Deep Learning Forecaster-Based Controller for SVC: Wind Farm Flicker Mitigation

One of the major challenges in grid-connected wind farms is flicker emission caused by reactive power extremely short time variations. A general used solution installing a static volt-ampere (VAR) compensator (SVC) as most cost-saving device, which still holds risk incomplete compensation subject to operating delay time. To this end, an accurate extreme short-term forecasting structure essential. Although there are many studies related long, medium, and terms, however, prediction only addressed few studies. The main aim article develop method based on convolutional neural network (CNN) directly learn nonstationary complex features from raw farm series contribute predictive controller mitigate voltage through SVC connected farm. However, high sensitivity typical mean-squared-based loss functions might lead considerable errors forecasting. resolve issue, modified function proposed enhance performance and, consequently, improved CNN presented. evaluation actual recorded data Manjil, Iran. modeling current source, records its amplitude phase angle change every 0.01 s, utilized model numerical results terms sensation perceptibility (Pst) measurement verify comparison with without common control system.