基于改进对偶动态规划的产销者点对点交易全分散式出清方法

Fully Decentralized Clearing for Peer-to-peer Trading of Prosumers Based on Improved Dual Dynamic Programming

分布式资源的增长使得电力用户逐步从消费者转变为产销者。主要针对产销者点对点(peer-to-peer,P2P)交易全分散式出清问题开展研究。首先,以产销者总成本最小为目标,综合考虑过网费与网络约束,建立产销者P2P交易模型。其次,基于多阶段优化理论将P2P交易模型转化为全分散式优化模型,并以节点累积偏移电压和线路累积使用容量为状态变量,清晰地刻画每一对交易对配电网运行的影响情况。接着,利用对偶动态规划(dual dynamic programming,DDP)进行求解。为了提高求解效率,采用帕累托最优割技术对DDP进行改进,以减少冗余Benders割的产生,在保证算法精度的同时提高其收敛速度。最后,通过算例分析验证了所提方法的有效性。

The growing distributed resources has gradually transformed the power users from consumers to prosumers.In this paper,a method of fully decentralized clearing for peer-to-peer trading of the prosumers is studied based on the improved dual dynamic programming.First,by considering the network utilization fees and the network constraints,a P2P transaction model for the prosumers is established to minimize the total cost.Second,based on the multi-stage optimization theory,the peer-to-peer transaction model is transformed into a fully decentralized optimization model.By using the node cumulative voltage offset and the cumulative capacity of the lines as the state variable,the impact of each pair of transactions on the distribution network is clearly described.Then,the dual dynamic programming is used to solve the constructed model.In order to improve the solution efficiency,the Pareto optimal cut technique is used to reduce the generation of redundant benders cuts,thus improving the convergence speed of the dual dynamic programming with its accuracy guaranteed.Finally,the effectiveness of the proposed method is verified by some simulation analyses.