FAULT DETECTION AND ISOLATION IN POWER NETWORKS USING MACHINE LEARNING

Abstract

The increasing complexity and scale of modern power networks demand fast, accurate, and reliable fault detection and isolation mechanisms to ensure system stability and uninterrupted power supply. Conventional protection schemes rely on fixed thresholds and rule-based logic, which often struggle to adapt to dynamic operating conditions, distributed generation, and non-linear fault characteristics. This paper presents a machine learning–based framework for fault detection and isolation in power networks. The proposed approach leverages intelligent data-driven models to analyze voltage, current, and frequency signals for identifying fault type, location, and severity in real time. By integrating supervised learning techniques with real-time measurement data, the system enhances fault diagnosis accuracy and reduces response time. MATLAB/Simulink-based analysis demonstrates improved detection speed, classification accuracy, and robustness compared to traditional protection methods, validating the effectiveness of machine learning for intelligent power system protection.