The spatial and temporal distribution characteristics of electric power accidents and their relationship with weather factors such as thunderstorm, gale, and daily mean temperature are analyzed by using the weather data of thunderstorm and gale and historical data of electric power accidents. By using the method of REEP (regression estimation of event probability) and Logistic regression, an early warning model of the relationship between the probability of power accident and thunderstorm, gale, and daily mean temperature in Rizhao is obtained. The results are listed as follows. 1) Thunderstorms and strong winds are important meteorological factors contributing to the power accidents in Rizhao. 2) Although thunderstorms, strong winds, and high temperature have impacts on the occurrence of electric power accidents, there is a big gap between their influence ability. 3) The two regression models both have good fitting effects. The REEP model is more intuitive, and the Logistic regression analysis method is more objective and more applicable. 4) The results of the regression analysis are based on objective data and the regression model is accurate and practical, which can provide theoretical and technical support for the power accident early to warning system.