ANN based IMC scheme for CSTR

Abstract

This paper demonstrates that neural networks can be used effectively for control of nonlinear dynamical systems. The proposed control scheme is based on the artificial neural network and is applied to an isothermic continuous stirred tank reactor (CSTR). In this paper we have tested the internal model control (IMC) strategy based on neural networks for process systems. This approach of control uses two Feed Forward Neural networks (FFNN), one as an identifier and the other as a controller. Multilayer neural network has been used for forward modeling and the inverse model of the process which has been determined off line using input output data of process, as controller. The modified back propagation algorithm has been used to train the neural networks. Neural network based IMC scheme has been implemented for both set point and regulatory control action and the comparison have been made for a set of constant momentum term.

Artificial neural network based modeling and controlling of distillation column system

“…There are several results of research concerning the Neural Network -Internal Model Control (NN-IMC), which are implemented on different process (Rivals et al, 2000, Changjie et al, 2005, Varshney et al, 2009). These papers show that the advantages of Internal Model control systems are their robustness with respect to a model mismatch and to disturbances.…”