Abstract:High penetration of renewable energy results in significant forecast errors of generation and load. Existing multi-timescale scheduling suffers from insufficient coordination between day-ahead and intraday dispatch as well as frequent full-period regulation without deviation threshold constraints. To tackle these drawbacks, a deviation-aware multi-timescale collabora-tive optimization framework is proposed. Flexible resources are modeled uniformly, and regulation budgets are determined based on historical forecasting deviations. A closed-loop mechanism consisting of day-ahead budget reservation, thresh-old-triggered intraday modification and real-time rapid regulation release is established to realize deviation-driven scheduling. In the day-ahead stage, the reference vector-guided multi-objective dung beetle optimizer combined with the minimum spanning tree is adopted to solve the multi-objective optimization model. Model predictive control is utilized in the intraday stage to track day-ahead scheduling plans iteratively, while real-time dispatch adopts droop control of energy storage assisted by power flow backtracking to guarantee power balance and voltage security. Simulations on the modified IEEE 33-bus test system verify that the proposed framework improves renewable energy accommodation, operational economy and voltage security under ordinary, extreme and persistent deviation scenarios, with favorable engineering practicability.
