The use of a partially simulated exothermic reactor to test nonlinear algorithms

Two nonlinear control algorithms for controlling nonlinear systems include the receding horizon method and the nonlinear neural network inverse model methods. These methods have been found to be useful in dealing with difficu lt-to-control nonlinear systems, especially in simulated systems. However a lthough much simulation work has been performed with these methods, simulat ion only is inadequate to guarantee that these algorithms could be successf ully implemented in real plants. For this reason, a relatively low cost and simple online experimental configuration of a partially simulated continuo us reactor has been devised which allows for the realistic testing of a wid e range of nonlinear estimation and control techniques i.e. receding horizo n control and neural network inverse model control methods. The results sho w that these methods are viable and attractive nonlinear methods for real-t ime application in chemical reactor systems.