Predictive Control of A Fixed-Bed Reactor via Karhunen-Loeve Expansion and Neural Network

摘要

A model combined an artificial neural network and the K-L transformation is used to control a wall-cooled, fixed-bed reactor. The performance the KL-NN model based predictive control in setpoint tracking and disturbance rejection is investigated by simulation of a two-dimensional pseudo-homogeneous model. The results are compared with those of PID controllers. It is concluded that although the structure of the neural network and K-L transformation combined model is very simple, it can be successfully implemented in control of a system with distributed variables.