随着地理信息系统（Geographic Information System，GIS）技术已经在城市建设的规划管理、土地管理、管线管理等行业都有大量成功的应用，该技术也开始应用在铁路建设、铁路规划和铁路运营管理中，并有效地提高了高速铁路规划、建设与管理的决策科学化、管理现代化，保障基础设施发挥最大的经济效益。

当前我国高速铁路的建设规模逐年增大，设施建设发展迅速，在全国区域内已形成大规模的、错综复杂的基础设施系统。而高速铁路的建设距离长、范围广、地形复杂多变，使用倾斜摄影测量技术，利用无人机可快速获取三维地理空间信息。将GIS技术、计算机技术和测绘基础的行业规范结合起来，突出三维模型管理的便利。当前在高速铁路的可视化研究当中，大多采用建筑信息模型（Building Information Modeling，BIM）而运用倾斜摄影测量技术的研究相对较少。为此，本文以高速铁路路基沉降监测数据与高速铁路区段内的倾斜摄影数据作为研究的实验数据，采用B/S架构以及Cesium三维可视化框架，创建基于倾斜摄影模型的三维可视化平台，并且实现三维模型的可视化分析，提高铁路建设及管理过程中的预测分析能力和数据管理效率。本文研究的主要内容以及研究成果包括：

（1）主要介绍了路基的结构，由路基的结构总结分析沉降产生的原因。针对路基沉降的监测方案编制，从监测网的建立、监测点的布设以及评估控制指标3个方面说明了路基沉降监测的方法。

（2）论述实景三维模型构建的过程。针对长距离倾斜摄影模型中地球曲率对模型的影响，研究倾斜摄影建模流程，并基于ICP（lterative Closest Point）算法对点云模型进行拼接。针对Web中三维模型加载速度慢的问题，基于QEM（Quadric Error Metrics）算法对倾斜摄影模型进行网格优化，减少模型的占用内存，提升了模型加载速度。改进实景三维模型的服务发布方法，加快数据加载速度并且提升数据的安全性。

（3）基于福厦铁路某区段内路基沉降监测数据，分别使用规范双曲模型、指数曲线模型、灰色预测模型、多项式拟合法以及BP神经网络进行路基沉降预测分析，以观测标0178200L1为算例并对预测精度进行研究。针对路基沉降特点，改进了BP神经网络分析算法中的激活函数，有效提升拟合相关系数以及预测精度。其中Sigmoid函数适合处理二元分类问题，ReLU函数在输入值小于0时会出现失效问题。相较于Sigmoid函数与ReLU函数对输入值有较高要求，Swish函数可更好的对多类型数据值进行拟合预测。

（4）简要总结了WebGIS（Web Geographic Information System，网络地理信息系统）及Web客户端开发技术，重点介绍Cesium开源地球平台和框架，分析虚拟场景的渲染机制的原理，研究使用低分辨率加载模型提高模型加载速度。基于Cesium框架搭建三维WebGIS可视化平台，采用B/S架构实现了高速铁路数据可视化应用。可视化平台包括信息查询模块、预测分析模块和三维分析共三个模块。信息查询模块包含了高速铁路变形监测中常用的监测信息；预测分析模块将多种预测方法与B/S系统相结合，实现了对沉降监测数据进行预测分析功能，使用可视化界面对预测结果进行可视化分析。而三维分析模块包括辅助三维GIS分析功能：淹没分析功能、剖面分析功能、光照分析功能、通视分析功能、可视阈分析功能以及飞行轨迹设计功能等。

本文的研究成果针对高速铁路及其相关的变形监测工程项目，提供了一套三维可视化信息管理系统。该系统具备信息管理、查询和可视化分析等多种功能，以便于工程师和决策者更直观地了解项目的实时动态信息和关键数据。此外，本研究还能为规划和管理过程中的科学决策和技术支持提供有力保障，从而提高项目的可控性和实施效率。

As Geographic Information System (GIS) technology has been successfully applied in urban construction planning management, land management, pipeline management and other industries, the technology has also begun to be applied in railway construction, railway planning and railway operation management. And effectively improve the planning, construction and management of high-speed railway decision-making scientific, management modernization, to ensure that the infrastructure to maximize economic benefits.

