Dictionary Learning for Short-term Prediction of Solar PV Production

Prediction of power generated from renewable energy resources such as solar photo-voltaic (PV) is a crucial task for stabilization of grids with high renewable penetration levels. Short-term prediction of these resources allow for preemptive regulation of injected power fluctuations. In this paper, a new algorithm based on dictionary learning for prediction of solar power fluctuations is introduced. This algorithm is effective on systems with structural regularities. In this method, a dictionary is trained to carry various behaviors of the system. Prediction is performed by reconstructing the tail of the upcoming signal using this dictionary. After introduction of the proposed algorithm, experimental results are provided to evaluate the prediction mechanism.