CN-121982584-A - Intelligent positioning method and system for 3D inspection of bridge disease area based on unmanned aerial vehicle

CN121982584ACN 121982584 ACN121982584 ACN 121982584ACN-121982584-A

Abstract

The invention relates to the technical field of bridge disease detection, in particular to an intelligent positioning method and system for 3D inspection of a bridge disease area based on an unmanned aerial vehicle. The disease detection system comprises a data acquisition unit, a 3D modeling unit, a disease identification and life cycle tracking unit, a precise positioning fusion unit and a data storage and output unit. According to the invention, the unique lasting avatar identification based on the component topological relation, the detection time stamp and the contour check code is allocated for identifying the diseases for the first time, and the scale invariance descriptors are generated by combining the multi-scale curvature features extracted by the curvature pyramid with the weighting of the texture and the depth features, so that the accurate matching of the diseases in the inter-period is realized, and the repeated or missing identification is avoided. Meanwhile, the disease size and the environmental parameters are integrated, the evolution law is analyzed through an LSTM network and an attention mechanism, a trend representation containing the expansion rate and the direction is generated, a full life cycle data chain is formed, the disease process is continuously tracked, and a comprehensive dynamic data support is provided for bridge predictive maintenance.

Inventors

WANG SHENGMING
ZHANG TIANTIAN
JI KECHENG
WANG DANFENG
LIU JI
WU SHENGHUI
WANG ZIMING
BAI SHUAI

Assignees

温州信达交通工程试验检测有限公司

Dates

Publication Date: 20260505
Application Date: 20260115

Claims (10)

