Neural-Network-Based Constrained Output-Feedback Control for MEMS Gyroscopes Considering Scarce Transmission...

Area of Science:

• Control Systems Engineering
• Neural Networks
• Microelectromechanical Systems (MEMS)

Background:

• Microelectromechanical system (MEMS) gyroscopes face challenges with scarce transmission bandwidth and lumped disturbances.
• Model uncertainties, dynamic coupling, and environmental factors contribute to control complexities in MEMS gyroscopes.

Purpose of the Study:

• To develop a neural-network-based constrained output-feedback control for MEMS gyroscopes.
• To address limitations of scarce bandwidth and lumped disturbances.
• To enhance tracking accuracy and system stability.

Main Methods:

• A hybrid quantizer was designed for adjustable communication rates and quantization density.
• Nonlinear mapping was employed to constrain state variables within predefined time-varying asymmetric functions.
• An echo-state network-based minimal learning parameter neural observer was developed to estimate unmeasurable states and disturbances.

Main Results:

• The proposed hybrid quantizer reduces chattering and enhances tracking accuracy.
• State variables are effectively constrained without stringent feasibility conditions.
• The neural observer successfully recovers unmeasurable states and disturbances, enabling an output-feedback solution.
• Lyapunov stability and nonsmooth analysis confirmed ultimate uniform boundedness of all closed-loop system signals.

Conclusions:

• The developed neural-network-based control strategy is effective for MEMS gyroscopes under bandwidth constraints and disturbances.
• The controller demonstrates improved tracking accuracy and robust stability.
• The approach offers a viable solution for enhancing the performance of MEMS gyroscopes in practical applications.