Data-Model Hybrid-Driven Multi-Timescale Low-Carbon Economic Dispatch for Manufacturing-Park Virtual Power Plants

INTRODUCTION: Manufacturing parks combine energy-intensive production equipment, auxiliary systems, distributed renewables, and flexible loads, creating coupled fluctuations between production schedules and electricity demand. OBJECTIVES: This study develops an auditable data-model hybrid method that links empirically updated uncertainty scenarios with a production-constrained, low-carbon, three-stage dispatch model for manufacturing-park virtual power plants. METHODS: The data module estimates wind-speed, irradiance, manufacturing-load, and price distributions from rolling historical records, generates joint scenarios, and updates operating states. The model module solves a coupled day-ahead-intraday-real-time stochastic program with process-feasibility, carbon-flow, storage, and inter-park constraints; reduced scenarios and measured deviations form the interface between the two modules. RESULTS: Case studies show a 22.6% reduction in operating cost, an 18.0% reduction in carbon emissions, a renewable-energy absorption rate of 94.8%, and a 48.8% reduction in power-fluctuation standard deviation compared with single-time-scale dispatch. CONCLUSION: The proposed framework coordinates energy and production decisions while maintaining production feasibility and improves economic, low-carbon, and operational performance under high renewable penetration.