ML Surrogate Ranks Criticality in Power-Communication Networks

Sohini Roy, Xheni Hylviu· July 13, 2026 View original

Summary

This paper develops a machine learning model to rapidly predict contingency severity and rank component criticality in interdependent power-communication networks, overcoming the computational cost of high-fidelity simulations. The Gradient Boosting surrogate achieves high correlation with ground-truth simulations, enabling efficient resilience planning.

Cyber-physical power systems face significant risks from cascading failures due to their tight integration with communication infrastructure. Traditional methods for evaluating these failures, especially across a large number of potential N-k contingencies, are computationally intensive and impractical for real-time resilience planning. Researchers have introduced a machine learning surrogate model designed to accelerate this process. Utilizing a Gradient Boosting algorithm, the model predicts the severity of contingencies and ranks the criticality of individual components within these complex networks. This approach allows for a much faster assessment, making it feasible to identify and prioritize components for hardening against potential failures. The surrogate model demonstrated strong performance on the IEEE 118-bus system, achieving high Spearman correlations for both contingency severity prediction and component criticality ranking. Its effectiveness is attributed to its ability to leverage inter-layer dependency information, significantly outperforming simpler topological centrality measures. This suggests a two-stage workflow where the ML model quickly identifies high-risk areas, with more detailed simulations reserved for verification.

Why it matters

Professionals in critical infrastructure management can use this ML surrogate to significantly speed up resilience planning and identify vulnerabilities in complex power and communication systems. This allows for more proactive and cost-effective hardening strategies against cascading failures.

How to implement this in your domain

  1. 1Integrate the ML surrogate model into existing infrastructure planning software to automate initial vulnerability assessments.
  2. 2Prioritize hardening efforts on components identified as highly critical by the surrogate to maximize resilience impact.
  3. 3Develop a two-stage workflow where the ML model provides rapid initial rankings, followed by high-fidelity simulations for selected critical scenarios.
  4. 4Collect and preprocess relevant structural and interdependency data from your power and communication networks to train and validate similar surrogate models.

Who benefits

EnergyUtilitiesTelecommunicationsCritical InfrastructureGovernment

Key takeaways

  • Machine learning can significantly accelerate the assessment of critical component vulnerabilities in interdependent networks.
  • A Gradient Boosting surrogate model achieved high accuracy in predicting contingency severity and ranking component criticality.
  • The model's advantage stems from its ability to process inter-layer dependency information effectively.
  • This approach enables a more efficient two-stage workflow for resilience planning, combining rapid ML assessment with targeted high-fidelity simulation.

Original post by Sohini Roy, Xheni Hylviu

"arXiv:2607.08918v1 Announce Type: new Abstract: Cyber-physical power systems are vulnerable to cascading failures caused by tight interdependencies between power and communication infrastructures. Evaluating these failures over large N-k contingency sets with a high-fidelity simu…"

View on X

Originally posted by Sohini Roy, Xheni Hylviu on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Research

AI ResearchAI Engineering & DevTools

Data-Efficient Deep Learning Guidelines for Inertial Sensor Classification

This study provides empirical guidelines for estimating the minimum training set size needed for deep learning models in inertial sensor classification tasks. It reveals that accuracy follows a consistent logarithmic growth pattern, allowing for data-efficient planning of recording campaigns and achieving practical stability with fewer samples than traditionally assumed.

Ofir Kruzel, Itzik KlienJul 13, 2026
AI Engineering & DevToolsAI Research

On-Device Adaptive AI Boosts EV Battery Power Prediction

Researchers developed a novel approach for on-device learning in electric vehicles (EVs) that continuously adapts pretrained battery power prediction models to new data. This method significantly improves forecasting performance, reducing mean absolute errors by up to 14.88% with offline adaptation and 7.49% with online adaptation.

Avik Bhatnagar, Anton Paule, Tobias Schuermann, Sebastian Reiter, Oliver BringmannJul 13, 2026
AI ResearchAI Engineering & DevTools

New Differentiable Logic Networks Outperform Fixed-Connection Models

Researchers introduce a novel method for optimizing connections in deep differentiable logic gate networks (LGNs) and lookup table networks (LUTNs), achieving superior performance with significantly fewer gates. The approach allows for parallel learning of optimal gate types and LUT entries, demonstrating improved accuracy on benchmarks like MNIST.

Wout Mommen, Lars Keuninckx, Matthias Hartmann, Werner Van Leekwijck, Piet WambacqJul 13, 2026