Friction estimation and compensation for rotational system using unscented Kalman filter

• Main Authors: Boonsri Kaewkham-Ai, Kasemsak Uthaichana
• Format: Conference Proceeding
• Published: 2018
• Subjects: Mathematics
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=84871689510&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/49256
• Institution: Chiang Mai University

A number of techniques in literatures including the extended Kalman filter (EKF) have been studied and proposed to solve the estimation problem of the states and parameters in friction models. The EKF technique provides acceptable performance on the friction estimation. However, difficulties associated with the EKF include Jacobian matrix computation requirement, and relatively large numerical errors in the true posterior mean and covariance for general nonlinear systems. This paper presents an application of the unscented Kalman filter (UKF), based on the so called "unscented transform" (UT) to solve the estimation problem. The performance of the state and parameter estimator for friction compensation based on the UKF technique is shown superior to those of the EKF based estimator in this study. The baseline performance comparison between the conventional PD controller with both compensators is set forth in this study. Both EKF and UKF based controllers are to track various position profiles, i.e., sinusoidal, square wave, and triangle. The simulation results indicate that the tracking performance of the UKF based controller is comparable to that of the EKF based without additional effort on controller parameter tunings.