Dr BU Siqi, Member of the Otto Poon Charitable Foundation Research Institute for Smart Energy (RISE), Associate Professor and Associate Head of the Department of Electrical and Electronic Engineering, and his teammates have developed a novel deep reinforcement learning-based interpretable framework considering outliers for the accurate prediction of photovoltaic (PV) power. Their study was recently published in the journal Sustainable Energy Technologies and Assessments (https://www.sciencedirect.com/science/article/pii/S2213138824002261?dgcid=coauthor#appSB).
The team utilised cutting-edge deep reinforcement learning-based interpretable models to predict the occurrence of PV power outliers and improve PV power prediction accuracy. The proposed framework not only distinguishes outliers in forecast data, but also obtains complete outlier information with correct classification of PV power types, which can help engineers take emergency measures such as participating in demand response, increasing energy reserves, and optimising bids.
The framework was verified against actual operation data from the PV plant in Northwest China. The prediction performance of the proposed framework was analysed and compared with benchmark methods. The research results demonstrated the great potential of the proposed PV power prediction framework for application in future electric energy systems.
