透過您的圖書館登入
IP:18.117.111.245
  • 期刊

Mixed-lnteger Hybrid Differential Evolution for Synthesis of Chemical Processes

結合整數混合差值進化於化工程序合成之應用

摘要


本文提出一種結合整數混合差值進化(MIHDE)法則,應用於非絕對分離之蒸餾序列與熱交換網路的程序合成問題。文中提出一種廣義的結合整數非線性規畫(MINLP)模式來描述蒸餾氫列的超結構。此廣義的MINLP模式更適合執行進化計算。為了求解程序合成之限制條件問題,我們提出一種適應處罰參數之乘積子更新於MIHDE法則。從計算例子,得知本方法可以容許較寬廣的啟始處罰參數設定,並且得到幾乎一致的最佳設計。

關鍵字

無資料

並列摘要


In this study, mixed-integer hybrid differential evolution is applied to solve the synthesis problems of the nonsharp distillation sequence and heat exchanger network. A general mixed-integer nonlinear programming (MINLP) model is proposed to formulate the superstructure of the distillation sequence Problem. The proposed MINLP model is more suitable for use with an evolutionary algorithm. We introduce mixed-integer hybrid differential evolution along with a multiplier updating method, including adaptive penalty parameters (MIHDE-MUMAPP), to solve constrained synthesis problems. Computational examples reveal that nearly identical minimum solutions can be obtained using MIHDE-MUM-APP even under wide variation of the initial penalty parameters.

延伸閱讀