Model predictive current control of a three-level five-phase NPC VSI using simplified computational approach

The paper proposes a computationally efficient finite state model based current control of a three-level neutral point clamped (NPC) five-phase voltage source inverter. The space vector model of a five-phase voltage source inverter (VSI) yields 243 space vectors, with 240 active and three zero vectors. The presented technique utilizes a modified switching algorithm to reduce the computation time while incorporating all the 243 switching state vectors. Search is made by the algorithm to find the vector which minimizes the chosen cost function. Capacitor voltage balancing is also done by incorporating an additional cost function. The performance of the current control heavily depends upon the choice of the cost function, the number of vectors used and the sampling time. The developed technique is tested for an RLE load using simulation and an experimental approach.