At present, the construction scale of high-speed railway in China is increasing year by year, and the construction of facilities is developing rapidly, and a large-scale and complicated infrastructure system has been formed in the whole country. However, the construction distance of high-speed railway is long, the range is wide, the terrain is complex and changeable, the tilt photogrammetry technology is used, and the three-dimensional geospatial information can be quickly obtained by using UAV. It combines GIS technology, computer technology and industry norms of surveying and mapping foundation to highlight the convenience of 3D model management. At present, in the visualization research of high-speed railway, Building Information Modeling (BIM) is mostly used, and the research using tilt photogrammetry is relatively few. To this end, this paper takes the high-speed railway subgrade settlement monitoring data and the tilt photography data in the high-speed railway section as the experimental data for research, adopts B/S architecture and Cesium 3D visualization framework to create a 3D visualization platform based on the tilt photography model, and realizes the visual analysis of the 3D model. Improve the forecasting and analysis ability and data management efficiency in the process of railway construction and management. The main contents and research results of this paper include:

(1)This paper mainly introduces the structure of roadbed, and summarizes and analyzes the cause of settlement. Aiming at the monitoring scheme of subgrade settlement, the methods of subgrade settlement monitoring are explained from three aspects: the establishment of monitoring network, the layout of monitoring points and the evaluation and control index.

(2)This paper discusses the process of constructing 3D model of real scene. Aiming at the influence of earth curvature on long-distance oblique photography model, the modeling process of oblique photography is studied, and the Point cloud model is spliced based on ICP (lterative Closest Point) algorithm. To solve the problem of slow loading speed of 3D model in Web, grid optimization of oblique photography model was carried out based on Quadric Error Metrics (QEM) algorithm, which reduced the memory consumption of the model and improved the loading speed of the model. Improve the service publishing method of realistic 3D model, speed up data loading speed and improve data security.

(3)Based on the monitoring data of subgrade settlement in a section of Fuzhou-Xiamen Railway, the normal hyperbolic model, exponential curve model, grey prediction model, polynomial fitting method and BP neural network are used to predict the subgrade settlement. The observation standard 0178200L1 is taken as an example and the prediction accuracy is studied. According to the characteristics of subgrade settlement, the activation function of BP neural network analysis algorithm is improved to effectively improve the fitting correlation coefficient and prediction accuracy. The Sigmoid function is suitable for binary classification, and the ReLU function will fail when the input value is less than 0. Compared with Sigmoid function and ReLU function, which have higher requirements on input values, Swish function can better fit and predict multi-type data values.

(4)This paper briefly summarizes WebGIS (Web Geographic Information System) and Web client development technology, focuses on Cesium open source earth platform and framework, and analyzes the principle of virtual scene rendering mechanism. The low resolution loading model is used to improve the loading speed of the model. The 3D WebGIS visualization platform is built based on Cesium framework, and the high-speed railway data visualization application is realized by using B/S architecture. The visualization platform includes three modules: information query module, prediction analysis module and 3D analysis module. The information query module contains the monitoring information commonly used in high-speed railway deformation monitoring; The prediction and analysis module combines a variety of prediction methods with B/S system to realize the prediction and analysis function of settlement monitoring data, and uses the visual interface to visualize the prediction results. The 3D analysis module includes auxiliary 3D GIS analysis functions, such as flood analysis, profile analysis, light analysis, visibility analysis, visual threshold analysis and flight path design.

The research results of this paper provide a set of 3D visualization information management system for high-speed railway and its related deformation monitoring engineering projects. The system has multiple functions such as information management, query and visual analysis, so that engineers and decision makers can more intuitively understand the real-time dynamic and key data of the project. In addition, this study can provide a strong guarantee for scientific decision-making and technical support in the planning and management process, so as to improve the project control and implementation efficiency.

[1]熊嘉阳,沈志云. 中国高速铁路的崛起和今后的发展[J].交通运输工程学报,2021,21(05):6-29.

[2]王爽. 高速铁路防灾安全智能监测单元的研究与实现[D]. 兰州交通大学, 2014.

[3]张泉. 高速铁路货运动车组开行可行性研究[D].华东交通大学,2017.

[4]张思远. (京沪)高速铁路路基沉降预测方法适用性研究[D]. 西南交通大学, 2012.

[5]张宇昕. 运营高速铁路监测数据评估管理与可视化研究[D].西南交通大学,2019.

