Event-Triggered Secure Control Design Against False Data Injection Attacks via Lyapunov-Based Neural Networks

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Robotics

Background:

• False data injection (FDI) attacks pose significant threats to the integrity and reliability of control systems.
• Existing secure control methods may not be resource-efficient, leading to unnecessary computational overhead and control actions.

Purpose of the Study:

• To develop a secure and resource-efficient control framework resilient to FDI attacks.
• To integrate a hybrid observer with an event-triggered control (ETC) strategy for enhanced operational performance.

Main Methods:

• A hybrid observer combining a Kalman filter and a neural network (NN) was designed for real-time FDI detection and compensation.
• Lyapunov-based update laws were derived for NN weights to guarantee system stability.
• An event-triggered control (ETC) mechanism was implemented to minimize control actions and resource usage.
• Linear matrix inequality (LMI) conditions were formulated for stability analysis and triggering threshold determination.

Main Results:

• The proposed framework effectively detected and compensated for various FDI attack scenarios.
• The event-triggered mechanism significantly reduced control redundancy and computational load.
• Lyapunov-based stability analysis confirmed the boundedness of estimation and system errors.

Conclusions:

• The developed secure control framework offers a robust and resource-aware solution for safety-critical applications.
• The integration of hybrid observers and ETC provides resilience against FDI attacks while optimizing system performance.