摘要
针对传统预测模型在样本不足的情形下,无法实现高精度月度供电量预测的问题,提出了一种基于改进的生成对抗网络数据增强方法,能够将大粒度的月度统计信息,同分布分解为按天统计的供电量信息,实现了数据增强与样本集的有效扩充。基于该样本集,运用深度残差网络构建用于月度供电量预测的深层模型。算例分析标明,所提出的方法能够在原有同类型预测方法的基础上,有效提升预测精度。
In order to solve the problem that the traditional forecasting model can not achieve high-precision power supply forecasting in the case of insufficient samples,an improved generation countermeasure network data enhancement method is proposed,which can decompose the monthly statistical information with large granularity into the daily statistical power supply information with the same distribution,and realize the effective expansion of data enhancement and sample set. Based on the sample set,a deep-seated model for monthly power supply forecasting is constructed by using the deep residual network. An example shows that the proposed method can effectively improve the prediction accuracy on the basis of the original prediction methods of the same type.
作者
尹力
周琪
YIN Li;ZHOU Qi(State Grid Wuhan Power Supply Company,Wuhan 430070)
出处
《计算机与数字工程》
2022年第2期448-452,共5页
Computer & Digital Engineering
关键词
小样本数据
供电量预测
生成对抗网络
数据增强
样本集扩充
深度残差网络
small sample data
power supply prediction
generation countermeasure network
data enhancement
sample set expansion
deep residual network
