A box girder pushing construction device and method based on machine vision technology

Abstract

The application relates to the technical field of box girder construction, in particular to a box girder pushing construction device and method based on machine vision technology, which has the beneficial effect that dynamic deviation data of steel box girder pushing is collected in real time and globally through machine vision, multi-dimensional deviation accurate analysis and self-adaptive deviation correction regulation are realized by combining intelligent algorithms, manual frequent intervention is not needed, pushing construction deviation correction accuracy, real-time performance and working condition adaptability are greatly improved, deviation accumulation hidden dangers are eliminated, the linear precision and construction safety of steel box girder pushing are ensured, and intelligent, refined and visualized control of steel box girder pushing construction is realized.

Description

Technical Field

[0001] This invention relates to the field of box girder construction technology, specifically to a box girder jacking construction device and construction method based on machine vision technology. Background Technology

[0002] The steel box girder launching construction technology is widely used in the construction of long-span continuous steel box girder bridges, intercity railway bridges, and urban viaducts due to its advantages such as not requiring large-scale scaffolding, short construction period, and minimal disruption to traffic and the environment below the bridge. During the launching construction process, the steel box girder needs to be slid into place segment by segment along a pre-set track. Its axial offset and vertical height difference deviation are the core control indicators affecting the bridge construction accuracy, structural safety, and forming quality.

[0003] Currently, the construction of steel box girder jacking and correction generally adopts a work mode of manual monitoring, fixed-point scale observation, and manual adjustment of jacking equipment. Some existing technologies use fixed sensors to collect displacement data at fixed points to achieve simple correction, but these technologies have many inherent defects in actual construction.

[0004] The monitoring accuracy is low and the real-time performance is poor. Manual monitoring relies on the experience of construction personnel and can only achieve intermittent fixed-point detection. It cannot continuously capture the minute offsets, dynamic deviations, and turbulence deviations during the jacking process of the steel box girder. In addition, there is a lag in data acquisition, recording, and feedback, which can easily lead to the accumulation of deviations. When the offset exceeds the threshold and correction is not performed, it can easily cause problems such as local stress concentration in the steel box girder, wear of the slide rail, and support offset.

[0005] The correction adjustment is lagging and has poor adaptability. Existing semi-automatic correction technology mostly uses control logic that triggers correction with a fixed threshold. It can only correct a single horizontal offset and cannot simultaneously identify vertical height difference, torsional deviation and dynamic slip deviation. It cannot achieve adaptive dynamic adjustment for complex working conditions such as changes in jacking speed, uneven settlement of the slideway and environmental wind interference. The correction accuracy is difficult to match the requirements of large-span and long-distance jacking construction.

[0006] The traditional construction method involves a large amount of manual intervention and carries high construction risks. It requires dedicated personnel to monitor and manually adjust the jacking cylinders and correction devices throughout the process. Manual operation results in large errors and high labor intensity. Moreover, in adverse construction environments such as at night, in strong winds, and in rainy weather, the monitoring field of vision is limited, and the difficulty of operation increases dramatically. Problems such as untimely correction and over-adjustment are very likely to occur, which can lead to the steel box girder exceeding the axis and structural deformation. In severe cases, it can affect the overall alignment of the bridge and construction safety.

[0007] Poor data traceability and low level of intelligence. Traditional construction lacks a full-process visualized data collection and recording system. The process of deviation generation and correction adjustment parameters lack complete data support, making it impossible to achieve traceability of the construction process, condition analysis and parameter optimization, which is difficult to meet the industry development needs of modern bridge intelligent construction and refined quality control.

[0008] In summary, the existing steel box girder jacking and correction technology has technical shortcomings such as discontinuous monitoring, non-adaptive correction, weak adaptability to working conditions, and low level of intelligence. There is an urgent need for a high-precision, real-time, and adaptive intelligent correction construction technology to solve many of the defects of the existing construction process. Summary of the Invention

[0009] The purpose of this invention is to provide a box girder jacking construction device and method based on machine vision technology to solve the problems mentioned in the background art.

[0010] To achieve the above objectives, the present invention provides the following technical solution: a box girder jacking construction device based on machine vision technology, comprising a machine vision acquisition module, a data processing and control module, an execution correction module, and a status monitoring module. The four modules achieve bidirectional data interaction through an industrial bus and a wireless transmission unit, forming a closed-loop collaborative control architecture.

