カルマンフィルタを用いて摩擦を補償するサーボ系の構成

抄録

Solid friction in mechanical control systems causes undesirable effects such as a jerking motion. Taking account of negative gradient characteristics from a static friction to a kinetic one, we propose an extended Kalman filter algorithm for solid friction estimation. An application of this estimation technique to a servo system is proposed in this paper. The proposed controller consists of a conventional linear controller and the extended Kalman filter. The real-time estimate given by the extended Kalman filter is fed back so as to cancel the solid friction. Using hybrid simulation where a plant with a solid friction is simulated by an analogue computer and the controller is simulated by a digital computer, we compare the performance of the proposed controller with that of a controller which neglects the negative gradient characteristics of the friction. The result shows that the proposed controller provides better performance.