Prediction of wind farm reactive power fast variations by adaptive one-dimensional convolutional neural network

One of the prominent problems in wind farms is voltage flicker emission. To prevent emission or mitigate impact as best possible, a static VAr compensator (SVC) great candidate both economically and technically. However, SVCs cannot completely compensate fast-changing reactive power due to delays caused by calculation unit triggering fire angle SVC. This paper proposes predictive control system for SVCs, merging an additional block into conventional system. It constructed based on deep neural networks, namely adaptive one-dimensional convolutional network (1D-CNN). The training process conducted learning weights enhance prediction accuracy computational complexity 1D-CNN. Numerical results actual dataset farm Manjil, Iran, have verified forecasting mitigation proposed controller.