Three-vector control based on current gradient adaptive updating for permanent magnet synchronous motor

Abstract

To address significant current ripple and current gradient stagnation in conventional model-free predictive current control for permanent magnet synchronous motors, this article proposes an enhanced three-vector control strategy with adaptive current gradient updating. The method overcomes the limitation of fixed vector amplitude and direction per control cycle inherent in conventional model-free predictive current control using a combination of two adjacent active vectors and a zero vector within one period, thereby expanding the achievable output voltage range. The voltage equation is concurrently reformulated based on the current gradient principle. Next, the current gradient error corresponding to the vector application times is determined. These errors are then used by a moving least squares algorithm to adaptively update the current gradients associated with the applied vectors. Critically, this process facilitates the update of gradients for all potential voltage vectors within a single control cycle, effectively eliminating gradient update stagnation. Furthermore, the voltage sector division is optimized, and the optimal voltage vector combination is screened through cost function ranking, achieving accurate vector selection with minimal computation. Experimental results validate the proposed strategy, demonstrating its ability to reliably update the online current prediction model, reduce prediction errors and current ripple significantly, and achieve enhanced transient response and steady-state performance.