Resilient Consensus Control of Nonlinear Multiagent Systems Under Hybrid Cyberattacks: A Disturbance Observer-Based...

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Artificial Intelligence

Background:

• Multiagent systems (MASs) are vulnerable to hybrid cyberattacks, including false data injection (FDI) and denial-of-service (DoS) attacks.
• External disturbances and inherent nonlinear dynamics further challenge the stability and consensus of MASs.

Purpose of the Study:

• To develop a novel observer-based adaptive neural network control strategy for nonlinear leader-following MASs.
• To address the combined effects of hybrid cyberattacks, external disturbances, and system nonlinearities.

Main Methods:

• A dimension expansion methodology was used to model and compensate for false data injection (FDI) attacks.
• Denial-of-service (DoS) attacks were probabilistically characterized using Bernoulli variables.
• A cascaded observer was designed for state and disturbance estimation, incorporating disturbance decoupling.
• An adaptive neural network was employed to approximate nonlinear dynamics, enhancing robustness.

Main Results:

• The proposed control method effectively mitigates hybrid cyberattacks, including FDI and DoS attacks.
• The cascaded observer accurately estimated system states and external disturbances.
• The adaptive neural network compensated for nonlinearities, improving system resilience.
• Simulation results validated the achievement of resilient consensus in leader-following MASs.

Conclusions:

• The developed observer-based adaptive neural network control approach provides a robust solution for achieving resilient consensus in nonlinear MASs.
• The strategy effectively handles complex scenarios involving hybrid cyberattacks, disturbances, and nonlinear dynamics.