A two-stage distributionally robust optimization operation scheduling model for solar PT-PV systems based on integrated load consumption prediction

At present, comprehensive utilization systems of multiple energy sources, with solar energy as the primary source, have been widely implemented. However, these systems face significant operational uncertainties, such as fluctuations in renewable energy supply and load demand. To address these challenges, this study proposes a two-stage distributionally robust optimization operation scheduling model that integrates load prediction and accounts for multiple uncertainties. First, the feature engineering process in load prediction is analyzed, and an integrated prediction load model incorporating the SHAP method is developed, effectively improving prediction accuracy. Second, a distributionally robust optimization ambiguity set based on joint constraints of the 1-norm and ∞-norm was constructed, and an effective integrated demand response model was established to improve the utilization of renewable energy. Finally, the proposed method is solved using the C&CG algorithm and validated through a case study on Tibet's solar photothermal-photovoltaic comprehensive utilization system. The results demonstrate that the proposed SHAP-based load prediction model improves the prediction accuracy of power and thermal loads by 10.19 %–32.72 % and 11.35 %–27.15 %, respectively. The two-stage DRO model reduces system operation costs by 8.62 %–57.43 % while significantly lowering computational expenses. Moreover, after incorporating the integrated demand response mechanism, the system's operation cost is further reduced by 19.32 %.