公路工程

地形地质复杂作为山区公路的主要特点，使山区公路在建设中具有大量高陡边坡。高陡边坡易发生滑坡、崩塌、倾倒等危害，须采取针对性的防护措施以减少边坡危害的发生。然而，高陡边坡地质信息的获取，对边坡防护措施和边坡稳定性分析具有重要的指导意义。因此，利用马尔科夫随机场理论和贝叶斯机器学习技术，建立了马尔可夫随机场(MRF)地质模型。该MRF模型能够利用稀疏钻孔数据进行模拟并获取非钻孔处的地质信息，同时利用机器学习技术进行多次模拟迭代，通过分析迭代结果准确获取高陡边坡地质信息和量化地层不确定性。此外，基于模型提出了一种能够准确获取山区公路高陡边坡地质信息的钻孔分布方法。该方法以量化的地层不确定性为驱动，在实际工程中能够指导钻孔的设置和分布。研究结果表明，基于新建MRF模型的钻孔分布方法可以高效准确地获取地质信息，从而为公路工程提质增效。

Complex terrain and geology is the main feature of mountain road, which leads to a large number of high and steep slopes in the construction of mountain road. High and steep slope is prone to landslide, collapse, toppling and other hazards, so it is necessary to take specific protective measures to reduce the occurrence of slope hazards. However, the acquisition of geological information of high and steep slope has important guiding significance for slope protection measures and slope stability analysis. In this paper, Markov Random Field(MRF) geological model is established by using Markov Random Field theory and bayesian machine learning technique. The MRF model can use sparse borehole data to simulate and obtain geological information at non-borehole sites. At the same time, machine learning technology is used to conduct several simulation iterations, and by analyzing the iterative results, the geological information of high and steep slopes can be accurately obtained and the stratigraphic uncertainty can be quantified. In addition, based on the model, a borehole distribution method which can accurately obtain the geological information of high and steep slope of mountain road is proposed. Driven by quantified stratigraphic uncertainty, this method can guide the setting and distribution of boreholes in practical engineering. The results show that the borehole distribution method based on the new MRF model can obtain geological information efficiently and accurately, so as to improve the quality and efficiency of highway engineering.