1. Intelligent positioning system is patrolled and examined in bridge disease area 3D based on unmanned aerial vehicle, its characterized in that includes: The system comprises a data acquisition unit (1), wherein the data acquisition unit (1) is used for carrying a laser radar and a high-definition camera through an unmanned aerial vehicle, acquiring bridge point cloud data and surface image data in a span period, synchronously acquiring unmanned aerial vehicle attitude data and environment parameters, and transmitting the unmanned aerial vehicle attitude data and environment parameters to a 3D modeling unit (2) and a disease identification and life cycle tracking unit (3); the 3D modeling unit (2) is used for receiving span primary data of the data acquisition unit (1), generating a high-precision bridge three-dimensional model and camera pose parameters after fusion processing, and transmitting the high-precision bridge three-dimensional model and camera pose parameters to the disease identification and life cycle tracking unit (3); The disease identification and life cycle tracking unit (3) is used for distributing unique and durable disease identification based on the topological relation of bridge members when the disease is identified for the first time, constructing a curvature pyramid, extracting contour curvature characteristics of each scale layer, generating scale invariance descriptors by combining surface texture characteristics and three-dimensional space depth information, performing cross-period disease characteristic matching based on the scale invariance descriptors, performing space integral operation on a disease contour through a distortion correction model constrained by geometric characteristics of a bridge structure according to three-dimensional point cloud data and camera pose parameters transmitted by the 3D modeling unit (2), converting pixel coordinates in a two-dimensional image into three-dimensional space physical coordinates, analyzing a disease evolution rule through a time sequence neural network based on the environmental parameters acquired by the data acquisition unit (1), generating a disease development trend characterization comprising an expansion rate vector and a direction parameter, and outputting disease core information to the accurate positioning fusion unit (4); The accurate positioning fusion unit (4), the accurate positioning fusion unit (4) receives the disease core information output by the disease identification and life cycle tracking unit (3), combines the unmanned aerial vehicle positioning data, fuses the high-precision bridge three-dimensional model generated by the 3D modeling unit (2) with the unmanned aerial vehicle real-time positioning data, and outputs three-dimensional coordinates and position parameters of a disease area; the data storage and output unit (5), the data storage and output unit (5) receives the data processed by the data acquisition unit (1), the 3D modeling unit (2), the disease identification and life cycle tracking unit (3) and the accurate positioning fusion unit (4), and the data is stored through the distributed encryption database to output a predictive maintenance decision report and a data interface.
2. The intelligent positioning system for 3D inspection of bridge defect areas based on unmanned aerial vehicle according to claim 1, wherein the data acquisition unit (1) comprises a bridge data acquisition module (11), an unmanned aerial vehicle state acquisition module (12) and an environmental parameter acquisition module (13), wherein: The bridge data acquisition module (11) acquires bridge point cloud data and surface image data according to span times at preset time intervals based on a laser radar and a high-definition camera carried by the unmanned aerial vehicle; The unmanned aerial vehicle state acquisition module (12) synchronously acquires attitude data in the unmanned aerial vehicle flight process, and takes the attitude data as auxiliary information for data calibration; the environmental parameter acquisition module (13) acquires environmental data around the bridge; Output data of the bridge data acquisition module (11), the unmanned aerial vehicle state acquisition module (12) and the environment parameter acquisition module (13) are synchronously transmitted to the 3D modeling unit (2) and the disease identification and life cycle tracking unit (3).
3. The intelligent positioning system for 3D inspection of bridge defect areas based on unmanned aerial vehicle according to claim 2, wherein the 3D modeling unit (2) comprises a point cloud preprocessing module (21), an image registration module (22) and a three-dimensional fusion modeling module (23), wherein: The point cloud preprocessing module (21) receives bridge point cloud data of the data acquisition unit (1), removes point cloud noise by adopting a filtering algorithm, and completes registration of inter-period secondary point clouds by iteration nearest point algorithm; The image registration module (22) receives bridge surface image data of the data acquisition unit (1), extracts image feature points through a scale-invariant feature transformation algorithm and completes stitching of inter-period images; The three-dimensional fusion modeling module (23) fuses the spliced images of the registration point cloud of the point cloud preprocessing module (21) and the image registration module (22), generates a high-precision bridge three-dimensional model and corresponding camera pose parameters, and synchronously transmits the high-precision bridge three-dimensional model and the corresponding camera pose parameters to the disease identification and life cycle tracking unit (3).
4. The intelligent positioning system for 3D inspection of bridge defect areas based on unmanned aerial vehicle according to claim 3, wherein the defect identification and life cycle tracking unit (3) comprises a defect identification module (31), a feature description and matching module (32), a profile distortion correction module (33) and a defect evolution analysis module (34), wherein: When the disease identification module (31) identifies the disease for the first time, unique and durable disease identification is distributed based on the topological relation of bridge members; the feature description and matching module (32) constructs a curvature pyramid, extracts contour curvature features of each scale layer, and generates scale invariance descriptors by combining surface texture features and three-dimensional space depth information to complete cross-period disease feature matching; the contour distortion correction module (33) performs distortion correction and three-dimensional coordinate conversion on the disease contour according to the three-dimensional point cloud data and the camera pose parameters of the 3D modeling unit (2); The disease evolution analysis module (34) is used for analyzing a disease evolution rule and generating trend characterization based on a multi-period matching result and environmental parameters, and outputting disease core information to the accurate positioning fusion unit (4).
5. The intelligent positioning system for 3D inspection of bridge diseased areas based on unmanned aerial vehicle according to claim 4, wherein the process of generating the disease identification by the disease identification module (31) comprises the following steps: s31.1, extracting a bridge component corresponding to the first identification disease, and generating a 12-bit component topological code based on the component number of the bridge design drawing The codes are mapped with the physical positions of the components one by one; s31.2, recording disease first detection time information, and generating an 8-bit first detection time stamp ; S31.3, MD5 hash operation is carried out on the initial outline point set of the disease, and the first 4 bits of the operation result are taken as outline feature check codes ; S31.4, will And (3) with Splicing to obtain disease identity And establishing unique identity association for inter-period disease tracking.
