Abstract: Power load forecast is essential in government management on electric power operation. With the rapid development of “Internet+” smart energy, it is a trend to substitute artificial load forecast into automatic forecast based on information system. In this paper, daily highest temperature is chosen as weather characteristic in prediction model. Considering temperature accumulation effect, revision of highest temperature on prediction day is presented, and correlation of power load and temperature is analyzed. In the end, a prediction method of highest power load of next work day is given based on decision tree model.