Optimization and validation of a non-intrusive indoor temperature measurement method for buildings using UAV-based infrared thermography

Efficient control of building heating systems relies on accurate and spatially comprehensive indoor temperature monitoring. However, conventional sensor deployments suffer from limited spatial coverage, high cost, and occupant intrusiveness. This study proposes a non-intrusive indoor temperature estimation method using unmanned aerial vehicle-based infrared thermography (UAV-IRT). Firstly, optimal UAV measurement parameters—including solar radiation, wind speed, UAV-to-window distance, and measurement location—were systematically determined. Subsequently, representative reference rooms within apartment units were identified based on orientation and room size, revealing that south-facing large rooms provided the highest inference accuracy. Principal component analysis was further applied to determine baseline user units, demonstrating that horizontal temperature variability was minimal within floors, whereas significant vertical temperature gradients required at least one representative user per floor. Field experiments validated that the proposed method, relying on infrared measurements from only 8 % of the building's rooms, achieved a mean absolute error of 0.70 °C, with over 95 % of room-level predictions falling within ±2 °C. This study provides a practical and efficient monitoring solution, offering significant potential for scalable indoor environmental assessment and intelligent heating management in buildings.