Mid-long term load forecasting is of great significance for guiding the planning and construction of the power grid and improving the optimal allocation of power system resources. At present, the time scale of the medium and long term load forecasting is too large and the prediction precision is limited. To overcome this defect, this paper uses time decomposition technology to respectively analyze the long-term trend and short-term characteristics of power load, and then proposes a new prediction method, which will shorten the time scale of mid-long term load forecasting to hours and achieve mid-long term hourly load forecasting. The case study shows that the performance of the proposed model is better than the existing similar models. It has high global accuracy and stability. At the same time, it can effectively present the probability density characteristics and extreme value characteristics of power load. The proposed method is expected to provide reference for mid-long term planning of power grid. This work is supported byR & D Consulting Program of Economic and Technological Research Institute of State Grid Fujian Electric Power Company Limited:Establishment of Residential Electricity Model (No. SGFJJY00GHJS1700062) and Science and Technology Project of State Grid Corporation of China:Electric Energy Alternative Planning Model, Policy Simulation and Application Research (No. XM2017020034670).