A DSVM-Based MPC Strategy With Magnetizing Inductance Online Identification Method for Thrust Ripple Suppression in Linear Induction Machine

The conventional discrete space vector modulation-based model predictive control (DSVM-MPC) strategies in linear induction machines (LIMs) have two problems, one is that the current harmonics and thrust ripple are large as accuracy of the voltage vector (VV) is insufficient under the limited computational burden, and the other is that the parameter robustness is poor. To solve these problems, an improved DSVM-MPC strategy with magnetizing inductance online identification method for LIMs is proposed. First, an improved virtual VV discretization method and an optimal VV searching method are proposed, which can obtain more accurate VV with less computational burden compare to conventional DSVM-MPC strategies. Then, to improve the parameter robustness, a magnetizing inductance online identification method based on back electromotive force (EMF) is proposed, and a sliding mode observer (SMO) is designed to observe back EMF, which effectively avoids the current differential term. Furthermore, an adaptive compensation algorithm for cascaded low-pass filters is conceived to effectively suppress the inherent chattering of the SMO, and the amplitude and phase of the fundamental wave are appropriately compensated. Finally, the superiority of the proposed method is verified through comprehensive simulations and experiments on a 3kW LIM prototype.