Power load forecasting is the foundation of power system in full consideration of the characteristics and the natural, social conditions, using the effective method, based on historical data to determine the technology in the future when the load value. Short term power load forecast is the sun, week load. As an important basis for ensuring network security, economic operation, power short-term load forecasting accuracy is arranged on, the plan (including the units start and stop, hydro thermal coordination, tie line power, economic load distribution, reservoir operation and equipment maintenance etc.) the premise and the foundation, can improve the electric power enterprise economic, social benefits. According to the power load is difficult to predict accurately the problem, this paper introduces the interval type-2 fuzzy logic method to reduce the prediction error, presents an interval type-2 fuzzy logic model for the time series of one hour of power load forecasting, and adopted the first modeling process model structure, and then use back propagation algorithm to adjust the model parameters are determined by simulation. The results show that, the model has higher prediction accuracy and practical value to a certain extent, the performance is better than that of the corresponding type fuzzy logic model, proved more efficient processing idea uncertainty of type-2 fuzzy logic method.