Enhancing DoS Detection in SmartGrid: Leveraging ML Using PCA and Explainable AI

Grid systems established integrated networks in every hierarchical level of power system for extensive smart automation. To ensure the availability and reliability of the SmartGrid system, fortification against cyber threats is crucial. Amidst other anomalies, Denial of Service (DoS) disrupts normal network operations by overwhelming the systems with excessive unauthorized traffic. To assure cybersecurity and undisrupted services, this article focuses on the IEC 60 870-5-104 protocol concerning 12 classes of DoS attack command detection. This anomaly detection mechanism utilizes a scrupulously indexed IEC 60870-5-104 intrusion detection dataset through panoramic preprocessing and subsequently introduces feature engineering and dimension reduction by Principal Component Analysis (PCA). Succeeding training of ML models achieved a raised and definitive accuracy of 98.709%. The introduction of SHAP analysis presents unambiguous and informative insight into the model's decision-making with the most significant features. This study establishes the groundwork for crafting robust security protocols that ensure the integrity and operational stability of integrated network systems in the face of constantly evolving cybersecurity threats.