Two-stage collaborative optimization for low-carbon park energy systems: day-ahead planning and intra-day rolling with dynamic carbon emission factors

Against the backdrop of increasing renewable energy penetration and flexible resources in low-carbon parks, energy systems face heightened uncertainties in multi-energy coupling and dynamic responses. This study develops a carbon-oriented two-stage collaborative optimization framework for integrated energy systems, consisting of day-ahead cost-minimization scheduling and intra-day rolling optimization for dynamic carbon mitigation. The day-ahead stage optimizes cooling/heating sources and thermal storage under forecasted conditions, while the intra-day stage performs carbon-oriented and demand-response adjustments by flexible resources including battery storage and controllable loads. A comprehensive simulation model of a typical low-carbon park, combined with ultra-short-term forecasting and rolling-horizon optimization is established to verify the performance under various supply-demand conditions. Robustness against operational uncertainties is ensured through representative intra-day disturbance scenarios. Results show that the strategy reduces carbon emissions by 2.1–9.34%, lowers average carbon intensity by 6.8%, decreases operational costs by 7.5%, and increases renewable energy utilization by 15.3%. These findings demonstrate that incorporating dynamic carbon signals into decision-making provides an effective and practical approach for low-carbon operation of park-level integrated energy systems.