The traditional ARIMA time series analysis method is based on the linear technology to predict the time series, while its processing of nonlinear data is not reasonable with poor effect. There are many factors influencing the demand of power supply, and most of the material sequences usually contain both the linear time series and the nonlinear time series. In this paper, based on the ARIMA forecast, the BP neural network is combined with error correction to extract the composite features in the material sequence in order to improve the forecast precision of the electric power materials. The experimental results show that the accuracy of power supply forecasting with error correction can be improved significantly, which can provide important data support for material procurement plan.