Implementation of neural network for internal model control and adaptive control

In this paper a nonlinear internal model control (NIMC) strategy based on neural network model is proposed. The NN model is identified from input output data using three layers feed forward network trained with back propagation algorithm. The NIMC controller consists of a model inverse controller and robustness filter with a single tuning parameter. Also a scheme has been developed using the feedback linearization on technique for indirect adaptive control. In this method online estimation of nonlinear function using neural network has been carried out. In order to overcome the difficulties associated with the optimal selection of the learning rate parameters, a new method for adaptive learning rate has been developed. The proposed strategy has been implemented and its performance is evaluated on simulated process. The robustness of the system has been confirmed for the set point tracking. The performance comparison of these neural controller configurations is also given in terms of ISE and IAE.