摘要
隧道施工由于隐蔽性和未知因素多的施工及结构特性，而使得作业环境风险性大，施工条件极为恶劣。然而在我国加快基础设施建设的大背景与快节奏下，隧道的修建与施工难以避免的越来越多。受隧道施工过程中独特的因素影响，为保障施工安全，隧道的超前地质预报在隧道施工中就显得不可或缺。在隧道或采矿等地下洞室施工过程中，隐伏岩溶及其所造成的突泥涌水问题也日益突出。从2010年底通车的宜万铁路的马鹿箐隧道、野三关隧道到2014年宣布全线贯通的关角隧道再到2019年在建的云南安石隧道，这些穿越喀斯特地貌等复杂地质条件的长大隧道中，所隐藏在地表以下的隐伏岩溶所造成的突泥涌水严重威胁着施工人员的安全，加强超前地质预报探测的工作已刻不容缓。随着TSP等超前地质预报在国内技术的不断成熟，由于操作方法简便、占用时间少、对隧道施工干扰性小等特点，TSP这一方法在包括电法、地质雷达探测法以及红外探水技术等超前预报方法中出类拔萃，然而TSP及其所附带的专属解译系统TSPwin解译所得数据波速图由于存在特征不易识别，存在多解性等不足之处，本文针对其这一缺陷，在TSP初解译过程中，运用特定的波场分离、参数选取与设置方法，得出人工神经网络所需的原始数据图像，建立并将其导入人工神经网络模型试验，提取模型并识别预测，从而解决了TSP波速图特征不易识别、难以判译的问题，为未来TSP波速图的再解译提供了可行性。另外，本文认为隐伏岩溶探测和深度解译的最终目的均应服务于隐伏岩溶防治这一工程实际需求，为此，本文深入研究各类型隐伏岩溶并建立相关工程地质模型，通过参考其地质模型，来解释相关地质岩性及成因并与工程实际相结合，以期为岩溶防治提供超前控制与防治准备措施。牛萌萌
关键词：隐伏岩溶；TSP解译；人工神经网络；图形识别；工程地质模型；灾害防治
Abstract
Tunnel construction has the characteristics of large concealment, many unknown factors, harsh operating environment, and high risk. In recent years, more and more tunnels have been built under the background and rapid pace of China's accelerated infrastructure construction. However, due to the unique factors in the construction of the tunnel, the advance geological forecast of the tunnel becomes indispensable in the construction of the tunnel, so that the life and property safety of the tunnel constructors are properly protected. During the construction of underground tunnels such as tunnels or mining, the problem of hidden karst and the water and sand surge caud by it are also becoming increasingly prominent. From the Malujing Tunnel and the Yesanguan Tunnel of the Yiwan Railway, which opened to traffic at the end of 2010, to the Guanjiao Tunnel, which was announced to be open in 2014, and then to the Yunnan Anshi Tunnel under construction in 2019, the muddy gushing water caud by the hidden karst hidden below the ground in the long tunnels that pass through complex geological conditions such as karst landforms riously threatens the safety of construction workers. It is urgent to strengthen the exploration of advance geological forecast. With the advancement of advanced geological forecasting technologies such as TSP in China, due to the
simple operation method, less time occupation, and less interference with tunnel construction, the TSP method is outstanding in advanced forecasting methods including electrical methods, geological radar detection methods, and infrared water detection technologies. However, TSP and its accompanying exclusive interpretation system TSPwin's interpreted wave velocity diagrams have disadvantages such as difficulty in identifying the features and multi-resolution, etc. In view of this shortcoming, this article deeply rearches various types of hidden karsts and establishes related engineering geological models. By referring to its geological models, the relevant geological lithology and genesis are explained and combined with engineering practice. Using specific wave field paration, parameter lection and tting
methods, an artificial neural network model test is established, and the model is extracted for identification and prediction, thereby solving the problem that the TSP wave velocity map features are difficult to identify and interpret. It provides ideas for the reinterpretation of the TSP wave velocity map in the future and fills in the blanks in this field. In addition, this paper believes that the ultimate purpo of hidden karst detection and deep interpretation should rve the actual needs of the hidden karst prevention and control project. To this end, this article studies in-depth various types of hidden karst and establishes relevant engineering geological models. By referring to its geological m
odels, this article explains the relevant geological lithology and genesis and combines them with the actual engineering, in order to provide advanced control and prevention preparation measures for karst prevention .
