Short-term load forecasting based on higher order partial least squares (HOPLS)

Abstract

The load forecasting plays a more and more important role in the operation of the power system and the demand side. However, artificial intelligence techniques are complex in the short-term forecasting. For this purpose, this paper proposes the load forecasting model based on higher order partial least squares, which is much simpler than artificial intelligence techniques in complexity. Considering the nonlinear relationship between dependent variables and independent variables, an extended input tensor is employed in the load forecasting model. Finally, load data of year 2015 in the ISO New England is used to verify the rationality and feasibility of proposed method. Simulation results of four days that belong to four seasons separately have shown that the proposed model is very suitable for short-term load forecasting.