Observer-Based NN Control for Nonlinear Systems With Full-State Constraints and External Disturbances.

Area of Science:

• Control Systems Engineering
• Nonlinear Dynamics
• Robotics

Background:

• Full-state constrained nonlinear systems present challenges in control design due to input saturation.
• Unmeasurable states and external disturbances further complicate achieving accurate tracking control.
• Existing methods often struggle with the 'explosion of complexity' in backstepping designs.

Purpose of the Study:

• To design an output-feedback tracking controller for nonlinear systems with input saturation, unmeasurable states, and disturbances.
• To ensure semiglobal uniform ultimate boundedness of all closed-loop system signals.
• To achieve tracking control without violating state constraints.

Main Methods:

• A composite observer combining state and disturbance observers is proposed.
• An auxiliary system with approximate coordinate transformation addresses input saturation.
• Radial basis function neural networks (RBF NNs) and barrier Lyapunov functions (BLFs) are employed within a dynamic surface control (DSC) framework.

Main Results:

• The proposed controller guarantees semiglobally uniformly ultimately bounded signals in the closed-loop system.
• Tracking error is effectively regulated by saturated input error and design parameters.
• State constraints are not violated during the tracking process.

Conclusions:

• The developed output-feedback controller effectively handles input saturation, unmeasurable states, and disturbances in constrained nonlinear systems.
• The combination of composite observers, RBF NNs, BLFs, and DSC provides a robust solution.
• Simulations on a robot arm demonstrate the controller's practical effectiveness.