基于遗传编程的火电厂主汽温系统建模研究  

Modeling Main Steam Temperature System of Power Plant Using Genetic Programming

作　　者：白焰[1] 蒋毅恒[1] 朱耀春[1] 董玲[2] 李霄飞 

机构地区：[1]华北电力大学自动化系,北京102206 [2]华北电力大学数学系,北京102206 [3]中国水利电力物资有限公司,北京100045

出　　处：《系统仿真学报》2008年第4期1076-1079,共4页Journal of System Simulation

摘　　要：遗传编程是基于自然进化理论的一种全局最优智能搜索技术,将它应用于系统辨识,可以实现同时辨识系统的结构和参数,并且得出的辨识结果是显示的。提出了一种新的多目标遗传编程算法,将进化的目标设定为偏差的平方和、表达式的复杂程度和最大动态偏差的综合最小。应用该算法辨识火电厂主汽温系统,仿真结果说明该算法是有效的,并且比常规遗传编程算法优越。Genetic Programming is a kind of searching technique, which is intelligent, and it works by emulating natural evolution. When applied to modeling a system, genetic programming can search out the frame and the parameters of the system at the same time, and the searching results are explicit, A novel multiobjective genetic programming, which searching aim is to minimize the sum of squares of deviations, the complexity and the maximal dynamic deviation, was put forward to model the main steam temperature system of power plant. The results from simulation show that it is valid and it is better than the familiar genetic programming.

关 键 词：多目标遗传编程 传递函数 系统辨识 火电厂主汽温系统 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]