A power system comprises generation, transmission, distribution, and utilization components, forming the core of any electrical infrastructure. Faults in a power system occur when abnormal electric currents are present, which can lead to severe equipment damage and even system collapse. This work aims to automatically classify power system faults into one of eleven categories, covering both balanced and unbalanced fault types. A synthetic dataset representing eleven types of faults on a 100 km overhead transmission line is generated using MATLAB/Simulink. Supervised machine learning techniques are applied for fault classification, implemented in Python using the scikit-learn library. Three widely used classification algorithms—Decision Tree (DT), K-Nearest Neighbor (KNN), and Support Vector Machine (SVM)—are evaluated and compared. Among them, the SVM model demonstrates superior performance, achieving a test accuracy of 91.6% on the generated dataset. The resulting predictive model enhances the intelligence and reliability of power system monitoring, contributing to more robust and uninterrupted power supply.