Polynomial Surface Model-Based Balancing Market Bid Planning and Evaluation for Multi-Site PV Plants

The Seventh Basic Energy Plan sets an ambitious target of expanding solar power generation beyond the current 7%, highlighting a strong national commitment to renewable energy. However, the large-scale deployment of photovoltaics (PV) faces two critical challenges: the inherent uncertainty of power generation due to weather variability, and the decline in profitability associated with daytime price drops under high renewable energy penetration. This study explores the creation of new value for PV systems through the provision of reserve power to address these challenges. Accurate management of prediction errors is essential to unlock this value and improve dispatchability. Building on our previous work, we applied kernel density estimation to model the probability distribution of historical prediction errors and constructed a polynomial surface model that incorporates PV-predicted values and the clearness index as features, with cumulative probability serving as the error threshold. In this study, the model is extended to multiple sites to demonstrate its general applicability and to evaluate the smoothing effect of aggregated prediction errors. The results show that aggregating PV reserve power can effectively reduce imbalances in balancing market operations, underscoring its potential to support the sustainable, large-scale integration of PV into the power grid.