Machine Learning Based Framework for Optimizing Electrical Fault Detection and Classification in Power Systems
Students & Supervisors
Student Authors
Supervisors
Abstract
Fault detection and classification in power systems is a major challenge because faults can cause equipment damage, supply interruption, and system instability. Reliable and fast detection is necessary for safe operation. This research applies machine learning methods to identify faults and classify different fault types using two datasets. One dataset was used to detect faults or no faults with 12001 samples and 9 features. Another dataset was used to classify fault types with 7861 samples and 10 features. Random Forest (RF) delivered the best results among all models. In binary fault detection, it achieved 99.67% accuracy with a ROC AUC of 0.9995. In multi-class classification, it reached 99.81% accuracy with a macro F1 score of 0.9978. These results show that ensemble models can capture the complex relations in fault signals with very high reliability, while simple linear models, such as Logistic Regression, are not suitable. The analysis also confirms that current-based features are more useful than voltage features for classification. This research demonstrates that machine learning can offer robust and accurate solutions for real-time monitoring and protection of power systems, reducing the risk of blackouts and improving system stability.
Keywords
Publication Details
- Type of Publication:
- Conference Name: International Conference on Computer and Information Technology (ICCIT 2025)
- Date of Conference: 19/12/2025 - 19/12/2025
- Venue: Long Beach Hotel, Cox’s Bazar, Bangladesh
- Organizer: IEEE