FAULT CONDITION PREDICTION IN POWER TRANSMISSION LINES

This project presents a real-time fault diagnosis system for Series-Compensated Power Transmission Lines (SC-PTL) using a Flask-based web interface. It integrates ensemble learning particularly Stacking and Voting classifiers along with Random Forest, XG-Boost, Ada-Boost, and Light-GBM for high accuracy and computational efficiency. Gradient Boosting Feature Selection is used to enhance performance. SMOTE is applied to balance class distribution. The system processes six key electrical features, achieving 99.79% accuracy with strong noise resilience. Results are logged for analysis, making the system suitable for smart grid applications and real-time power system monitoring.

Keywords: Fault Diagnosis, Series-Compensated Transmission Lines, Ensemble Learning, SMOTE, Gradient Boosting, Feature Selection, Real-Time Monitoring, XG-Boost, Noise Resilience, Logistic Regression , Stacking Classifier, Voting Classifier