A Deep Spiking Neural Network Anomaly Detection Method.

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Industrial Control Systems

Background:

• Increasing frequency and sophistication of cyber-attacks on industrial control systems (ICS).
• Significant financial and real-world implications of successful attacks, including corporate espionage and political targeting.
• Vulnerability of specialized systems like vibration monitoring in petroleum infrastructure to malicious interventions causing severe accidents.

Purpose of the Study:

• To examine digital security threats to vibration monitoring systems in petroleum infrastructure.
• To present a novel deep spiking neural network anomaly detection method for enhanced protection.
• To improve the accuracy and reliability of anomaly detection in critical infrastructure.

Main Methods:

• Utilizing a deep spiking neural network (SNN) to model spike sequences and information processing.
• Employing an innovative form of the Galves-Löcherbach Spiking Model (GLSM) with intrinsic stochasticity.
• Enhancing the GLSM with confidence intervals and stochastic variable-length memory chains.

Main Results:

• Achieved very high accuracy in detecting anomalies within vibration analysis systems.
• Successfully modeled complex spatiotemporal situations inherent in industrial environments.
• Demonstrated the effectiveness of the SNN approach in identifying malicious interventions.

Conclusions:

• The proposed deep spiking neural network offers a robust solution for securing critical petroleum infrastructure.
• Advanced SNN models, like the enhanced GLSM, are crucial for defending against sophisticated cyber-attacks.
• Accurate anomaly detection in vibration monitoring is vital for preventing catastrophic failures and environmental damage.