Adaptive Backstepping-Based Neural Tracking Control for MIMO Nonlinear Switched Systems Subject to Input Delays

Area of Science:

• Control Systems Engineering
• Artificial Intelligence in Engineering
• Nonlinear Dynamics

Background:

• Complex nonlinear switched systems present significant control challenges due to uncertainties and input delays.
• Existing control methods often struggle with the conservativeness introduced by common coordinate transformations in subsystem analysis.

Purpose of the Study:

• To develop a novel adaptive neural network-based output tracking control scheme.
• To address disturbed, multiple-input multiple-output (MIMO), uncertain nonlinear switched systems with input delays.
• To overcome limitations of existing control designs by employing distinct coordinate transformations.

Main Methods:

• Utilized radial basis function neural networks (NNs) for their universal approximation capabilities.
• Employed adaptive backstepping recursive design combined with an improved multiple Lyapunov function (MLF) scheme.
• Implemented different coordinate transformations for subsystems to reduce design conservativeness.

Main Results:

• Achieved semiglobally uniformly ultimately boundedness for all closed-loop system variables under various switching signals.
• Demonstrated that the system output successfully follows the desired reference signal.
• Validated the controller's practicability through application to a mass-spring-damper system.

Conclusions:

• The proposed adaptive neural output tracking control scheme is effective for a class of complex nonlinear switched systems.
• The novel design approach enhances robustness and tracking performance.
• The method offers a practical solution for real-world control applications.