GMDH-Based Monitoring in an Atmospheric Distillation Process

In atmospheric distillation processes, the stabilization of processes is required in order to optimize the crude-oil composition corresponding to the product market conditions. However, the process control systems have sometimes fallen into unstable states in the case where unexpected disturbances are put into them, and the unusual phenomena have given undesirable influence on products. On the other hand, the useful chemical engineering model has not been established for these phenomena yet. This is still painful problem in the atmospheric distillation process now. This paper describes a new modeling scheme to predict the unusual phenomena in the atmospheric distillation process using the GMDH (group method of data handling) network which is one of network models. According to the GMDH network, the model structure can be determined systematically. However, the least squares method has been usually utilized in determining weight coefficients (model parameters). Then, the estimation accuracy is not expected so much, because the sum of squared errors between the measured values and estimates is evaluated. Therefore, instead of evaluating the sum of squared errors, the sum of absolute value of errors is introduced and the Levenberg-Marquardt method is employed in order to determine model parameters. The effectiveness of the proposed method is evaluated on the foaming prediction in the crude oil switching operation in the atmospheric distillation process