Abstract:Based on power load information and weather data from 2005 to 2006, it is conclu ded that the annual and seasonal variations of power load in agricultural region correspond statistically to the cycle of agricultural activity. The time of sec ond peak power load of daily variation is consistent with the peak of daily ma ximum temperature and minimum air humidity. Relationship is analyzed not only be tween daily average power load and atmospheric pressure, temperature, humidity, wind, precipitation, but variation of power load and meteorological element as w ell. Rainy day and clear days are counted in sections. The results show that var iations of precipitation, air humidity, temperature have better correlation with the power load. Four different models are established to simulate fluctuation o f power load by using stepwise regression equation. The better simulated result is element-difference model after error analysis, so the forecast of power load fluctuation is tested and error is analyzed in May of 2007. The result is t hat?in rainy day the error in?common power load forecast method error is 63.2% , but the forecast error in the element-difference model is 18.7%. In addition t he monthly average forecast error of power load for the common power load foreca st model and the element-difference model are 26.0% and 13.4%, respectively.