非线性回归模型辨识的混合计算智能算法

A Hybrid Computational Intelligent Algorithm for Identifying Nonlinear Regressive Models

通过定义牛顿算子、选择算子、混合数据结构以及适应度，得到可结合遗传算法和牛顿法两者长处，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）的非线性参数辨识算法。数值计算结果表明该方法显著优于遗传算法和牛顿法。

In this paper, a Newtonian operator, a selecting operator, a mixed numerical structure, and a fitness are defined and then a hybrid computational intelligent algorithm for nonlinear parameter identification is obtained.The new algorithm with the faster convergence and the greater probability for the global solution (consistently convergence estimates) combines the advances of both the genetic algorithm and Newtonian algorithm. The numerical computing results show that the algorithm is distinctly superior to the genetic algorithm and the Newtonian algorithm.