Post-fire assessment of heating temperatures experienced by concrete using short video imaging, hyperspectral imaging and laser-induced breakdown spectroscopy

This study develops rapid post-fire analysis methods to predict the heating temperature to which the concrete was exposed. Short video imaging (SVI), hyperspectral imaging (HSI) and laser-induced breakdown spectroscopy (LIBS) were used for the first time to obtain spectra of concrete after high-temperature exposure (100–800 °C). The differences in colour levels and spectroscopic signals due to varying temperatures were observed. To handle the complex relationship between spectra and temperatures, machine learning models were used to extract meaningful information from spectra for the quantification of temperature. Furthermore, domain knowledge related to concrete composition and LIBS signal was integrated into machine learning to improve quantification performance. The highest coefficients of determination for prediction achieved based on SVI, HSI and LIBS measurements were 91.4, 94.8 and 98.6, respectively. These results demonstrate that SVI, HSI and LIBS can be fast and viable methods for assessing the fire temperature that the concrete has experienced.