A kalman-filter-based tuning method for input shapers in servo systems

This paper presents a new tuning method for input shapers in industrial servo systems. The input shaper is an FIR filter for removing endpoint vibration effects of a control system. The input shaper has a sequence of impulses to produce a response for canceling out the endpoint vibration. To determine the values of impulses of the input shaper, prior knowledge about system characteristics or endpoint vibrations is required. In this paper, we proposed a new automatic tuning method for the input shaper using the decaying sinusoidal signal model and an extended Kalman filter. The proposed method does not require any iteration or repetitive motion to estimate the vibration parameters. Furthermore, the method uses the stable filter to estimate the vibration parameters. Experiment results for an industrial servo system show that the input shaper which is tuned by using the estimated parameters of the proposed method successfully removes the endpoint vibration of the system.