Abstract: The access of a large amount of renewable energypromotes the fluctuation and diversity of the energy Internet and puts forward the higher requirements for the accuracy and stability of the cooling and heating power load forecasting, which is an important prerequisite for the operation optimization of the energy Internet and of great significance for the demand side analysis. Based on the weather information, a combined forecasting method of short-term cooling and heating power load is proposed. This method includes two steps: the regional weather forecast and the combined forecast of cooling and heating power load. In the regional weather forecasting, first of all, this method makes full use of the historical weather information, the measured weather situations and the weather forecasts to forecast the weather in a designated area with the adjustment error method. Then this method uses the historical load data, the historical weather data and the weather forecast data, and adopts the Genetic Algorithm to optimize BP Neural Network (GA-BP) prediction algorithm to jointly predict the cooling and heating power load. The simulation results show that this method can effectively improve the accuracy of the cooling and heating power load prediction.