AI-Enhanced IoT-Based Solar Panel Fault Detection Device Using Edge AI, Machine Learning, and Digital Twin Technology for Real-Time Photovoltaic System Monitoring

Abstract

Solar photovoltaic (PV) systems require continuous monitoring to maintain energy efficiency, minimize power losses, and detect operational faults at an early stage. This paper presents an AI-enhanced Internet of Things (IoT)-based solar panel fault detection device that performs real-time monitoring, intelligent fault diagnosis, and edge-based data processing. The proposed system integrates multiple environmental and electrical sensors with an ESP32 microcontroller to collect voltage, current, temperature, irradiance, and other operating parameters. A lightweight Decision Tree machine learning model, optimized using TensorFlow Lite, is deployed on the edge device to classify common solar panel faults, including dust accumulation, partial shading, temperature imbalance, electrical anomalies, and panel degradation. Edge AI processing significantly reduces detection latency, minimizes cloud dependency, and enables rapid fault identification in remote environments. The developed prototype was validated using both synthetic datasets and real-time sensor measurements obtained from an operational solar panel setup. Experimental evaluation demonstrated a fault detection accuracy ranging from 92% to 100% with an average response time of less than one second. A digital twin-enabled web dashboard was also developed to provide real-time visualization, remote monitoring, historical data analysis, and fault alerts for enhanced system management. The proposed AI-based IoT solution offers a low-cost, scalable, energy-efficient, and reliable platform for intelligent photovoltaic system monitoring, predictive maintenance, and smart renewable energy applications, making it suitable for residential, commercial, and remote solar installations.