Real-time identification and compensation of asymmetric friction using unscented Kalman filter

This is the authors' abstract. We don't add key points for this paper.

This paper presents a method to design the contour control of the ball screw system. The friction is one of the main factors to deteriorate the positioning performance. There exists the nonlinearity in the friction such as the Stribeck effect and the hysteresis property. The friction has the position dependent and asymmetric characteristics which change the friction size depending on the position and the direction of object. In addition, the friction varies depending on temperature and wear. From the above, it is necessary to successively estimate the friction. In these conditions, we propose the method to estimate and compensate the friction in real time. The LuGre model is applied to describe the dynamics of friction, which is able to represent the characteristics by a few parameters. The unscented Kalman filter is used to estimate the friction which is applicable to nonlinear systems. The estimated variables are used to compensate the friction in real time. The effectiveness is illustrated by simulations and experiments.