Multi-timescale Multi-type Electricity and Carbon Emission Market Simulation by Hierarchical Quantum-classical Decision Making Framework

Electricity market simulations are designed to model the interactions and decision-making behaviors of market participants, such as GENCOs, under a set of predefined market rules. These simulations assist policymakers and market designers in assessing how changes in market mechanisms will impact the behavior of market participants, the efficiency of energy distribution, and the fairness of market outcomes. However, current simulation tools exhibit notable limitations, such as challenges in fully modeling GENCOs' decision-making processes across multi-market environments, poor convergence in multi-timescale simulations, and fail to accurately capture the correlations and irrationality in GENCOs' behaviors. In this paper, we introduce an innovative quantum-classical decision simulation framework, utilizing quantum computing advantages to overcome these challenges. We first construct a hierarchical Markov decision process (H-MDP) model to simulate GENCOs' decision-making across different time scales, such as long-term and spot markets, and across diverse market types, including electricity and carbon emission auction markets. Building on this H-MDP model, we propose a quantum-enhanced multi-agent soft actor-critic (Q-MASAC) algorithm, employing a variational quantum circuit (VQC) in place of conventional deep neural networks to optimize GENCOs' decision-making. We exploit quantum computing's parallel processing capabilities to accelerate simulations and significantly improve convergence performance. Furthermore, by leveraging the superposition and entanglement properties of quantum states, our framework more effectively captures the correlations and irrationality in GENCOs' decisions. The case study results demonstrate that our algorithm achieves superior convergence, with simulation outcomes more closely mirroring real-world market dynamics.