Optimal Gain Selection Strategy in Back EMF Observer for Position Sensorless Operation of BLDC Motors

Abstract

This article presents the design and realization of a new unknown input back EMF observer for position sensorless operation of brushless DC motor. In the proposed observer, the back EMF is considered to be an unknown interaction input from the system structure and has been estimated as a state. In the observer structure, a combination of linear and nonlinear error feedback has been used. The observer gain is the key determining factor for the fast convergence and stability of the observer. A bio-inspired optimization algorithm called Differential Evolution Krill Herd, an efficient and novel algorithm rarely used in drives application, is further developed in this work to determine the optimal value of the observer gain. Observers with optimized gains have enhanced performance in reducing the absolute mean and peak errors in the estimated speed compared to conventional back EMF observers. This can ensure a more accurate and error-free operation of the closed-loop controllers used in the process. The operating performances are verified by ways of simulation and experiment to demonstrate the convergence, robustness, as well as the effectiveness of the observer. The sensitivity of the observer with parameter variation has also been verified, and the results favour the suitability of the observer in sensorless operation.