隧道不良地质识别: 方法、现状及智能化发展方向

随着隧道施工对于不良地质识别精度要求的不断提高以及人工智能技术的发展, 融合多源信息的不良地质智能化识别已成为发展趋势。本文首先阐述了常见的6种隧道不良地质类型及其地质成因, 回顾分析了隧道主要的不良地质识别方法及现状, 详细介绍了笔者在不良地质智能化识别方面的探索性研究: 基于机器学习利用图像识别技术对隧道围岩岩性与裂隙特征进行智能识别; 融合图像和光谱特征进行不良地质识别; 将地化分析融入到传统的超前钻探中, 融合随钻参数和地化信息进行不良地质随钻识别, 既可以发挥超前钻探在感知岩体质量和地层信息变化方面的优势, 又可以发挥地化分析在岩性和不良地质异常识别方面的优势; 基于地质与物探联合反演进行不良地质识别, 旨在实现掌子面前方不良地质体“形”(位置、形态、规模)和“性”(性质和类型)的精确识别。最后, 对隧道不良地质智能化识别的发展趋势进行了展望。

Adverse Geology Identification in Tunnel: Method, Research Status and Intelligent Development Direction

With the continuous improvement of the accuracy requirements for adverse geology identification in tunnel construction and the development of the artificial intelligence technology, the adverse geology intelligent identification with multi-source information has become a development trend. This study illustrated six common types of adverse geology in tunnels and their geological causes, review the main identification methods and current situation of adverse geology in tunnels, and introduce in detail the exploratory research on intelligent identification of adverse geology. Our research includes: (1) image identification technology is used to identify the lithology and fracture characteristics of tunnel surrounding rock intelligently; (2) image and spectral features fusion method is used to identify adverse geology; (3) geochemical analysis integrated into the traditional advanced drilling is used to identify adverse geology by integrating drilling parameters and geochemical information, which can not only give play to advantages of advanced drilling in perceiving changes of rock mass quality and stratum information, but also give play to advantages of geochemical analysis in identifying lithology and adverse geological anomaly; (4) the joint inversion of geology and geophysical exploration is used to identify the adverse geology, which aims to realize the accurate identification of the “shape” (position, shape, and scale) and “nature” (character and type) of unfavorable geological bodies in front of the working face. Finally, we prospect the development trend of intelligent identification of tunnel adverse geology.