Two-Stage optimization based finite state predictive torque control for 3L-NPC inverter Fed IM drives

The computational burden in Finite State Predictive torque control (FS-PTC) increases with the number of voltage vectors and number of variables in the cost function. The conventional FS-PTC for three level neutral-point clamped voltage source inverter (3L-NPC VSI) fed motor drive has a disadvantage of long execution time since all available vectors are evaluated in the prediction stage. This paper proposes a new method for reducing the number of voltage vectors from 27 to 17 in the prediction stage, which reduces the computational burden of conventional FS-PTC. The performance of the proposed voltage selection method is investigated for an induction motor drive in terms of electromagnetic torque and stator flux responses (transient and ripple), total harmonic distortion of stator currents, neutral point voltage variations, average switching frequency reduction and robustness of the drive. Experimental results confirm that the execution time is reduced by 30% compared to conventional FS-PTC while similar dynamic and steady-state performances are preserved.