近些年，我国开始实施城市更新行动，大批既有建筑面临改造更新，但由于我国既有建筑数量和规模十分庞大，仅靠人力检测存在困难，既有建筑安全检测与鉴定技术的自动化和可视化有待进一步发展。本文根据传统的建筑物检测与鉴定方法，结合三维激光扫描技术、点云数据处理技术和建筑信息模型技术 (BIM)等，提出了基于点云的既有建筑安全检测与鉴定方法，具体成果如下：
（1）通过对既有建筑安全检测与鉴定问题的研究，确立了建筑安全鉴定指标体系与鉴定流程，建立了以 BIM 技术为背景、以三维激光扫描技术为工具、以综合评估理论为支撑的整体框架；
（2）基于既有建筑安全检测与鉴定框架，首先利用编程软件 Python 和点云处理软件 Cyclone、Cloud Compare 对点云进行预处理，然后提出了一套基于建筑点云数据的缺陷特征提取方法，提取建筑的裂缝、挠度和倾斜等特征并获取其量化数据，同时，以测量距离和入射角为自变量进行室内实验，对点云裂缝特征提取的影响因素和精度进行了研究；
（3）基于点云数据以及获取的缺陷信息，利用建模软件 Revit 对既有建筑进行BIM 建模，并利用参数化建模插件 Dynamo 将缺陷信息集成到 BIM 模型中。在此基础上，在 Dynamo 中建立了基于模糊 AHP- DEMATEL-TOPSIS 理论的建筑安全鉴定数字化平台，并通过实际案例进行验证，实现了既有建筑安全鉴定的高效化、可视化以及科学化。
本文对基于点云的既有建筑缺陷特征提取及数字化安全鉴定进行了研究，提出了一种快速、可视化的鉴定方法，为类似工程的建设提供了新的参考。

关键词：既有建筑；安全检测与鉴定；建筑缺陷；特征提取；三维激光点云技术；建筑信息模型 (BIM)

In recent years, urban renewal actions have been initiated in China, requiring extensive renovations of existing buildings. However, due to the vast quantity and scale of existing buildings, manual inspection alone faces difficulties. The automation and visualization of safety detection and identification techniques for existing buildings need further development. This thesis proposed a point cloud-based method to detect and identify the safety for existing buildings, based on 3D laser scanning technology, point cloud data processing techniques, Building Information Modeling (BIM) technology and comprehensive assessment theory, etc. The specific achievements of this thesis include the following 3 aspects:
(1) Through research on the safety detection and identification for existing buildings, a system of building safety identification and a corresponding process were established. A comprehensive framework was developed based on BIM technology, 3D laser scanning technology, and comprehensive assessment theory.
(2) Based on the framework of safety detection and identification for existing buildings, the point cloud was first preprocessed through Python and point cloud processing software such as Cyclone and Cloud Compare. Then, a method for extracting the defect characteristics of existing building was proposed based on building point cloud data. With the help of this method, the quantitative data can be obtained by extracting the features such as cracks, deflections, and inclinations of existing builds. Additionally, indoor experiments with some independent variables including distance measurement and incident angle were conducted to study the potential and limitations of this method.
(3) Based on point cloud data and obtained defect data, the existing buildings were modeled through BIM method in modeling software Revit , and defect data was integrated into the BIM model using the parametric modeling plugin Dynamo. Besides, a digital identification platform for building safety levels was established in Dynamo. This platform was based on the fuzzy AHP-DEMATEL-TOPSIS theory, and achieved efficient, visual, and scientific safety identification of existing building.
This thesis conducts research on the extraction of existing building defect features and the digitalized identification of safety levels based on point clouds, proposing a rapid and visual identification method, providing new references for similar engineering projects.