Computer Engineering and Applications ›› 2007, Vol. 43 ›› Issue (32): 218-220.

• 工程与应用 • Previous Articles     Next Articles

Short-term load forecasting based on twi-time-series neural network model

LIN Hui1,HAO Zhi-feng2,CAI Rui-chu2   

  1. 1.Guangdong Power Grid Group Co.Ltd.Huizhou Power supply Branch,Huizhou,Guangdong 516000,China
    2.School of Compute Sciences and Engineering,South China University of Technology,Guangzhou 510640,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-11-11 Published:2007-11-11
  • Contact: LIN Hui

基于双时间序列神经网络的短期电网负荷预测

林 辉1,郝志峰2,蔡瑞初2   

  1. 1.广电集团公司 惠州供电分公司,广东 惠州 516000
    2.华南理工大学 计算机科学与工程学院,广州 510640
  • 通讯作者: 林 辉

Abstract: Short-term load forecasting is an important procedure for safe and economical running of the power-grid.The low forecasting accuracy and complicated structure are the main disadvantages of the existed method.After analyzing of the periodic variety of the short-term load,a novel twi-time-series neural network model is proposed.A new preprocessing procedure is also used to handle the missing value of the temperature data.The application of this model in a city of Guangdong province shows that it can obtain high forecasting accuracy both in normal and special days.

Key words: neural network, time series, short-term load-forecasting

摘要: 短期电网负荷预测是电网安全运行和经济调度的基础。现有预测方法存在对节假日预测不准确,不利于系统化等问题。根据短期负荷周期性变化的特点,创造性地提出双时间序列神经网络模型。同时为了克服实际温度数据缺失问题,提出一种新的温度量化方法。在广东省某地区的实际应用表明,该方法对于普通日和特殊日都取得了有较好的预测精度。

关键词: 神经网络, 时间序列, 短期负荷预测