At present, the accuracy of domestic mid and long term power forecasting is hard to be further improved. Because of its own characteristics, traditional GM（1,1）no longer meets the new law of total electricity consumption in recent years. Based on the total electricity consumption data of various industries in provincial power companies, the changing law of electricity consumption is analyzed. Through mining various factors that influence the change of power consumption in each industry, multi variable gray models are established to forecast the whole society’s electricity consumption respectively in Jiangsu Province from 2016 to 2018,and the result is reliable. Finally, the characteristics of the model application are summarized.