The identification and categorization of electrical faults are essential for ensuring the safety and reliability of modern power systems. Disruptions such as line-to-ground and line-to-line faults can result in severe effects, including hazardous conditions, equipment destruction, and power failures. Conventional detection techniques frequently fall short due to their high prices, inefficiencies, and poor accuracy. This work explores the use of deep learning models and advanced machine learning (ML) to address these problems effectively. A variety of models, including Support Vector Machines (SVM), Decision Trees (DT), Random Forests (RF), and Artificial Neural Networks (ANN), were employed to recognize and classify different types of electrical abnormalities. The research included a diverse dataset, and the models underwent a thorough assessment employing precision, recall, F-score, and accuracy metrics. The results demonstrate that the ANN model surpassed the other models, attaining an F-score of 98.48% and a testing accuracy of 98.50%. This indicates that the ANN model is the most reliable for identifying electrical faults. This work demonstrates how important machine learning is to improving contemporary power systems’ efficiency, safety, and robustness and provides a viable path forward for the field’s future development.