Secure predictor-based neural dynamic surface control of nonlinear cyber-physical systems against sensor and actuator...

Area of Science:

• Cyber-Physical Systems
• Control Theory
• Artificial Intelligence

Background:

• Cyber-physical systems are vulnerable to sophisticated sensor and actuator deception attacks.
• Existing control methods may struggle with nontriangular system dynamics and unknown uncertainties.
• Ensuring system security and stability under adversarial conditions is critical.

Purpose of the Study:

• To develop a secure predictor-based neural dynamic surface control (SPNDSC) strategy.
• To address deception attacks on both sensors and actuators in nontriangular cyber-physical systems.
• To guarantee the stability and boundedness of the closed-loop system.

Main Methods:

• Utilizing partial states as neural network inputs to avoid algebraic loops.
• Employing neural networks for approximating unknown system dynamics.
• Introducing nonlinear gain functions and attack compensators to mitigate attack effects.
• Developing compensation terms for neural network approximation errors.

Main Results:

• Successfully alleviated adverse effects of intelligent adversaries.
• Demonstrated ultimate boundedness of all signals in the closed-loop system.
• Validated the proposed control strategy through two illustrative examples.

Conclusions:

• The proposed SPNDSC is effective in securing cyber-physical systems against deception attacks.
• The method ensures system stability and signal boundedness under adversarial conditions.
• This approach offers a robust solution for enhancing the security of critical infrastructure.