AI-Based Digital Twin for Distribution Transformer Health Prediction and Remaining Useful Life Estimation Using an ESP32-Based Multi-Sensor Monitoring System

Abstract

Distribution transformers are essential components of power distribution systems, but continuous condition monitoring is often limited. This paper presents an AI-based digital twin for transformer health prediction and remaining useful life (RUL) estimation using an ESP32-based multisensor monitoring system. The prototype monitors AC voltage, load current, oil level, oil moisture, temperature, and vibration. Sensor readings are displayed on an LCD and transmitted via Wi-Fi to a Python-based digital twin. The digital twin combines a thermal ageing model with data-driven analysis to estimate the transformer health index, condition class, and RUL. The proposed low-cost system supports real-time monitoring, early fault detection, and predictive maintenance.