AI Based Fault and Anomaly Detection in Power Systems

Artificial -based infiltration system detection systems (IDs) are emerging as an important defense mechanism for modern power systems, which are rapidly weak for cyber-attacks due to comprehensive digitization, integration of IOT devices and dependence on comprehensive field communication networks. Traditional rules-based IDS methods often fail to detect sophisticated hazards such as false data injection attacks, refusal-service (DOS), and load-transport infiltration, especially large scale, smart grids were distributed. The AI-operated IDS leverage machine learning, deep learning, and hybrid models to detect discrepancy, adapted to develop the attack pattern, and supported real-time status awareness. This review examines the approach to detect state-of-the-art AI-based infiltration for electrical systems, which focus on detection methods, datasets, evaluation matrix and implementation challenges. Special emphasis is given to explain capacity, scalability and integration with supervisory control and data acquisition (SCADA), Phasor Measurement Units (PMU), and distributed energy resources (DERS). Finally, open research intervals and future instructions are highlighted, including federated learning, graphs in graph neural network-based detections include graph neural network-based detections and digital twin-capable cyber-flexibility.