基于结构和参数自适应的改进遗传算法

目的研究一种新的自适应寻优机制，以避免遗传算法早熟．方法在参数自适应基础上引入结构自适应思想。利用种群早熟判别的熵准则，当种群收敛于局部最优时，转换算法结构为先变异后交叉，同时辅之以参数自适应方法使之跳出局部最优．结果数值实验表明，算法的结构自适应可有效避免早熟，与参数自适应配合使用，能明显提高算法的全局寻优能力．结论基于结构和参数自适应的改进遗传算法是可行的、有效的．

Improved Genetic Algorithm Based on Structure and Parameters Self-adjustment

In order to prevent premature problems in GA,the paper introduces the idea of structure self-adjustment to parameters self-adjustment GA.The diversity of population is measured by entropy,and the entropy criterion of population premature is obtained.When the population becomes premature,the mutation is actualized before crossover and parameters self-adjustment are executed at the same time,which makes the algorithm jump out of the local-optimization.The experiment demonstrates that this structure self-adjustment algorithm can prevent becoming premature effectively,and the global-optimization ability of GA is improved evidently when it cooperates with the parameters self-adjustment.