• CSCD核心库收录期刊
  • 中文核心期刊
  • 中国科技核心期刊

电力建设 ›› 2022, Vol. 43 ›› Issue (2): 81-88.doi: 10.12204/j.issn.1000-7229.2022.02.010

• 智能电网 • 上一篇    下一篇

基于行业聚类电量曲线分解的中期负荷预测

钟士元1, 张文锦2(), 罗路平1, 王伟1, 肖异瑶2, 廖志伟2   

  1. 1.国网江西省电力有限公司经济技术研究院,南昌市 330096
    2.华南理工大学电力学院,广州市 510641
  • 收稿日期:2021-07-09 出版日期:2022-02-01 发布日期:2022-03-24
  • 通讯作者: 张文锦 E-mail:1023775294@qq.com
  • 作者简介:钟士元(1978),男,高级工程师,主要研究方向为电力系统规划;
    罗路平(1984),男,高级工程师,主要研究方向为电力系统规划;
    王伟(1989),男,高级工程师,主要研究方向为电力系统规划;
    肖异瑶(1997),女,硕士研究生,主要研究方向为电力系统运营及控制;
    廖志伟(1973),男,博士,副教授,主要研究方向为电力系统运营与控制技术、人工智能及大数据应用。
  • 基金资助:
    国家自然科学基金重点项目(51437006)

Medium-Term Load Forecasting Based on Industry Clustering Electricity Curve Decomposition

ZHONG Shiyuan1, ZHANG Wenjin2(), LUO Luping1, WANG Wei1, XIAO Yiyao2, LIAO Zhiwei2   

  1. 1. Institute of Economy and Technology, State Grid Jiangxi Electric Power Co., Ltd., Nanchang 330096, China
    2. School of Electric Power Engineering, South China University of Technology, Guangzhou 510641, China
  • Received:2021-07-09 Online:2022-02-01 Published:2022-03-24
  • Contact: ZHANG Wenjin E-mail:1023775294@qq.com
  • Supported by:
    National Natural Science Foundation of China(51437006)

摘要:

传统电量序列分解方法难以有效结合地区行业发展趋势分析,为此文章提出一种基于行业发展趋势的行业聚类电量曲线分解中期负荷预测模型。首先,采用动态时间规整算法计算行业电量周期性,从而分类发展趋势有无变化的行业;其次,通过k-means算法按照用电特性相似聚类预分类行业,并通过季节分解算法分解聚类行业电量序列;最后,针对各电量子序列建立支持向量回归模型,并以江西省某市电量数据作算例分析。算例分析结果表明,文章方法可以分离不同用电特性的行业电量,有助于分析当地行业经济发展状况,并提高地区中期负荷预测准确性。

关键词: 中期负荷预测, 行业分类, 动态时间规整算法, 分解预测

Abstract:

The traditional decomposition method of power consumption sequence is difficult to effectively combine with the analysis of regional industry development trend. Therefore, this paper proposes a medium-term load forecasting model based on the decomposition of the clustering industry electricity curve which combines with the industry development trend. Firstly, the dynamic time warping algorithm is used to calculate the periodicity of the industrial power consumption, to classify the industry which has changed development trend. Secondly, the k-means algorithm is used to cluster pre-classified industries according to similar electricity consumption characteristics, and the seasonal decomposition algorithm is used to decompose the power consumption sequence of the clustering industries. Finally, the support vector regression model is established for each power consumption sub-sequence, and the electricity data of a city in Jiangxi province is taken as an example. The results show that the proposed method can separate the industry power consumption with different electricity consumption characteristics, help to analyze the local industry economic development, and improve the accuracy of regional medium-term load forecasting.

Key words: medium-term load forecasting, industry classification, dynamic time warping, decomposition forecast

中图分类号: