基于遗传算法和最速下降法的非线性动态系统辨识的混合算法  

A Hybrid Algorithm for Identifying Nonlinear Dynamical Systems Based on Genetic Algorithm and Steepest Decent Algorithm

作　　者：赵明旺[1] 

机构地区：[1]武汉冶金科技大学自动化系,武汉430081

出　　处：《高技术通讯》1997年第6期31-36,共6页Chinese High Technology Letters

基　　金：武汉市科委"晨光计划"资助

摘　　要：通过定义最速下降算子、选择算子、混合数据结构以及适应度函数，得到非线性动态系统非线性参数的加权最小二乘（WLS）辨识混合计算智能算法。该算法结合了遗传算法和最速下降法两者的长处，既有较快收敛性，又能以较大概率求得全局解（一致收敛估计）。数值计算表明该方法显著优于遗传算法和最速下降法。In this peper, a steepest decent operator, a selecting operator, a mixed numerical structure, and a fitness function are defined and then a hybrid computation-intelligent algorithm of the weighted leastsquares identification for nonlinear parameters of nonlinear dynamical systems is obtuined. The new algorithm with the faster convergence and the greater probability for the glob8l solution (consistently convergence estimates) combines the advances of both the genetic algorithm and the steepest decent algorithm. The numerical computution shows that the algorithm is distinctly suPerior to the genetic algorithm and the steepest decent algorithm.

关 键 词：计算智能 遗传算法 参数辨识 非线性系统 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程] TP391.9[自动化与计算机技术—控制科学与工程]