Research of Annual Electricity Demand Forecasting Based on Kernel Partial Least Squares Regression

An important feature of the smart grid electricity is using prediction of high-precision power consumption for intelligent deployment, highly accurate forecasting of the power information is a key indicator of intelligent network. In this paper, we implement Gaussian kernel function to transform nonlinear regression of a low-dimensional to the linear regression in high-dimensional space. The power load forecasting model based on kernel partial least squares regression, can overcome the adverse effects of the nonlinear factors on the prediction model. Application of Jiangsu Province from 2006 to 2008 industrial electricity consumption data were verified, showing that the power load forecasting based on kernel partial least squares regression compared to linear partial least squares regression, has better prediction performance.