Comparitive performance analysis of various training algorithms for control of CSTR process using narma-L2 control

In recent years, there has been an expansive growth in the study and implementation of neural networks over a spectrum of research domains. The NARMA model is an exact representation of the input-output behaviour of finite dimensional non-linear discrete time dynamical systems in the neighborhood of the equilibrium state. To implement neural network based NARMA-L2 control, first step is modeling of the process for system identification and the second step is the controller design. Neural network based NARMA-L2 controller is implemented for a CSTR process using Levenberg-Marquardt algorithm, Scaled Conjugate Gradient algorithm and their performance are compared.