Filtering Based Multi-Stage Recursive Least Squares Parameter Estimation Algorithm for Input Nonlinear Output-Error Autoregressive Systems

A filtering based multi-stage recursive estimation method is presented in this article. The system to be identified is called Hammerstein model, in which the output is described by a pseudo-linear regressive form of all unknown parameters based on the key term separation. Filtering the input and output data and separating the original unknown parameter vector into a few low-dimensional vectors, then interactively identifying each of the vectors is the basic thought of the proposed algorithm. Because the dimensions of the involved covariance matrices are smaller than those in the recursive generalized least squares algorithm, the discussed method has a lower calculational burden. The numerical experiment results demonstrate the validity of the presented method.