[0011] The machine vision acquisition module is deployed throughout the entire area of ​​the box girder jacking construction to acquire real-time dynamic displacement images of the box girder and obtain high-definition image data of the box girder outline and the benchmark target.

[0012] The data processing and control module is used to receive visually acquired images and status monitoring data, complete image preprocessing, feature extraction and multi-dimensional deviation calculation, and run an adaptive correction algorithm to generate precise control commands.

[0013] The execution correction module is used to respond to control commands and simultaneously complete the horizontal axis correction, vertical height difference leveling and sliding speed control of the box girder;

[0014] The status monitoring module is used to collect parameters such as box girder sliding displacement, tilt angle and hydraulic cylinder thrust, and to perform bidirectional verification with machine vision recognition results to eliminate the potential error of single vision recognition.

[0015] Preferably, the machine vision acquisition module includes high-definition industrial cameras deployed in multiple regions, an adaptive supplementary lighting source, and a reference calibration target. The high-definition industrial cameras are respectively installed at the front end of the box girder jacking, the outer sides of the two side flanges, and the rear end face. The camera field of view covers the entire construction area of ​​the box girder jacking, with no blind spots. The reference calibration target is fixed on the surface of the preset construction reference axis point and the slide reference elevation point. The target surface is set with high-contrast circular recognition feature points, which serve as the absolute reference for image recognition and coordinate conversion. The adaptive supplementary lighting source is equipped with a photosensitive sensor, which can automatically adjust the brightness and start / stop status according to the ambient light intensity, adapting to low-light conditions such as day and night construction, rainy and foggy days, and avoiding interference from ambient light with image recognition accuracy.

[0016] Preferably, the data processing and control module comprises a high-performance industrial industrial control computer, a multi-channel image acquisition card and a 5G data transmission unit; the image acquisition card supports synchronous access to image data from more than 8 industrial cameras, and the transmission delay of a single frame of image is less than 20ms; the industrial industrial control computer is built-in with an image preprocessing program, a feature recognition algorithm and a fuzzy PID adaptive deviation correction program, which can complete the parameter calculation of the box girder axis position, vertical elevation, torsion angle and sliding speed in real time; the industrial control computer is equipped with a 10-inch touch display terminal, which can realize real-time visual display of the construction status, and is also equipped with an acousto-optic alarm device, which immediately triggers an over-limit warning when the deviation of any dimension exceeds a preset threshold.

[0017] Preferably, the execution and deviation correction module comprises a plurality of groups of distributed arranged main pushing oil cylinders, transverse fine adjustment deviation correction oil cylinders, vertical leveling jacks and electro-hydraulic proportional flow regulating valves; the main pushing oil cylinders are arranged at equal intervals along the longitudinal direction of the bottom of the box girder, and are used for controlling the overall pushing sliding speed and cumulative displacement of the box girder; the transverse fine adjustment deviation correction oil cylinders are symmetrically installed at the slideway positions on both sides of the box girder, the stroke of a single cylinder is set to ±50mm, which is used for accurately correcting the horizontal axis offset of the box girder; the vertical leveling jacks are arranged at the four corner positions of the bottom of the box girder, which are used for synchronously correcting the vertical height difference deviation at the four points of the box girder; the electro-hydraulic proportional flow regulating valves are connected with each group of oil cylinders in one-to-one correspondence, which can dynamically adjust the expansion and contraction rate and stroke precision of the oil cylinders according to the control instructions, so as to realize millimeter-level fine adjustment.

[0018] Preferably, the state monitoring module comprises a high-precision pull-wire displacement sensor, a dual-axis inclination sensor and a high-frequency pressure sensor; the displacement sensor is installed at the end of the oil cylinder to collect the actual expansion and contraction displacement data of the oil cylinder in real time; the dual-axis inclination sensor is fixed at the center of the top surface of the box girder to collect the parameters of the longitudinal inclination angle and the plane torsion angle of the box girder; the pressure sensor is installed on the oil inlet pipeline of the oil cylinder to collect the jacking force value of the oil cylinder in real time. The sampling frequency of all monitoring data is not less than 100Hz, which is compared and verified with the machine vision recognition result in real time, and the secondary recognition process is automatically triggered when the data deviation exceeds 3%.

[0019] Preferably, the reference calibration target comprises a base plate, a target mark is fixed on the front side of the base plate, a mounting frame is slidably connected to the back side of the base plate, a guide groove is formed on the surface of the mounting frame, a bolt is inserted into the guide groove, and the bolt penetrates through the base plate and then is locked.

