Neuro-learning-based adaptive control for state-constrained strict-feedback systems with unknown control direction

Area of Science:

• Control Theory
• Artificial Intelligence
• Nonlinear Systems

Background:

• Strict-feedback nonlinear systems often face challenges with state constraints and unknown system parameters.
• Existing control methods struggle with asymmetric constraints and unknown control gain signs, requiring complex structural changes.

Purpose of the Study:

• To develop a unified adaptive control policy for strict-feedback nonlinear systems with asymmetric full-state constraints and unknown gain directions.
• To handle both symmetric and asymmetric constraints within a single adaptive structure.
• To address unknown control gain signs and system uncertainties effectively.

Main Methods:

• A neural networks (NNs)-based learning policy is proposed.
• A unified state-constrained function is introduced to manage various constraint types.
• Nussbaum gain technique and NN approximation are employed to handle unknown gain signs and uncertainties.

Main Results:

• The proposed control policy unifies the treatment of systems with or without constraints.
• Asymmetric and symmetric constraints are managed without altering adaptive structures, avoiding discontinuous actions.
• Tracking errors are proven to be semi-globally uniformly ultimately bounded (SGUUB) using Lyapunov theory.

Conclusions:

• The developed NN-based adaptive control strategy effectively manages strict-feedback nonlinear systems with asymmetric constraints and unknown gain directions.
• The unified state-constrained function offers a flexible approach to handling diverse constraint scenarios.
• Numerical simulations confirm the proposed scheme's effectiveness and robustness.