CMAC-based supervisory control for nonlinear chaotic systems

This study proposes a supervisory control system based on a cerebellar-model-articulation-controller (CMAC) to resolve the tracking control problem of nonlinear chaotic systems. This CMAC-based supervisory control (CSC) system is composed of a CMAC and an H∞ supervisor. CMAC is a main tracking controller that is investigated to mimic a perfect control law and the H∞ supervisor is a supervisory controller that is developed to attenuate the effect of the approximation error between CMAC and the perfect control law. The gradient descent method and H∞ control technique are utilized to derive the on-line parameter tuning laws to achieve favorable tracking performance. Finally, the proposed CSC system is applied to control a Duffing–Holmes chaotic system and a chaotic Chua’s circuit. A comparison between a sliding-mode control and the proposed CSC for these nonlinear chaotic systems are presented. Simulation results show that the proposed CSC scheme can effectively achieve favorable control performance for nonlinear chaotic systems.