A microgrid power trading framework based on blockchain and deep reinforcement learning

Yangsheng Lou,Xinning Jia,Yanhong Liu,Xianzhi Zeng,Jie Xu,Zehui Xu
DOI: https://doi.org/10.1145/3622896.3622927
2023-08-25
Abstract:With the development of renewable energy technologies and the emergence of distributed power generation devices, traditional centralized power trading markets no longer meet people's transactional needs. Peer-to-peer (P2P) electricity trading within microgrids has become the future direction. However, in distributed trading scenarios, both parties in P2P transactions lack a foundation of trust and incentive mechanisms. To address these issues, we have designed an electricity trading framework based on blockchain and deep reinforcement learning. Users utilize deep reinforcement learning for load prediction and power planning to maximize their own interests. We propose a proof-of-work (POW) consensus algorithm based on reputation values to further improve consensus efficiency and block creation time. By introducing the InterPlanetary File System (IPFS), offloaded transactions are uploaded to IPFS for storage, enabling the unloading of unnecessary information and improving the resource utilization efficiency of blocks. Our framework is beneficial for reducing the costs for blockchain users, implementing credit management for P2P e-commerce transactions, and thereby enhancing the stability and efficiency of transactions.
Engineering,Environmental Science,Computer Science
What problem does this paper attempt to address?