Optimal State Reference Computation with Constrained MTPA Criterion for PM Motor Drives

This research proposes a procedure that maps a PMSM torque request onto optimal state (current) references. Combining the procedure with a dynamic (current) controller yields a torque controller. The maximum torque per ampere (MTPA) criterion is used to minimize conduction and switching losses. This research extends the concept to field-weakening operation to obtain high efficiency at any machine speed. The resulting constrained MTPA criterion is formalized as an optimization problem. Since it is difficult to solve directly, the maximum and intersection torque subproblems are identified. An algorithm is obtained that maps a torque onto an optimal state reference, and it is sufficiently efficient for real-time implementation. This method is compatible with a variety of state (current) controllers with/without PWM, SPM and IPM machines with saliency and reverse saliency, and a variable dc-link voltage. The proposed procedure relies on a sufficiently accurate torque model that may not be provided using rated machine parameters. Thus, an approach to compute locally optimized machine parameters is proposed that takes magnetic saturation into account. The concept is developed on a software-in-the-loop platform and evaluated on an experimental test bench.