[6]徐佳沅,文薪荐,王思敏等. 一站式地理大数据智能化平台构建[J].测绘通报,2020,No.525(12):132-137.

[7]王兴玲. 基于XML的地理信息Web服务研究[D].中国科学院研究生院（遥感应用研究所）,2002.

[8]徐立,孙群,徐明世等. 一种基于信息层次的地图投影选择方法[J].测绘工程,2019,28(02):41-44+51.

[9]杨凡,方爱平,魏立茹. 地理信息产品的信息安全要求及认证模式研究[J].信息安全研究,2019,5(05):430-433.

[10]郭啟倩. 基于B/S的地震分析预报信息管理与服务系统设计与实现[D]. 中国地震局地震研究所, 2019.

[11]魏军,魏铁鑫,陈莎,等. 基于GIS的石家庄市暴雨内涝数学模型与应用[J].干旱气象,2018,36(04):701-708.

[12]林雄. 汕头三维仿真网络城市系统的设计与应用[D].电子科技大学,2011.

[13]薛明,刘春,肖学年. 基于WebGIS的复杂城市交通网络信息服务[J].测绘通报,2011,No.417(12):64-68.

[14]戴瑶,段增强,艾东. 基于GeoServer的国土空间规划野外调查辅助平台搭建与应用[J]. 测绘通报, No.526(01):121-123+147.

[15]朱骏侠.基于WebGL的三维智慧管网系统的设计与实现[J].测绘通报,2021,No.529(04):146-150.

[16]郭泽,谭衢霖,秦晓春等.基于Cesium的铁路线路三维场景空间信息服务[J].测绘通报,2022,No.548(11):144-148.

[17]黄欣然,江文萍,刘一飞,等.Cesium环境中三维场景数据的加载优化策略[J].测绘地理信息,2022,47(S1):94-98.

[18]李义先. 关于倾斜摄影测量技术的应用及发展前景[J].安徽地质,2017,27(04):305-307.

[19]Kearns F R, Kelly M, Tuxen K A. Everything happens somewhere: using webGIS as a tool for sustainable natural resource management[J]. Frontiers in Ecology and the Environment, 2003, 1(10): 541-548.

[20]Werts J D, Mikhailova E A, Post C J, et al. An integrated WebGIS framework for volunteered geographic information and social media in soil and water conservation[J]. Environmental management, 2012, 49: 816-832.

[21]N. Nizamuddin, I. Meutia and A. Ardiansyah, "Development of WebGIS Application for Updating Spatial Data of Paddy Fields," 2018 International Conference on Electrical Engineering and Informatics (ICELTICs), Banda Aceh, Indonesia, 2018, pp. 139-144

[22]X. Ma, X. Zhang and L. Shi. "Research on the Algorithm of Building Object Boundary Extraction Based on Oblique Photographic Model," 2018 IEEE 3rd Advanced Information Technology, Electronic and Automation Control Conference (IAEAC), Chongqing, China, 2018, pp. 1909-1912.

[23]杨国东,王民水. 倾斜摄影测量技术应用及展望[J].测绘与空间地理信息,2016,39(01):13-15+18.

[24]曾政祥,何晶晶,张婧.融合激光点云的城市级高精度建模技术[J].测绘通报,2023,No.552(03):133-138.

[25]马洪成,张玉驹,刘为民. 基于开源Cesium框架的智慧街道三维可视化平台的研究与应用[J].测绘与空间地理信息,2019,42(08):121-123.

[26]卢秀山,邢恺强,刘如飞,等. 一种三维激光点云中建筑物立面渐进分割方法[J].测绘科学,2019,44(12):7-13.

[27]刘洋. 无人机倾斜摄影测量影像处理与三维建模的研究[D].东华理工大学,2016.

[28]王京,郭瑞隆,马明义. 基于无人机倾斜摄影测量的三维重建及精度分析[J].测绘标准化,2017,33(03):32-33.

[29]周晓敏,孟晓林,张雪萍,等.倾斜摄影测量的城市真三维模型构建方法[J].测绘科学,2016,41(09):159-163.

[30]谭仁春,李鹏鹏,文琳等. 无人机倾斜摄影的城市三维建模方法优化[J].测绘通报,2016,No.476(11):39-42.

