Abstract:Based on the meteorological observation data including daily mean temperature, the maximum temperature, the minimum temperature, relative humidity, sunshine hours and the maximum wind speed of solar greenhouse in Juye during winter of 2012-2013,the main meteorological factors affecting the minimum temperature of solar greenhouse were determined by using correlation analysis, the forecast model of the minimum temperature was established by using stepwise regression analysis and principal component analysis, and the difference of the two forecast models was tested with the winter meteorological data during 2013-2014. Results show:(1) a significant correlation existed among the minimum temperature, temperature, relative humidity, sunshine of the day and the last day inside and outside the solar greenhouse.(2)All forecast models have all passed the significance and verification test(p<0.01). The correlation coefficient, average absolute error, the average relative error of forecast models using stepwise regression and principal component analysis were 0.82 and 0.58, 0.9℃ and 1.8℃, 9% and 15%, respectively, which indicated that the higher forecast accuracy of stepwise regression makes it meet the business requirement for forecast the minimum temperature of solar greenhouse in winter.