改进遗传算法在非线性变参数估计中的应用

Improved Genetic Algorithms for Varying Parameter Estimation in Nonlinear System

针对解决非线性系统模型变参数估计问题要求运算速度快、效率高的情况 ,提出了实数编码遗传算法的改进算法 ,即分阶段设置收敛判断条件及择优操作等操作步骤。分阶段设置收敛判断条件是指在交叉操作完成之后增加了判断是否收敛的操作步骤 ;择优操作是指用交叉操作后形成的解群中的的适应值最大的优秀个体替代适应值最小的最差个体 ,使参加变异的优秀个体数目增加 ,使得算法在优秀个体局部邻域内的搜索机会增加。通过实例对改进算法与原算法的运算性能进行了比较 ,收敛速度平均提高了 2～ 3倍。实验结果表明 ,改进算法对提高遗传算法的运算速度是可行和有效的。

Aiming at the request of rapidly calculating velocity by solving problems of varying parameter estimation in nonlinear system, a real-coded genetic algorithms is proposed, that is, adding two operations of phase-divided-setting convergence judging condition and selecting the best individual. One of improved operations means that convergence judging conditions are respectively set after the crossing and the mutation operations are finished. Another improved operation means that the individual with maximum adaptive value will replace another individual, which has mininum adaptive value before mutating, so excellent individual number involving in the mutation and their local searching chance can be increased. The properties of improved algorithms and unimproved algorithms are compared with an actual example, the velocities of improved method can increase 2～3 times quickly as unimproved one. Experimental results show that improved genetic algorithms are effective and practicable to enhance the calculating velocity.