Nonsmooth estimation and adaptive control with application to automotive brake torque

Estimation and adaptive control has been successfully used when system parameters are unknown or slowly time varying. One requirement of such algorithms is that in order for the parameters to converge, the signal must be “persistently excited”. When the signal is not rich enough to meet this requirement, it is intuitively possible to self excite it by using a switching control input. In this paper, a formal analysis of such a method is performed. Its application to the estimation and adaptive control of automotive brake torque is also presented. Experimental results are used to corroborate the theory developed