Data-driven NODE based multirate sampled data state feedback control.

Area of Science:

• Control Systems Engineering
• Nonlinear System Dynamics
• Robotics

Background:

• Multirate sampled data systems enhance performance and adaptability in control systems, particularly with mixed signal types.
• Existing controllers for these systems can be overly complex and lack clear guidance on sampling interval selection.
• Ensuring stability and optimizing performance in nonlinear multirate systems remain significant challenges.

Purpose of the Study:

• To develop a robust stability criterion for multirate sampled data systems.
• To optimize controller parameters for enhanced system performance.
• To address limitations of existing controllers in nonlinear multirate systems.

Main Methods:

• A novel approach integrating system stability and practical considerations for selecting multiple sample periods.
• Utilizing Neural Ordinary Differential Equations (NODE) for optimal parameter selection in linear practical controllers.
• Designing and implementing linear stabilizers using multirate sampled data across three sampling levels.

Main Results:

• A criterion for selecting multiple sample periods that guarantees system stability was established.
• Optimal controller parameters were identified using NODE, maximizing performance based on a predefined index.
• Stability and controller design were successfully analyzed at three distinct sampling levels.

Conclusions:

• The proposed strategy effectively guarantees system stability in nonlinear multirate sampled data systems.
• The method provides an optimal approach to controller parameter selection, enhancing system performance.
• Demonstrated effectiveness through simulations and a real-world single-link robot system application.