Abstract: Based on kernel ridge regression (KRR) inferential estimator,a novel inferential control strategy is proposed to implement on-line examination and control of aviation kerosene dry point in an atmospheric distillation column.Firstly,regression methods of the support vector machine (SVM) and the least squares support vector machine (LS-SVM) are analyzed,and a kernel ridge regression method is presented by solving the optimization problem directly.Secondly,the kernel ridge regression method is used to set up a esitmator model for predicting the dry point of aviation kerosene through the collected samples of the primary and secondary variables.Finally,simulation is made using the same samples,and the results show that the proposed modeling method needs less adjusting parameters and obtains a higher estimation accuracy than the RBF neural networks and the SVM regression method.