[31]魏铼,胡卓玮,陈天博,等. 单机倾斜摄影方式的无人机航线设计[J].测绘科学,2018,43(06):147-155.

[32]A. Kamal, A. Y. Javaid, V. K. Devabhaktuni, D. Kaur, J. Zaientz and R. Marinier, "A Novel Approach to Air Corridor Estimation and Visualization for Autonomous Multi-UAV Flights," 2020 IEEE International Conference on Electro Information Technology (EIT), Chicago, IL, USA, 2020, pp. 370-377.

[33]Boulos M K , Al-Shorbaji N M . On the Internet of Things, smart cities and the WHO Healthy Cities[J]. International Journal of Health Geographics, 2014, 13(1):10.

[34]Craglia M , Ostermann F , Spinsanti L . Digital Earth from vision to practice: making sense of citizen-generated content[J]. International Journal of Digital Earth, 2012, 5(5):398-416.

[35]李红霞,赵新华,迟海燕,等.基于改进BP神经网络模型的地面沉降预测及分析[J].天津大学学报,2009,42(01):60-64.

[36]马锦,张浩生.基于GIS空间分析的山地城市弃土场选址研究[J].测绘通报,2021(S2):214-217.

[37]冯绍权, 花向红, 陶武勇,等.一种基于GA-BP-MC神经网络的高铁桥墩沉降预测模型[J]. 测绘通报,No.508(07):50-53+82..

[38]乐世华,张煦,张尚弘等. 基于Cesium的WebGIS流域虚拟场景搭建[J].水利水电技术,2018,49(05):90-96.

[39]周贺. 基于卷积神经网络的脱机手写汉字识别与目标检测研究[D].燕山大学,2020.

[40]郭敏钢,宫鹤. 基于Tensorflow对卷积神经网络的优化研究[J].计算机工程与应用,2020,56(01):158-164.

[41]谈琼,张涛,丁乐乐,等.基于Cesium的多源三维数据可视化与分析平台设计与开发[J].测绘与空间地理信息,2022,45(06):120-123.

[42]李小敏. WebGIS三维实景地图在城市规划管理中的应用研究[D].西安科技大学,2019.

[43]罗杨. 基于微服务框架的地理空间数据服务平台设计与实现技术研究[D].西安电子科技大学,2019.

[44]孟令奎,史文中,张鹏林. 网络地理信息系统原理与技术[M]. 科学出版社, 2005.

[45]杜鹏,屈帅. 基于WebGL技术的网络三维地理信息系统平台[J].科技资讯,2015,13(30):26+72.

[46]张福友. 基于大疆无人机的Context Capture三维实景建模探讨[J].广东水利水电,2017,No.258(08):88-91+103.

[47]张文刚. 基于数字化校园综合安防管理系统的设计[D].西北工业大学,2007.

[48]王京,郭瑞隆,马明义. 基于无人机倾斜摄影测量的三维重建及精度分析[J].测绘标准化,2017,33(03):32-33.

[49]何厚学. 银西高速铁路路基沉降监测数据的预处理与预测方法研究[D].兰州交通大学,2020.

[50]彭杰. 基于切片地图Web服务的地理信息发布技术研究[D]. 浙江大学, 2011.

[51]李陈. 基于TensorFlow的卷积神经网络优化研究[D].华中科技大学,2021.

[52]田娟,李英祥,李彤岩. 激活函数在卷积神经网络中的对比研究[J].计算机系统应用,2018,27(07):43-49.

[53]李泉洲. 基于Cesium的倾斜摄影三维模型单体化研究与实现[D]. 长安大学, 2019.

[54]王宇,邢希凯,赵胜楠. 基于无人机倾斜摄影三维模型构建及网页发布[J].地理空间信息,2022,20(06):17-19+23.

[55]Larsen K W, DeWolfe A W, Nguyen D, et al. 3D Planetary Data Visualization with CesiumJS[C]//AGU Fall Meeting Abstracts. 2017, 2017: IN31D-04.

[56]Petrie G. Systematic oblique aerial photography using multiple digital cameras[J]. Photogrammetric Engineering & Remote Sensing, 2009, 75(2): 102-107.

[57]Zhang R, Li H, Duan K, et al. Automatic detection of earthquake-damaged buildings by integrating UAV oblique photography and infrared thermal imaging[J]. Remote Sensing, 2020, 12(16): 2621.

