The sales of electricity have great effect on operation decisions and budget arrangement of power grid companies directly, so the accuracy of electricity sale forecasting is quite crucial. Since the sales of electricity data has obvious increase trend with time and seasonal variation characteristics, single ARIMA model cannot get ideal forecasting results. According to the monthly electricity sales data of Z Province from 2011 January to 2014 October, this paper builds SARIMA model and get high forecast precision. In addition, since the sales of electricity is affected by temperature obviously and there are many temperature anomaly days in Z Province in 2014 compared with previous years, this paper processed the temperature correction and further improved the prediction accuracy.

No abstract