Detection and Elimination of Denial-of-Service Attack Using Supervised Machine Learning Considering Heterogeneous Measurements

Cyber-physical power systems (CPPS) rely on cyber infrastructure for reliable and efficient operation, making state estimation essential for distribution systems. However, these systems are vulnerable to cyberattacks that interupts data communication and compromise stability. This paper introduces a Forecasting-Aided State Estimation (FASE) algorithm, integrating heterogeneous meter data such as voltage magnitudes and their angles, as well as currents from the IEEE 33-bus system. Additionally, a novel framework is proposed to detect and mitigate Denial-of-Service (DoS) attacks using supervised machine learning (ML) techniques. The methodology uses a decision tree (DT)-based model to estimate attack intervals and reduce their impact on state estimation accuracy. The results demonstrate the effectiveness of the method in improving the resilience of the CPPS while maintaining accurate and efficient Distribution System State Estimation (DSSE). This study underscores the potential of ML-based approaches in mitigating cyber threats in modern power systems.