16 November 2023 in IEEE Transactions on Industrial Electronics
Model-free predictive current control (MFPCC) completely separates from the motor model, and only needs to read the current variations in the lookup tables to predict future currents, thus having strong parameter robustness. However, the stagnation of current variations is a core issue that needs to be addressed in existing MFPCCs. Therefore, this article proposes a more effective MFPCC strategy of permanent magnet synchronous motor. First, an advanced strategy is proposed to estimate current variations using a variant of the ultralocal model, where the local variables in the model are provided in real-time by sliding mode observer. This method can easily extract the current variations corresponding to all voltage vectors from the measured current variations caused by the voltage vectors applied in the previous period. Secondly, an anti-stagnation mechanism with higher reliability and better performance was designed. This mechanism effectively improves prediction accuracy by utilizing a pair of antiphase vectors and corresponding current variations, avoiding update stagnation. Since non-optimal vectors are not forced to be applied, this anti-stagnation mechanism has no impact on control performance. Finally, through comparative experiments on multiple predictive current control schemes, it is demonstrated that the proposed MFPCC has improved both steady-state and dynamic performance.
2 September 2025 in IEEE Transactions on Energy Conversion
The performance of the conventional deadbeat predictive current control (DPCC) method heavily depends on the accuracy of the motor model. To overcome this problem, a model-free deadbeat predictive current control (MFDPCC) technology based on measuring current variations with reliable prediction accuracy is proposed. The proposed method takes the natural response component of current variation as an adaptive variable, and the measured current variation as a constraint term for each component, thereby achieving the goal of separating the current variations components. Furthermore, the forced response component of current variation is utilized to calculate the unit voltage current variation and predict the reference voltage vector. To ensure prediction accuracy, the integral effect of current prediction error is designed as the adaptive rate of the natural response component, and the average filtering principle is used to suppress the impact of current sampling noise on performance. Compared to conventional DPCC and MFDPCC based on ultralocal models, the proposed strategy only involves sampling information without any observer or motor parameters, thus completely eliminating the dependence on the motor model. The experiments have shown that the steady-state performance of the proposed strategy is superior to conventional DPCC and MFDPCC based on ultralocal models.