Refined instrumental variable method for Hammerstein-Wiener continuous-time model identification

This study presents the first attempt of direct continuous-time model identification using instrumental variable method for Hammerstein-Wiener systems from sampled data. Under the assumption of monotonic function for the Wiener part, the whole non-linear model is first estimated as an augmented multiple-input single-output linear model, from which the model parameters are then extracted by singular value decomposition. A refined instrumental variable method is proposed to consistently identify this non-linear system acting in a coloured noisy environment. Monte Carlo simulation analysis is presented to illustrate the effectiveness of the proposed method.