CN-121999197-A - Intelligent concrete crack detection method and system based on unmanned aerial vehicle platform

CN121999197ACN 121999197 ACN121999197 ACN 121999197ACN-121999197-A

Abstract

The invention relates to the technical field of concrete crack detection, and discloses an intelligent concrete crack detection method and system based on an unmanned aerial vehicle platform, wherein the method comprises the steps of firstly setting unmanned aerial vehicle flight parameters, and generating an initial inspection path which is free of omission and keeps a preset safety distance by combining GPS positioning data and target concrete structure geometric parameters; the unmanned aerial vehicle is controlled to fly along a path through a depth camera to collect image/video data in real time, a path planning algorithm is utilized to dynamically adjust the track and synchronously avoid static and dynamic obstacles, then YOLOv model improved by CA attention mechanism is adopted to identify and position crack information in real time, the identification result is transmitted to a ground station through a digital image transmission module, and a log containing detection time, position and crack parameters is generated and visually displayed. The invention greatly improves inspection efficiency and detection precision, ensures operation safety, provides accurate data support for operation and maintenance of a concrete structure, and is suitable for various scenes such as bridges, tunnels and the like.

Inventors

XIONG LV
Xiao Cilin
XU HUANCHUN
Qiu Yuanzhao
DU MINGTAO
QIU NANFENG
LI ZEXIONG
Zhong Qiaoqiao
SONG BIN

Assignees

广东交通职业技术学院

Dates

Publication Date: 20260508
Application Date: 20251128

Claims (10)

1. The intelligent concrete crack detection method based on the unmanned aerial vehicle platform is characterized by comprising the following steps of: Setting the flying speed, hovering precision and data acquisition frequency of the unmanned aerial vehicle, generating an initial inspection path by combining GPS positioning data and geometric parameters of a target concrete structure, ensuring that the path covers a detection area without omission and keeping a preset safety distance with the concrete structure; controlling the unmanned aerial vehicle to fly according to the initial inspection path, acquiring image or video data of the concrete structure in real time by using a depth camera, dynamically adjusting the flight path through a path planning algorithm in the flying process, and synchronously completing static and dynamic obstacle avoidance; Performing real-time processing on the image or video data by utilizing a YOLOv target detection model improved by a CA attention mechanism, identifying and positioning concrete cracks in the image or video data, and extracting minimum width, maximum width and area occupation ratio parameters of the cracks; and transmitting the identification result containing the crack position and size information to the ground station through a digital image transmission module, generating a detection log containing detection time, position and crack parameters, and performing visual display through a ground station system interface.
2. The method of claim 1, wherein the YOLOv target detection model modified by CA attention mechanism is a YOLOv s-CA model with CA attention mechanism module embedded in a back bone layer of YOLOv5, and the processing comprises: extracting local features of an input image through convolution operation to generate a feature map; Distributing attention weights for the feature map sub-channels and the space positions by using a CA attention mechanism, strengthening key local features and inhibiting irrelevant information; And multiplying the feature map with the corresponding attention weight, and fusing the local information and the global information through pooling or summation operation to obtain the weighted feature representation containing the local and global important information.
3. The method of claim 1, wherein the implementation of the path planning algorithm comprises: With improvements The algorithm takes the flying spot of the unmanned aerial vehicle as a starting point and the boundary of the detection area as an end point, and combines the barrier distribution information of the known environment and the concrete structure layout to generate a global optimal inspection path; Based on a pre-trained DQN algorithm, environmental data acquired by a depth camera are received in real time, random static obstacles in a flying path are detected, and a local optimal round-the-fly path is re-planned according to the positions and the sizes of the obstacles; and constructing an environment map through Gmapping algorithm in SLAM technology, and constructing a real-time environment map by combining unmanned aerial vehicle GPS positioning data and depth camera point cloud data, wherein the real-time environment map is used for verifying global path rationality and optimizing local path planning efficiency.
4. A method according to claim 3, wherein the obstacle avoidance process during the flight of the unmanned aerial vehicle comprises: Acquiring environmental point cloud data through a depth camera, and primarily identifying an area where an obstacle is located by combining a target detection algorithm; clustering point cloud data by using a DBSCAN algorithm, and separating different barrier types; And adjusting flight parameters of the unmanned aerial vehicle according to the type of the obstacle, the relative distance between the unmanned aerial vehicle and the dynamic characteristics.
5. The intelligent concrete crack detection system based on the unmanned aerial vehicle platform is characterized by comprising the unmanned aerial vehicle platform and a ground station, wherein: The unmanned aerial vehicle platform is provided with a flight control module, an electric adjustment motor, a wireless communication module, a depth camera, a digital image transmission module, a power module, an onboard sensor, a communication interface module and an onboard computer, wherein the depth camera is used for collecting images or video data of a concrete structure, the onboard computer is used for executing the intelligent concrete crack detection method according to any one of claims 1-4, and the digital image transmission module is used for realizing real-time data interaction between the unmanned aerial vehicle and a ground station; The ground station is used for receiving the detection data and the unmanned aerial vehicle state data sent by the digital image transmission module, configuring detection parameters, storing detection logs and realizing visual display and inquiry of detection results.
6. The system of claim 5, wherein the ground station is provided with an intelligent detection and assessment system for concrete cracks, comprising: The parameter setting module is used for configuring a confidence coefficient threshold value, an intersection ratio threshold value, a display segmentation result and a display detection frame and label of the detection model; The detection result display module is used for displaying the total number of the cracks, the total dividing area occupation ratio and the position and size information of each crack in real time; and the operation module is used for controlling data import, starting the unmanned aerial vehicle to detect in real time, exporting a detection result and inquiring a history log.
7. The system of claim 5, wherein the flight control module integrates an inertial measurement unit, a gyroscope, an accelerometer, a magnetometer, and an barometer, supports multiple PWM servo outputs, multiple universal serial ports, a GPS port, and a USB interface for unmanned aerial vehicle attitude control, position determination, and flight trajectory correction.
8. An electronic device, comprising: The intelligent concrete crack detection method based on the unmanned aerial vehicle platform comprises a memory and a processor, wherein the memory and the processor are in communication connection, the memory stores computer instructions, and the processor executes the computer instructions, so that the intelligent concrete crack detection method based on the unmanned aerial vehicle platform is executed.
9. A computer-readable storage medium having stored thereon computer instructions for causing a computer to perform the unmanned platform-based concrete crack intelligent detection method of any one of claims 1 to 4.
10. A computer program product comprising computer instructions for causing a computer to perform the unmanned platform-based concrete crack intelligent detection method of any one of claims 1 to 4.

