An adaptive prediction method for mechanical properties deterioration of sandstone under freeze–thaw cycles: a case study of Yungang Grottoes

Researchers have identified that the number of freeze–thaw cycles is the dominant factor influencing the degradation of sandstone mechanical properties at UNESCO World Heritage Site Yungang Grottoes, with porosity playing a minor but notable role. The study introduces an adaptive multi-output gradient boosting decision trees (AMGBDT) algorithm, which outperforms traditional models like neural networks and random forests in predicting how these factors affect rock strength and elasticity. Unlike conventional methods, AMGBDT adapts to nonlinear relationships between physical properties and degradation patterns, offering precise insights into the failure mechanisms of sandstone under freeze–thaw conditions. This approach enhances understanding of environmental impacts on cultural heritage preservation.

Liu, C., Liu, Y., Ren, W., Xu, W., Cai, S., & Wang, J. (2021). An adaptive prediction method for mechanical properties deterioration of sandstone under freeze–thaw cycles: a case study of Yungang Grottoes. Heritage Science, 9(1). https://doi.org/10.1186/s40494-021-00628-8