Abstract：A certain area of Central China Power Grid is taken as the research object to analyze the relationship between the electric load and meteorological elements. A rolling prediction model is established by stepwise regression and BP neural network by electric load and refined meteorological data. We quantitatively analyze the main contribution of meteorological factors to electric power load forecasting. It is found that the daily load has a good correlation with the historical load， and the daily temperature and the previous day temperature also have a greater impact on the load. Meteorological factors have positive contributions of load prediction accuracy by stepwise regression and neural network， with contribution rates of 0.28% to 17.87% and 0.97% to 17.78%， respectively. Especially in the case of turning weather conditions， the improvement of the accuracy of short-term load forecasting by the refined meteorological factors is particularly important.
2020, Vol. 54 