6. The intelligent positioning system for 3D inspection of bridge defect areas based on unmanned aerial vehicle according to claim 5, wherein the process of generating scale invariance descriptors and completing span-order matching by the feature description and matching module (32) comprises the following steps: S32.1 based on scale factors Generating a multi-scale space, sequentially performing Gaussian smoothing and Laplacian-Gaussian operation on the disease contour point set, and extracting contour curvature characteristics of each scale layer ; S32.2, extracting surface texture features and three-dimensional space depth features of the disease area, and Respectively normalized and weighted fused to generate scale invariance descriptors Determining the contribution degree of the weight to the characteristic variance of the disease sample through principal component analysis; S32.3, calculating invariance descriptors of different periods of disease scale Similarity of (2) Judging whether the same disease exists according to a preset threshold value, and matching the result with the disease identity of S31.4 Binding to form a cross-period characteristic association chain.
7. The intelligent positioning system for 3D inspection of bridge defect areas based on unmanned aerial vehicle according to claim 6, wherein the process of finishing defect profile distortion correction and three-dimensional coordinate conversion by the profile distortion correction module (33) comprises the following steps: S33.1, extracting inherent geometric characteristics of a component where the disease is located from a high-precision bridge three-dimensional model of the 3D modeling unit (2), and determining a component design datum point as a reference point ; S33.2, calling camera pose parameters output by the 3D modeling unit (2) and three-dimensional point cloud depth values corresponding to the disease areas; S33.3, performing distortion correction on the two-dimensional pixel points of the disease outline firstly, and then combining Converting camera parameters and point cloud depth values into three-dimensional physical coordinates ; And S33.4, performing interpolation fitting on the coordinate conversion result to obtain a continuous disease three-dimensional contour, and providing three-dimensional data support for three-dimensional space depth feature extraction, contour curvature feature accurate calculation and cross-period disease feature matching in the generation process of the scale invariance descriptor.
8. The intelligent positioning system for 3D inspection of bridge defect areas based on unmanned aerial vehicle according to claim 7, wherein the process of completing the defect evolution analysis and trend characterization by the defect evolution analysis module (34) comprises the following steps: s34.1, integrating three-dimensional physical size parameters of the multi-period diseases and environmental parameters of the data acquisition unit (1), and normalizing to form a network input matrix; S34.2, inputting the input matrix into the LSTM network, and calculating the weight of each period of data through an attention mechanism After weighting the network hiding state, mapping the network hiding state into disease feature vectors through a full connection layer ; S34.3, calculating disease expansion rate vector based on the difference value of the multi-period disease size parameter and the corresponding inspection time interval Combining disease feature vectors Generating a trend vector field containing the expansion rate and direction; s34.4, integrating the three-dimensional outline and trend vector field of the disease into disease core information, outputting the disease core information to the accurate positioning fusion unit (4), and simultaneously integrating the disease core information with the disease identification mark of S31.4 Binding to form a full life cycle data chain of an identity-scale invariance descriptor-three-dimensional coordinate-development trend.
9. The intelligent positioning system for 3D inspection of bridge defect areas based on unmanned aerial vehicle according to claim 8, wherein the accurate positioning fusion unit (4) comprises a positioning data receiving module (41), a data fusion processing module (42) and a coordinate parameter output module (43), wherein: the positioning data receiving module (41) is used for respectively receiving disease core information of the disease identification and life cycle tracking unit (3), a high-precision bridge three-dimensional model of the 3D modeling unit (2) and real-time positioning data of the unmanned aerial vehicle, and carrying out format unified processing on the received data; The data fusion processing module (42) is based on unmanned aerial vehicle real-time positioning data, combines with a space reference of a high-precision bridge three-dimensional model, performs multi-source comparison fusion with three-dimensional contour data in disease core information, and completes data calibration; The coordinate parameter output module (43) maps the fusion calibration result to a bridge three-dimensional model coordinate system, generates three-dimensional coordinates of a disease area and corresponding position parameters, and transmits the three-dimensional coordinates and the corresponding position parameters to the data storage and output unit (5).
10. The intelligent positioning method for the 3D inspection of the bridge defect area based on the unmanned aerial vehicle is based on the intelligent positioning system for the 3D inspection of the bridge defect area based on the unmanned aerial vehicle, which is characterized by comprising the following steps: S1, multi-source data span-time acquisition, namely carrying a laser radar and a high-definition camera on an unmanned aerial vehicle, acquiring bridge point cloud data and surface image data according to a preset time interval span-time, synchronously acquiring attitude data and bridge surrounding environment parameters in the flight process of the unmanned aerial vehicle, and transmitting the bridge point cloud data, the surface image data, the unmanned aerial vehicle attitude data and the environment parameters to S2 and S3; S2, high-precision bridge three-dimensional modeling, namely receiving acquired span-time primary data, removing noise from bridge point cloud data by adopting a filtering algorithm, finishing span-time registration by adopting an iterative nearest point algorithm, extracting characteristic points from bridge surface image data by adopting a scale-invariant feature transformation algorithm, finishing span-time splicing, fusing registered point cloud and spliced images, generating a high-precision bridge three-dimensional model and corresponding camera pose parameters, and transmitting to S3; S3, intelligent identification and full life cycle tracking of diseases, namely distributing unique identity marks consisting of bridge component topological codes, first detection time stamps and disease initial contour feature check codes for the first identified diseases, generating a multi-scale space based on scale factors, extracting multi-scale contour curvatures, surface textures and three-dimensional depth features of the diseases, weighting and fusing to generate scale invariance descriptors, completing cross-period disease matching and binding with the identity marks, combining geometrical features of a bridge three-dimensional model and camera pose parameters, correcting and converting two-dimensional pixel coordinates of the diseases into three-dimensional space physical coordinates, integrating the multi-period disease parameters and environment data, analyzing evolution trend through an LSTM network and an attention mechanism, forming a full life cycle data chain and outputting the full life cycle data chain to S4; S4, accurate positioning and fusion of the disease area, namely receiving disease core information, a high-precision bridge three-dimensional model and real-time positioning data of the unmanned aerial vehicle, unifying data formats, completing multi-source data comparison and calibration by combining a space reference of the bridge three-dimensional model on the basis of the real-time positioning data of the unmanned aerial vehicle, mapping a fusion result to a bridge three-dimensional coordinate system, generating three-dimensional coordinates of the disease area and corresponding position parameters, and transmitting to S5; And S5, data storage and decision output, namely receiving various data processed in the steps S1-S4, storing the data through a distributed encryption database, generating a predictive maintenance decision report based on the stored data, and providing a standardized data interface for subsequent application calling.

