一种Volterra级数模型的简化辨识方法及其应用

针对基于Volterra级数模型的非线性系统辨识中存在的“维数灾难”问题,提出了一种基于分块最小均方(BLMS)滤波器的简化辨识方法.该方法利用影响指数的概念,在保证一定辨识精度的前提下,根据每个Volterra核函数对辨识结果贡献的大小对其进行筛选,然后用筛选出的有效核函数作为对原系统的近似,从而达到降低辨识维数,减少计算量的目的.最后,文章用一个工程实例验证了该算法的有效性.

Study on a simplified Volterra series identification method and its application

Volterra series identification often surfers from the so called 'dimension disaster',(because) the scale of Volterra kernels expands exponentially with the increasing of system's order and degree.In order to solve this difficulty,a simplified identification method that based on Block Least Mean Square(BLMS) filter is proposed in this paper.By means of this method,the concept of the effect index is introduced in order to reduce the scale of Volterra kernels by selecting them(according) to the degree of their effectiveness in identification.This method is suitable for Volterra(identification) on line and its effectiveness has been validated by an engineering instance.