Abstract: Using the daily electrical load and meteorological data of Huaibei in Anhui province from 2012 to 2016, statistical methods such as correlation analysis, regression analysis, and curve fitting were used to analyze the seasonal variation and weekend/holiday effect of electrical load.Main meteorological influence factors were extracted.The effect of temperature (1 ℃) on electrical load and the sensitivity of electrical load to maximum temperature were also analyzed.The trend load and trend equation were determined in this study.The methods of applying weekend/holiday effects to different forecasting models and scientific methods of extracting meteorological load were introduced.The multivariate regression equation and curve-fitting equation of daily electrical load forecasting were established using the trend method.Considering the weakness of the trend method, a 2-day increment method was proposed.The corresponding forecasting model was established.Among them, the historical fitting rate of the 2-day increment forecasting model and the accuracy of the trial forecasting in 2017 both reach 96%-97% which is 2%-3% higher than those with the trend method, and 4%-5% higher than the current assessment requirements.In conclusion, a 2-day increment method improves the accuracy of electrical load forecasting.