Neural Network Model Predictive Control System for Fluid Catalytic Cracking Unit

The fluid catalytic cracking unit (FCCU) is one of the most important and complicated process in the petroleum refining industry. The catalysts performance and advanced control of the cracking catalytic plant contribute to increase the profit and gasoline production. One concept of the advanced control is represented by Neural Network Model Based Predictive Control System. The aim of this study is to develop and investigate the performance of the neural network model predictive control structure applied to the fluid cracking catalytic unit. Industrial data from a Romanian petroleum refinery was used to develop, train and validate two neural networks, for simulation and control the process. The neural networks as a process model are used to develop two neural network predictive controllers for the cracking catalytic process. In the final part will be outlined the performance that can be obtained using neural network model predictive algorithm for controlling a fluid cracking catalytic unit.