Considering that there are many factors having highly nonlinear effect on the Murphree efficiency of column in petrochemical industry and that the obtained accurate value of the Murphree efficiency plays an important role in the performance of the column model, a novel hybrid model method which combines mechanism with least square support vector machine (LS-SVM) is proposed to model solvent dehydratic distillation column (SDDC). In the hybrid model, the mechanism method is used to develop the whole model of SDDC, and LS-SVM is employed to describe the relationship between the factors and the Murphree efficiency due to the nonlinear representing ability of LS-SVM. Further, the correlation between the factors and the Murphree efficiency is analyzed to select the input variables of LS-SVM, and leave-one-out cross-validation method is employed to demonstrate the performance of LS-SVM. The simulation result shows that the hybrid model has robust and good predicting performance and can simulate different conditions of SDDC.
