A Parallel Hybrid Forecasting Framework for Economic Dispatch: With Applications for China's Electricity Market Operations

China's “Dual Carbon” targets demonstrate a strategic commitment to addressing energy security and climate change. To achieve these decarbonization goals, China has implemented two initiatives: accelerated deployment of variable renewable energy sources (VRES) and a nationwide emissions trading system (ETS) for the power sector. The demand fluctuations, carbon price volatility, and inherent variability of VRES generation introduce complex challenges for economic dispatch (ED) in the developing electricity markets. This study alleviates the uncertainties by proposing a parallel hybrid dynamic self-learning forecasting (PHDSLF) framework that balances prediction accuracy with computational efficiency. The impacts of VRES generation forecasting are analyzed by utilizing a random error generation method. Incorporating the dynamic battery storage state and ancillary services, this study elucidates the operational interdependencies between the electric energy markets (EEMs) and the ancillary service market (ASM) by developing a dynamic economic dispatch (DED) model for the day-ahead market and a revised DED model for the real-time market. The robustness of the proposed forecasting-optimization framework is verified based on the China-based node system and the real carbon price series. Key findings reveal: (1) Ancillary service expansion alone cannot sufficiently mitigate dispatch cost escalations caused by significant VRES forecasting errors; (2) Strategics between energy market bidding and ancillary service participation are critical for battery storage cost recovery; (3) Utilities of carbon pricing on emission reductions are more evident when there are lower VRES penetration rates, more competitive battery cost, and larger storage capacities.