Multi-Loop Recurrent Neural Network Fractional-Order Terminal Sliding Mode Control of MEMS Gyroscope

Abstract

This paper proposes a fractional-order nonsingular terminal sliding mode control of a MEMS gyroscope using a double loop recurrent neural network approximator. For the system stability, a nonsingular terminal sliding mode controller is formulated to guarantee the convergence. For higher accuracy and faster convergence, the fractional-order (FO) calculus is employed with additional degree of freedom. For the system robustness, the neural network is designed to approximate the lumped uncertainty. The inner recurrent loop and external recurrent loop are employed to provide a feedback signal to obtain satisfactory approximation accuracy. For higher adaptability of the neural network, the dynamic function is formulated and the updating law of the parameter is given. Furthermore, the Lyapunov stability theorem is employed to verify the asymptotical stability and convergence of system. Simulations for a MEMS gyroscope are studied to exhibit the superiority of the proposed control strategy.