Prediction-Error-Driven Position Estimation Method for Finite-Control-Set Model Predictive Control of Interior Permanent-Magnet Synchronous Motors

This paper proposes a prediction-error-driven position estimation method based on the current prediction errors of the finite-control-set model predictive control (FCS-MPC) for the position-sensorless control of interior permanent-magnet synchronous motors, which operate in the whole speed range, including zero speed. The method is customized for FCS-MPC with which switching devices are directly controlled without modulators, whereas the conventional sensorless control methods usually presume the use of modulators. In the low-speed region, the proposed method fully exploits the inherent excitation of the discrete switching of the FCS-MPC to extract the position information. Under the low-current condition, an active vector injection technique is designed to automatically maintain sufficiently rich excitation for the position estimation while exerting a minimum impact on the fundamental control. The resultant current ripples and acoustic noises are lower compared with conventional injection methods. A novel initial polarity detection method based on magnetic saturation effect and inductance identification is also proposed to prevent pole ambiguity. The high-speed region position estimation uses the excitation from the back electromotive force, which is also reflected in the current prediction errors, permitting the use of one simple linear estimator in the whole speed range. Finally, the effectiveness and benefits of the proposed method are validated by experimental results.