[0020] Preferably, a sliding groove is formed on the back side of the base plate, the sliding groove is a "convex"-shaped groove, a connecting strip is inserted into the sliding groove, the connecting strip is a "convex"-shaped strip, the connecting strip is fixed on the surface of the mounting frame, the mounting frame is an L-shaped plate structure, and a bolt hole is formed at one end of the mounting frame.

[0021] Preferably, there are two guide grooves, which are symmetrically distributed about the connecting strip. Two through holes are opened on the surface of the substrate, and the through holes correspond to the guide grooves one by one. Two rubber pads are fixed on the surface of the mounting bracket. After the bolts are tightened, the rubber pads are clamped between the substrate and the mounting bracket and deform.

[0022] Preferably, the surface of the substrate has four sets of air vents, which are distributed on the outer sides of the four corners of the target mark.

[0023] A box girder jacking construction method based on machine vision technology includes the following steps:

[0024] Step 1: Pre-construction calibration and system initialization. After the box girder prefabrication and assembly are completed and the slide is laid in place, fix all benchmark calibration targets, calibrate the intrinsic parameters, extrinsic coordinates and pixel scale of the industrial camera, establish the global benchmark coordinate system for box girder jacking construction, debug the communication links of each module, preset the jacking speed threshold, axis offset threshold, vertical height difference threshold and torsion angle threshold, and complete the system accuracy verification.

[0025] Step 2: Synchronous acquisition of dynamic images across the entire area. After the box girder jacking construction is started, multiple sets of industrial cameras continuously acquire dynamic images of the entire box girder jacking process at a frequency of more than 15 frames per second. The supplementary light source automatically adjusts the output brightness according to the ambient light. Simultaneously, auxiliary data such as cylinder pressure, box girder tilt angle and sliding displacement are collected through the status monitoring module.

[0026] Step 3: Image preprocessing and precise feature recognition. The industrial control computer performs Gaussian denoising, lens distortion correction and histogram enhancement on the original image in sequence to eliminate environmental noise and equipment imaging errors. The edge detection algorithm extracts the outline features of the box girder and the target feature points, matches the global reference coordinate system, and calculates the core parameters of the box girder's real-time axis position, vertical elevation, torsion angle and slip velocity.

[0027] Step 4: Multi-dimensional deviation classification judgment. The real-time identification parameters are compared with the preset benchmark parameters, and the horizontal axis offset, vertical height difference deviation, plane torsion angle deviation and speed fluctuation deviation are calculated respectively. According to the proportion of the deviation to the threshold, the working condition is divided into three levels: minor deviation, normal deviation and over-limit deviation.

[0028] Step 5: Intelligent matching of adaptive correction strategy. Based on the fuzzy PID algorithm, the corresponding correction scheme is dynamically generated according to the deviation level, deviation size and deviation change rate: for minor deviation conditions, a micro-scale slow fine-tuning mode is adopted, with a single correction amount not exceeding 2mm to avoid excessive correction causing structural stress fluctuations; for normal deviation conditions, a uniform speed precise correction mode is adopted, which synchronously adjusts the corresponding hydraulic cylinder at a constant rate to quickly eliminate the deviation within the preset stroke; for excessive deviation conditions, an audible and visual warning is immediately triggered, and the jacking speed is automatically reduced to less than 30% of the original speed. A graded progressive correction mode is adopted, which gradually corrects the deviation in 3-5 steps to prevent sudden changes in local stress in the box girder.

[0029] Step Six: Closed-loop dynamic control for correction. The industrial control computer converts the correction parameters into electro-hydraulic control commands, synchronously linking the horizontal correction cylinder, the vertical leveling jack, and the main jacking cylinder to complete horizontal offset correction, vertical height difference leveling, and sliding speed calibration in parallel. During the correction process, the machine vision module continuously collects images in real time and dynamically provides feedback on the correction effect, forming a closed-loop control process of "collection-analysis-correction-feedback" until all deviation parameters return to the preset safety threshold range.

