A permanent magnet synchronous generator (PMSG) is a generator where the excitation field is provided by a permanent magnet instead of a coil. In this work space vector modulation (SVM) method using artificial neural network (ANN) on the stator side converter of a 6KW PMSG controlled by direct power control (DPC). In this article, the converter is controlled by a new SVM technique in order to minimize the total harmonic distortion (THD) and powers ripples (active and reactive). The validity of the proposed strategy applied on the PMSG is verified by Matlab/Simulink. On the other hand, the reactive power, stator current, active power is determined and compared with conventional command strategy.