Error Dynamics Diagonalization Technique-Based Model-Free Filter Embedding LPF for Servo Drive Applications

Abstract

The proposed technique systematically incorporates the low-pass filter (LPF) and disturbance observer (DOB) into the model-free order reduction filter as auxiliary systems to improve the accuracy of speed and acceleration estimations for the servo drive applications. The resultant filter also makes the tuning procedure more convenient through error dynamics diagonalization property. This paper results in several practical contributions: (a) the Luenberger observer, serving as the main filtering mechanism from position measurement, enables the filtering estimation error dynamics to be diagonalizable according to the order reduction characteristics; (b) the low-pass filter acting as the first assistant extracts the fundamental component of the speed from the pure differentiation of the position measurement; (c) the disturbance observer-like system as the second assistant improves the disturbance attenuation capability of the entire filtering system against the high-frequency noises. A prototype dynamometer incorporating a 500W brushless DC motor as the test servo drive, showcases the filtering performance of the proposed solution by demonstrating an improvement in the feedback system's performance.