基于GEP和GA技术的非线性系统辨识研究  

Nonlinear System Identification Based on GEP and GA Techniques

作　　者：朱耀春[1] 白焰[1] 蒋毅恒[1] 

机构地区：[1]华北电力大学自动化系,北京102206

出　　处：《信息与控制》2007年第5期592-596,603,共6页Information and Control

摘　　要：给出了利用基因表达式编程(GEP)进行非线性系统辨识的方法,弥补了传统辨识方法需要过多预知信息的不足,有着比遗传编程(GP)更简洁有效的系统模型结构表达方式.利用改进的遗传算法(GA)并行地进行模型参数进化,可以在有限的给定数据内得到合适的模型.关于模型适应度的定义,综合考虑了精确性和复杂性因素,能够获取一种比较折中的辨识结果.仿真结果表明,这种方式可以快速、准确地获取非线性模型.　A method based on gene expression programming（GEP） for identifying the nonlinear system model is presented,which makes up the insufficiency that traditional identification methods need much a priori information,and has a tidier and more efficient system model expression mode than genetic programming（GP）.It uses the improved genetic algorithm（GA） to carry out the model parameter evolution in a parallel mode,and the appropriate models can be obtained with limited given data.The definition of model fitness considers fully the accuracy and complicacy factors,and can get a trade-off identification solution.The simulation result indicates that the presented method can obtain nonlinear model in a quick and accurate way.

关 键 词：基因表达式编程 遗传算法 非线性系统 系统辨识 多目标优化 NARMAX模型 

分 类 号：TP15[自动化与计算机技术—控制理论与控制工程]