Low-carbon optimization scheduling of IES based on enhanced diffusion model for scenario deep generation
Danhao Wang, Daogang Peng, Dongmei Huang, Huirong Zhao, Bogang Qu
Abstract
With the widespread integration of renewable energy sources into IES (Integrated Energy Systems), the uncertainties in renewable energy outputs and the volatility of user loads present new challenges for energy scheduling optimization. To effectively address these uncertainties, this paper proposes a scheduling method for IES based on an enhanced DM (Diffusion Models). Firstly, a mathematical model of a regional IES encompassing energy supply devices, energy coupling devices, and energy storage devices is constructed. Secondly, an enhanced DM optimized with a VAE (Variational Autoencoder) is developed. By efficiently generating data in the latent space, this model enhances the representation capability of uncertainty factors, enabling precise modeling of fluctuations in wind and solar power outputs as well as heating, cooling, and electrical loads. Finally, leveraging the improved DM, a large number of high-quality error scenarios are generated, and scheduling optimization experiments are designed with the objectives of minimizing economic costs and carbon emissions. Experimental results demonstrate that the proposed method not only achieves more efficient energy distribution and lower carbon emissions but also maintains the continuous and stable operation of the system when faced with uncertainties such as renewable energy generation fluctuations and changes in load demand.