A hybrid model for predicting product sulfur concentration of diesel hydrogen desulfurization process

摘要

In this paper, a hybrid sulfur predictor is developed for the hydrodesulfurization unit and validated using lab-scale reactor and industrial data. The proposed sulfur predictor is configured to estimate product sulfur concentration under unknown feed composition changes. The hybrid structure has an offline feed sulfur estimator based on a mechanistic model for the hydrogen desulfurization reactor. The online predictor is based on support vector regression. The developed soft sensor is validated against an industrial data set. A Matlab-based graphical user interface (GUI) is developed for easy deployment of the developed hybrid online sulfur predictor.