Description

Intelligent concrete crack detection method and system based on unmanned aerial vehicle platform Technical Field The invention relates to the technical field of concrete crack detection, in particular to an intelligent concrete crack detection method and system based on an unmanned aerial vehicle platform. Background In the field of civil engineering, crack detection of concrete structures (such as bridges, tunnel liners, sleepers, etc.) is a key element for guaranteeing safe operation and maintenance of infrastructure. Along with the expansion of the construction scale of the infrastructure in China, the traditional concrete crack detection mode gradually exposes the problems of low efficiency, poor safety, insufficient precision and the like, the currently mainstream manual and machine inspection mode depends on field operation detection equipment of technicians, the single detection range is limited, the personnel safety risk is high under the complex scenes such as high altitude, tunnels, river-crossing bridges and the like, meanwhile, the manual judgment cracks are easily influenced by subjective experience and illumination conditions, the problems of micro-crack omission inspection and large dimension measurement error exist, the data acquisition efficiency is only 3-5m <2 >/h, and the rapid inspection requirement of the large-scale infrastructure is difficult to meet. In order to improve detection efficiency, partial research is introduced into an unmanned plane platform carrying camera to collect images, but the existing unmanned plane inspection scheme still has obvious technical shortboards, namely, firstly, the path planning capability is insufficient, the traditional A-type algorithm, the particle swarm algorithm and the like can only deal with the known static environment, when the existing unmanned plane inspection system faces to sudden random static barriers (such as outstanding concrete components and accessory equipment) in inspection, local paths cannot be re-planned in real time, inspection interruption is easily caused, secondly, the precision of a crack detection algorithm is limited, the spatial position sensitivity of a main stream target detection model such as YOLOv and the like to concrete cracks is low, particularly, the detection rate is less than 85% under the scene of complex illumination (such as backlight and shadow) or micro cracks (with the width of less than 0.2 mm), key operation and dimension data such as crack size, area ratio and the like cannot be synchronously output, thirdly, the data management and application are disjointed, and the existing system can only store original images, and manual secondary analysis is required, and real-time visual display and historical trace log functions are lacked, so that the detection result is difficult to support and decision are fast. In addition, the existing scheme is insufficient in research on dynamic obstacle avoidance (such as flying birds in the air and temporary construction equipment), the obstacle avoidance logic is not formed and is complete, the close-range collision risk of the unmanned aerial vehicle is high, and the real-time detection requirement cannot be met. Therefore, development of a set of unmanned aerial vehicle detection scheme integrating efficient path planning, high-precision crack recognition and intelligent data management is needed, so that the pain point of the traditional detection mode is solved, and the security, efficiency and precision of inspection of a concrete structure are improved. Disclosure of Invention In order to solve the problems of low efficiency, high safety risk and insufficient precision of the traditional concrete crack detection mode, the invention provides an unmanned plane platform-based intelligent concrete crack detection method and system, which can realize full-flow automation, high detection precision and operation safety of inspection of a concrete structure. In a first aspect, the invention provides an intelligent concrete crack detection method based on an unmanned aerial vehicle platform, which comprises the following steps: Setting the flying speed, hovering precision and data acquisition frequency of the unmanned aerial vehicle, generating an initial inspection path by combining GPS positioning data and geometric parameters of a target concrete structure, ensuring that the path covers a detection area without omission and keeping a preset safety distance with the concrete structure; controlling the unmanned aerial vehicle to fly according to the initial inspection path, acquiring image or video data of the concrete structure in real time by using a depth camera, dynamically adjusting the flight path through a path planning algorithm in the flying process, and synchronously completing static and dynamic obstacle avoidance; Performing real-time processing on the image or video data by utilizing a YOLOv target detection model improved by a CA attention mechanism, i