CN-122024095-A - Unmanned aerial vehicle measuring method for dynamic vector change of railway bridge

CN122024095ACN 122024095 ACN122024095 ACN 122024095ACN-122024095-A

Abstract

The invention discloses an unmanned aerial vehicle measuring method for dynamic vector change of a railway bridge, which comprises the following steps of hovering and collecting an unmanned aerial vehicle to obtain a high-definition uncompressed sequence image, sequentially carrying out binarization processing and edge detection on the sequence image to obtain a continuous edge contour, establishing a piecewise polynomial model with continuity constraint for the continuous edge contour to obtain smooth edge contour geometric expression, obtaining a pixel displacement sequence representing dynamic deformation of the bridge based on the smooth edge contour geometric expression, establishing a perspective transformation model to obtain dynamic conversion coefficients of pixels of a current frame and real length in the image, obtaining an accurate deflection value of dynamic vertical displacement of the bridge under the action of a train load based on the obtained pixel displacement sequence and the dynamic conversion coefficients, and repeating the steps to extract deflection when the train passes. The invention can measure the deflection of any point of the girder without erecting instruments and targets at fixed points under the bridge, and solves the problem that the dynamic line shape of the railway cannot be measured.

Inventors

SUN ZONGLEI
DI HAO
WANG CHENGYUE
SU WEI
CUI MIAOMIAO
MENG FANZENG
CHEN HOUJUN
LI YAN
ZHUO YI
SHI YANCHAO
SONG SHUFENG

Assignees

中国铁路设计集团有限公司
中国铁路经济规划研究院有限公司

Dates

Publication Date: 20260512
Application Date: 20260119

Claims (8)

1. The unmanned aerial vehicle measuring method for the dynamic vector change of the railway bridge is characterized by comprising the following steps of: A. hovering and collecting the unmanned aerial vehicle to obtain a high-definition uncompressed sequence image; B. Sequentially carrying out binarization processing and edge detection on the sequence images to obtain continuous edge contours; C. Establishing a piecewise polynomial model with continuity constraint for the continuous edge contour to obtain a smooth edge contour geometrical expression; D. obtaining a pixel displacement sequence representing dynamic deformation of the bridge based on the geometric expression of the smooth edge profile; E. establishing a perspective transformation model to obtain dynamic conversion coefficients of pixels of the current frame and the real length in the image; F. Based on the obtained pixel displacement sequence and dynamic conversion coefficient, obtaining an accurate deflection value of the bridge dynamic vertical displacement under the action of train load; G. Repeating the steps, and extracting deflection when the train passes through.
2. The unmanned aerial vehicle measurement method for dynamic vector change of the railway bridge, which is disclosed by claim 1, is characterized in that the unmanned aerial vehicle hovering acquisition is carried out in the step A, and high-definition uncompressed sequence images are obtained, wherein the method comprises the following specific processes: Firstly, selecting a high-resolution unmanned aerial vehicle carrying a stability-increasing cradle head, and hovering to a preset optimal observation position of the side surface of a main beam; Then, high-definition sequence images of the whole process from the bridge entering to the complete bridge leaving of the train are acquired at a specific frame rate per second.
3. The unmanned aerial vehicle measuring method for railway bridge dynamic vector change according to claim 1, wherein the step B is characterized in that binarization processing and edge detection are sequentially carried out on the sequence images to obtain continuous edge contours, and the specific process is as follows: firstly, binarizing the obtained sequence image by adopting an Ojin method; and then, applying a Canny edge detection algorithm to the binarized image, and obtaining a continuous edge contour through non-maximum suppression and dual-threshold hysteresis connection means.
4. The unmanned aerial vehicle measuring method for railway bridge dynamic vector change according to claim 1, wherein step C is characterized in that a piecewise polynomial model with continuity constraint is established for continuous edge contours to obtain smooth edge contour geometric expression, and the method comprises the following specific steps: Firstly, establishing a piecewise polynomial model with continuity constraint aiming at complex deformation characteristics of continuous edge contours; And then, adopting Sequential Quadratic Programming algorithm, and repeatedly iterating and optimizing the coefficients of each piecewise polynomial by using the overall fitting error between the actual coordinates of the minimized edge pixel points and the calculated coordinates of the model until the convergence condition is met, so as to obtain the high-precision and smooth geometric expression of the edge contour.
5. The unmanned aerial vehicle measuring method for dynamic vector change of a railway bridge according to claim 1, wherein the step D is based on smooth edge contour geometric expression, and a pixel displacement sequence for representing dynamic deformation of the bridge is obtained, and the method comprises the following specific steps: Firstly, in each frame of image, determining pixel points corresponding to fixed supports at two ends of a bridge based on a fitted piecewise polynomial model, and connecting to form a reference string; Then, aiming at the selected to-be-measured point, accurately calculating an image vector value by solving a pixel coordinate difference value between the vertical projection of the to-be-measured point on the frame polynomial curve and a reference chord line; and finally, executing frame by frame, and generating a pixel displacement sequence representing the dynamic deformation of the bridge.
6. The unmanned aerial vehicle measuring method for railway bridge dynamic vector change according to claim 1, wherein the method is characterized in that the step E is to establish a perspective transformation model to obtain dynamic conversion coefficients of pixels of a current frame and the real length in an image, and the specific process is as follows: firstly, aiming at perspective projection deformation and interframe micro-shaking existing in unmanned aerial vehicle shooting, establishing a perspective transformation model by identifying a preset calibration object with known physical size in each frame of image and utilizing the corresponding relation between pixel coordinates and real coordinates of the calibration object; then, adopting direct linear transformation to solve the conversion parameters of the model frame by frame; finally, the dynamic conversion coefficient of the pixel and the real length in the current frame is accurately obtained, and projection and jitter errors are effectively eliminated.
7. The unmanned aerial vehicle measuring method for dynamic vector change of a railway bridge according to claim 1, wherein the step F is based on the obtained pixel displacement sequence and dynamic conversion coefficient, and the accurate deflection value of the dynamic vertical displacement of the bridge under the action of train load is obtained, and the specific process is as follows: Firstly, after obtaining an image vector value and a dynamic conversion coefficient in a pixel displacement sequence frame by frame, carrying out data fusion on the image vector value and the dynamic conversion coefficient; Then, calculating an instantaneous vector value of the to-be-measured point corresponding to the real physical scale in each frame through scalar multiplication; Then, in order to obtain a final deflection time-course curve, selecting an initial frame sequence of the train in an unloaded state through a front bridge, and taking an average vector value as a reference; then subtracting the reference value from the image vector values of all subsequent frames to effectively eliminate the influence of the initial line shape of the bridge; and finally, outputting an accurate deflection value reflecting the dynamic vertical displacement of the bridge under the action of the train load.
8. The unmanned aerial vehicle measuring method for dynamic vector change of a railway bridge according to claim 1, wherein the step G is repeated, deflection extraction is carried out when a train passes through, and the method comprises the following specific steps: and D, extracting deflection of different trains when passing through the train by repeating the steps A to F through the images of the different trains passing through the time sequence through the field actual measurement.

