Design and parameter identification of a general Hammerstein model

This paper deals with parameter identification of Hammerstein model having a unified model of several discontinuous nonlinearities containing hysteresis, saturation, preload and dead-zone. This model contains different parameters the choice of which may generate nine different nonlinearities. Contrary to the nonlinearity block structure which is unknown, the structure of linear block is assumed to be known. The estimation of the nonlinearity selection parameters as well as the linear model parameters is ensured by recursive least squares method. This latter is tuned, so that it enables the estimation of internal variables relative to the selected nonlinearity. An illustrative example is presented to raise the efficiency of the proposed nonlinearity unified model.