Based on Zhundong’s power load information and the forecast products of T213 from April to September in 2005-2006, the effective timeserial data of daily average are established. Correlations of power load and T213 main forecast factors are analyzed. 700 hPa water vaporflux, 850 hPa water vaporflux and precipitation are selected as those important forecast factors, and forecast equation of daily average power load variation is obtained. Testing forecast to year 2007, power load change tendency is consistent with reality, but there are some forecast errors sometimes when load makes fluctuate evidently. The reason is that T213 makes mistakes for forecasting rain weather. An improving scheme is combining T213 model with weather forecaster experience. The forecast equation is rebuilt by using T213 important forecast factors and new precipitation factors which are selected from weather observation elements. The forecast result is more accurate. After analyzing forecast errors from April to September in 2007, the conclusion is that the planed power load’s average error is 11.5%, T213 model’s is 8.2%, and the optimizing model’s is 6.4%. In rain days, the planed power load’s error is 22.3%, T213 model’s is 13.4%, and the optimizing model’s error is reduced to 8.9%.