On the development of an intelligent controller for neural networks: a type 2 fuzzy and chatter-free approach for variable-order fractional cases

The rise of artificial intelligence has revolutionized all aspects of today's life. Neural networks, which are a stepping stone in the search for artificial intelligence, considerably affect the pace of advances in this field. Therefore, developing trustable tools for their modeling and control is of crucial importance. Motivated by this, we propose an intelligent controller for neural networks. Although the proposed approach is based on the sliding mode concept, it is chatter-free and provides smooth results. A type 2 fuzzy observer is applied to estimate the unknown function of the chaotic neural networks and enhance the performance of the proposed controller. This way, the proposed controller will act smartly in unseen conditions. The offered disturbance observer possesses an updating mechanism that modifies the type 2 fuzzy observer’s weights. Using the Lyapunov stability method, the analysis of stability is performed, and it is guaranteed that the proposed control scheme is asymptotically stable. Finally, the simulation results are presented to show the effectiveness of the offered method for the chaotic variable-order fractional neural networks under uncertainties and unknown external disturbances.