Temperature control of CST process using Gaussian neural network with adaptive learning rate

This paper deals with an indirect adaptive control scheme using the feedback linearization technique. In this method an online estimation of nonlinear functions using neural network was carried out. The estimated functions were then used to construct a control law to compute the control signal. The neural network with Gaussian potential functions in the hidden layer was used here. In order to overcome the difficulties associated with the optimal selection of the learning rate parameter, a new method for adaptive learning rate was developed. Further, to overcome the limitations of online identification and control in the neural network based adaptive control system, in addition to the control signal obtained from the estimated parameters, a linear control law was obtained. This proposed strategy was implemented and its performance for real time application was evaluated on the experimental setup of a continuously stirred tank (CST) process.