基于遗传编程的非线性系统辨识

Nonlinear System Identification Based on Genetic Programming

为了实现对非线性系统的辨识,能够对目标系统的结构和参数进行同步辨识,将遗传编程(Genetic Programming,GP)作为辨识工具。使用基本遗传编程算法对非线性静态系统进行辨识—对电厂钢球磨煤机存煤量与产粉量之间的特性关系曲线进行辨识;使用一种改进的遗传编程算法对非线性动态系统进行辨识—对一个二阶离散非线性差分方程进行辨识。所有辨识都取得了满意的结果。遗传编程进化过程中,目标系统的结构与参数同时准确辨识,证明遗传编程非常适合于解决非线性系统辨识问题,并在算法上实现了结构辨识和参数辨识的统一。

To identify both structures and parameters of nonlinear systems simuhaneously, the genetic programming（GP） algorithms are employed. The basic GP algorithm is used to identify a nonlinear static system of a coal pulverizer in power plant. And an improved GP algorithm is used to identify a nonlinear dynamic system of a second-order nonlinear difference equation. The satisfied results are achieved with accurate and simultaneous identification of both structures and parameters of object systems. The results show that GP can evolve good mathematical models for both static and dynamic nonlinear systems, and can identify structures and parameters at the same time.