关键词: 遗传规划, 进化策略, 函数拟合, 最小二乘误差, 均方差

Abstract: In this paper, function fitting can be considered as optimal search processes of model structures and parameters. A new genetic programming modeling method, combining evolution strategy, was proposed for hybrid identification of model structure and model parameters by performing global optimal search in the complex solution space where the structures and parameters coexist and interact. The method which determining the structures of the function and solving parameter at the same time conquer the traditional disadvantage which rely on data, lower precision and determining the structures first and solving parameters. Application results proved the high precision and generalization capacity of the fitting function model obtained by the new method.

Key words: genetic programming, evolution strategy, function fitting, least square error, square deviation