Multi-Agent LSTM Optimal Strategy of Microgrid-Distribution Layered Network Considering High Proportion of Renewable Energy

• Published in: International Conference on Electrical Machines and Systems (Online) pp. 2339 - 2343
• Main Authors: Lu, Wenqi, Shen, Yu, Kong, Xiangyu, Hu, Wei, Sun, Fangyuan
• Format: Conference Proceeding
• Language: English
• Published: KIEE & EMECS 31.10.2021
• Subjects: Costs; dispatch strategy; energy blockchain; Fluctuations; Generators; Microgrids; multi-subject optimization; power market; Real-time systems; renewable energy resource; Renewable energy sources; Smart grids
• ISSN: 2642-5513
• DOI: 10.23919/ICEMS52562.2021.9634622

With high proportion of renewable energy and the improvement of sensing and communication equipment in the smart grid, the autonomy of the multi agents to participate in market transactions has increased. But it is difficult to smooth power's fluctuation and achieve a fair distribution of costs and benefits. Based on the Internet of Things (IoT) and energy blockchain technology, this paper proposes a real-time multiagent power transaction framework and optimization model. The method firstly predicts the real-time demand response (DR) in electrical power with the price changes using the Long Short-Term Memory (LSTM) network. Then, an optimizition model is built to pursue multiple subjects' respective benefit and optimize the microgrids' internal economy. Further, the distribution network operator monitors the system security, returns the corrected results to the microgrid for circular optimization. Finally, the improved particle swarm algorithm is used to solve the established model. A case study proves that the model can effectively smooth the power fluctuation of new energy, stimulate users to participate in the market DR, and realize the safety, economy, and fairness of smart grid operation in the new environment.