A Neural Network Current Predictive Control Of Statcom For Non-Linear Loads

This paper presents a novel current predictive control scheme based on Artificial Neural Network of STATCOM for non-linear loads which have sudden changing current. The current harmonics and power of non-linear loads, especially, uncontrolled rectifier is introduced by the instantaneous power theory. The track issue of the traditional hysteresis current control strategy of STATCOM is analyzed. In order to resolve this issue, a novel hysteresis current control scheme using the Artificial Neural Network Predictive Current Pulse Compensator is designed to predict compensating reference current waveform of the STATCOM. The method could change the STATCOM output current in advance, make the output current quickly track reference current, and then, minimize harmonics of grid current, suppress the sag of micro-grid voltage. Simulation results verified that the proposed control scheme can effectively compensate reactive power of the grid, and suppress grid current harmonics, even in sudden changing current of non-linear load.