半月刊

ISSN 1000-1026

CN 32-1180/TP

+高级检索 English
基于深度确定策略梯度算法的主动配电网协调优化
作者:
作者单位:

1.上海电力大学电气工程学院;2.国网上海市电力公司

作者简介:

通讯作者:

基金项目:

国家自然科学基金青年科学基金资助项目 (51607112)。


Coordinated Optimization of Active Distribution Network Based on Deep Deterministic Policy Gradient Algorithm
Author:
Affiliation:

1.Department of Electric Power Engineering, Shanghai University of Electric Power, Shanghai 200090, China;2.State Grid Shanghai Electric Power Company, Shanghai 200438, China

Fund Project:

This work is supported by National Natural Science Foundation of China (No. 51607112).

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    针对提升新一代人工智能在智能电网和能源互联网中的应用,实现高比例可再生能源及时有效接入电网,文中基于深度学习中的深度确定策略梯度(DDPG)算法进行主动配电网优的化运行。首先,构造了含多微网主动配电网优化模型的DDPG回报函数,使配电网的节点电压总偏差和线损最小,最大限度上降低微电网功率调节量的变化以减小对微电网运行的影响,同时维持联络线功率平衡以减小对配电网的影响。接着,分析了主动配电网优化控制的DDPG样本数据处理、回报函数设计、模型训练和学习过程。最终通过改进IEEE 14节点算例仿真验证了DDPG算法的有效性。

    Abstract:

    In order to promote the application of the new generation of artificial intelligence in smart grid and Energy Internet, and achieve high-penetration renewable energy access to the power grid in a timely and effective manner, the deep deterministic policy gradient (DDPG) algorithm is applied in the optimization operation of active distribution network (ADN) in this paper. Firstly, DDPG return function of the multi-microgrid active distribution network optimization model is constructed. We can minimize the total node voltage deviation and line loss of ADN. We also can minimize the variation of the power regulation of microgrid(MG) to reduce the impact on the operation of MGs, and maintain tie-line power blance to reduce the impact on the distribution network. Secondly, DDPG sample data processing, design of return function, model training and learning process of ADN optimization control are analyzed. Finally, the effectiveness of the DDPG algorithm is verified by the improved IEEE 14-node example simulation.

    表 1 Table 1
    图1 主动配电网的系统优化结构Fig.1 System optimized structure of active distribution network
    图2 主动配电网的DDPG优化学习过程Fig.2 DDPG optimal learning process of active distribution network
    图3 DDPG算法训练过程中的回报值Fig.3 Return value in process of DDPG algorithm training
    图4 DDPG算法训练过程中的损失函数值Fig.4 Loss function value in process of DDPG algorithm training
    图5 S1至S4的优化目标值Fig.5 Optimization target values of S1 to S4
    图6 S1至S4模式下的平均计算时间Fig.6 Average calculation times with Mode S1 to Mode S4
    图 14节点系统图Fig. System Structure of 14-bus distribution network
    图 光伏和风电发电量的百分比Fig. Percentage of photovoltaic and wind power generation
    图 4个微网的等效输出负荷Fig. Equivalent output load of four microgrids
    图 配电网的系统总负荷Fig. System total load of distribution network
    图 4种模式的配电网成本Fig. 4 modes of distribution network costs
    图 4种模式的配电网联络线功率Fig. Tie-line power of distribution network in 4 modes
    图 4种模式的节点1电压Fig. Voltage of bus 1 in 4 modes
    图 4种模式的节点5电压Fig. Voltage of bus 5 in 4 modes
    图 4种模式的配电网网损Fig. Power loss of distribution network in 4 modes
    图 4种模式的配电网等效负荷Fig. Equivalent load of distribution network in 4 modes
    参考文献
    相似文献
    引证文献
引用本文

龚锦霞,刘艳敏.基于深度确定策略梯度算法的主动配电网协调优化[J].电力系统自动化. DOI:10.7500/AEPS20190321010.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2019-03-21
  • 最后修改日期:2019-11-13
  • 录用日期:2019-10-08
  • 在线发布日期:
  • 出版日期: