Kalman filtering strategies utilizing the chattering effects of the smooth variable structure filter

The Kalman filter (KF) remains the most popular method for linear state and parameter estimation. Various forms of the KF have been created to handle nonlinear estimation problems, including the extended Kalman filter (EKF) and the unscented Kalman filter (UKF). The robustness and stability of the EKF and UKF can be improved by combining it with the recently proposed smooth variable structure filter (SVSF) concept. The SVSF is a predictor–corrector method based on sliding mode concepts, where the gain is calculated based on a switching surface. A phenomenon known as chattering is present in the SVSF, which may be used to determine changes in the system. In this paper, the concept of SVSF chattering is introduced and explained, and is used to determine the presence of modeling uncertainties. This knowledge is used to create combined filtering strategies in an effort to improve the overall accuracy and stability of the estimates. Simulations are performed to compare and demonstrate the accuracy, robustness, and stability of the Kalman-based filters and their combinations with the SVSF.

► SVSF gain causes a phenomenon known as chattering that may be used to determine system changes.
► Paper presents new filters based on combining the EKF and UKF with elements of the SVSF.
► Results show improved stability and overall estimation accuracy.