摘要
负荷时间序列的波动性和非线性特征的加剧对负荷预测方法提出了更高的要求,而常规组合预测方法针对海量负荷数据存在应用局限性问题。为此,提出了配电网负荷预测中时序分解方法和预测模型组合的双层优化策略。首先针对某一负荷预测数据,在时序信号分解层配置权重,以负荷均方根误差最小寻优各分解方法的权重系数,进而获得各时序信号分解方法的最优组合;在此基础上,在预测模型层进行组合方案寻优,通过配置权重系数以获得各预测模型的最优组合,进一步提升负荷预测的精度。仿真结果表明,所提策略可根据预测对象的特征优化组合各信号分解方法和预测模型,降低了配电网负荷序列的非平稳性对预测精度的影响。
The increasing volatility and nonlinear characteristics of load time series put forward higher requirements for load forecasting methods.The conventional combined forecasting method has application limitations for massive load data.Therefore,the two-layer optimization strategy is proposed for time series decomposition method and prediction model combination in load forecasting of distribution network.Firstly,for a certain load forecasting data,the weights are configured in the time series signal decomposition layer,optimized by minimizing the root mean square error of the load,to obtain the optimal combination scheme of time series signal decomposition method.On this basis,the proposed strategy optimizes the combination scheme at the prediction model layer,and obtains the optimal combination of each prediction model by configuring the weight coefficient to further improve the accuracy of load forecasting.The simulation results show that the proposed strategy can optimize the combination of signal decomposition methods and prediction models according to the load characteristics,and reduce the influence of non-stationarity load sequence on prediction accuracy in distribution network.
作者
张扬
ZHANG Yang(China Railway Siyuan Survey and Design Group Co.,Ltd.,Wuhan 430000,China)
出处
《智慧电力》
北大核心
2024年第9期104-111,共8页
Smart Power
基金
国家重点研发计划资助项目(2021YFB2600400)。
关键词
配电网
预测模型
时序信号分解
双层优化
组合预测
distribution network
prediction model
time series signal decomposition
two-layer optimization
combined forecasting
