Empirical Modelling and Advanced Control of an Industrial Distillation Column

摘要

This article presents some aspects of implementing empirical modelling and advanced control strategies of an industrial propylene/propane splitter, including a softsensing model of the distillate composition based onNeural-Network-Based-Fuzzy-System (NNBFS) and Principal Component Analyisis (PCA) from the selected secondary measurements, some advanced regulatory schemes and a multivariable predictive control scheme of top product quality and bottom product quality,etc. Industrial application results show that the empirical model and advanced control can increase the splitter tower stability, improve the distillate product quality and increase recovery of distillate product while reducing energy consumption.