Description

Intelligent positioning method and system for 3D inspection of bridge disease area based on unmanned aerial vehicle Technical Field The invention relates to the technical field of bridge disease detection, in particular to an intelligent positioning method and system for 3D inspection of a bridge disease area based on an unmanned aerial vehicle. Background Along with the development of unmanned aerial vehicle, sensor and artificial intelligence technique, unmanned aerial vehicle carries check out bridge inspection by detection equipment and has become the industry mainstream, but prior art still faces disease positioning accuracy inadequately, stride the problem such as the time tracking difficulty, evolution trend is difficult to predict, has restricted preventive maintenance level's promotion. In the prior art, related patents have been explored in the field of bridge disease unmanned aerial vehicle detection. The invention discloses a multi-sensor collaborative detection method for a bridge disease unmanned aerial vehicle, which relates to the technical field of bridge monitoring and mainly comprises the steps of generating corrosion layered thickness distribution through a spectrum matching algorithm, acquiring three-dimensional point cloud data through a laser radar to generate deformation distribution data, calculating a local stiffness degradation coefficient matrix, inputting the local stiffness degradation coefficient matrix into a finite element model to obtain predicted deformation data, comparing actual measurement with the predicted deformation data to generate residual data, further generating a dynamic detection priority map, and executing a missed approach planning, thereby solving the problem of single sensor detection one-sided performance. In another example, chinese patent CN202411755388.6 discloses a bridge disease detection system based on an unmanned aerial vehicle and a detection method thereof, including an unmanned aerial vehicle system and a ground station, wherein the unmanned aerial vehicle system comprises a rotor power, a navigation control unit, a communication unit and an intelligent detection unit, the navigation control unit is responsible for environment sensing, positioning and three-dimensional point cloud model generation, the intelligent detection unit captures bridge images and identifies disease information, the ground station supports route planning and disease result display, and the problem of inspection when the unmanned aerial vehicle is difficult to obtain effective GPS positioning information is solved. Although the technical scheme has corresponding design advantages, the technical scheme also has the technical defects that firstly, the capability of disease cross-period secondary accurate tracking and full life cycle management is lacked, chinese patent CN202510639575.6 can only generate a retest priority map through multi-sensor data comparison, CN202411755388.6 can only identify the disease type and position in single inspection, a unique association mechanism aiming at single disease is not established, the same disease cannot be accurately matched in multi-period inspection, the development change rule of the disease is deeply analyzed without integrating multi-period disease data and environmental parameters, the continuous tracking of the disease is difficult, the complete process from occurrence to expansion cannot be provided for maintenance decision, secondly, the disease positioning accuracy is insufficient, the accurate mapping from two dimensions to three dimensions is not realized, chinese patent CN202510639575.6 side deformation distribution data and a prediction model are only capable of marking the approximate position of the disease, the distortion influence in the disease imaging process is not considered, the intrinsic geometric feature of a point cloud component is not combined for coordinate conversion, the disease positioning is difficult to stay in the complete process of the extended complete process, the two-dimensional coordinate positioning is difficult to form a two-dimensional coordinate accurate and the two-dimensional coordinate position is difficult to form, and the accurate maintenance accuracy is difficult to form a two-dimensional coordinate position. In view of the above, we provide an intelligent positioning method and system for 3D inspection of bridge defect areas based on unmanned aerial vehicles. Disclosure of Invention The invention aims to provide an intelligent positioning method and system for 3D inspection of a bridge disease area based on an unmanned aerial vehicle, which are used for solving the problems that the defect span secondary accurate tracking and full life cycle management capability and defect positioning accuracy are insufficient and accurate mapping from two dimensions to three dimensions is not realized in the background art. In order to solve the above technical problems, one of the purposes of