Recursive identification of Hammerstein systems with dead-zone input nonlinearity

Abstract

A robust recursive identification algorithm is proposed for Hammerstein systems with dead-zone input nonlinearity and unknown dynamic disturbances. The dynamic disturbance is considered as a slow time-varying parameter to be estimated by the recursive forgetting tracking estimation technique. Based on the decoupling estimation theory and overparameterized method, two improved recursive least squares algorithms are constructed to estimate the model parameters and disturbance sequence. The variable forgetting factor strategy is also used to improve the tracking estimation performance of the disturbance and the estimation accuracy of the model parameters. Besides, the asymptotic convergence property of the proposed algorithm is also illustrated in detail. One example is used to demonstrate the superiority of the proposed algorithm.