Monthly electricity sales contain a lot of information of both power supply and demand

and accurate forecast is helpful to the coordination of demand side management and grid operation. Firstly

the factors that drive the monthly fluctuation of electricity sales are analyzed from two aspects of necessity and contingency. It is determined that the prediction should be the superposition of inertial and non-inertial prediction. Secondly

the combination forecasting model is established

that is

the"logistic model"is defined as the inertial forecasting sub-model based on the annual growth of electricity sales

and the"AHP-Fuzzy Clustering Model"is defined as the non-inertial forecasting sub-model based on the similarity and repeatability of the data samples along the time axis. Finally

the case of a certain power supply company is tested.