两种非线性自适应滤波方法及其在建模中的应用

本文研究了基于Ｖｏｌｔｅｒｒａ级数模型和双线性模型的两种非线性自适应滤波，这两种滤波分别基于Ｖｏｌｔｅｒｒａ核和双线性递归差分方程，具有类似于ＦＩＲ和ＩＩＲ的结构。我们以Ｗｉｅｎｅｒ模型为代表，讨论了应用这两种自适应滤波进行非线性系统建模的问题，并考虑了ＭＳＥ和广义ＭＳＥ准则，分析了各种算法的复杂性，文中给出了一组仿真实例。

TWO TYPES OF NONLINEAR ADAPTIVE FILTERING WITH APPLICATIONS IN MODELLING

This paper is concerned with two types of nonlinear filtering which are based on the Volterra series and bilinear model. The two types are described using the Volterra kernels and recursive bilinear difference equations and possess F1R-like and IIR-like structures respectively. Adaptive modelling of nonlinear systems represented by the Wiener model using the two methods are explored. Two criteria, MSE and generalized MSE, are both considered for the purpose of comparison with the complexities of different algorithms being analysed. A collection of simulation results are exhibited.