The mean of power consumption elasticity coefficient is an important indicator that reflects the relationship between power con sumption and economic development, and used for power demand forecasting. At present, the calculation of the mean in China mostly adopts the arithmetic mean method, which is more susceptible to extreme year data fluctuations,and cannot avoid significant errors. To ob tain a more accurate mean, a time series fitting regression method based on the double logarithmic model is proposed. which selects“Gross Domestic Product”and“Electricity Consumption of the Whole Society”as samples to establish time series. and performs regression fitting analysis on them under the framework of linear relationships. By conducting tests on the regression model and comparing it with the geo metric mean method example, it is statistically demonstrated that the mean error of the power consumption elasticity coefficient calculated based on the double logarithmic model is smaller and more reliable.