幾年來啟發式演算法大量被設定在解決多目標最佳問題,然而文獻當中提到不當的參數設定會造成結果顯著的劣勢。本研究利用粒子群演算法(PSO)結合最佳資源配置(OCBA)的方法以有效率的方法找尋五個測試標準例題之最佳解,利用數個設計的方法找尋最佳設計,並以有效率且準確的方式找到一組較佳解。之後利用下水道之專案問題尋求最小成本,其結果為PSO結合OCBA之方法可以達到精確且有效率的方式得到最佳設計與較佳值。透過本研究之方法,可以使在不同環境與結構下時,利用OCBA有效率的動態找尋最佳設計並利用PSO強力在空間內搜尋的機制下,能使加快整體模擬效率與接近母體平均值。
In recent years, Particle Swarm Optimization (PSO) heuristic algorithm has been widely used for combinatorial optimization problems like project scheduling, resource allocation, etc. However, PSO has the issue of how to set suitable parameter values to find better solution, as indicated in the literature. In this study, a smarter way to deal with the parameter setting issue in PSO is suggested. The parameter setting issue is re-stated as a design problem in that a specific set of parameter values is found to be the best design if the optimization problem has also the best solution. Since the design set may be large and brutal force simulation for PSO per design could be time-consuming, Optimal Computing Budget Allocation (OCBA) is incorporated into PSO to provide an efficient way to select a design that assists PSO in finding the best solution. Five standard testing examples were used to show that PSO-OCBA has better solution quality and more efficient than PSO only. PSO-OCBA also applied to a time-cost trade-off problem for a sewer project to find the minimum cost given the project duration. The project specialist can have the benefit of using PSO without handling the setting of PSO parameters. The results from PSO-OCBA are equally good as using only PSO.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。