ABSTRACT

In this work the control algorithm developed by Bandyopadhyay et al. [4] is extended when no on-line continuous concentration measurements are available. The dynamic evolution of a state variable, the temperature, is reconstructed with a neural network model, which allow us to calculate the feedback controller gain without the use of the Poincare` map. The method is applied to the case of two irreversible reactions occurring in a nonisorthermal CSTR, which exhibits a chaotic behavior. The performance of the developed control algorithm is tested either in a regularization problem, period-1 and period-2 cycle, and regulating problem.