本研究以遺傳演算法(genetic algorithms)結合運算樹(operation tree),發展出一套能產生高性能混凝土強度(high-performance concrete, HPC)自組織公式的方法。文中以五種已發表的方法來比較本方法的準確度及模型的解釋能力。比較的方法包括倒傳遞網路(back-propagation networks, BPN)、迴歸分析(regression analysis, RA)、巨觀進化遺傳規劃(macroevlutionary genetic programming, MEGP)、語法式進化遺傳演算法(grammar evolution genetic algorithms, GEGA)以及遺傳演算法整合迴歸分析(genetic algorithms of regressionanalysis, GARA)五種已經被提出的方法。從預測的準確度及模型可解釋的能力可知,遺傳演算法結合運算樹的方法確實是一個可以產生自組織公式的方法,且模型的準確度僅低於倒傳遞網路。
This study used genetic algorithms combined with operation trees (GAOT) to produce self-organized formulae for the strength of High- Performance Concrete, and compared the accuracy and explanation ability of the results with five existing methods, including back-propagation networks, regression analysis, macro-evolutionary genetic programming, grammar evolution genetic algorithms and genetic algorithms combined with regression analysis. The results showed that GAOT certainly could produce rather accurate self-organized formulae, and it is more accurate than other methods only excepting back-propagation networks.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。