A risk uncertainty–based flexible dispatch method for pumped storage under extreme weather

The large-scale integration of renewable energy into new power systems presents significant challenges in terms of controllability and predictability due to its inherent randomness and volatility. The uncontrollability is further compounded by the sudden impacts of extreme weather events, making multi-scale power balance increasingly difficult. This paper proposes a two-stage flexible dispatch method (TFDM) for measuring uncertainty under extreme weather. The model constructs probabilistic disaster impact assessments for wind farms and transmission lines using fuzzy sets and captures the associated uncertainties through chance constraints and risk costs. It provides more flexible disaster risk reduction methods by integrating uncertainty prediction errors and unit commitment plans into the interaction between two-stage dispatch. An accelerated algorithm that combines historical scene learning with improved K-nearest neighbours (KNN) dispatch is proposed. It is employed to obtain initial solutions for binary unit commitment variables, accelerating the resolution of frequent unit combination switching problems. Tests on the IEEE 39-bus and IEEE 118-bus systems show that the proposed method can effectively utilize the regulating capacity of pumped storage units to significantly improve the resilience of the system to cope with extreme weather. The improved KNN algorithm has stronger convergence and efficiency for large-scale power grid scenarios.