Kalman filter based observer design for real-time frequency identification in motion control systems

Abstract

The paper deals with a systematic observer design for a robust frequency identification of electromechanical systems. The observer estimates the frequency response of the system which is excited by a swept harmonic signal with a variable frequency, amplitude and DC component. The paper presents a computationally efficient algorithm based on the recursive discretization of a Kalman filter. An application of the proposed method is demonstrated on a problem of motion control of a compliant mechanism.