Parametric Nonlinear System Identification of Brushless DC Motors for Aerobot Drive

Abstract

This paper presents an effective identification framework to build a more accurate nonlinear model for brushless DC motors, which are frequently used as drive systems of micro electromechanical unmanned aerial vehicles. The identification method uses a two-step procedure to obtain a fully parametric model. First, the initial model of system is estimated. Next, a coefficient shrinkage method is used for model structure selection. There are two main highlight processes in this paper. One is the use of Hammerstein series models for identification, which leads to a trade-off between model accuracy and complexity. Another is the use of modified nonnegative garrote method for model reduction and finding the true model order. The proposed method is validated on simulated system and finally used to identify a brushless DC motor system. The obtained nonlinear model of brushless DC motor is of high performance and controllable model complexity