Adaptive observer and nonlinear control strategy for chemical reactors via neural networks

An adaptive observer and nonlinear feedback control strategy with constraints on control action are developed by using a supervised learning rule of a neural network and the theory of functional-link networks. The convergence of the adaptive observer and the stability of the control system are proven. They are applied to the control of an exothermic stirred-tank reactor. It is shown that an adaptive observer for concentration can be constructed for a reaction system when only temperature measurements are available on line. An adaptive observer is used to identify the pre-exponential Arrhenius constant and to provide on line estimation of the unmeasured reactant concentration for a nonlinear state-feedback controller. Simulations show that the combined observer/controller provides satisfactory closed-loop behaviour, fast responses and strong robustness. Estimated and actual concentration are in good agreement. A nonlinear feedback controller can provide effective feedback control over a wide range of operating conditions.