Weathering assessment approach for building sandstone using hyperspectral imaging technique

Abstract Weathering is one of the most common causes building sandstone damage. The evolution in various weathering behaviors critical for research. An intelligent assessment approach classifying degree a humid environment presented this study. This synthesis method relates to three parts: microscopic observation characteristics, hyperspectral acquisition weathered samples, and machine learning technology classification model. At first, process divided into initial stage, accelerated stable stage according mechanisms weathering. Secondly, novel proposed based on stage. Then, mapping relationship between characteristics image shedding samples can be established visible near-infrared spectral ranges (400–1000 nm) change law absorption feature. Next, data with different degrees are classified using Random Forest Furthermore, hyperparameters model optimized by Gray Wolf Optimizer algorithm better performance. trained finally applied evaluate large-scale walls quantitatively. whole worth recommending diagnosing monitoring sandstone.