Abstract: Middle and longterm load forecasting is the foundation of distribution network planning, precise load forecasting can improve the reliability and economy of distribution network planning. In this paper, the annual electricity demand forecasting is used as research object, and forecasting value of annual electricity demand is obtained by annual, quarterly, monthly and industry electricity consumption, based on which a linear combination model for load forecasting is established by its developing sequence forecasting. Then the Theil coefficient and IOWHA operator are used to evaluate the weight coefficients. The model can overcome such disadvantages as constant weighted average deficiencies and single error index for measuring forecasting precision using traditional combination forecasting method. The example shows that the new model can improve forecasting precision effectively, reduce the risk of forecasting, and possess better practicability.