[58]Salvati P , Balducci V , Bianchi C , et al. A WebGIS for the dissemination of information on historical landslides and floods in Umbria, Italy[J]. Geoinformatica, 2009, 13(3):305-322.

[59]Huang J , Huang R , Ju N , et al. 3D WebGIS-based platform for debris flow early warning: A case study[J]. Engineering Geology, 2015, 197:57-66.

[60]Zhang Y , Wu L . Stock market prediction of S&P 500 via combination of improved BCO approach and BP neural network[J]. Expert Systems with Applications, 2009, 36(5):8849-8854.

[61]赵宏,杜明成,吴俐民,等.基于倾斜摄影测量技术的智慧城市5D产品制作工艺实现[J].测绘工程,2016,25(09):73-76.

[62]黄丽. BP神经网络算法改进及应用研究[D]. 重庆师范大学, 2010.

[63]陈善雄. 高速铁路沉降变形观测评估理论与实践[M]. 中国铁道出版社, 2010.

[64]叶茂. 京沪高速铁路沉降监测数据处理与分析[D]. 西南交通大学, 2011.

[65]王明慧, 张桥. 基于GM(1,1)模型群的高速铁路路基沉降预测研究[J]. 数学的实践与认识, 2013, 43(19):108-117.

[66]汪延彬. B/S架构京沪高速铁路沉降信息共享平台的研究与实践[D]. 西南交通大学, 2010.

[67]张献州,莫春,马下平.高速铁路沉降观测评估预测系统设计与实现[J].铁道勘察,2010,36(05):1-4.

[68]黄泽纯, 张献州, 汪延彬,等. 京沪高速铁路沉降观测数据管理系统研究[J]. 中国铁路, 2009(12):52-55.

[69]许双安,任晓春,袁永信.高速铁路沉降数据处理与评估系统设计与实现[J].测绘地理信息,2014,39(04):54-57.

[70]朱栩逸,苗放.基于Cesium的三维WebGIS研究及开发[J].科技创新导报,2015,12(34):9-11+16.

[71]王逸凯, 徐萌, 罗建松,等. 基于Cesium的WebGIS倾斜三维平台的实现[J]. 测绘与空间地理信息, 2019, 42(4):2.

[72]王晓艳,陈宪冬.基于Cesium的三维模型可视化[J].地理空间信息,2022,20(01):115-117+9.

[73]许星. 三维模型网格简化技术研究[D].华中科技大学,2019.

[74]侯宝明,刘雪娜,鄂旭.基于QEM以方差作为权值的3D网格模型简化[J].计算机技术与发展,2013,23(11):50-53+58.

[75]李红波,刘昱晟,吴渝等.基于二次误差度量的大型网格模型简化算法[J].计算机工程与设计,2013,34(09):3158-3162.

[76]解则晓,徐尚.三维点云数据拼接中ICP及其改进算法综述[J].中国海洋大学学报(自然科学版),2010,40(01):99-103.

[77]黄源,达飞鹏,陶海跻.一种基于特征提取的点云自动配准算法[J].中国激光,2015,42(03):250-256.

[78]吕颖,陈志明,杨天天,等.基于改进特征匹配的点云数据无缝拼接研究[J].计算机仿真,2020,37(10):200-205.

[79]王丽辉.三维点云数据处理的技术研究[D].北京交通大学,2011.

[80]孙晓鹏,李华.三维网格模型的分割及应用技术综述[J].计算机辅助设计与图形学学报,2005(08):1647-1655.

[81]邱春霞,董乾坤,刘明.倾斜影像的三维模型构建与模型优化[J].测绘通报,2017,No.482(05):31-35.

[82]侯沛宏.三维模型表面纹理的无缝拼接技术研究[D].中国科学院大学(中国科学院工程管理与信息技术学院),2017.

[83]黄扬子.基于NodeJS平台搭建REST风格Web服务[J].无线互联科技,2015,No.68(16):57-59.

[84]李大松. 基于NodeJS与云存储的文件管理平台的设计与实现[D].华中科技大学,2015.

[85]刘爱华,韩勇,张小垒,等.基于WebGL技术的网络三维可视化研究与实现[J].地理空间信息,2012,10(05):79-81+7.

[86]张帆. 基于WebGL的大规模三维场景可视化技术研究[D].国防科学技术大学,2014.