| Abstract |
This contribution deals with an improvement of the Predictive Direct Power Control (P-DPC). The standard D-PDC strategy a) establishes a constant switching period, b) based on the nearest three-vector (NTV) strategy, the method selects three voltage-vectors to be concatenated during the switching period, building a symmetrical 3+3 switching patern c) by taking into account the time-derivates of the active and reactive powers, the vector-application times that minimize the final tracking error are computed. The selection of three vectors based on the NTV strategy minimizes the current ripple and the THD but does not offer the entire available transient dynamic. When hard transients are required, incoherent application times are obtained, i.e., control saturation occurs. This paper presents a method in order to avoid this saturation. When saturation is detected, the algorithm changes to a three-voltage vectors sequence built by two far and strongly active voltage vectors followed by a near and poorly active voltage-vector. Though the solution (application times) of the standard P-DPC problem is obtained by the minimization of a weight function, this contribution shows that it is possible to compute the exact application times for both steady and transient vector sequences. In comparison to other state-of-the-art control methods under constant switching frequency, the exact D-PDC strategy offers the faster transient behavior and similar steady-state performance. Simulated results match with the expected behavior and reinforce the choice of the P-DPC as an attractive candidate for the control of MV grid connected converters. |