[0030] Step 7: Full-process data archiving and construction completion. The original image data, deviation records, correction parameters and equipment operation data are stored in real time throughout the entire process of box girder jacking. Traceable construction log reports are generated. After a single box girder segment is jacked into place, the overall alignment accuracy of the box girder is manually verified. After confirming that it meets the requirements of bridge construction specifications, all equipment operation is stopped, the single jacking correction construction is completed, and the next box girder jacking process begins.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] The box girder jacking construction device and method proposed in this invention, based on machine vision technology, collects dynamic deviation data of the steel box girder jacking in real time across the entire domain using machine vision. Combined with intelligent algorithms, it achieves accurate multi-dimensional deviation analysis and adaptive correction and control, eliminating the need for frequent manual intervention. This significantly improves the accuracy, real-time performance, and adaptability of the jacking construction, eliminates the hidden dangers of deviation accumulation, ensures the alignment accuracy and construction safety of the steel box girder jacking, and realizes intelligent, refined, and visual management and control of the steel box girder jacking construction.

[0033] Through continuous imaging monitoring across the entire domain using machine vision, the deviation identification accuracy can reach ±0.5mm, which is far higher than the accuracy standard of ±3mm for traditional manual monitoring. Moreover, there is no monitoring lag, and it can capture dynamic minute deviations in real time, completely solving the problems of accumulated deviations and difficulty in subsequent correction in traditional construction, and effectively ensuring the alignment accuracy of the steel box girder during jacking.

[0034] Abandoning the traditional manual correction and fixed threshold passive correction mode, it dynamically adapts to different construction conditions through intelligent algorithms and automatically matches the optimal correction strategy. It can effectively cope with complex conditions such as slide settlement, wind interference, and jacking speed fluctuations, significantly improving the stability and adaptability of correction and eliminating problems such as over-correction and incomplete correction.

[0035] It significantly reduces on-site personnel for monitoring and operation, thereby reducing labor intensity and human error. It can still operate stably and accurately in harsh construction environments, avoiding construction safety hazards caused by limited visibility and operational errors during manual monitoring, and reducing the incidence of quality defects such as deformation of steel box girder structures, damage to sliding tracks, and support displacement.

[0036] The entire process automatically collects, stores, and uploads construction data and image information, providing real-time visualization of the construction status and enabling full traceability of the construction process. This facilitates subsequent condition analysis and process optimization, aligning with the industry trend of intelligent construction and refined construction in bridge engineering, and significantly improving the quality of bridge construction and structural durability.

[0037] Dynamic correction can be completed without interrupting the jacking construction, avoiding the delays caused by the traditional correction process that requires machine shutdown for debugging, effectively shortening the single-section jacking construction cycle, improving the overall efficiency of steel box girder jacking construction, and shortening the project duration. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the structure of the present invention;

[0039] Figure 2 This is a schematic diagram of the connection structure between the substrate and the mounting bracket of the present invention;

[0040] Figure 3 This is a schematic diagram of the substrate structure of the present invention;

[0041] Figure 4 This is a schematic diagram of the mounting bracket structure of the present invention.

[0042] In the figure: substrate 1, air vent 101, perforation 102, slide groove 103, target mark 104, mounting bracket 2, connecting strip 201, bolt hole 202, rubber pad strip 203, guide groove 204, bolt 3. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the present invention clear and complete, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only some, not all, embodiments of the present invention, and are merely illustrative of the embodiments of the present invention. They are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Example 1: This invention provides a technical solution: a box girder jacking construction device based on machine vision technology, comprising a machine vision acquisition module, a data processing and control module, an execution correction module, and a status monitoring module. These modules work collaboratively, resulting in a simple structure, convenient installation, and strong adaptability.

[0045] 1. Machine Vision Acquisition Module: This module includes multiple sets of high-definition industrial cameras, supplementary lighting sources, and benchmark calibration targets. The industrial cameras are deployed at the front, sides, and rear of the steel box girder during the jacking process, covering the entire construction area of ​​the steel box girder. The benchmark calibration targets are fixed at the preset construction benchmark axis and slide rail benchmark points, serving as references for image recognition and deviation calculation. The supplementary lighting sources are adaptable to day and night conditions as well as low-light conditions such as rain and fog, ensuring image acquisition clarity and avoiding interference from ambient light on recognition accuracy.

[0046] 2. Data Processing and Control Module: This module includes an industrial control computer, an image acquisition card, and a data transmission unit. It primarily receives visually acquired image data, performs image preprocessing, feature extraction, and deviation calculation, runs an adaptive correction algorithm, outputs precise correction control commands, and simultaneously provides data storage, real-time visualization, and over-limit alarm functions.

