A Model Predictive Current Control Based on Adaline Neural Network for PMSM

The parameter mismatch seriously affects the control performance of the model predictive current control (MPCC). Aiming at improving the parameter robustness of MPCC, a MPCC method based on Adaline neural network for permanent magnet synchronous motor (PMSM) is established. First, the parameters sensitivity analysis of PMSM incremental current predic-tion model is carried out that eliminates effects of rotor flux linkage mismatch and resistance mismatch. Therefore, a new incremental current prediction model is built, which does not require rotor flux and resistance parameters. Second, based on the above model, Adaline neural network strategy is introduced to identify the inductance parameters which has a variable momentum to improve accuracy. Then, the strategy proposed in this paper has strong robustness to parameter mismatch. Finally, experiment results verify that the proposed method can effectively improve the parameter robustness.