文章摘要
戴月,刘明波,王志军,等.高光伏渗透配电网分散式最优潮流的割平面一致性算法[J].电力系统自动化. DOI: 10.7500/AEPS20170907002.
DAI Yue①,LIU Mingbo②,WANG Zhijun③, et al.Cutting Plane Consensus Algorithm for Optimal Power Flow of Distribution Networkswith High Photovoltaic Penetration[J].Automation of Electric Power Systems. DOI: 10.7500/AEPS20170907002.
高光伏渗透配电网分散式最优潮流的割平面一致性算法
Cutting Plane Consensus Algorithm for Optimal Power Flow of Distribution Networkswith High Photovoltaic Penetration
DOI:10.7500/AEPS20170907002
关键词: 主动配电网  高光伏渗透  最优潮流  割平面一致性算法  分散式优化
KeyWords: active distribution networks  high photovoltaic penetration  optimal power flow  cutting plane consensus algorithm  decentralized optimization
上网日期:2018-07-09
基金项目:国家重点基础研究发展计划(973计划) (2013CB228205)
作者单位E-mail
戴月 华南理工大学电力学院 13450243047@163.com 
刘明波 华南理工大学电力学院 epmbliu@scut.edu.cn 
王志军 华南理工大学电力学院 epzjwang@163.com 
谢敏 华南理工大学电力学院 minxie@scut.edu.cn 
摘要:
      本文采用割平面一致性算法对高光伏渗透主动配电网的最优潮流问题进行分散式求解。首先对潮流方程进行线性近似,将节点电压表示为节点注入功率线性函数,搭建包含网络损耗和弃光惩罚费用的二次规划模型。然后利用拉格朗日对偶松弛技术将集中式模型解耦,将配电网络和各光伏逆变器分别作为独立的代理,对每个代理建立主问题对原问题进行近似,在相邻代理间仅传递割平面约束,最终一致地求出各代理的最优解。各代理相互独立,保密性好,不需要上层协调器,并且在发生割平面信息丢失时仍然可以收敛到最优解。最后,通过对一实际屋顶光伏系统和33节点系统进行测试,证明文中所提方法的正确性和有效性。
Abstract:
      ABSTRACT:In this paper, an improved cutting plane consensus algorithm is adopted to optimize the optimal power flow problem in the high photovoltaic active distribution network.Firstly, the power flow equation is approximated linearly, i.e., the node voltage is expressed linearly by node injection power, and hencea quadraticprogramming model including network loss and penalty cost of the curtailment of photovoltaics are constructed. Then, using Lagrange dual relaxation technology to decouple the centralized model, take the distribution network and the PV inverters as independent agents. For each agent, build a master problem to approximate the original problem and only the cutting plane constraints are transferred between agents. At last, on the premise of satisfying the coupling constraint between each agent, the optimal solutions can be obtained for each agent. The agents are independent of each other, with good confidentiality, do not need an upper level coordinator, and can converge to the optimal solution when the cutting plane information is occasionally lost. Finally, an actual rooftop photovoltaic system and a 33-bus system are tested to demonstrate the correctness and effectiveness of the proposed method.
查看全文   查看/发表评论  下载PDF阅读器