Adaptive Neural-Network Control of MIMO Nonaffine Nonlinear Systems With Asymmetric Time-Varying State Constraints.

Area of Science:

• Control Systems Engineering
• Nonlinear Dynamics
• Robotics

Background:

• Designing controllers for unknown nonaffine nonlinear systems with asymmetric time-varying state constraints is challenging.
• Existing methods often require complex offline computations or struggle with unknown control gains.

Purpose of the Study:

• To propose a novel robust adaptive barrier Lyapunov function (BLF)-based backstepping controller.
• To address interconnected, multi-input-multi-output (MIMO) unknown nonaffine nonlinear systems with asymmetric time-varying (ATV) state constraints.

Main Methods:

• Utilizing a neural-network-based online approximator for uncertain system dynamics.
• Developing a novel adaptive law based on the Hadamard product for weight tuning.
• Incorporating Nussbaum gain to handle unknown control gain in nonaffine systems.
• Proposing a theorem to pre-determine bounds on virtual control signals, simplifying feasibility checks.

Main Results:

• The proposed controller ensures system stability under asymmetric time-varying state constraints.
• The novel adaptive law effectively tunes neural network weights.
• The new theorem eliminates the need for tedious offline computations for virtual controller feasibility.
• Simulation results on a robot manipulator demonstrate the controller's effectiveness.

Conclusions:

• The developed BLF-based backstepping controller offers a robust and efficient solution for complex nonlinear systems.
• The methodology simplifies controller design by avoiding extensive offline analysis.
• The approach is validated through practical application in robot manipulator control.