I. Introduction
Operation optimization problems in industrial production have become increasingly important in the face of increasing global competition and environmental requirements, and they aim to determine the best values of the variables (e.g., temperature and material flow rate) that control the operating condition of real plant equipments based on production constraints [1]. Distillation unit (DU), a common separating unit used in refineries [2], separates feedstock into a series of intermediate products (hereafter called “products”). In the DU, the feedstock properties and the setting values of the operating variables (e.g., steam flow) control the tray temperatures that influence distillate yields and profitability of oil refineries. Optimizing the DU operating variables with consideration of the feedstock properties and production constraints is thus a challenging research problem of considerable importance for decreasing the production costs and improving the production benefit.