我国光伏发展已由“集中式强、分布式弱”的格局转变为集中式与分布式并举的趋势。在未来以可再生能源为主体的新型电力系统中，高比例的分布式光伏极有可能成为农村、牧区、山区以及城市社区、商业区的重要甚至主要电力来源。然而，大量分布式光伏的分散接入会引起配电网电压越限等问题，导致配电网控制需求增强、控制数量增加、控制难度增大。针对不同电压等级配电网的特点，本文将高比例分布式光伏接入的配电网按中压和低压进行分层研究。对低压配电网进行分区控制以快速响应电压越限风险并增强控制可靠性，对中压配电网进行集中优化以提高配电网运行的经济性和稳定性。 利用聚类思想对低压配电网进行分区。在定义电气距离以量化配电网节点间电气联系的基础上，提出了对电气距离矩阵进行特征值分析的分区数确定方法，并联合使用凝聚层次聚类和K均值算法将配电网的所有节点进行区域划分。 低压配电网以区域自治和区间协调为思想进行能量管理。基于低压配电网线路参数的特点，在分析有功调节和无功调节对于网损变化的影响之后，提出了一种以调节储能和光伏逆变器无功功率为主的电压控制策略，并根据天气情况设置光伏储能荷电状态的初值。仿真结果表明，该低压配电网能量管理模型可有效保证节点电压不越限，同时降低配电网线路上的功率损耗。 中压配电网将高比例分布式光伏接入的低压配电网视作广义负荷，将线路网损和削减光伏出力构成的电能损失最小与节点电压偏移最小设定为优化目标。中压配电网和低压配电网以不同调度时间间隔协调运行，低压配电网通过分区管理保证低压层各节点电压不越限，中压配电网通过集中优化减少电能损失和改善电压质量。最后通过对IEEE33节点算例进行仿真分析，验证该中压配电网能量管理模型能有效减少全天的电能损失，并改善各节点的电压偏移情况。

China's photovoltaic (PV) development has changed from the pattern of “strong in centralized PV and weak in distributed PV” to the trend of both centralized and distributed PV. As the country works toward embracing renewables as the main source of new power system, the high-proportion distributed PV is rising to become a key or main source of electricity for not only rural communities, pastoral and mountainous areas but also urban and commercial areas. However, the decentralized access of a large number of distributed photovoltaics will cause problems such as the voltage violation of the distribution network (DN), leading to an increase in need for control of DN, the number of the control variables, and the difficulty of the control. According to the characteristics of DN in different voltage levels, different control strategies are proposed between medium-voltage DN and low-voltage DN. The management of low-voltage DN is aimed to quickly respond to the risk of voltage and enhance the reliability of DN’s control with the partition strategy, while the management of medium -voltage DN is aimed to improve the economy and stability of DN’s operation with centralized optimization.Dividing low-voltage DN into several partitions by the clustering idea. On the base of defining the electrical distance to quantify the electrical connection between the nodes of DN, the optimal number of partitions is obtained by eigenvalue analysis of the electrical distance matrix, after which hierarchical clustering and K-means algorithm are performed on all nodes to get DN partition results.Low-voltage DN carries out energy management with the idea of regional autonomy and partition coordination. Based on the characteristics of the line parameters of the low-voltage DN, after analyzing the influence of active power regulation and reactive power regulation on the change of power loss, a voltage control strategy based on regulating energy storage and reactive power of PV inverter is proposed, and the initial value of the state of charge of PV energy storage is set according to the weather conditions. The simulation results show that the energy management model of low-voltage DN can effectively ensure that the node voltage keep in the limit and reduce the power loss on the line of DN.The low-voltage DN connected with a high proportion of distributed PVs is regarded as a generalized load of medium-voltage DN, and the minimum power loss caused by line network loss and reduction of photovoltaic output and the minimum node voltage offset are set as the optimization objectives. The medium-voltage DN and the low-voltage DN operate coordinately at different dispatch time intervals. The low-voltage DN ensures its each node keeping in voltage limit through partition management, while the medium-voltage DN reduces power loss and improves voltage quality through centralized optimization. Finally, the simulation results of the IEEE33 node example verifies that the energy management model of the medium-voltage DN can effectively reduce the power loss throughout the day and improve the voltage offset of each node.