本文主要在提出一種新的混合式高階啟發式演化法(HS-DLM),期可有效地搜尋複雜、多極值之非線性最佳化問題的近似全域解,並應用於求解結構最佳化設計問題。本文所提出之混合搜尋法是以和聲搜尋法(Harmony search method, HS)作為全域多點搜尋的工具,並藉由離散拉格朗日法(Discrete Lagrangian method, DLM)來補強局部搜尋的能力和提供處理束制函數的機制。HS-DLM混合搜尋法即在搜尋空間交互運用,藉由HS在設計空間進行跳躍式搜尋,找尋近似全域解,而DLM處理束制函數和強健的局部搜尋能力,將使HS-DLM獲得真正最佳解的機率大為增加。本文藉由包含離散與混合變數之非線性、非凸集合的桁架和構架設計問題,來展示HS-DLM混合搜尋法的性能。
In this paper, a novel hybrid meta-heuristic algorithm, HS-DLM, is proposed to find the optimal solution of complex and highly multimodal systems, with particular emphasis on structural optimization. The hybrid optimization approach proposed in this paper uses harmony search (HS) method to perform global search, while uses discrete Lagrangian method (DLM) to enhance the local search capacity of the algorithm and to repair violated constraints in the design process. The global and local optimizers work in concert to efficiently locate quality design points better than either could alone. The proposed hybrid algorithm are applied to solve for problems of trusses and frames optimization that are characterized by non-linearity, non-convexity and by discrete, mixed design variables.