艙口圍緣之鍵槽端肘板構件最佳化之研究

Abstract

In designing a product, the optimum design has paved the way to achieve the prospective design performing effectively as well as being manufactured economically. In the paper we presented two self-developed optimization design tools, the successive Taguchi design of experiments (STDE) for discrete variables [17], and the Eagle-Foraging Algorithm (EFA) for continuous variables [25]. How to employ the edges of both different tools in designing a device is described through a match-stick-hole end-bracket device of a hatch coaming generally designed in ship containers. The EFA is a population-based evolutionary algorithm based on the particle swarm optimization (PSO). Other two evolutionary algorithms developed by modifying the PSO, PSOI and PSOII, are also introduced in the paper. The comparisons among performances from EFA, PSO, PSOI, and PSOII in solving the case are presented in the work. Finally, the evolution data of the case study from the PSO is analyzed by using the statistical histogram in an attempt to classify the qualities of the device.在產品設計開發中，最佳化設計是滿足產品高性能、低成本與高品質指標的方法之一。本文以貨櫃船艙口圍緣之鍵槽端肘板構件作為設計例，以本文循序式田口法[17]與鷹群掠食演算法[25]進行設計，期降低圍綠底角處的應力集中現象，以提高船體結構之安全與空問材料佈置之效益。循序式田口法為離散變數之實驗設計方法。鷹群掠食演算法係基於粒子群演算法所開發之進化演算法。本文採用5種不同最佳化技術，包括循序式田口法、群掠食演算法、及三種粒子群演算法，進行一系列實驗，所得結果比較檢討之。最後，本文以直方圖統計方法，對鍵槽端肘板構件的演化資料，進行構件品質分級與討論。