TRANSMISSION FAILURE PREDICTION USING AI AND STRUCTURAL MODELING INFORMED BY DISTRIBUTION OUTAGES

ABSTARCT

Transmission systems are critical infrastructures where failures can result in widespread power disruptions and financial losses. Traditional failure detection methods rely heavily on manual inspections and post-event analysis, leading to delayed responses and increased risk. This system addresses these challenges by utilizing artificial intelligence techniques to predict four major transmission failure classes—G, C, B, and A—based on sensor readings of currents and voltages across three phases.

The backend architecture integrates multiple machine learning models, including Random Forest, Gradient Boosting, and Decision Tree classifiers, combined through a Stacking Classifier to enhance predictive accuracy and robustness. Following extensive evaluation, the Random Forest model is deployed on the frontend for real-time prediction due to its high interpretability and efficiency. A Flask-based web interface enables users to input real-time sensor data and instantly receive diagnostics identifying potential failures. The system uses MySQL for secure user data management and session handling. By providing early warnings based on live input data, it empowers transmission operators to take preventive measures, minimize downtime, and improve grid reliability. The integration of AI-driven modeling with insights informed by distribution outages offers a scalable and proactive solution for intelligent and resilient energy infrastructure management.

Keywords: Transmission Failure, Multi-class Classification, Random Forest, Stacking Classifier, Gradient Boosting, Decision Tree, Structural Modeling, Distribution Outages, Predictive Maintenance, Flask Web Application.