Description

Unmanned aerial vehicle measuring method for dynamic vector change of railway bridge Technical Field The invention belongs to the field of railway bridge deflection measurement, and particularly relates to an unmanned aerial vehicle measurement method for railway bridge dynamic vector change. Background By the end of 2024 years, the railway operation mileage of China reaches 16.2 ten thousand kilometers, wherein the high-speed rail reaches 4.8 ten thousand kilometers. Wherein the number of simply supported beam bridges exceeds 60 ten thousand holes, the number of continuous beam bridges is 8000 and the number of large-span bridges is 140. In the bridge detection process in special areas such as severe cold, dangerous mountain areas and large river crossing areas, the conventional detection means have limitations and detection blind areas, and particularly in severe weather, the conventional detection difficulty is increased, so that the detection efficiency is low, the difficulty is high, the risk coefficient is high, and meanwhile, the detection fineness and effect are also influenced. The number of simply supported girder bridges with service life of over 40 years in China is approximately 2 ten thousand, and the number of simply supported girder bridges is increased year by year, so that under the condition that the current transportation capacity and load are continuously increased, how to rapidly detect and evaluate the service performance of the bridge, and ensuring the safe operation of railways is an important challenge for the operation and maintenance of the railway bridge. The static and dynamic alignment of railroad bridges is an important indicator for evaluating their performance. At present, the measuring method for bridge linearity and deflection comprises a displacement meter, a photoelectric deflection meter, a microwave radar, a total station, an inclinometer, a static level gauge and the like, but for bridges such as cross-river, mountain canyon, cross-road overpass and the like, the implementation of the traditional measuring method is difficult due to the fact that the bridge mid-span position is difficult to reach, the cost is high, and development of research on railway bridge static and dynamic linearity measuring technology based on unmanned aerial vehicles and computer vision is needed to provide accurate deformation data for bridge health monitoring and safety performance evaluation. The unmanned aerial vehicle is used for measuring dynamic linearity and deflection of the railway cable-stayed bridge, so that the contradiction between a large visual field and high precision exists, and the problem of shaking of the unmanned aerial vehicle is solved, and the pixel noise in the measuring process is required to be solved. At present, the unmanned aerial vehicle for measuring the deflection of the bridge mainly comprises the following steps: firstly, a fixed datum point compensation method is used for measuring deflection of a to-be-measured point after the posture of the unmanned aerial vehicle is compensated by setting a pier as a datum point or setting a target at a beam end as a datum point. However, the method needs to set targets on the beam body or the bridge pier, and can not realize true target-free telemetry; secondly, a laser projection point compensation method. According to the method, the attitude of the unmanned aerial vehicle is compensated by using a fixed laser point projected onto the bridge below the bridge as a fixed point. However, the laser point adopting the mode has drift due to air shake, atmospheric refraction, ground vibration and the like, and a new error is introduced; Thirdly, shooting the splicing method in a partition mode. According to the method, after the unmanned aerial vehicle or the fixed camera is adopted to carry out approaching shooting on the main beam, the cameras are spliced, so that a high-resolution image of the main beam is obtained. However, the method requires unmanned aerial vehicle translation shooting, and cannot meet the high-frequency data acquisition requirement when a train passes. In summary, developing a practical method suitable for unmanned aerial vehicle to measure railway bridge deflection is still a key problem to be solved in the technical field of bridge detection. Disclosure of Invention The invention provides an unmanned aerial vehicle measuring method for dynamic vector change of a railway bridge, which aims to solve the problems existing in the prior art. The technical scheme of the invention is that the unmanned aerial vehicle measuring method for the dynamic vector change of the railway bridge comprises the following steps: A. hovering and collecting the unmanned aerial vehicle to obtain a high-definition uncompressed sequence image; B. Sequentially carrying out binarization processing and edge detection on the sequence images to obtain continuous edge contours; C. Establishing a piecewise polynomial model with co