Adaptive Neural Tracking Control for Interconnected Switched Systems With Non-ISS Unmodeled Dynamics

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Nonlinear Dynamics

Background:

• Interconnected switched systems often exhibit unmodeled dynamics, complicating control design.
• Some subsystems may not satisfy the input-to-state stable (ISS) condition, posing stability challenges.
• Existing control methods struggle with systems featuring non-ISS unmodeled dynamics and rapid switching.

Purpose of the Study:

• To develop an adaptive neural network tracking control strategy for interconnected switched systems with unmodeled dynamics.
• To address stability issues arising from non-ISS subsystems.
• To design a novel switching signal scheme that enhances system stability and tracking performance.

Main Methods:

• Utilizing neural networks for approximating composite unknown nonlinear functions.
• Employing dynamic surface control for decentralized tracking controller design.
• Developing a new switching signal scheme to ensure semi-global boundedness and prescribed performance.

Main Results:

• The designed decentralized controllers stabilize some subsystems.
• A novel switching scheme guarantees overall system stability (semi-global boundedness) despite non-ISS dynamics.
• Tracking errors converge to predefined residual sets with a prescribed performance index.
• The proposed switching scheme allows for a higher switching frequency than traditional average dwell time methods.

Conclusions:

• The adaptive neural network control approach effectively handles interconnected switched systems with challenging unmodeled dynamics.
• The novel switching signal scheme provides robust stability and performance guarantees.
• The method demonstrates superior performance and flexibility compared to existing techniques.