2012-0227-基于AEA算法的自适应惩罚函数求解约束优化及其在丁烯烷化过程的应用.docVIP

基于AEA算法的自适应惩罚函数求解约束优化及其在丁烯烷化过程的应用 桑志祥，李绍军，张杰 (华东理工大学 化工过程先进控制和优化技术教育部重点实验室， 上海 200237) 摘要：本文提出了一种基于AEA算法处理约束问题的自适应惩罚函数法。该算法通过统计迭代种群中个体对每个约束条件违反的次数，判定各约束的强弱地位，动态自适应地调整各个约束的惩罚系数，对于强约束给予较大的惩罚系数。同时对目标函数做出了相适应区分修改，使得可行解和不可行解的目标函数值出现一定的区分，目标函数项和惩罚项趋于平衡，避免了惩罚力度过大或过小，有利于算法前期快速进入可行解区域，后期寻找最满意解。通过标准测试函数试验结果与DE+AMP、SSaD 关键字：自适应惩罚函数；约束优化；AEA；丁烯烷化过程 中图分类号：TP301 文献标识码：A A New Adaptive Penalty Function Based AEA Algorithm for Solving Constrained Optimization Problems and Its Application in Process Optimization of Butene Alkylation SANG Zhi-xiang, LI Shao-jun, ZHANG Jie (Key Laboratory of Advanced Control and Optimization for Chemical Processes, Ministry of Education, Shanghai, East China University of Science and Technology, 200237, China) Abstract: A new adaptive penalty function method based on the AEA algorithm is proposed to process constrained problems in this paper. It can determine the intensity of each constraint by the conditions breach number of each constraint in iterative population and adjust the punishment coefficient of various constraints dynamically and adaptively. And greater punishment coefficient will be given for a stronger constraint. In addition, the objective function is made a change accordingly, so that it contributes to distinguishing the objective function values between feasible solution and unfeasible solution. The improvements of objective function and penalty are helpful for balancing them and avoiding the penalty is too large or too small. The algorithm would quickly enter the feasible solution area in earlier stage, and find the most satisfied solution later. Comparisons of the standard test functions results with DE+AMP, SSaDE algorithm show that the proposed method has good applicability and global optimization performance. Furthermore, this method was applied to the process optimization of butene alkylation and satisfactory result was achieved. Keywords: adaptive penalty function; constrained optimization; A