A novel filtering techniques-based approach for parameter estimation of fractional Hammerstein-Wiener systems

This work treats the identification issue of fractional-order Hammerstein-Wiener systems with ARMA noise. The considered system consists of a fractional-order linear subsystem embedded between two nonlinear blocks. The estimation problem concerns all system parameters, including the noise model parameters. A novel recursive based least squares algorithm is proposed using filtering techniques. Starting from filtered input and output data, the auxiliary model principle is used to overcome the problem of unknown parameters and unmeasured intermediate signals. To prove the consistency of the algorithm, a convergence analysis is developed. The results of the simulation and the benchmark demonstrate the effectiveness of the proposed method to provide convergent parameter estimates with high accuracy and rapid convergence.