Key words: hidden karst, TSP interpretation, artificial neural network, pattern recognition, engineering geological model, disaster prevention
目录
第一章绪论 (1)
1.1选题背景和意义 (1)
1.1.1 隐伏岩溶存在的普遍性 (1)
1.1.2 隐伏岩溶的危害 (5)
1.1.3 对TSP探测图形进行再解译的必要性 (7)
1.2国内外研究现状 (17)
1.2.1 隐伏岩溶的研究现状 (17)
1.2.2 图形识别与分类的研究现状 (19)
1.2.3 TSP图形解译技术的研究现状 (21)
1.2.4 TSP解译中存在的问题 (21)
1.2.5 针对TSP探测隐伏岩溶解译过程中所得图形特征不易识别的研究 24 1.3本章小结 (25)
第二章TSP超前预报的初解译 (27)
2.1工程地质概况 (27)
2.2预报数据采集 (28)
2.3数据的提取与分离 (29)
2.3.1 TSPwin有效反射波波场分离 (29)
2.3.2 TSPwin深度偏移图提取 (32)
水仙花花语
2.3.3 TSPwin反射面提取 (33)
2.3.4 TSPwin 的2D成果提取 (34)
满分作文网2.4本章小结 (35)
第三章TSP图形识别分类方法与选择 (36)
3.1图像识别分类基本概念 (36)
3.2主要方法概述 (36)
3.2.1 基于神经网络的图像识别与分类 (36)
3.2.2 基于小波矩的图像识别与分类 (37)
3.2.3 基于分形特征的红外热成像图的识别与分类 (38)
3.3常见的深度学习技术和神经网络模型 (38)
3.3.1 多层感知机 (39)
3.3.2 卷积神经网络 (40)
3.3.3 循环神经网络 (41)
3.3.4 对抗式生成网络 (42)
3.3.5 区域卷积神经网络 (43)
双十一活动宣传
3.4TSP图像识别方法的选择 (43)
3.5本章小结 (44)
第四章TSP图形的再解译技术 (45)
4.1基于T ENSOR F LOW框架的深度学习TSP波形图识别算法 (45)
4.1.1 卷积神经网络的TensorFlow框架 (46)
4.1.2 基于TensorFlow的TSP波速图识别算法的实现 (48)
左肾结晶
4.2隐伏岩溶的图谱分析验证 (67)
4.3隐伏岩溶的TSP再解译应用 (68)
4.4 本章小结 (69)
第五章隐伏岩溶的工程地质特征及其综合防治 (70)
5.1一般岩溶工程地质特征分类 (70)
5.2背斜接触带型岩溶 (71)
5.2.1 背斜接触带型岩溶特征 (71)
5.2.2 工程实例分析——云雾山隧道 (72)
5.2.3 工程地质模型的建立 (72)
5.2.4 TSP的探测解译与再解译 (73)大自然的画
5.3向斜承压水型岩溶 (74)
门可罗雀是什么意思5.3.1 向斜承压水型型岩溶特征 (74)
会议服务礼仪培训
5.3.2 工程实例分析——圆梁山隧道 (75)
5.3.3 工程地质模型的建立 (76)
5.3.4 向斜承压水型岩溶的防治与治理——以圆梁山隧道为例 (77)
5.4节理密集带型岩溶 (78)
5.4.1 节理密集带型岩溶特征 (78)
5.4.2 工程实例分析——黄草岭隧道 (79)
5.4.3 工程地质模型的建立 (79)
5.4.4 TSP的探测解译与再解译 (81)
5.5断层破碎带型岩溶 (82)
5.5.1 断层破碎带型岩溶特征 (82)
5.5.2 工程实例分析——关角隧道 (82)
5.5.3 工程地质模型的建立 (83)
