研究目的
To improve operational efficiency and customers' satisfaction in smart microgrids by proposing a distributed cooperative energy management system that addresses the stochastic nature of renewable energy and privacy concerns.
研究成果
The proposed DCEMS effectively reduces energy purchasing costs by over 30% compared to heuristic methods, closely matches centralized EMS performance, avoids single point of failure, and preserves privacy by sharing only aggregated data. Future work includes enhancing prediction accuracy with advanced deep learning and extending to electricity price prediction.
研究不足
The study assumes the utility does not buy back power from users and that electricity price is known one-day ahead, which may not hold in real scenarios. The simulation is based on emulation, not real-world deployment, and the solar prediction accuracy could be improved.
1：Experimental Design and Method Selection:
The study uses a two-stage approach: first, each SMG optimizes its operation based on solar energy prediction using LSTM RNN; second, SMGs share energy exchange information with neighbors using a distributed ADMM algorithm to optimize global cost.
2：Sample Selection and Data Sources:
Historic hourly solar energy data from 1991-1993 from the National Renewable Energy Laboratory is used for training the LSTM model.
3：List of Experimental Equipment and Materials:
A distributed SMGs simulation platform based on CORE and GridLAB-D is used for emulation.
4：Experimental Procedures and Operational Workflow:
The algorithm initializes randomly, iterates to convergence using ADMM, and compares performance with centralized and heuristic EMS.
5：Data Analysis Methods:
Performance is evaluated based on monetary cost reduction and convergence error using mean squared error for prediction and primal/dual residues for ADMM.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
