透過您的圖書館登入
IP:3.144.99.209
  • 期刊

以基因演算法求解單原片方形物件排列問題

Solving Two-Dimensional Packing Problems by Using a Genetic Algorithm

摘要


工業界莫不亟思要如何精減人力以提昇效率、全面降低生產成本。然而尚有部分產業仍利用人工方式來處理原物料切割/排列等相關作業。此類複雜之「物件切割」問題需要有合適之演算法來迅速求得原物料切割計畫。過去文獻中,不乏因為使用遺傳基因演算法(genetic algorithm, GA)而獲致不錯演算績效之研究,因此本研究採用GA之演算架構,但繼續加強在上述文獻中尚未被著墨之「菁英策略」,並使用不同之「編碼」設計,以期超越文獻之GA演算結果。在例題測試階段,我們首先以部分例題對所提出之GA演算法之參數進行實驗,以獲取參數組合建議。接著經由建議之參數組合,對38題國際文獻演算例題進行測試,再與國際文獻演算結果進行比較。結果顯示本研究所提出之GA演算法,在中、小題型之單原片方形物件排列問題中幾乎皆優於國際文獻最佳結果,某些例題甚至超越文獻結果將近10%之使用率,演算績效可謂非常傑出。

並列摘要


In this study, a genetic-based algorithm (GA) is proposed for nesting two-dimensional rectangular parts in a material plate. In the literature GA has been adopted for solving the same problem, fairly good results having been obtained. A new method for designing the coding of the chromosomes, accompanied by a proposal for an elitism strategy, is presented in this research, aiming at further extending the problem-solving capability of the traditional GA. Promising computational results are obtained and reported by running examples from the literature.

延伸閱讀