Improved Fully Adjusted Neural Network based Recursive Terminal Sliding Mode Control for MEMS Gyroscopes

Abstract

This paper proposes an adaptive recursive terminal sliding mode control (ARTSMC) using an improved fully tuned RBF neural network for Micro-Electro-Mechanical System (MEMS) gyroscopes. First, a mathematical model of a MEMS Z-axis vibrating gyroscope is introduced. Then, an ARTSMC is constructed with a recursive fast nonsingular terminal sliding surface to guarantee finite-time tracking error convergence. In addition, to release the dependence of the proposed controller on system parameters and to correctly estimate the angular velocity, an improved fully tuned RBF neural network is used to approximate gyroscope parameters. Simulation studies are conducted to verify the effectiveness of the proposed scheme.