The application of neurocomputing on space vector modulation for current source converters

This paper presents the application of neurocomputing on space vector modulation (SVM) for current source converters. The proposed technique takes advantages of a modified Kohonenpsilas competitive layer to calculate the dwelling time of the adjacent switching state vectors. By using the algorithm, the SVM does not require much calculation to guarantee exact positioning of the switching instants and their dwelling time, so the hardware and software complexity is reduced, and the maximum attainable switching frequency and thus the bandwidth of the control system is increased. When compared to the conventional implementations of SVM techniques, the proposed method has more advantages, which is verified by the Matlab simulation.