CN-122023692-A - Unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance auxiliary method

CN122023692ACN 122023692 ACN122023692 ACN 122023692ACN-122023692-A

Abstract

The invention discloses a dangerous large project dynamic acceptance auxiliary method for unmanned aerial vehicle oblique photography and BIM fusion, which particularly relates to the technical field of intelligent construction informatization, and aims to realize synchronous acquisition of multi-source data by collecting multi-view images and laser point cloud data of a dangerous large project area through an unmanned aerial vehicle carrying an RTK-PPK dual-mode positioning system, a five-lens oblique photography camera and a laser radar, automatically extract edge endpoint characteristics of a BIM model and point cloud data through an improved ICP-NDT algorithm to perform coarse registration, realize sub-pixel level fine registration through iterative optimization, introduce a digital twin model, compare BIM design parameters with measured data through a multi-criterion acceptance decision tree based on fuzzy logic, automatically generate an acceptance report, return key data to a cloud platform through a 5G network, solve the problem of model deletion of an occlusion area through multi-source data fusion, ensure the integrity of a three-dimensional model, realize millisecond level defect identification through edge intelligent decision, and improve the safety hidden danger discovery timeliness.

Inventors

Qin Kaiwen
PAN WENJUN
YUAN YONG
ZHOU ZIDONG
LI FULONG
WU SHUANG
ZHANG MAOPENG
Cheng Zhennian
HU LONGBO
CAO YANGYANG
ZHOU XIONG
CHEN PEIJI
YE CHUAN
LI YANCHUN

Assignees

中铁广州工程局集团有限公司
中铁广州工程局集团第三工程有限公司

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (10)

1. The unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance assisting method is characterized by comprising the following steps of: S1, acquiring multi-source data of a dangerous large engineering area through an unmanned aerial vehicle carrying an RTK-PPK dual-mode positioning system, a five-lens oblique photographic camera and a laser radar, wherein the multi-source data comprises multi-view images and laser point cloud data, so that ground data are synchronously acquired; s2, carrying out fusion processing on the multi-source data to construct a dynamically updated digital twin model; s3, the unmanned aerial vehicle is further provided with an onboard edge computing unit, the onboard edge computing unit is provided with a lightweight CNN model, real-time defect identification is carried out on the multi-source data after registration, and the defect identification comprises concrete cracks, steel bar spacing deviation and formwork deformation; S4, comparing BIM design parameters with measured data, automatically generating an acceptance report based on a fuzzy logic-based multi-criterion acceptance decision tree, and returning key data to the cloud platform through a 5G network.
2. The unmanned aerial vehicle oblique photography and BIM fusion critical engineering dynamic acceptance assisting method of claim 1, wherein in step S2, the fusion processing comprises the following steps: a1, automatically extracting edge endpoint features of a BIM model and point cloud data by the multisource data through an improved ICP-NDT algorithm to perform coarse registration, and then realizing sub-pixel level fine registration through iterative optimization; A2, introducing Kalman filtering to fuse GNSS, IMU, laser radar and image data, and constructing the digital twin model.
3. The unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance assisting method of claim 1 is characterized in that in the step S1, the specific mode of multi-scale data acquisition comprises the steps of synchronously acquiring images of 1 vertical view angle and 4 oblique view angles through a five-lens oblique photography system, fusing laser radar point cloud data and supplementing missing data of an image shielding area.
4. The unmanned aerial vehicle oblique photography and BIM fusion critical engineering dynamic acceptance assisting method according to claim 1, wherein in step S2, the specific implementation mode of the improved ICP-NDT algorithm comprises the steps of automatically extracting the BIM model, performing rough registration on intersection points of points Yun Zhongliang columns and corner edge endpoint characteristics, and performing iterative optimization by combining an ICP algorithm and normal distribution transformation.
5. The unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance assisting method of claim 1, wherein in step S3, the edge intelligent decision is used for realizing millisecond defect identification through a lightweight CNN model, and a decision tree based on fuzzy logic is used for automatically judging whether the verticality and the interval deviation of a formwork are beyond an allowable range or not and triggering grading early warning.
6. The unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance assisting method is characterized in that a feature fusion module based on an attention mechanism is adopted in the lightweight CNN model, texture features from oblique photography images and geometric features from laser point clouds can be processed in parallel, and the module adaptively fuses multi-source features by calculating channel attention weights, so that recognition robustness to cracks and deformation under illumination change and local shielding environments is improved, and omission ratio is reduced.
7. The unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance assisting method of claim 1 is characterized in that in step S4, the acceptance report is generated to automatically output a dynamic report containing a deviation color chart, a data table and an acceptance conclusion according to built-in standards, then multi-party visual collaborative management is realized through a cloud platform, and early warning information is pushed to related responsible persons in real time.
8. The unmanned aerial vehicle oblique photography and BIM fusion critical large project dynamic acceptance assistance method of claim 1, wherein in step S1, high-precision ground laser scanner supplementary data acquisition is adopted in a complex shielding area, so that the integrity of the point cloud is ensured.
9. The unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance assisting method of claim 1 is characterized in that in step S2, the digital twin model supports dynamic comparison analysis of multi-time sequence point cloud data, the dynamic comparison analysis automatically superimposes and compares the acquired real scene point cloud data with a point cloud model in a history period, deformation trends of critical parts of the dangerous large project are automatically identified and quantified through calculating Euclidean distance variation of point cloud coordinates, and when the accumulated displacement exceeds a preset threshold, a trend early warning chart is automatically generated in an acceptance report.
10. The unmanned aerial vehicle oblique photography and BIM fusion critical large project dynamic acceptance auxiliary method of claim 1, wherein in step S2, the improved ICP-NDT algorithm integrates an adaptive parameter adjustment strategy based on point cloud density, and the adaptive parameter adjustment strategy dynamically adjusts the nearest point search radius and NDT voxel grid size in ICP matching according to the distribution density of point cloud in different areas to ensure registration accuracy and optimize calculation efficiency.