[0047] 3. Correction Module: This module includes multiple sets of main jacking cylinders, lateral fine-tuning correction cylinders, vertical leveling jacks, and flow control valves. The main jacking cylinders control the overall jacking and sliding speed and displacement of the steel box girder. The lateral correction cylinders are used to correct the horizontal axis deviation of the steel box girder. The vertical leveling jacks are used to correct vertical height differences. The flow control valves dynamically adjust the cylinder extension and retraction rates and strokes according to control commands, achieving fine-tuning.

[0048] 4. Status monitoring module: including displacement sensor, tilt sensor and pressure sensor, to help collect data on the sliding displacement, tilt angle and hydraulic cylinder thrust of the steel box girder, and to perform bidirectional verification with machine vision recognition data to improve the accuracy of deviation detection and avoid the error risks of single vision recognition.

[0049] Example 2, based on Example 1, proposes the following to facilitate adjusting the distance between the installation of the benchmark calibration target and the front end of the box girder jacking:

[0050] The reference calibration target includes a base plate 1, a target mark 104 is fixed on the front side of the base plate 1, a mounting frame 2 is slidably connected to the back side of the base plate 1, a guide groove 204 is formed on the surface of the mounting frame 2, a bolt 3 is inserted into the guide groove 204, and the bolt 3 penetrates through the base plate 1 and then is locked; a sliding groove 103 is formed on the back side of the base plate 1, the sliding groove 103 is a "convex"-shaped groove, a connecting strip 201 is inserted into the sliding groove 103, the connecting strip 201 is a "convex"-shaped strip, the connecting strip 201 is fixed on the surface of the mounting frame 2, the mounting frame 2 is an L-shaped plate structure, and one end of the mounting frame 2 is provided with a bolt hole 202; there are two guide grooves 204, the two guide grooves 204 are vertically symmetrically distributed about the connecting strip 201, two through holes 102 are formed on the surface of the base plate 1, the through holes 102 are in one-to-one correspondence with the guide grooves 204, two rubber cushion strips 203 are fixed on the surface of the mounting frame 2, and after the bolts 3 are locked, the rubber cushion strips 203 are clamped between the base plate 1 and the mounting frame 2 and deformed; four sets of air vents 101 are formed on the surface of the base plate 1, and the four sets of air vents 101 are distributed outside the four corners of the target mark 104.

[0051] Use an electric drill to drill a hole at the front end of the box girder for incremental launching through the bolt hole 202, then fix the mounting frame 2 at the front end of the box girder for incremental launching with bolt members, then fit the sliding groove 103 on the back side of the base plate 1 onto the connecting strip 201, push the base plate 1 to slide along the mounting frame 2 to a specified position, then pass the bolt 3 through the through hole 102 and the guide groove 204 and lock it. When the position of the base plate 1 needs to be adjusted again, just loosen the bolt 3, move the base plate 1 by pushing, and then screw the bolt 3 back to lock it.

[0052] Embodiment 3 is based on Embodiment 1, and proposes a box girder incremental launching construction method based on machine vision technology, which includes the following steps:

[0053] This embodiment is applied to the incremental launching construction of a high-speed long-span steel box girder, which crosses the South-to-North Water Diversion Project. The steel box girder is 144m long, 18m wide, and has a self-weight of 5640t. A multi-point synchronous incremental launching construction process is adopted, the preset incremental launching speed is 0.5m / min, the allowable axis offset threshold for construction is ≤2mm, the vertical height difference deviation threshold is ≤1mm, and the torsion angle threshold is ≤0.1°.

[0054] 1. Equipment layout

[0055] One 4K high-definition industrial camera is respectively arranged on the left and right sides of the front end of the steel box girder and at the center of the tail, and the camera frame rate is set to 30fps to ensure high-frequency image acquisition; high-contrast calibration targets are fixed at the reference axis of the incremental launching slideway and the reference position of the end of the steel box girder; displacement, inclination and pressure sensors are arranged at each incremental launching point, and the equipment is connected to the industrial personal computer through a wired + wireless dual-mode communication to ensure stable data transmission; the lateral deviation correction oil cylinder and the vertical leveling jack are correspondingly matched with each incremental launching point to meet the requirements of fine adjustment.

