基于分段积温效应的夏季负荷组合预测方法

Combined summer load forecast with the break of accumulated temperature effect

分析了积温效应的2种表现形式:多日积温效应和两日积温效应,提出了考虑积温效应的夏季负荷组合预测方法。该方法充分考虑积温效应的2种表现形式,建立了3种温度的修正模型;为了提高预测精度,采用基于多元线性回归法、BP神经网络和支持向量机的组合预测方法。以江苏某地区的负荷数据作为历史数据,采用基于最小二乘法优化的模拟退火法求解最优参数对温度进行修正,并将修正之后的温度代入组合预测模型中预测负荷,结果表明,预测精度高,可以满足系统调度人员的需要。

The multi-day accunmlated effect and the two-day accumulated effect, the two forms of accumulated temperature effect is analyzed and the load forecast using combined forecast is proposed with accumulated temperature effect. This method is fully considered the two forms of accumulated temperature effects and the correction model of temperature is established. In order to improve prediction accuracy, combined forecast based on multiple linear regressions, neural network and support vector machine is adopted to forecast load. Finally, the load data of a region of Jiangsu is used as history data and simulated annealing based on least squares is adopted to obtain optimal parameters for temperatnre correction, and the correction of temperature is used into combined forecast to forecast load. The results show high prediction accuracy, meet the needs of the system operators.