Neurodynamic Approximation-Based Quantized Control with Improved Transient Performances for MEMS Gyroscopes: Theory and Experimental Results

This paper investigates a neurodynamic quantized control problem with improved transient performances for microelectromechanical system gyroscopes subject to lumped disturbances induced by parameter uncertainties, dynamic coupling, and external disturbances. A modified prescribed performance control mechanism based on a hyperbolic cosecant performance function is proposed to specify the transient behavior of tracking errors with an arbitrarily small overshoot. To eliminate the lumped disturbances with a better identification property, a novel echo state network approximator that uses the estimation error to learn neural weights is designed with the aid of a minimal learning parameter technique, which not only can exclude the poor transient behaviors occurring extensively in the existing neural control with a large adaptive gain, but also dramatically reduce the number of parameters to be updated online. Furthermore, contrasting to the available tracking results assuming a continuous control updating, a hysteresis quantizer is introduced to provide discrete values of control signals, supporting the use of digital devices in MEMS gyroscope implementations to realize a closed-loop error stabilization. Finally, ultimately uniformly bounded stability of closed-loop system is proved, while not only simulations but also experimental results are performed to validate the effectiveness of proposed scheme