Description

Unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance auxiliary method Technical Field The invention relates to the technical field of intelligent construction informatization, in particular to a dynamic acceptance auxiliary method for a dangerous large project fused by unmanned aerial vehicle oblique photography and BIM. Background Currently, in the field of building engineering, acceptance for dangerous projects mainly depends on the traditional manual measurement and BIM model comparison method. With the technical progress, unmanned aerial vehicle oblique photography technology is gradually applied to engineering mapping and three-dimensional modeling due to the characteristics of high efficiency and high authenticity. According to the technology, a plurality of sensors are carried on the same flight platform, images are collected from a plurality of angles such as vertical angles and inclined angles, and a real-scene three-dimensional model reflecting ground object information can be quickly built. Meanwhile, BIM technology is used as an important tool for digitizing building engineering, and a parameterized model of the BIM technology can describe the properties of building components in detail. In the prior art, attempts have been made to combine live-action models generated by oblique photography with design BIM models for auxiliary planning and acceptance; However, through extensive analysis, the prior art has the significant disadvantage that conventional oblique photography relies primarily on optical images to generate a point cloud through multi-view geometric matching. In the existing method, the rough registration is carried out by manually selecting characteristic points or arranging physical targets, the degree of automation is low, the registration precision is easily influenced by human factors, and the sub-centimeter high-precision registration is difficult to realize. Therefore, a dangerous large project dynamic acceptance assisting method for fusion of unmanned aerial vehicle oblique photography and BIM is provided for solving the problems. Disclosure of Invention In order to overcome the defects in the prior art, the invention provides a dynamic acceptance assisting method for a dangerous large project fused by unmanned aerial vehicle oblique photography and BIM, which aims to solve the problems in the prior art. In order to achieve the purpose, the invention provides the following technical scheme that the unmanned aerial vehicle oblique photography and BIM fusion dangerous large project dynamic acceptance assisting method comprises the following steps: S1, acquiring multi-source data of a dangerous large engineering area through an unmanned aerial vehicle carrying an RTK-PPK dual-mode positioning system, a five-lens oblique photographic camera and a laser radar, wherein the multi-source data comprises multi-view images and laser point cloud data, so that ground data are synchronously acquired; s2, carrying out fusion processing on the multi-source data to construct a dynamically updated digital twin model; s3, the unmanned aerial vehicle is further provided with an onboard edge computing unit, the onboard edge computing unit is provided with a lightweight CNN model, real-time defect identification is carried out on the multi-source data after registration, and the defect identification comprises concrete cracks, steel bar spacing deviation and formwork deformation; S4, comparing BIM design parameters with measured data, automatically generating an acceptance report based on a fuzzy logic-based multi-criterion acceptance decision tree, and returning key data to the cloud platform through a 5G network. Preferably, in step S2, the step of the fusion process is as follows: a1, automatically extracting edge endpoint features of a BIM model and point cloud data by the multisource data through an improved ICP-NDT algorithm to perform coarse registration, and then realizing sub-pixel level fine registration through iterative optimization; A2, introducing Kalman filtering to fuse GNSS, IMU, laser radar and image data, and constructing the digital twin model. Preferably, in step S1, the specific mode of the multi-scale data acquisition includes synchronously acquiring images of 1 vertical view angle and 4 oblique view angles through a five-lens oblique photography system, fusing laser radar point cloud data, and supplementing missing data of an image shielding area. Preferably, in step S2, the specific implementation mode of the improved ICP-NDT algorithm comprises the steps of automatically extracting the BIM model, performing rough registration on the points Yun Zhongliang column intersection points and corner edge endpoint features, and performing iterative optimization by combining the ICP algorithm and normal distribution transformation. Preferably, in step S3, the edge intelligent decision realizes millisecond defect identification through