Improved Deadbeat Predictive Thrust Control for Linear Induction Machine With Online Parameter Identification Based on MRAS and Linear Extended State Observer

Abstract

This paper proposes an improved deadbeat predictive thrust control (IDPTC) for linear induction machine (LIM) drives based on online parameter identification. First, to achieve fast thrust dynamic response with low thrust ripples, a DPTC method is induced based on the reference primary flux vector calculator and discrete-time LIM model. Second, a novel online magnetizing inductance identification is designed based on back electromotive force model reference adaptive system (MRAS) and linear extended state observer. Compared to the conventional MRAS identification strategy, there is no pure integration and differential operation in both reference and adaptive models for the proposed method, so that integral initial values, dc bias and high-frequency-noise amplification problems can be solved. Then, to improve the robustness of DPTC against magnetizing inductance mismatch, the proposed online parameter identification is further combined with the DPTC method, in which the flux can also be estimated as an intermediate variable without conventional parameter-based flux observer. Finally, comprehensive simulation and experiments have been conducted on one 3 kW arc induction machine, showing that the proposed method can effectively eliminate the influence of magnetizing inductance mismatch on the control performance and significantly improve the parameter robustness compared to the existing DPTC methods.