Abstract: The line loss rate is an important indicator to measure the economics of the power grid.In order to realize the time series prediction of the line loss rate of the power grid, based on the autoregressive integrated moving average（ARIMA） model, combined with the support vector regression（SVR） model that has excellent nonlinear data processing effects, we propose a combined regression model of line loss rate based on the time series. The SVR model optimized by the gray wolf optimier（GWO） algorithm is used to reconstruct the residual data of the ARIMA prediction and re-model the prediction, and finally combine the prediction results of the two models to obtain the final prediction result.After an example analysis, the mean square error and the mean absolute error are all better than the single ARIMA model and the commonly used ARIMA-SVR model, which has excellent prediction effects.