Distributed Cooperative Energy Management in Smart Microgrids with Solar Energy Prediction

Abstract

Smart Microgrid (SMG), integrated with renewable energy, energy storage system and advanced bidirectional communication network, has been envisioned to improve efficiency and reliability of power delivery. However, the stochastic nature of renewable energy and privacy concerns due to intensive bidirectional data exchange make the traditional energy management system (EMS) perform poorly. In order to improve operational efficiency and customers’ satisfaction, we propose a distributed cooperative energy management system (DCEMS). We adopt recurrent neural network with long short-term memory to predict the solar energy generation with high accuracy. We then solve the underlying economic dispatch problem with distributed scalable Alternating Direction Method of Multipliers (ADMM) algorithm to avoid single point of failure problem and preserve customers’ privacy. In the first stage, each SMG optimizes its operation decision vector in a centralized manner based on one-day ahead solar energy generation prediction. In the second stage, all SMGs share their energy exchange information with directly connected neighboring SMGs to cooperatively optimize the global operation cost. The proposed DCEMS is deployed in our distributed SMGs emulation platform and its performance is compared with other approaches. The results show that the proposed DCEMS outperforms heuristic rule-based EMS by more than 30%. It can also protect customers’ privacy and avoid single point of failure without degrading performance too much compared to centralized EMS.