Reinforcement Learning-Based Optimization for Interval Type-2 Fuzzy Unknown Nonlinear System: An Zero-Sum Control...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Fuzzy Logic Systems

Background:

• Nonlinear systems with unknown dynamics pose significant control challenges.
• Interval Type-2 Fuzzy (IT2F) models offer enhanced uncertainty handling capabilities.
• Traditional control methods often require complete system parameter knowledge, limiting their applicability.

Purpose of the Study:

• To develop a novel model-free reinforcement learning (RL) algorithm for optimal control of IT2F systems.
• To address the challenge of unknown system dynamics in nonlinear control applications.
• To achieve robust control with guaranteed stability and performance indices.

Main Methods:

• Formulated optimal control as a zero-sum game between control input and external disturbance.
• Developed a model-based policy iteration (PI) algorithm within an RL framework to solve system equations.
• Designed a model-free fuzzy control algorithm leveraging RL, requiring only state and input information.

Main Results:

• Achieved optimal control of IT2F systems without prior dynamic parameter information.
• Ensured asymptotic stability and H-infinity performance using Lyapunov functions.
• Demonstrated the algorithm's effectiveness on a semi-car active suspension model (SCASM).

Conclusions:

• The proposed model-free RL algorithm effectively controls unknown nonlinear IT2F systems.
• The method provides a robust solution for real-world systems where parameters are difficult to obtain.
• The approach guarantees stability and performance, validated by a practical application.