constraint”. The former problem, which attempts to minimize the fluctuations of resource demand, also called “resource leveling” problem. The genetic algorithm has been applied to the problem under multi-mode situation. However, the parameters of GAs haven’t been researched. This thesis tries to verify the better solution can be found using the “Parameter Design” of Taguchi method.