Abstract:Peer-to-peer energy trading offers a promising solution for enhancing renewable energy utilization and economic benefits within interconnected microgrids. However, existing real-time P2P markets face two key challenges: high computational complexity in trading mechanisms, and suboptimal participant decision-making under diverse uncertainties. Existing prediction-based decision-making methods rely heavily on accurate forecasts, which are typically unavailable for microgrids, while prediction-free methods suffer from myopic behaviors. To address these challenges, this paper proposes an improved double auction mechanism combined with an adaptive step-size search algorithm t…
Abstract:Peer-to-peer energy trading offers a promising solution for enhancing renewable energy utilization and economic benefits within interconnected microgrids. However, existing real-time P2P markets face two key challenges: high computational complexity in trading mechanisms, and suboptimal participant decision-making under diverse uncertainties. Existing prediction-based decision-making methods rely heavily on accurate forecasts, which are typically unavailable for microgrids, while prediction-free methods suffer from myopic behaviors. To address these challenges, this paper proposes an improved double auction mechanism combined with an adaptive step-size search algorithm to reduce computational burden, and a data-driven dual-reference online optimization (DDOO) framework to enhance participant decision-making. The improved mechanism simplifies bidding procedures, significantly reducing computational burden and ensuring rapid convergence to the market equilibrium. Additionally, the prediction-free DDOO framework mitigates myopic decision-making by introducing two informative reference signals. Case studies on a 20-microgrid system demonstrate the effectiveness and scalability of the proposed mechanism and approach. The improved mechanism significantly decreases the computational time while increasing local energy self-sufficiency periods from 0.01% to 29.86%, reducing reverse power flow periods from 24.51% to 3.96%, and lowering average operating costs by 19.20%. Compared with conventional approaches such as Lyapunov optimization and model predictive control, the DDOO framework achieves a 10%-13% reduction in operating costs with an optimality gap of only 5.76%.
| Subjects: | Systems and Control (eess.SY) |
| Cite as: | arXiv:2510.02985 [eess.SY] |
| (or arXiv:2510.02985v1 [eess.SY] for this version) | |
| https://doi.org/10.48550/arXiv.2510.02985 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Kaidi Huang [view email] [v1] Fri, 3 Oct 2025 13:22:51 UTC (1,584 KB)