多目標基因演算法於鋼筋混凝土結構設計之應用

Abstract

結構設計過程是一種複雜程序且設計方案需滿足各方面需求，隨著社會需求的日益嚴苛，根據需求將結構功能特性區分為五項，分別為安全性、耐久性、經濟性、環境性與適用性，設計目標除了滿足這些需求外還必須進一步量化評估以迎合社會需求，故設計結構時往往需同時考量到多方面的要求。因此本研究就其安全與經濟兩方面進行鋼筋混凝土構件設計探討，在安全耐震部分依據美國ATC-40耐震性能評估法進行考量，其中引入位移指標對結構的耐震性能進行控制概念。而安全與經濟又是處於衝突狀態，這類問題稱為多目標問題，一般無法精確地定義出所謂的最佳解，而是必須透過Pareto Front有關非支配解的搜尋，尋求問題之可行方案。故利用多目標基因演算法（Multiobjective Optimization Genetic Algorithm，MOGA）來尋求耐震性能設計與最適成本，此方法為有效的全域搜尋技術，可以處理離散變數（discrete variables）的最佳化問題及具有克服局部最小化的能力，並且找出適當Pareto front。經由數值案例測試，證實多目標基因演算法可以進行耐震設計並找出最適成本。

Design of structures that satisfy all kinds of requirements is a very complicated process. Requests of designed structure performances are categorized as five issues: safety, durability, economy, environment and comfort. Following, design objectives of structures’ performance are not only satisfying the aforementioned issues but also can be quantified and evaluated. Also, design of structures is considered as a multiobjective optimizatioal problem. This work aims to study design of reinforced concrete buildings with two objectives, safey and economy. Herein, safey is consided for seismic assessment of reinforced concrete buildings for ATC-40 regulation. It is based on displacement based design. These two objectives, safey and economy, are conflict explictly. In this work, a multiobjective optimization genetic algorithm (MOGA) is employed to solve the problem. Rather than unquie optimal solution for single objective optimization problem, a set of solution, called Pareto front, can be found and all are consided as feasible optimal solutions. Two cases, a two-story snd a seven-story RC buildings, are adapted to verify the performance of the porposed approach. The results of two cases reveal that MOGA can find a set of optimal solutions for reinforced concrete buildings satisfying safty and minimum cost.