Abstract: The data of the light,temperature and precipitation in the growing area of sugarcane(Saccharum officinarum L.) in Guangxi Autonomous Region was combined to the data with different period of the time by each ten days by using the expansion technique.The predicted factors with the correlative coefficient at 0.01 level of the significance were selected by using the method of correlative pervasive investigation.The regression model was established by the stepwise regression to predict the sugarcane yields.The prediction model of BP Neural Network was established by using the same factors of the stepwise regression at the same time.The precision of prediction and simulation results by the model of BP Neural Network was higher than that by the stepwise regression model.

Key words: BP Neural Network, BP Neural Network, Stepwise regression, Sugarcane(Saccharum officinarum L.), Yield prediction