Robust Deadbeat Predictive Current Sensorless Control Scheme for Permanent Magnet Synchronous Machine Considering Parameter Mismatch

In position sensor-based cases, the parameter-robustness issue of the deadbeat predictive current control (DPCC) is solved by estimating the parameter mismatch effect (PME) from the current errors. However, present methods fail to figure out such a problem in sensorless applications where the unknowable back electromagnetic force also needs to be observed by current errors. Hence, in this article, PME disturbance suppression methods are investigated in sensorless applications with respect to dynamic compensation and steady-state identification. In a dynamic process, the knowable part of PME is deduced and then compensated instantly. The proposed DPCC is validated as even more robust than proportional-integral current control under sensorless control. Although the robustness of sensorless DPCC with PME is guaranteed by the above compensation term, the unmatched parameters still bring problems of degradation of efficiency and overshoot in a dynamic process. To eliminate the PME for steady states, an R-statistic-based PME detection method is proposed to trigger inductance identification appropriately. The possible false or missed detections of the present methods are avoided by taking the variance of the current prediction errors as the process variable. Based on the above two improvements, a parameter robust sensorless DPCC with an overall PME compensation scheme is developed. Experiments are designed to validate the abovementioned theory.