一种新的遗传算法求解有等式约束的优化问题  

New genetic algorithm for equality constrained optimization problem

作　　者：刘伟[1] 蔡前凤[1] 王振友[1] 

机构地区：[1]广东工业大学应用数学学院,广东广州510090

出　　处：《计算机工程与设计》2007年第13期3184-3185,3194,共3页Computer Engineering and Design

基　　金：广东工业大学青年基金项目(052039)

摘　　要：针对有等式约束的优化问题,提出了一种新的遗传算法。该算法是在种群初始化、交叉、变异操作过程中使用求解参数方程的方法处理等式约束,违反不等式约束的个体用死亡罚函数进行惩罚设计出的实数编码遗传算法。数值实验结果表明,新算法性能优于现有其它算法;它不仅可以处理线性等式约束,而且还可以处理非线性等式约束,同时提高了收敛速度和解的精度,是一种通用强、高效稳健的智能算法。A new genetic algorithm is presented to solve equality-constrained optimization problems. Parametric equation method is taken to keep particles satisfying with equality constraints, and death penalty method is used to handle inequality constraints during the process of population initiation, crossover and mutation, and a new real-code genetic algorithm is proposed. The experiment results demonstrate that the new genetic algorithm is superior to some other techniques; the proposed algorithm is a general, effective and robust method, it can handle not only linear equality constraints, but also nonlinear equality constraints, furthermore, the speed ofconvergence and the precision are improved.

关 键 词：参数方程 等式约束 遗传算法 死亡罚函数 约束优化 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]