Holiday Load Power Forecasting Based on SOIF-PAMAM method

Abstract

Xunyang Wang, Dingmei Wang, Zhicheng Ma and Ruixiao Zhang

Focusing on the particularity of holiday load, in this paper, a periodic autoregressive moving average model (PAMAM) algorithm based on selecting optimal input features (SOIF) is proposed to predict the short-term holiday power load. In short-term load forecasting models, there are few researches on feature selection (FS). However, as more and more intelligent hybrid models are used in real-time load forecasting, FS has become a key factor affecting the forecasting accuracy. Based on the idea of SOIF, PAMAM model is proposed to improve the influence of FS factors, and the holiday equations are combined into periodic autoregressive moving average model, so as to improve the short-term forecasting. In order to simplify the calculation, in this paper, the probability distribution is used to calculate the FS, and the autoregressive spline algorithm is used to establish the nonlinear solar radiation and temperature effect model. Based on the statistics of solar radiation intensity, temperature and other data during the Spring Festival, in this paper we analyze the influence of the above factors on the short-term power load forecasting during holidays. Experimental results show that SOIF-PAMAM algorithm in which temperature and other weather conditions are considered can significantly improve the prediction accuracy, the average absolute error is 2.45%, and the root mean square error is 2.61%.

HTML PDF