Threat-Aware Intrusion Detection and Prevention for IEDs in Smart Grids Using Edge Computing and Hybrid ML Models

Abstract

An edge computing-based and threat behavior-aware smart prioritization framework is designed for cybersecurity intrusion detection and prevention in intelligent electronic devices (IEDs) within smart grids. Using the Power System Intrusion Detection dataset, various machine learning algorithms, including SVM, One-Class SVM, Gradient Boosting, LightGBM, Random Forest, and a Stacking Classifier combining Random Forest, LightGBM, and ExtraTreesClassifier, are employed for binary and multi-class intrusion detection tasks. The framework leverages the integration of modified LightGBM and One-Class SVM models to enhance detection accuracy and adaptability to evolving threats. Results demonstrate that the Stacking Classifier achieves the highest performance, with an accuracy of 99.5% in binary classification and 87.2% in multi-class classification. By leveraging the strengths of individual models and optimizing their integration, the proposed framework significantly improves the accuracy and reliability of intrusion detection systems in smart grids, ensuring robust and efficient cybersecurity. This approach is particularly effective in identifying complex threat behaviors, prioritizing critical alerts, and enabling real-time prevention mechanisms within the edge computing paradigm.