Hammerstein systems identification in presence of nonparametric backlash nonlinearities

The problem of Hammerstein system identification is considered in presence of backlash nonlinearities with unknown border structure. The memory and nonparametric features of such class of nonlinearities make the identification problem particularly challenging. A multiple-experiment identification scheme is developed under minimal assumptions and shown to yield consistent estimates of the linear subsystem parameters as well as m points on both nonlinearity borders.