Fault accommodation for nonlinear systems using cerebellar model articulation controller

This paper presents a learning approach using cerebellar-model-articulation-controller (CMAC) to accommodate faults for nonlinear dynamic systems. A CMAC with Gaussian-type receptive-field basis function is proposed to estimate the unknown fault. Then, a robust fault accommodation scheme is derived based on Lyapunov stability theorem. Finally, the proposed fault accommodation system is applied to a jet engine compression system. Simulation results show that this method can effectively achieve the fault accommodation.