A synthetic approach to weathering degree classification of stone relics case study of the Yungang Grottoes

• Published in: Heritage science 2018-01, Vol.6 (1), p.1-7, Article 1
• Main Authors: Meng, Tianhua, Lu, Yuhe, Zhao, Guozhong, Yang, Chengquan, Ren, Jianguang, Shi, Yunlong
• Format: Article
• Language: English
• Subjects: Chemistry and Materials Science; Classification; Least squares support vector machine; Materials Science; Regression models; Regressive prediction model; Research Article; Stone; Support vector machines; THz-TDS; Transmission coefficient; Ultrasonic; Ultrasonic testing; Weathering
• ISSN: 2050-7445; 2050-7445
• DOI: 10.1186/s40494-017-0165-y

Classification of the weathering degree of large outdoor stone relics has long been a challenge in their protection engineering. An approach to classifying the weathering levels of the Yungang Grottoes is presented in this paper. This approach is based on combined terahertz (THz) spectra and ultrasonic velocity and the use of a least squares support vector machine (LS-SVM). A regression model for prediction of the weathering level was established using the optimal values of parameters determined by the double cross-validation (D-CV) method. This SVM regression prediction model (SVMRPM) predicts the weathering levels of the grottoes with relative errors of 8.16% or less. The SVMRPM can also be used to predict the weathering levels of some areas where ultrasonic testing is extremely difficult; the THz spectra can easily be obtained using approximately 0.2 g of sample material. This method is a highly efficient and economical technique for determining the degree of weathering of large exposed stone relics.