Bilinear models-based short-term load rolling forecasting of smart grid

Aiming to strong nonlinear variations of smart grid, this paper presents a method of bilinear models-based short-term load rolling forecasting. Firstly, daily loads in adjacent days are defined to be input signals and daily loads at the same day in adjacent weeks are defined to be output signals, which result in the bilinear mathematic models of short-term load. Secondly, the fading memory recursive least square method is used to update the parameters in the model based on measurements of daily loads. Then, the updated measurements of daily loads are used to forecast short-term load of smart grid rolling. Finally, the competition load data of European Intelligent Technology Network (ENUNITE) are exploited to illustrate the effectiveness of the method proposed here.