The sintering temperature in rotary kiln is usually hard to be stable because of complex industrial environment. In this paper, we present a prediction model based on Bayesian classification algorithm to predict the trend of the amount of feed coal for controlling the sintering temperature in rotary kiln. To avoid the influence of attribute independence assumption of Bayesian classification algorithm, the FastICA algorithm used to find the independent components of the working condition data set in rotary kiln. Then we use AdaBoost algorithm to find a best classifier. The final simulation results show that the model has better control performance.