Predictive Calculation Technique for Electric Power Demand Based on Analysis of Semi-parametric Regression

Market economy highly requires more precise prediction of electric power demand.The advantage and disadvantage of prediction models previous used both at home and abroad were briefly introduced.In order to make the prediction more precise and feasible,a semi-parametric model for predicting the electric power demand composed of linear parameter and non-linear parameter was established.Linear parameter reflected the understandable regularity of load prediction while the non-linear parameter reflected the impact of uncertain factors of the load prediction.The semi-parametric model was estimated by the method of partial residual deviation.Estimation was obtained by two steps.The part of linear parameter was estimated first followed by the estimation of the part of non-linear parameter.The exemplified calculation showed that the estimation deviation by semi-parametric model regression is rather small,and even smaller than that obtained from binary linear regression.The result showed that this method is more precise,and easy-to-calculate with good adaptability.