Deep Learning Based Behavior Modeling in Coupled Electricity and Carbon Market

In response to global efforts toward carbon neutrality and the pressing need for sustainable energy systems, the coupled electricity and carbon markets have become critical for optimizing both economic and environmental objectives. This paper proposes a deep learning-based behavior modeling framework to analyze the interactions between these two interconnected markets. By integrating real-world data and hybrid experimental learning, we simulate strategic offering behaviors using generative models, particularly Variational Autoencoders (VAE). Our approach accounts for the complexities of joint market dynamics, including the impact of carbon pricing on electricity generation strategies. A nationwide case study using data from over 1,600 market participants across China demonstrates the model’s efficacy in replicating market conditions and offers insights into market participant behaviors. The results underscore the significant influence of trading behavior diversity on market outcomes, providing a valuable tool for policymakers and stakeholders in achieving sustainability goals through informed decision-making.