带约束的微分进化算法及应用  

Constrained Differential Evolution Algorithm and Application

作　　者：闵涛[1] 邢星[1] 耿蓓[1] 

机构地区：[1]西安理工大学理学院,西安710054

出　　处：《科技通报》2014年第1期1-8,共8页Bulletin of Science and Technology

基　　金：国家自然科学基金重大项目(51190093);国家自然科学基金(51179151)

摘　　要：针对经典的微分进化算法难以求解约束优化,特别是大规模复杂约束优化,并且对于多峰值优化无法一次求出多个全局最优解等问题,本文提出了一种改进的微分进化算法。该算法采用一种简单有效的函数对其约束进行处理,并利用全局-局部微分进化算法进行演化。大量测试函数的实验结果表明,这种改进的算法能有效地解决约束优化问题得到全局最优解,并且对于多峰问题能一次得到其多个全局最优解,而且比传统演化算法具有更高的精度和收敛速度。In view of the classic differential evolution algorithm is difficult for solving constrained optimization, in particular large-scale, complex constrained optimization, and it can not find the multiple solutions for multi-peak optimization and other issues. The improved differential evolution algorithm is presented in this paper. This algorithm uses a simple and effective function to process its constraints and uses Global- local differential evolutionary algorithm to evolution .A large number of test functions experimental results show that this improved differential evolution algorithm can solve constrained optimization problem effectively and obtain global optimal solution and can obtain multiple solution once for multi-peak problem. It has higher accuracy and convergence rate than traditional methods.

关 键 词：微分进化 约束 优化 多峰 

分 类 号：O29[理学—应用数学] TP18[理学—数学]