12-09-2017, 10:41 AM
This MPC strategy is novel from the point of view of the power supply (a three-phase inverter) since, instead of using a single voltage vector as a control action during a complete sampling period of the controller, as is done in the standard DTFC strategies, it is proposed to use an appropriately selected sequence of voltage vectors during the same period. This latter property allows us to synthesize smoother control actions, which considerably reduces the ripple component of the electromagnetic variables (current, flux, electromagnetic torque). The selection of the voltage vector sequence at each sampling time is carried out by minimizing a quadratic cost function that penalises errors between magnitude and flux pair and their respective desired references over a prediction horizon. This is standard in MPC. However, in contrast to the usual form of MPC, we propose to apply the entire optimization sequence (of duration equal to the sampling period) instead of applying the first vector of the sequence in a backward horizon. In addition, by appropriate parameterisation of the sequence of the voltage vector on the horizon, we will provide the proposed strategy with the same implementation strategy as that used in the spatial vector modulation technique. This property achieves a fixed switching frequency for the voltage source inverter that powers the motor. This is well known to ensure good inverter performance. The results of the simulation show that the proposed strategy considerably exceeds the standard MPC strategy with a very moderate computational cost increase.