Policy-Iteration-Based Active Disturbance Rejection Control for Uncertain Nonlinear Systems With Unknown Relative...

Area of Science:

• Control Systems Engineering
• Nonlinear Dynamics
• Artificial Intelligence

Background:

• Uncertain nonlinear systems pose significant control challenges, particularly in achieving precise real-time tracking.
• Traditional control methods often require detailed system models and knowledge of parameters like relative degree.
• Active Disturbance Rejection Control (ADRC) offers a framework for handling uncertainties, but adapting it to unknown relative degrees remains complex.

Purpose of the Study:

• To propose a policy-iteration-based active disturbance rejection control (ADRC) for uncertain nonlinear systems.
• To achieve real-time output tracking performance irrespective of the system's specific relative degree.
• To integrate reinforcement learning (RL) for adaptive control parameter tuning.

Main Methods:

• A partial control input generator is combined with a policy-iteration-based reinforcement learning (RL) agent.
• The RL agent iteratively refines policies to adjust degree weights, enhancing the influence of the correct partial control input.
• Lyapunov stability theorem and affinely quadratically stable properties are used to ensure closed-loop system stability.

Main Results:

• The proposed ADRC method successfully achieves real-time output tracking for uncertain nonlinear systems.
• The RL agent adaptively adjusts degree weights, improving control performance without prior knowledge of the relative degree.
• Simulations and experimental results on a permanent magnet synchronous motor demonstrate the method's effectiveness and robustness.

Conclusions:

• The policy-iteration-based ADRC with RL offers a model-free and relative-degree-agnostic approach for controlling uncertain nonlinear systems.
• The method guarantees semi-global uniformly ultimately boundedness of all closed-loop signals, ensuring stability.
• This approach significantly reduces the need for detailed system information, making it broadly applicable.