Adaptive Reconstruction-Based Model Predictive Control for Networked Stochastic Systems Under False Data Injection...

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Networked Systems

Background:

• Networked systems are vulnerable to false data injection (FDI) attacks, compromising control performance.
• Existing resilient model predictive control (MPC) methods often exhibit conservatism and high resource consumption.
• Stochastic MPC frameworks have not been adequately explored for addressing FDI attacks.

Purpose of the Study:

• To propose a novel resilient stochastic MPC method for networked systems facing FDI attacks.
• To mitigate the conservatism and reduce resource demands of current resilient control strategies.
• To enhance the security and reliability of networked control systems under cyber threats.

Main Methods:

• Development of an adaptive input reconstruction mechanism to relax FDI attack energy assumptions.
• Co-design of adaptive prediction horizon and terminal constraints to minimize computational complexity.
• Transformation of hard constraints into stochastic constraints to alleviate conservatism.

Main Results:

• Sufficient conditions are derived to ensure recursive feasibility and closed-loop stability.
• The proposed method effectively reconstructs feasible control inputs under FDI attacks.
• Simulations on a DC-DC converter system demonstrate the method's effectiveness.

Conclusions:

• The proposed resilient stochastic MPC method offers a robust solution against FDI attacks.
• The adaptive input reconstruction and constraint design reduce conservatism and computational load.
• This framework advances the security and performance of networked stochastic systems.