為能提高傳統土木營建產業之競爭能力，並推廣節能減碳與永續發展的概念，本研究主要針對橋梁基礎中之樁基礎與淺基礎進行最佳化設計之研究。首先摘要回顧各種多點隨機搜尋之最佳化方法，以及其混合最佳化技術，並針對目前樁、淺基礎最佳化設計之相關文獻進行剖析探討。依此本研究建立起嚴謹且詳實之基礎設計分析模式，並整合各混合最佳化技術至實數編碼遺傳演算法、和聲搜尋法、差分演化法、粒子群法與螞蟻群法之中，並提出基於螞蟻群法之自適應參數技術，期能達到少參數化之目標。本研究另提出以最佳化與平行運算之技術來提昇最佳化參數研究工作之深度與廣度，並透過七個實際樁基礎與一個淺基礎設計案例進行最佳化效能測試。分析結果顯示，於全部所率定出之最佳化參數中，自訂參數模式下以混合和聲搜尋法（HHS/M）表現最為良好，而自適應參數模式下則以混合差分演化法（HDE/A）表現最為優良。且此兩演算法之率定參數於各種案例搭配不同年度單價之情況下，皆能有效找出全域最佳解，且與最佳解差距也較其他演算法為低。若從較原設計解節省之工程造價來看，本研究之各最佳化方法平均節省幅度可達42%，顯示已能滿足工程低價化設計與經濟性之需求。最後，本研究基於物件導向之程式語言開發出完善之視覺化介面與輸出分析報表之功能，降低使用者操作難度並能快速展示分析結果，期能對工程實務設計有所助益。
Nine hybrid optimizations are presented for pile and shallow foundations and are based on the Real-coded Genetic Algorithm (RGA), Harmony Search (HS), Differential Evolution (DE), Particle Swarm Optimization (PSO), and Ant Colony Optimization (ACO). The design variable set of pile foundation is seven independent variables, including pile diameter, pile length, pile spacing of x and y direction, pile number of x and y direction, and thickness of pile cap and the set of shallow foundation includes size and depth of shallow foundation. To evaluate the efficiency of all hybrid methods, six test functions and eight design cases are used and compared with their global optimum solutions by parallel computing technology and operating optimization. The conclusions show that the performance of HHS/M and HDE/A are better than other algorithms in all case study; In addition, the result of saving costs from original case of all hybrid method is around 42%, and demonstrates the economy and efficiency obviously. Finally, this study develops a brilliant program with a friendly graphical user interface and a detailed reporter, and hopes to bring simplicity and convenience for engineers in the practical design of foundations.