[0056] ① System calibration and debugging: Establish a construction reference coordinate system, complete the ratio conversion between camera pixels and actual size through calibration targets, and calibrate the imaging accuracy of the equipment; input the preset construction threshold and standard jacking parameters into the industrial control computer, complete the no-load test run of the system, and verify the image recognition accuracy, equipment communication stability and correction execution sensitivity.

[0057] ② Initial state detection: Before the formal jacking, the initial position image of the steel box girder is acquired through the machine vision system, and the initial alignment and attitude of the steel box girder are verified in combination with sensor data. After confirming that there is no initial deviation, the automatic jacking correction program is started.

[0058] ③ Dynamic image acquisition and preprocessing: During the jacking construction, industrial cameras continuously acquire dynamic images of the entire steel box girder. The supplementary lighting source is automatically turned on in rainy or backlit environments. The industrial control computer performs Gaussian noise reduction, distortion correction, and contrast enhancement on the real-time images to remove environmental interference and ensure that the image features are clear and recognizable.

[0059] ④ Multi-dimensional deviation identification and calculation: The outer contour of the steel box girder and the target feature points are extracted by the Canny edge detection algorithm, matched with the reference coordinate system, and the horizontal axis offset, vertical elevation difference, plane torsion angle and real-time slip velocity of the steel box girder are calculated in real time; combined with tilt angle and displacement sensor data, bidirectional verification is performed to correct minor errors in visual recognition and ensure that the deviation data is accurate and reliable.

[0060] ⑤ Adaptive correction control: When the detected deviation value is less than 1mm (minor deviation), the system starts the micro-adjustment mode, controls the correction cylinder to gradually correct at low speed and small stroke, and maintains continuous jacking operation; when the deviation value is between 1-2mm (normal deviation), the correction adjustment rate is dynamically increased to quickly correct the deviation and avoid the deviation from expanding; when the deviation value is greater than 2mm (over-limit deviation), the system immediately issues an audible and visual warning, automatically reduces the jacking speed to 0.3m / min, adopts a graded progressive correction strategy, gradually eliminates offset, height difference and torsional deviation, and restores the normal jacking speed after the parameters return to the threshold.

[0061] ⑥ Closed-loop feedback control: After each correction is executed, the machine vision system collects and updates the posture data of the steel box girder in real time, compares the parameter changes before and after the correction, and dynamically optimizes the adjustment amount for the next round of correction, forming a closed-loop adaptive control to avoid over-correction or under-correction.

[0062] ⑦ Construction data archiving and completion: After the single-segment steel box girder is pushed into place, the system automatically generates a construction data report, which includes the entire deviation change curve, correction parameters, equipment operating status and other data; the alignment accuracy of the steel box girder is manually checked, and after confirming that all indicators meet the requirements of the "Technical Specifications for Highway Bridge and Culvert Construction", the equipment is shut down and the jacking construction is completed.

[0063] After applying the technical solution of this invention in this embodiment, the maximum offset of the steel box girder jacking axis is ≤1.2mm, the vertical height difference deviation is ≤0.8mm, and the torsion angle is ≤0.08°. All accuracy indicators are superior to traditional construction processes. No manual monitoring and correction are required throughout the construction process, reducing labor costs by more than 60%, increasing the efficiency of single-section jacking construction by 20%, and eliminating problems such as deviation accumulation and abnormal structural stress. Construction safety and forming quality are greatly improved.

[0064] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A box girder jacking construction device based on machine vision technology, characterized in that: It includes a machine vision acquisition module, a data processing and control module, an execution correction module, and a status monitoring module. The four modules achieve bidirectional data interaction through an industrial bus and a wireless transmission unit, forming a closed-loop collaborative control architecture. The machine vision acquisition module is deployed throughout the entire area of ​​the box girder jacking construction to acquire real-time dynamic displacement images of the box girder and obtain high-definition image data of the box girder outline and the benchmark target. The data processing and control module is used to receive visually acquired images and status monitoring data, complete image preprocessing, feature extraction and multi-dimensional deviation calculation, and run an adaptive correction algorithm to generate precise control commands. The execution correction module is used to respond to control commands and simultaneously complete the horizontal axis correction, vertical height difference leveling and sliding speed control of the box girder; The status monitoring module is used to collect parameters such as box girder sliding displacement, tilt angle and hydraulic cylinder thrust, and to perform bidirectional verification with machine vision recognition results to eliminate the potential error of single vision recognition.

