Semi-parametric identification of parallel Hammerstein systems

In this paper, a semi-parametric identification method for parallel Hammerstein systems is proposed. The static nonlinearities are represented as a linear combination of basis functions, the linear dynamic systems are represented by their nonparametric frequency response function (FRF). A three step procedure is used. In the first step, the FRF of the best linear approximation is estimated for different excitation levels. In the second step, the varying dynamics are decomposed over a number of parallel branches using a singular value decomposition (SVD). In the last step, the static nonlinearities are estimated using a linear least squares estimation. The method is illustrated on a series of simulations.