The current electricity consumption environment in China fluctuates nonlinearly and irregularly with the global economy. The medium and long-term electricity consumption prediction methods used in the past are difficult to meet the current precise planning requirements. Whether it is the traditional GM (1

1) or the MGM (1

n)

they are no longer adapted to the current new rules due to their natural attributes. Based on the changes in the total electricity consumption of different industries and the change trend of GDP in a provincial power grid

the logical relationship and change rules between the data are analyzed

and a background optimization function compatible with the current changes is constructed to improve the MGM to meet the current precise planning requirements for electricity consumption prediction. The total electricity consumption of the whole society in a province from 2010 to 2018 was predicted. The prediction results are highly reliable

and the prediction accuracy meets the current planning needs.