Overhead Transmission Line Insulator Failure Detection

Thesis Overview

This research evaluated the automated detection and classification of high-voltage transmission line insulator defects using computer vision and deep learning techniques.

Key Contributions

• Developed and trained a VGGNet-16 deep learning model in Python to classify insulator conditions as defective or healthy.
• Implemented an image-based predictive maintenance pipeline to catch structural faults early and minimize electrical utility grid downtime.
• Achieved 98% training accuracy, proving the viability of automated computer vision for remote electrical infrastructure monitoring.

Technologies

Python, TensorFlow, Keras, Computer Vision, Deep Learning, VGGNet-16