Bilevel Model for Electricity Market Mechanism Optimisation via Quantum Computing Enhanced Reinforcement Learning

TL;DR

This paper introduces a novel quantum computing enhanced bilevel optimization model for electricity markets, integrating reinforcement learning and multi-agent strategies to improve market mechanism design and bidding strategies.

Contribution

It presents a dynamic bilevel model combining RL and quantum-enhanced multi-agent learning, a first in applying quantum computing to electricity market optimization.

Findings

Effective market mechanism optimization demonstrated on IEEE 30-bus system.

Quantum-enhanced MADQN improves bidding strategy simulation accuracy.

Model captures complexities of modern electricity markets.

Abstract

In response to the increasing complexity of electricity markets due to low-carbon requirements and the integration of sustainable energy sources, this paper proposes a dynamic quantum computing enhanced bilevel optimization model for electricity market operations. The upper level focuses on market mechanism optimization using Reinforcement Learning (RL), specifically Proximal Policy Optimization (PPO), while the lower level models the bidding strategies of Generating Companies (GENCOs) using a Multi-Agent Deep Q-Network (MADQN) enhanced with quantum computing through a Variational Quantum Circuit (VQC). The three main contributions of this work are: (1) establishing a dynamic bilevel model with timely feedback between the upper and lower levels; (2) parameterizing and optimizing market mechanisms to derive the most effective designs; and (3) introducing quantum computing into the context of electricity markets to more realistically simulate market operations. The proposed model is tested on the IEEE 30-bus system with six GENCOs, demonstrating its effectiveness in capturing the complexities of modern electricity markets.