In the high temperature technology widely used today, high demand for high temperature materials, and fused magnesia is a very good high temperature material. The operation of the fused magnesium furnace is very important, so we use statistic method to monitor the abnormal condition. First, utilized statistic method to handle the collected data of the preprocess and extract the information of the missing values; then find the distributions of latent variables. Second, use the EM method to establish a nonlinear regression model which can describe the relationship between the output and extracted latent variables. In this way, the proposed method is able to handle the missing data and select the relevant samples more accurately. Finally, the application to the fused magnesium process shows the advantages of the proposed method.