基于基因表达式编程的Hammerstein模型的辨识

Identification of Hammerstein model based on gene expression programming

针对辨识Hammerstein模型的传统方法需假定系统结构或阶次,且过程复杂的问题,本文将基因表达式编程技术引入Hammerstein模型辨识,做了下列工作:①证明了Hammer-stein模型的渥尔特拉核,给出了系统稳定的充分必要条件;②形式化描述了结构体和参数体等新概念;③提出基于基因表达式编程的Hammerstein模型辨识的新方法(HMI-GEP),全自动地构造模型结构和决定参数。实验结果表明:该方法简单、有效,改善了辨识性能;模型结构采用不同的形式表达,参数估计的最低准确率约为93%。

In identification of Hammerstein model, traditional methods need the pre--assumption of system structure and ranks, and the procedures are complicated. To Overcome these shortcomings, the technique of gene expression programming is applied to the identification of Hammerstein model. The main contributions of this work include： ①proving the Volterra kernel of Hammerstein model and giving the necessary and sufficient condition for the system stability; ②formalizing the new concepts of structure individual and parameter individual, etc;③ proposing a novel method to identify Hammerstein model based on gene expression programming, which can construct model structure and determine parameters automatically. Experiment results show that the proposed method is simple and efficient, which improves the identification performance. Different expressions of the model structure are adopted, and the lowest accuracy of the parameters estimation is about 93%.