Non-destructive detection of surface damage in hollow gravity type concrete dam by using passive infrared thermography data with deep learning method

For large-scale infrastructure, such as concrete dams, a non-destructive and non-contact testing method is essential for estimating subsurface damage.Passive infrared thermography is an effective technique for detecting this type of damage, relying on heat capacity differences within the concrete.Since surface temperature distribution irregularities correlate with crack propagation and material degradation, passive infrared thermography can be applied to assess damage progression in concrete structures.This study investigates the relationship between thermal responses and subsurface damage using a combination of heat balance analysis and deep learning models, contributing to the field of fracture mechanics and structural integrity assessment.In this study, the thermal behavior of a concrete dam is analyzed using heat balance modeling and Long Short-Term Memory (LSTM) to assess subsurface damage.Measured and predicted surface temperatures at exudation and non-damaged points are compared to investigating the correlation between thermal distribution irregularities and exudation.The results indicate that a higher estimation accuracy is achieved by LSTM compared to heat balance analysis, as shadow effects, latent heat transfer and moisture dynamics are implicitly considered-factors affecting surface temperature distribution.Through these analyses, the potential of passive infrared thermography is validated as a nondestructive method for monitoring and assessing structural integrity in large-scale concrete infrastructure.