A State Estimator for Sandwich Hysteresis Systems

Abstract

A new rate-dependent hysteresis model and a nonsmooth Kalman filtering are proposed for the state estimation of the rate-dependent hysteresis in piezoceramic actuators disturbed by stochastic noise. In the modeling scheme, a sandwich model with rate-dependent hysteresis consists of both input and output linear subsystem. Firstly, according to the properties of this complex nonlinear system, a non-smooth stochastic state-space model is introduced. Then, an auto-switcher filter is introduced to solve the state estimation of the rate-dependent hysteresis. The filter can be switched automatically according to the operating conditions. Moreover, the convergence of the proposed novel filter method is studied. Finally, the simulation result shows that the rate-dependent hysteresis model is used to describe the properties of piezoelectric actuators more accurate than the static hysteresis model when the frequency of input voltage is increasing, and the estimation errors of the non-smooth filter can converge close to zero quickly.