Implementation of Neural-Network-Based Inverse-Model Control Strategies on an Exothermic Reactor

In recent years there has been a significant increase in the number of control system techniques that are based on nonlinear concepts. One such method is the nonlinear inverse-model based control strategy. This method is however highly dependent on the availability of the inverse of the system model under control, which are normally difficult to obtain analytically for nonlinear systems. Since neural networks have the ability to model many nonlinear systems including their inverses, their use in this control scheme is highly promising. In this work, we investigate the use of these neural-network-based inverse model control strategy to control an exothermic reactor. The use of the specialised method of training the inverse neural network model is demonstrated. The utilization of two different inverse-model schemes namely the direct inverse control and the internal-model control methods are shown for both set point and disturbance rejection cases. The overall results for set point tracking are good in both control strategies but the direct inverse control method had limitations when dealing with disturbances. Other important aspects relating to the use of neural networks for identification and controls are also discussed in this paper.