含高光伏渗透配电网最优潮流的分散式鞍点动态求解

Decentralized Saddle-point Dynamics Solution for Optimal Power Flow of Distribution Networks With High Photovoltaic Penetration

在高比例光伏接入配电网的情况下,研究其最优潮流问题愈加重要。基于鞍点动态法,提出一种解决高光伏渗透率配电网最优潮流问题的分散式优化方法。该算法从动态系统控制的角度出发,将凸优化问题寻优的过程转化为动态系统渐进稳定的过程。首先,将潮流方程进行线性化,建立包含网络损耗和弃光惩罚的二次规划模型,保证算法最终稳定在全局最优解。然后,将电力网络和各个光伏逆变器分别视为独立可控的单元,将集中式最优潮流模型转化为可分散式求解的模型。网络和各个光伏逆变器之间仅需要交换节点注入功率信息,各个光伏逆变器之间相互独立,保密性好,各单元的优化求解可同时进行,并具有即插即用的特点。最后,通过仿真计算验证潮流方程线性化模型的准确性和所提算法的有效性。

With high penetration of photovoltaic generation in distribution networks, the research on optimal power flow(OPF) becomes much more significant. Based on distributed saddle-point dynamics(DSPD) approach, a decentralized method for solving the OPF problem for distribution networks with high photovoltaic(PV) penetration was developed. From the viewpoint of the dynamic system control, this novel approach transforms the convex optimization problem into the asymptotically stable of dynamic systems. First, by leveraging a linear approximation of power flow equations, a quadratic programming(QP) model with the objective of minimizing real-power losses and the punishment for curtailed active power were formulate, so that a global optimal solution can be guaranteed. Then a decentralized model was obtained by regarding the electric network and photovoltaic inverters as independently controllable units. The decentralized approach can run in parallel pattern as the electric network and photovoltaic inverters just need to exchange nodal injection power, while keep completely independent and private among photovoltaic inverters. Furthermore, the proposed approach possesses the plug-and-play feature. Numerical simulations verify the accuracy of the linearized power flow equations, and demonstrate the effectiveness of the proposed approach.