考虑需求响应的电力系统短期负荷预测模型

Short-Term Load Forecasting Model Considering Demand Response of the Power System

针对常规预测方法难以准确预测负荷曲线产生的相应变化,本文建立了考虑需求响应的电力系统短期负荷预测模型。根据系统调度员(distribution system operators,DSOs)接收的需求响应信号,确定用户的实际需求响应,并以此作为建模的依据,构造出考虑需求响应的负荷时间序列,建立计及需求响应的径向基函数神经网络(radial basis function-neural networks,RBF-NN)预测模型,并通过实际负荷算例进行仿真分析。分析结果表明,若在RBF-NN预测模型中计及需求响应因素,平均绝对误差为4.439%;若不计及需求响应因素,平均绝对误差为12.784%;在预测模型中融入需求响应因素,可使平均绝对误差降低8.345%,预测准确度较高。因此,电力系统短期负荷预测模型中融入需求响应因素,能够达到更高的准确度。该研究具有较好的理论价值和实际应用价值。

On account of the fact that regular forecasting methods can hardly predict the changes of the load curve very precisely,this paper constructs a short-term load forecasting model of the power system,which takes demand response(DR)into account.According to the signals of demand response confirmed by distribution system operators,this paper seeks to figure out the users' actual demand response which serves as the ground for the model building,thus developing the corresponding load time series.It establishes the forecasting model considering demand response on the basis of Radial Basis Function-Neural Networks(RBF-NN)and conducts a stimulation analysis with an actual case.The analysis result exhibits that,if the demand response is taken into account in RBF-NN modeling,the mean absolute percentage error will be 4.439%,and otherwise,it will be 12.784%.The mean absolute percentage error can be decreased by 8.345%through integrating demand response into the short-term load forecasting model of power system and higher forecasting precision can be achieved.This work enjoys favorable theoretical value and practical significance.