Distributed optimization for network-constrained peer-to-peer energy trading among multiple microgrids under uncertainty

• Main Authors: Wang, Luhao, Wang, Zhuo, Li, Zhengmao, Yang, Ming, Cheng, Xingong
• Other Authors: School of Electrical and Electronic Engineering
• Format: Article
• Language: English
• Published: 2023
• Subjects: Engineering::Electrical and electronic engineering; Multiple Microgrids; Peer-to-Peer Energy Trading
• Online Access: https://hdl.handle.net/10356/172149
• Institution: Nanyang Technological University

This paper addresses network-constrained peer-to-peer (P2P) energy trading problems for multiple microgrids (MGs) under uncertainty. A bi-level distributed optimization framework is proposed to bridge the gap between physical power flows supervised by distribution system operators and logical P2P transactions among multiple MGs under uncertainty. At the upper level, a conditional optimal power flow model is formulated to minimize power losses and guarantee the operating security of local distribution networks. At the lower level, a stochastic programming-based P2P trading model for multiple MGs is formulated to pursue the flexibility of energy transactions among different entities. To realize the consistency of decision-making processes between the two levels and among different MGs, a nested bi-level distributed algorithm including a parallel analytical target cascading algorithm and an alternating direction multiplier method is designed to solve the proposed model in a distributed manner. Furthermore, an adaptive updating method for penalty parameters is adopted to decrease the sensitivity to the initialization. Finally, numerical tests are implemented in a modified IEEE 33-node distribution network with four MGs to testify to the validity of the proposed energy trading framework. The results confirm that the obtained P2P trading schemes can protect against uncertainties and satisfy network constraints, especially since the proposed parallel distributed algorithm has better computing performances compared to the traditional sequential distributed algorithm.