2. The box girder jacking construction device based on machine vision technology according to claim 1, characterized in that: The machine vision acquisition module includes high-definition industrial cameras deployed in multiple regions, an adaptive supplementary lighting source, and a reference calibration target. The high-definition industrial cameras are respectively installed at the front end of the box girder jacking, the outer sides of the two flange plates, and the rear end face. The camera field of view covers the entire construction area of ​​the box girder jacking, with no blind spots. The reference calibration target is fixed on the surface of the preset construction reference axis point and the slide reference elevation point. The target surface is set with high-contrast circular recognition feature points, which serve as the absolute reference for image recognition and coordinate conversion. The adaptive supplementary lighting source is equipped with a photosensitive sensor, which can automatically adjust the brightness and start / stop status according to the ambient light intensity, adapting to low-light conditions such as day and night construction, rainy and foggy days, and avoiding interference from ambient light with image recognition accuracy.

3. The box girder jacking construction device based on machine vision technology according to claim 2, characterized in that: The data processing and control module includes a high-performance industrial control computer, a multi-channel image acquisition card, and a 5G data transmission unit. The image acquisition card supports simultaneous access to image data from more than 8 industrial cameras, with a single-frame image transmission latency of less than 20ms. The industrial control computer has built-in image preprocessing programs, feature recognition algorithms, and fuzzy PID adaptive correction programs, which can calculate parameters such as the box girder axis position, vertical elevation, torsion angle, and sliding speed in real time. The control computer is equipped with a 10-inch touch screen display terminal, which can realize real-time visualization of the construction status. It is also equipped with an audible and visual alarm device, which immediately triggers an over-limit warning when any dimensional deviation exceeds a preset threshold.

4. The box girder jacking construction device based on machine vision technology according to claim 1, characterized in that: The execution correction module includes multiple distributed sets of main jacking cylinders, lateral fine-tuning correction cylinders, vertical leveling jacks, and electro-hydraulic proportional flow regulating valves. The main jacking cylinders are equidistantly arranged along the longitudinal direction of the bottom of the box girder to control the overall jacking and sliding speed and cumulative displacement of the box girder. The lateral fine-tuning correction cylinders are symmetrically installed on both sides of the box girder at the slide positions, with a single cylinder stroke set to ±50mm, to accurately correct the horizontal axis offset of the box girder. The vertical leveling jacks are arranged at the four corner points of the bottom of the box girder to synchronously correct the vertical height difference deviation of the box girder at the four points. The electro-hydraulic proportional flow regulating valve is connected to each set of cylinders in a one-to-one correspondence, and can dynamically adjust the cylinder extension and retraction rate and stroke accuracy according to control commands to achieve millimeter-level fine-tuning.

5. The box girder jacking construction device based on machine vision technology according to claim 1, characterized in that: The said state monitoring module comprises a high-precision cable displacement sensor, a dual-axis inclination sensor and a high-frequency pressure sensor; the displacement sensor is mounted at the end of the oil cylinder to collect the actual telescopic displacement data of the oil cylinder in real time; the dual-axis inclination sensor is fixed at the center of the top surface of the box girder to collect the parameters of the longitudinal inclination angle and the plane torsion angle of the box girder; the pressure sensor is installed on the oil inlet pipeline of the oil cylinder to collect the jacking force value of the oil cylinder in real time. The sampling frequency of all monitoring data is not lower than 100Hz, and the data is compared and verified with the machine vision recognition result in real time. When the data deviation exceeds 3%, the secondary recognition process is automatically triggered.

6. The box girder jacking construction device based on machine vision technology according to claim 2, characterized in that: The said reference calibration target comprises a base plate (1), a target mark (104) is fixed on the front side of the base plate (1), a mounting frame (2) is slidably connected to the back side of the base plate (1), a guide groove (204) is formed on the surface of the mounting frame (2), a bolt (3) is inserted into the guide groove (204), and the bolt (3) penetrates through the base plate (1) and then is locked.

7. The box girder jacking construction device based on machine vision technology according to claim 6, characterized in that: A sliding groove (103) is formed on the back side of the said base plate (1), the sliding groove (103) is a "convex"-shaped groove, a connecting strip (201) is inserted into the sliding groove (103), the connecting strip (201) is a "convex"-shaped strip, the connecting strip (201) is fixed on the surface of the mounting frame (2), the mounting frame (2) is in an L-shaped plate structure, and a bolt hole (202) is formed at one end of the mounting frame (2).

