Neural-network-based event-triggered adaptive secure fault-tolerant containment control for nonlinear multi-agent...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Network Security

Background:

• Nonlinear multi-agent systems (MASs) face challenges with sensor output triggering, multiple faults, and denial-of-service (DoS) attacks.
• Backstepping theory struggles with non-differentiable virtual control signals from triggered sensor outputs.

Purpose of the Study:

• To develop an event-triggered adaptive secure fault-tolerant containment control for nonlinear MASs.
• To address multiple faults and DoS attacks simultaneously in nonlinear MASs.

Main Methods:

• Utilizing a switched neural network estimator with intermittent output signals for first-order derivable state estimation.
• Constructing first-order differentiable virtual control laws using estimated states.
• Employing dynamic filtering technology to prevent repeated differentiation of virtual control laws.

Main Results:

• The designed controller effectively compensates for system faults and DoS attacks.
• Each follower agent converges to the dynamic convex hull defined by the leader agents.
• Simulation results validate the proposed control method's effectiveness.

Conclusions:

• The proposed neural-network-based event-triggered adaptive control is effective for secure fault-tolerant containment in nonlinear MASs.
• The method successfully handles sensor output triggering, multiple faults, and DoS attacks.