Nonsmooth dynamic filtering for mechanical servo-systems

In this paper, a nonsmooth dynamic filtering approach is proposed for stochastic noise filtering of mechanical servo-systems. In this approach, a stochastic nonsmooth state-space model with sandwich structure is developed to describe the nonsmooth dynamic feature of the system in random noisy case. Then, an approach for observability test of the nonsmooth system with sandwich architecture is proposed. Afterwards, a nonsmooth dynamic filtering method motivated by the structure of Kalman filter is proposed for noise suppression. The proposed nonsmooth Kalman filtering method is applied to random noise filtering of a servo-system with dead-zone feature.