Bi-Level Optimization for Capacity Configuration and Low-Carbon Scheduling of Multi-Energy Systems with Energy-Carbon Coupling

Coordinating capacity configuration and operational scheduling in multi-energy systems is inherently complex, given their distinct time scales and intertwined economic and carbon constraints. To address this challenge, a bi-level optimization model for capacity configuration and operational scheduling of multi-energy systems with energy-carbon coupling was developed within a unified Mixed-Integer Linear Programming (MILP) framework. The model incorporates Wasserstein Generative Adversarial Network (WGAN) and K-means++ algorithms for Photovoltaic (PV) and Wind Turbine (WT) output scenario generation and clustering, and constructs a carbon flow traceability matrix based on a load-proportional allocation method. The proposed approach was validated through a case study covering 40 typical and extreme scenarios across four seasons. Optimization results showed that compared to the original park configuration without PV, WT, and energy storage, this configuration reduced the annualized total cost by 34.2% and daily carbon emissions by 76.0%, while lowering daily carbon trading cost from 85.13 CNY to zero. The proposed model provides a systematic tool for coordinating capacity configuration and low-carbon operational scheduling in industrial park multi-energy systems under multiple scenarios.