Identification of Linear Systems with Hard Input Nonlinearities of Known Structure

Abstract

Hard input nonlinearities are common in engineering practice. These nonlinearities severely limit the performance of control systems. Therefore, robust controls are often used [8] to cancel or reduce the effect of these harmful nonlinearities. Those control designs require values of the parameters that represent the hard nonlinearities. Clearly, system identification constitutes a crucial part in such control designs if the parameters are unknown. The difficulty of identification for the system with a hard input nonlinearity is that the unknown parameters of the nonlinearity and the linear system are coupled. Moreover, the output of the hard nonlinear block may not be written as an analytic function of the input. Surprisingly, there is only scattered work reported in the literature on identification of systems with hard nonlinearities [4, 9], although robust control designs involving these hard nonlinearities become a very active research area in recent years. This chapter is based on [1] with permission from Automatica/Elsevier and all the proofs can be found in [1].