Improved solar photovoltaic energy generation forecast using deep learning-based ensemble stacking approach

An accurate solar energy forecast is of utmost importance to allow a higher level integration renewable into the controls existing electricity grid. With availability data in unprecedented granularities, there an opportunity use data-driven algorithms for improved prediction generation. In this paper, generally applicable stacked ensemble algorithm (DSE-XGB) proposed utilizing two deep learning namely artificial neural network (ANN) and long short-term memory (LSTM) as base models forecast. The predictions from are integrated using extreme gradient boosting enhance accuracy PV generation model was evaluated on four different datasets provide comprehensive assessment. Additionally, shapely additive explanation framework utilized study deeper insight mechanism algorithm. performance by comparing results with individual ANN, LSTM, Bagging. DSE-XGB method exhibits best combination consistency stability case studies irrespective weather variations demonstrates improvement R2 value 10%–12% other models.