8. The box girder jacking construction device based on machine vision technology according to claim 7, characterized in that: There are two said guide grooves (204), the two guide grooves (204) are distributed symmetrically up and down about the connecting strip (201), two through holes (102) are formed on the surface of the base plate (1), the through holes (102) are in one-to-one correspondence with the guide grooves (204), two rubber cushion strips (203) are fixed on the surface of the mounting frame (2), and after the bolt (3) is locked, the rubber cushion strips (203) are clamped between the base plate (1) and the mounting frame (2) to deform.

9. A box girder jacking construction device based on machine vision technology according to claim 8, characterized in that: Four sets of air ports (101) are opened on the surface of the said base plate (1), and the four sets of air ports (101) are distributed outside the four corners of the target mark (104).

10. A box girder jacking construction method based on machine vision technology, employing the box girder jacking construction device based on machine vision technology as described in any one of claims 1-9, characterized in that: It includes the following steps: Step 1: Calibration in the early stage of construction and system initialization. After the prefabrication and assembly of the box girder is completed and the slide rails are laid in place, all reference calibration targets are fixed, the internal reference parameters, external reference coordinates and pixel scale of the industrial camera are calibrated, the global reference coordinate system for the box girder jacking construction is established, the communication link of each module is debugged, the jacking speed threshold, axis offset threshold, vertical height difference threshold and torsion angle threshold are preset, and the system accuracy verification is completed; Step 2: Synchronous acquisition of global dynamic images. After the box girder jacking construction is started, multiple groups of industrial cameras continuously collect dynamic images of the whole process of box girder jacking at a frequency of more than 15 frames per second, the fill light source automatically adjusts the output brightness according to the ambient light, and the state monitoring module synchronously collects auxiliary data such as oil cylinder pressure, box girder inclination and sliding displacement; Step 3: Image preprocessing and precise feature recognition. The industrial control computer performs Gaussian denoising, lens distortion correction and histogram enhancement on the original image in sequence to eliminate environmental noise and equipment imaging errors. The edge detection algorithm extracts the outline features of the box girder and the target feature points, matches the global reference coordinate system, and calculates the core parameters of the box girder's real-time axis position, vertical elevation, torsion angle and slip velocity. Step 4: Multi-dimensional deviation classification judgment. The real-time identification parameters are compared with the preset benchmark parameters, and the horizontal axis offset, vertical height difference deviation, plane torsion angle deviation and speed fluctuation deviation are calculated respectively. According to the proportion of the deviation to the threshold, the working condition is divided into three levels: minor deviation, normal deviation and over-limit deviation. Step 5: Intelligent matching of adaptive correction strategy. Based on the fuzzy PID algorithm, the corresponding correction scheme is dynamically generated according to the deviation level, deviation size and deviation change rate: for small deviation conditions, a micro-scale slow fine-tuning mode is adopted, and the single correction amount does not exceed 2mm to avoid excessive correction that may cause structural stress fluctuations. In normal deviation conditions, a uniform speed and precision correction mode is adopted, which adjusts the corresponding oil cylinder synchronously at a constant rate to quickly eliminate deviation within the preset stroke. In the case of excessive deviation, an audible and visual warning is immediately triggered, and the jacking speed is automatically reduced to less than 30% of the original speed. A graded progressive correction mode is adopted, which corrects the deviation in 3-5 steps to prevent sudden changes in local stress in the box girder. Step Six: Closed-loop dynamic control for correction. The industrial control computer converts the correction parameters into electro-hydraulic control commands, synchronously linking the horizontal correction cylinder, the vertical leveling jack, and the jacking main cylinder to complete horizontal offset correction, vertical height difference leveling, and sliding speed calibration in parallel. During the correction process, the machine vision module continuously collects images in real time and dynamically provides feedback on the correction effect, forming a closed-loop control process of "collection-analysis-correction-feedback" until all deviation parameters return to the preset safety threshold range. Step 7: Full-process data archiving and construction completion. The original image data, deviation records, correction parameters and equipment operation data are stored in real time throughout the entire process of box girder jacking. Traceable construction log reports are generated. After a single box girder segment is jacked into place, the overall alignment accuracy of the box girder is manually verified. After confirming that it meets the requirements of bridge construction specifications, all equipment operation is stopped, the single jacking correction construction is completed, and the next box girder jacking process begins.

Images