基于两重门限GARCH模型的短期负荷预测

针对负荷时间序列的非线性和波动性特征,在研究负荷时间序列波动性门限特征的基础上,引入冲量门限的概念,提出了一种基于两重门限GARCH模型的短期负荷预测新方法.利用条件极大似然估计方法,估计了模型参数.同时,考虑到负荷时间序列波动的厚尾效应,将模型推广为服从非高斯分布假设下的情形,建立了2种基于厚尾假设的两重门限GARC...

Short term load forecasting based on double-threshold GARCH models

Considering the nonlinearity and volatility of load time series,threshold characteristics in load time series are analyzed.The concept of momentum threshold is employed and a novel double-threshold generalized auto-regressive conditional heteroskedasticity(DT-GARCH) model is proposed for short term load forecasting.By using the conditional maximum likelihood estimation(CMLE),the parameters are estimated.In addition,with fat-tail effect in volatility,the proposed models with non-Gaussian distributions are highlighted and estimated.Furthermore,the hybrid news impact surface is proposed to help analyze the impact of different shocks and momentums to the load time series.In case study,short term load forecasting is carried out based on the historical daily power consumption data of Nanjing,which validates the feasibility and effectiveness of the proposed model.Numerical results indicate that the DT-GARCH model with generalized error distribution provides satisfying forecasting results.