Performance-Guaranteed Fault-Tolerant Control for Uncertain Nonlinear Systems via Learning-Based Switching Scheme.

Area of Science:

• Control Engineering
• Nonlinear Systems Analysis
• Fault-Tolerant Control Systems

Background:

• Industrial systems often face actuator faults, necessitating fault-tolerant control (FTC).
• Traditional FTC relies on fault detection-isolation mechanisms, which can be complex and prone to errors.
• Unknown system dynamics and time-varying constraints pose significant challenges in FTC design.

Purpose of the Study:

• To develop an adaptive FTC scheme for unknown multi-input single-output (MISO) nonlinear systems.
• To guarantee system output constraints under actuator faults without explicit fault detection.
• To enhance the safety and reliability of industrial systems operating under fault conditions.

Main Methods:

• A learning-based switching function scheme for automatic actuator rotation.
• Error transformation techniques to enforce user-defined time-varying asymmetric output constraints.
• Neural networks for adaptive control of unknown system dynamics and dynamic surface control (DSC) for complexity reduction.
• Lyapunov stability analysis to ensure system boundedness.

Main Results:

• The proposed scheme effectively maintains system outputs within specified constraints during actuator failures.
• Elimination of the need for explicit fault detection mechanisms simplifies the control design.
• Demonstrated stability of the closed-loop system using Lyapunov methods, ensuring bounded signals.
• Simulation results verified the effectiveness of the adaptive FTC scheme.

Conclusions:

• The developed adaptive FTC strategy successfully guarantees output constraints for unknown MISO nonlinear systems with actuator faults.
• The learning-based switching and error transformation methods provide a robust and simplified approach to fault tolerance.
• The integration of neural networks and DSC enhances adaptability and reduces design complexity, ensuring system safety.