传统上基础设计均采用试误法程序來设计,虽能符合规範中.PDF

摘 要 傳統上基礎設計均採用試誤法的程序來設計，雖能符合規範中的要 求，卻無法保證造價的經濟性，且所得結果皆須調整為最靠近且方便施 工之設計值。本論文則是利用 遺傳演算法 來進行基礎結構物之降低成本 及自動化設計。所設計之基礎不僅可滿足規範的要求，亦可符合經濟性 及施工的考量。 本文的目標函數是基礎之總造價，包含混凝土與鋼筋量二部分；其 設計變數包括結構物斷面的尺寸、採用鋼筋號數與根數，且均以離散變 數來處理。限制式則依國內建築設計規範建立，主要考慮穿孔剪力、樑 式剪力、抗拉鋼筋、混凝土承壓應力以及基礎承載力。 本論文共設計三個常用基礎形式：獨立基腳、聯合基腳與筏式基 礎。為簡化起見，筏式基礎採用剛性法設計。 各設計案例在經遺傳演算法設計後，將所得之結果與文獻加以比 較，均獲得更佳之結果，顯示 遺傳演算法在設計基礎結構上有不錯的效 能。另外本論文亦探討族群大小對設計結果的影響。在編碼長度、交配 率、突變率與演化世代數不變之情況下，發現族群大小宜設定至一適當 範圍，則能兼顧正確性與效能。 I Abstract Traditionally, the foundation design usually uses Trial-and-Error procedure. Although it can conform to the code requirements, it is unable to guarantee the efficiency in terms of construction cost. In the thesis, genetic algorithm is used to carry on the foundation design to reduce the cost and to achieve design automation. Foundations designed by this method not only may conform to the code requirements, but also may satisfy to cost-efficiency and the construction consideration. The objective function is the cost of construction materials of foundation, comprising the cost of concrete and steel reinforcements. The design variables including the foundation sizes, the cross-sectional areas and the number of steel reinforcements are all in discrete format. Constrains are formulated based on the domestic reinforcement concrete design code, including punching shearing stress, beam shearing stress, tension stress of reinforcements, concrete bearing stress as well as foundation bearing capacity. Three types of commonly used foundations are designed, i.e.,