铁路选线地质灾害时空危险性评估理论与方法研究

The increasing costs of construction, operation and maintenance, which are caused by the geological hazard founded during railway construction, are highly related to lacking of effective methods to quantitatively risk assess the potential geological hazard for railway line. The project is based on the time-space geographical hazard mapping theory, and explores the essential theory and method of geographical hazards in the railway line region from the macro level. Conducting the methods of numerical simulaiton on the historical data of typical precipitation and simulating the rainfall-time distribution of several typical precipitation, the most unfavorable rainfall process and critical rainfall of every precipitation process were achieved when applying the rainfall-time distribution function. Based on the mechanism of geological hazard and history hazards information, a regional geological hazard sensibility assessment index system was established. Establishing a regional geological hazard sensiblity mapping and conducting a reliability analysis after applying artificial intellegence, statistics modal and GIS spatial analysis technology. With the theory of cluster analysis, preliminary assessing the time-space risk level of potential geographical hazards along the railway line after conductting the spatial analysis of the geological hazard sensibility mapping and the railway track plan and selectting evaluation index which can qualify geographical hazards in the railway line region.The research achievement provides the preliminary, intuitive and quantitative assessment basic theory and evaluation tool of geological hazard level, and improves the efficiency and quailty of the railway selection.

铁路施工过程中发现不少地质灾害，增加工程造价和运营维护成本，这与铁路地质选线阶段尚缺乏有效手段定量评估潜在地质灾害对线路的危险性有直接关系。课题基于区域地质灾害时空敏感性区划理论，从宏观层面探索线域地质灾害时间-空间危险性评估的关键理论与方法。结合历史典型降雨过程的数值仿真，模拟若干典型降雨历时雨量-时间分布，得到每个雨量-时间分布函数下最不利降雨历程和临界雨量。根据地质灾害机理和历史灾害信息，建立区域地质灾害敏感性评价指标体系，基于人工智能、统计学模型和GIS空间分析技术编制区域地质灾害时空敏感性区划并进行可靠性分析。将时空敏感性区划与线路平面方案进行空间分析，优选量化线域地质灾害危险性评估指标，应用聚类分析理论，初步定量评估沿线潜在地质灾害时空危险性等级。研究成果为铁路地质选线提供初步、直观、定量的路段地质灾害危险程度评价基础理论和评价工具，提高地质选线效率与质量。

定量评估潜在地质灾害对铁路安全影响是地质选线工作的重要内容。本研究基于区域地质灾害时空敏感性区划理论，从宏观层面探索线域地质灾害时间-空间危险性评估的关键理论与方法。结合国内外文献和研究区域的实际情况，建立了地质灾害空间易发性区划指标体系，包括高程、地形地貌、坡度、与河流距离、与构造线距离、岩溶类型、年均降雨量、与铁路距离、与公路距离等10个致灾因子；构建了研究区域铁路线域地质灾害时空危险性评价基础信息数据库和地质灾害易发性区划模型库（包括主客观权重、Logistic 回归、信息量、支持向量机、BP 神经网络、自动编码器-Kmeans和深度信念网络等7个模型）；利用区划评价指标和模型库，四川省空间地质灾害易发性分为少发区、偶发区、易发区和极易发区四个等级；根据四川省52个气候观测站点长达20年的降雨量原始观测数据，开展区域雨量-时间模拟，编制了研究区域地质灾害时-空敏感性区划图；综合选取铁路路段地质灾害易发性、历史灾点对铁路影响度、工程施工对原始地貌的破坏程度和铁路的重要程度等4个指标18个二级指标，建立了铁路线域地质灾害时-空安全评价体系；以物元-可拓学、灰色关联度和支持向量机(SVM)等模型对川藏铁路（雅安-康定段）进行地质灾害时-空安全性等级评估。研究成果可用于四川省境内和川藏线沿线的地质灾害防治，并为以后的铁路地质选线提供一个完整的支撑体系。