A Novel Control Method for DSTATCOM Using Artificial Neural Network

In this paper, a new control method for distribution static compensator (DSTATCOM) using artificial neural networks (ANN) instead of the conventional direct controller is presented. A feedforward three-layer neural network is adopted to control the output compensating currents. This is accomplished by producing the appropriate switching patterns of the IGBTs. A backpropagation algorithm trains this network. A practical case with PSCAD/EMTDC is presented to check the proposed control performance. The simulation results show a better source current waveforms and a near unity power factor operation is achieved. This technique offers an alternative to the multi-pulse techniques that require complex magnetic circuit arrangements as well as to the conventional PI controller that the parameters are hard to determine