Hybrid transformer overload protection scheme integrating real-time thermal imaging and numerical relay coordination

Abstract

Transformer thermal overload is a major threat to power system reliability, leading to insulation degradation, reduced service life, and unplanned outages. Conventional overload protection using numerical relays, such as the GE Multilin D60, relies primarily on current-based IEC 255–8 thermal models, which cannot directly monitor real-time temperature conditions and may delay fault detection. This study proposes a hybrid protection scheme in which real-time thermal imaging is used as the primary decision criterion, while the D60 numerical relay functions as a back-up overcurrent protection device. Surface temperature is measured using a FLIR E63900 infrared camera, and a Python-based image processing module with OpenCV and Tesseract OCR extracts temperature values from thermal images. These are processed through a heat transfer model to estimate internal core temperature. A Hybrid Protection Decision Controller (HPDC) receives both thermal and current data via Modbus TCP and issues a trip signal if either the thermal threshold is exceeded or the back-up relay detects overcurrent. Local alarms are triggered immediately upon thermal violations, and all events are logged to SCADA/EnerVista for diagnostics. Experimental validation on a 3 kVA single-phase transformer under varying load conditions demonstrated faster detection of overload conditions compared to relay-only protection, enabling proactive isolation and reducing the risk of insulation failure. The proposed scheme entails higher implementation costs than conventional relay-only solutions owing to the inclusion of thermal imaging hardware and advanced data processing systems; however, it remains a justified investment for critical transformers where the potential consequences of failure significantly outweigh the additional expenditure.