With the widespread application of intelligent energy meters, the demand for energy meters is becoming increasingly large. Using empirical manual estimation can easily lead to temporary shortages and inventory backlog of energy meters. Based on the historical installation data of electricity meters in the metering system, simple seasonal model, Winters addition model, and Winters multiplication model are used. And LSTM model is combined to compare and analyze the results, compare the advantages and disadvantages of different models, and determine the optimal method for predicting electricity meter demand. The empirical results indicate that the Winters multiplication model has the best prediction effect, with an average error of 0.96% in the predicted values, and the predicted trend is in line with the actual situation. Winters multiplication index smoothing method can scientifically and reasonably predict the trend of electricity meter demand changes, assist power supply companies in more efficient operation and management of electricity meters, and improve the efficiency of measuring asset utilization.