Parameter Estimation of Sandwich Systems with Backlash via Modified Kalman Filter

Accurate models for sandwich systems with backlash are very important for engineers to develop a technique to compensate the effect of backlash on the system and derive satisfactory performance. In this paper, an online modified Kalman filtering (MKF) algorithm for the parameter identification of stochastic sandwich systems with backlash is proposed. With the switch functions introduced to represent the effect of backlash, the pseudo-linear model with separated parameters is obtained to describe the sandwich system with backlash. Then, a stochastic state space model is constructed on account of the modeling residual is the Gaussian white noise sequence. Afterwards, the MKF algorithm is applied to estimate parameters of this model. Finally a simulation example is presented to evaluate the proposed scheme.