华电技术 ›› 2020, Vol. 42 ›› Issue (1): 35-40.

• 研究与开发 • 上一篇    下一篇

基于最小二乘支持向量机的周用电量预测方法

  

  1. 1.贵州乌江水电开发有限责任公司,贵阳〓550002;2.华电电力科学研究院有限公司,杭州〓310030
  • 出版日期:2020-01-25 发布日期:2020-03-20

Prediction method for weekly electricity consumption  based on LSSVM algorithm

  1. 1.Guizhou Wujiang Hydropower Development Company Limited, Guiyang 550002, China; 2.Huadian Electric Power Research Institute Company Limited, Hangzhou 310030, China
  • Online:2020-01-25 Published:2020-03-20

摘要: 随着电力体制改革的不断深化、电力市场的蓬勃发展,发电企业为合理制定发电计划及市场竞价策略,对社会用电量预测提出了更精细化的需求。将最小二乘支持向量机(LSSVM)算法与短期用电量预测需求相结合,提出了一种周用电量预测方法,在充分考虑电量变化的周期性及延续性特点的基础上,将周气象特征指标纳入模型输入。实际算例测试表明,采用该周用电量预测模型实现了较高预测精度和较快计算速度,弥补了传统电量预测模型仅考虑历史电量影响,而无法更精确预测气象变化较大季节期间短期电量变化趋势的不足,满足电力市场背景下对周用电量进行精细化预测需求,具有较强实用性。

关键词: 电力市场, 发电企业, 负荷预测, LSSVM算法, 周气象特征, 周用电量

Abstract: With the deepening of the reform of electric power system and the vigorous development of electric power market, a more refined electricity consumption forecast is required by power generation enterprises in order to make a reasonable powergeneration plan and a market bidding strategy. A prediction method for weekly electric power consumption is proposed based on Least Squares Support Vector Machine (LSSVM) and shortterm electricity consumption forecast.Fully considering the periodicity and continuity of the electricity demand, weekly meteorological characteristics are added into the model input,which makes up for the shortcomings of traditional power forecasting models that only consider historical electricity quantity and cannot accurately predict shortterm power trends during great weather changes. The model proposed can meet the demand of a more refined electric quantity prediction in commercialized electric power market.It is of great practicability.

Key words: electricity market, power generation enterprises, load forecast, LSSVM algorithm, weekly meteorological characteristics, weekly electricity consumption