AI-Based Arc Fault Detection in Electrical Distribution Systems

Abstract

Electrical arc faults are a significant cause of fires and system failures in low-voltage distribution networks, especially in environments where PVC-insulated wiring is widely used. Traditional protective devices such as fuses and circuit breakers are often incapable of detecting such faults due to their intermittent and low-current nature. This limitation poses serious safety risks in residential, industrial, and commercial installations. The proposed project aims to develop a low-cost, AI-based arc fault detection system specifically tailored for PVC wire-based systems. The solution integrates current and voltage sensors with a microcontroller that collects real-time signal data, which is then analyzed using machine-learning models trained to recognize the unique characteristics of arc faults. The system will be simulated using MATLAB and Proteus software and later implemented in hardware for validation. It is designed to trigger immediate alerts upon detecting fault conditions, significantly reducing the risk of fire and system damage. The expected outcome is an intelligent, adaptable, and efficient protection system that enhances electrical safety, especially in aging or overloaded PVC wiring networks.

KEYWORDS: Arc Fault, AI, Machine Learning, Electrical Safety