一种辨识Hammerstein模型的新方法

A NEW METHOD FOR THE IDENTIFICATION OF HAMMERSTEIN MODEL

本文结合参数估计理论和函数逼近论中的一些理论结果,提出了一种开环辨识Hammerstein模型的新方法。这种方法能够克服目前广泛采用的辨识Hammerstein模型的方法中的缺点,获得模型线性动态部分参数的渐近无偏估计及无记忆非线性特性,具有接近最佳逼近效果和较好收敛性的逼近多项式。

In this paper, a new method for open-loop identification of Hammerstein model is proposed by combining parameter estimation theory with some results in function approximation theory. The new method circumvents the defects of the currently widespread methods for Hammerstein model identification. Using the method, asymptotically unbiased parameter estimates of the model's linear dynamics and the memoryless nonlinearity's approximation polynomial of near the best approximation effect and good convergency can be obtained.