A Blind Approach to Identification of Hammerstein Systems

Hammerstein systems form a class of block-oriented nonlinear models, where a static nonlinearity precedes a linear dynamic system. There exist a large number of works on the topic of identification of Hammerstein systems in the literature. The methods of Hammerstein identification can be classified as the ten methods in Section 3.9 of [7] or the four groups in Chapter 1 of [8]. This chapter focuses on the blind approach that was formulated in [15] and further studied in [2, 17]. This type of approaches assumes inputs to be piece-wise constant for certain consecutive samples and aims at the main difficulty in identification of Hammerstein systems: the inner signal between the nonlinearity and linear system is unmeasurable. Such approaches have two important merits: (i) identification of Hammerstein systems is possible even without an explicit parametrisation of the nonlinearity, and (ii) the nonlinearity does not have to be static, but could be the