Cost-Function Extremum Seeking Based Model Predictive Torque Control of PMSM With Sensorless Position Estimation

In this article, a cost-function extremum seeking-based model prediction torque control (CES-MPTC) method of permanent magnet synchronous motor with sensorless position estimation is proposed. First, as the predictive torque and flux linkage are the functions of the DQ-axis voltage components, once the partial derivative of the cost function with the DQ-axis voltage components are zero, the extreme point of the cost function can be located and the optimal voltage vector can be directly represented by the calculated DQ-axis voltage components. Using the calculated DQ-axis voltage components as the inputs of the space vector pulsewidth modulation module, the inverter can be controlled with fixed switching frequency as the traditional vector control. Then, an improved slide model observer-based sensorless position estimation scheme is proposed with hyperbolic sinusoidal function, frequency adaptive complex coefficient filter and quadrature phase-locked loop integrated. The sensorless position estimation scheme can be integrated with the proposed CES-MPTC, which can ensure enhanced rotor position measurement accuracy and reliable sensorless operation. Finally, detailed experimental results are provided to verify the validity of the proposed scheme.