- 學位論文
自適應基因演算法應用於全球衣生產決策支援系統
Application of Self-Adaptive Genetic Algorithm to Decision Support Systems for Global Garment Production
摘要
在全球化的完全競爭巿場之中，為了要維持競爭優勢，快速且有效地訂定生產決策是必要的。傳統的決策制定方式，大多憑藉著決策者的經驗及腦力，決策結果往往容易受到決策者的情緒或決策時間太短等因素所影響。而且在全球化的巿場當中，影響生產決策的因素十分地龐大且複雜，若決策者需在短時間內考量所有的影響因素來訂定生產決策，這樣的決策經常會產生偏差，而偏差則會使得生產成本上升，在傳統製造業獲利有限的情況之下，成本的上升將會使得企業失去競爭能力。本研究以成衣製造商為研究個案，在參考相關文獻後，提出階層式生產決策制定流程，並使用自適應基因演算法做為分析工具，以建立一個全球決策支援系統，期望利用電子化資料分析方式，在短時間內計算出較佳的結果，以輔助決策者或高階經理人在短時間內將訂單安排至合適的工廠進行生產。 本研究使用自適應基因演算法，透過演算法可依照染色體分佈狀況自動調整運算參數，不但可以提高尋找全域最佳解的機會，而且可以使得每次計算出來的結果趨於穩定，可減少機會成本發生的次數。並且透過此一機制，避免使用者因為缺乏足夠的專業知識，在進行參數調整時，而產生不必要的錯誤。透過相關系統驗證及資料分析，此系統所產生之結果不但穩定，而且與精確最佳解十分地接近。在實際業者要求的時間內，可考量許多的影響因素，進而計算出具有參考價值之決策結果。而系統所計算出之結果，透過圖形的表示方式，可讓使用者在短時間內了解到每筆訂單及工廠之生產狀況，進而輔助決策者快速且正確地訂定生產決策。
並列摘要
It is very important for managers to make correct production decisions quickly in enhancing competitive capabilities. In the real companies, decision makers or senior managers only have few minutes for making production decisions. There are excess and complex influencing factors when senior managers make decisions. If senior managers have to make decisions with the influencing factors by their own experiences, some influencing factors would be omitted. Sometimes, it causes some mistakes and it also increases production cost. Because many global companies have limited profits and they can not cover extra cost, they need a useful tool to assist senior managers in order allocation into suitable factories within short time. This study arms to develop a global decision support system for garment production. Some past studies presented that using traditional genetic algorithm (or simple genetic algorithm) could analyze data within complex and excess influencing factors. However, there are some drawbacks in traditional genetic algorithm. The results created by traditional genetic algorithm are usually unstable, and it is impossible for managers to adjust genetic parameters. Thus, we used self-adaptive genetic algorithm as the analytic tool of global decision support system. Results from this study show that the global decision support system which used self-adaptive genetic algorithm as then analytic tool can get suitable decisions for the garment industry. Visual charts are used to present results, and they enable users to understand situations of orders easily. Characteristics of this system are easy to use, effective, and useful.