中文摘要： 專案選擇與排程問題至目前已被深入的探討，但探討的觀點卻只著重在單一部門的環境，至於多部門的環境至目前並沒有被深入的著墨。有鑑於此，此篇論文考慮在多部門環境中的專案選擇與排程問題，並以公司整體利益最大化為主要的目標函數，發展此問題的數學模型。 而專案選擇與排程問題本質上屬於一種組合問題，以解析式的方法來解決組合問題當問題的複雜度增加時，求解的時間必定以指數型的方式漸增。因此，我們試著提出可以解決組合問題的啟發式方法—基因演算法，來解決多部門環境中的專案的選擇與排程問題。 在模擬實驗的過程中，田口方法會被使用於所發展的基因演算法中主要參數的決定，而所模擬出來的結果，會在目標函數值與執行時間此兩個比較維度下與用解析式方法(AMPL modeling language)所得到的結果做比較。 Abstract: Before implementing projects, project selection is a very important proceeding work. An organization can’t possibly implement all coming projects because of some limitations such as budgets and human resources. Every project selection result sifting from candidate projects in an organization can be called a “project portfolio”. Further, after getting a project portfolio in an organization, deciding when to implement (schedule) these selected projects is also an important issue. By scheduling, we can make the resource consumption in each period satisfies the budget constraints. In reality, organizations in a firm such as departments are facing this problem. Multiple departments have multiple candidate projects to choose in a company. How to decide the project portfolio in each department so as to gain the overall maximum profit in a firm? This is a very complex problem in reality. Although project selection and scheduling problem has been discussed in depth and several related models have been proposed, none of them discussed this problem in a multiple-department environment. For this reason, our paper focuses on the project selection and scheduling problem in a multiple-department environment. Owing to project selection and scheduling problem belongs to a typical combination problem, we try to propose a problem-specific genetic algorithm in project selection and scheduling problem to find a satisfactory result in our paper. In addition, due to deciding the parameters of the proposed genetic algorithm are necessary, Taguchi method will be applied in the process for deciding the most suitable parameters.
