中国电力 ›› 2016, Vol. 49 ›› Issue (2): 67-72.DOI: 10.11930/j.issn.1004-9649.2016.02.067.06

• 电网 • 上一篇    下一篇

南京城市电网核心区短期负荷特性分析及预测

朱斌1,姜宁2,霍雪松1,王勇2,吴海伟1,孙凯祺3,胡爽3   

  1. 1.国网江苏省电力公司,江苏 南京 210024;
    2.国网南京供电公司,江苏 南京 210019;
    3. 山东大学 电气工程学院,山东 济南 250061
  • 收稿日期:2015-09-12 出版日期:2016-02-18 发布日期:2016-03-21
  • 作者简介:朱斌(1965-),男,江苏镇江人,高级工程师,从事调度运行与管理等研究。E-mail: zhubin@js.sgcc.com.cn
  • 基金资助:
    国家电网公司科技项目“城市核心区输电线路输送能力提升关键技术研究”(SGJS0000DKJS1500985)

Forecasting and Studies on Load Characteristics of Nanjing Center Area Power Network

ZHU Bin1, JIANG Ning2, HUO Xuesong1, WANG Yong2, WU Haiwei1, SUN Kaiqi3, HU Shuang3   

  1. 1. State Grid Jiangsu Electric Power Company, Nanjing 210024, China;
    2. State Grid Nanjing Power Supply Company,Nanjing 210019, China;
    3. School of Electrical Engineering, Shandong University, Jinan 250061, China
  • Received:2015-09-12 Online:2016-02-18 Published:2016-03-21

摘要: 城市电网核心区负荷变化复杂,影响因素众多,对电网调度部门的安全运行提出了挑战。基于南京电网实际负荷数据,分析了负荷变化特性及各类影响负荷变化的因素,同时针对预测方法中存在的边缘效应等问题,通过改进训练策略,提出了一种新的人工神经网络短期负荷预测模型。该模型采用多隐含层和动态神经元个数的预测方法,对不同神经元预测结果进行比较,以达到预测负荷的目的。预测结果表明,基于该方法建立的预测模型适用性强且能获得较高的预测精度,可为城市核心区的短期负荷预测提供可行方案。

关键词: 电网, 负荷特性, 短期负荷预测, Elman人工神经网络, 训练策略

Abstract: Load characteristics and various contribution factors in city distribution network are analyzed. In order to overcome edge effect problems in Elman neural network of load forecasting method, a new short-term forecasting model is proposed by training strategy improvement. This model adopts multiple-hidden-layer networks and dynamic neural networks element as forecasting method, generating results by comparing different forecasting results of neural networks element. The testing results prove the adaptability and accuracy of proposed method under different conditions. It provides a feasible alternative for short-term forecasting of city central area power network.

Key words: load characteristics, short-term forecasting, Elman neural network, training strategy

中图分类号: