Safety Impact Analysis Considering Physical Failures and Cyber-Attacks for Mechanically Pumped Loop Systems (MPLs)

Area of Science:

• Complex systems engineering
• Cyber-physical systems security
• Control systems analysis

Background:

• Mechanically pumped loop systems (MPLs) integrate physical and cyber components, making them susceptible to both physical failures and cyber-attacks.
• The impact of these threats on MPL operations and their severity assessment remains largely unexplored.
• Existing safety analysis methods may not adequately address the unique vulnerabilities of cyber-physical systems like MPLs.

Purpose of the Study:

• To develop and propose a novel safety impact analysis framework for MPLs.
• To quantitatively assess the severity of physical failures and cyber-attacks on MPL operations.
• To identify critical vulnerabilities and inform protective strategies for MPLs.

Main Methods:

• Utilized an Elman Neural Network (ENN) observer for system state estimation and anomaly detection.
• Employed the Gaussian Mixture Model (GMM) algorithm for robust failure and attack mode classification.
• Simulated and analyzed three common threat scenarios: sudden sensor hard failure, gradual sensor soft failure, and denial-of-service (DoS) attacks.

Main Results:

• Simulation data confirmed that both sensor failures and DoS attacks compromise MPL safety.
• Sensor failures, particularly gradual soft failures, were found to inflict more severe harm than DoS attacks.
• Sensors critical for global control were identified as requiring enhanced protection compared to those in local control.

Conclusions:

• The proposed ENN and GMM based framework effectively assesses the safety impact of physical and cyber threats on MPLs.
• Gradual sensor degradation represents a significant and potentially underestimated risk to MPL operational integrity.
• Prioritizing the security of globally critical sensors is essential for maintaining the overall safety and reliability of MPLs.