Model-Free Predictive Current Control of PMSM Drives Using Recursive Least Squares Algorithm

Model predictive control (MPC) has received tremendous attention and has been widely studied in academia due to its straightforward concept, easy implementation, and fast dynamic response. However, MPC suffers from performance degradation when the system parameters are mismatched, which hinders its widespread adoption. To tackle this challenge, the model-free predictive current control (MFPCC) strategy has been developed. Compared with conventional MPC methods, the MFPCC strategy can be implemented by utilizing the input and output measurement data of the system without prior knowledge of the system parameters. Therefore, the influence of parameter mismatch can be eliminated with the MFPCC method. However, the conventional MFPCC method has the problem of current stagnation updates, which will degrade control performance. In this work, we propose a novel MFPCC method for a permanent magnet synchronous machine (PMSM) based on the recursive least squares (RLS) algorithm. The proposed method first replaces the classical fundamental model of PMSM with an ultralocal model and then employs the recursive least squares method to identify the parameters of this ultralocal model. In addition, an oversampling technique is used in this work to obtain a more accurate slope of the stator current, which facilitates the resolution of the parameters by RLS. The effectiveness and superiority of the proposed MFPCC method have been verified by the experimental results.