CN116008284A - Inspection device and application thereof - Google Patents

Abstract

The application belongs to the technical field of traffic, and particularly relates to a patrol device and application thereof. The existing inspection device is difficult to meet the requirements of rapid and accurate inspection of hidden defects of the high-speed railway ballastless track structure with the length of tens to hundreds kilometers, and the engineering research of technical equipment is insufficient. The application provides a patrol and examine device, including interconnect's defect vision collection module and data processing module, defect vision collection module is connected with electric power and drive module, defect vision collection module is including the first defect vision collection subassembly, second defect vision collection subassembly and the third defect vision collection subassembly that connect gradually, electric power and drive module can drive the removal of second defect vision collection subassembly, electric power and drive module do defect vision collection module and data processing module power supply. Fills the blank of technical equipment for quick, accurate identification and quantitative measurement of hidden defects of the current ballastless track structure.

Description

Inspection device and application thereof

Technical Field

The application belongs to the technical field of traffic, and particularly relates to a patrol device and application thereof.

Background

The ballastless track is used as the railway track structure type which is paved most widely in China high-speed rail, the service process is subjected to complex and various loads, such as temperature cyclic load, high-frequency train load, rainwater freeze thawing load and the like, and various types of structure hidden defects gradually develop and evolve along with the increase of the service time, and if the ballastless track is not timely detected and remedied, the ballastless track can induce typical dominant diseases such as high Wen Zhangban, unsmooth track and sudden degradation and the like, and the durability of the high-speed running and ballastless track structure is seriously influenced. The ballastless track damage detection system of the high-speed railway in China consists of periodic dynamic detection and static work detection of a large-scale detection vehicle, a disease display and medical treatment management maintenance mode is adopted for long-term detection of the ballastless track structure damage, the existing detection technical method has good effects on management and control and repair of dominant diseases of the track structure, but accurate identification and detection cannot be achieved on hidden defects of the ballastless track structure in an early form of the dominant diseases, long-term high-standard service quality of the high-speed railway is difficult to ensure, and the requirement of 'early discovery and early diagnosis' maintenance operation on the defects of the track traffic infrastructure set forth in a fourteen-five period cannot be met. The precise detection and identification equipment for hidden defects of ballastless tracks is urgent to develop. In recent years, research and development results of hidden defects of ballastless track structures are reported, such as laser scanning, infrared imaging, ultrasonic air coupling, elastic waves, geological radars and the like, however, most of technologies are in a theoretical experiment stage, corresponding equipment is handheld or hand-pushed, the problems of serious clamping of detection speed and precision and the like are solved, the requirements of rapid and accurate inspection of hidden defects of the ballastless track structures of high-speed rails with the length of tens to hundreds of kilometers are difficult to adapt, and engineering research of technical equipment is insufficient.

Disclosure of Invention

1. Technical problem to be solved

Aiming at the problems that the technology and equipment research and development results of some ballastless track structures are reported in recent years, such as laser scanning, infrared imaging, ultrasonic air coupling, elastic waves, geological radars and the like, most of the technologies are in a theoretical experiment stage, corresponding equipment is handheld or hand-pushed, detection speed-precision restriction is serious and the like, the requirements of rapid and accurate inspection of the hidden defects of the ballastless track structures with the length of tens or hundreds of kilometers and the length of high-speed rail are difficult to adapt, and the engineering research of technical equipment is insufficient.

2. Technical proposal

In order to achieve the above-mentioned purpose, the application provides a patrol and examine device, including interconnect's defect vision collection module and data processing module, defect vision collection module is connected with electric power and drive module, defect vision collection module is including the first defect vision collection subassembly, second defect vision collection subassembly and the third defect vision collection subassembly that connect gradually, electric power and drive module can drive the removal of second defect vision collection subassembly, electric power and drive module do defect vision collection module and data processing module power supply.

Another embodiment provided herein is: the second defect vision acquisition assembly comprises a mobile platform, the mobile platform comprises a chassis, be provided with a plurality of first laser light sources, a plurality of first 2D camera arrays and a plurality of first light filling light sources on the chassis, the chassis with electric power and drive module are connected, first defect vision acquisition assembly includes first universal frame, first universal frame with the chassis is connected, be provided with second laser light sources, second light filling light sources, first 2D camera, first thick inspection 3D camera and first fine inspection 3D camera on the first universal frame, third defect vision acquisition assembly includes the second universal frame, the second universal frame with the chassis is connected, be provided with third laser light sources, third light filling light sources, second 2D camera, second thick inspection 3D camera and second fine inspection 3D camera on the second universal frame, first universal frame with be provided with first cloud platform between the chassis, the second universal frame with be provided with the second cloud platform between the chassis, be provided with on the third universal frame a plurality of first cloud platform a plurality of first vibration damping arrays and a plurality of first vibration damping discs.

Another embodiment provided herein is: the mobile platform comprises a plurality of wheel pairs, the wheel pairs are coupled with steel rails, and the wheel pairs are connected with the chassis through damping springs.

Another embodiment provided herein is: the first universal frame comprises a first universal arm, a first universal joint, a second universal arm, a second universal joint and a third universal arm which are sequentially connected, and the first universal arm is connected with the chassis; the second universal frame comprises a fourth universal arm, a third universal joint, a fifth universal arm, a fourth universal joint and a sixth universal arm which are sequentially connected, and the fourth universal arm is connected with the chassis.

Another embodiment provided herein is: the first universal arm, the second universal arm, the third universal arm, the fourth universal arm, the fifth universal arm and the sixth universal arm are all telescopic, and fixing clamping grooves are formed in the first universal arm, the second universal arm, the third universal arm, the fourth universal arm, the fifth universal arm and the sixth universal arm.

Another embodiment provided herein is: the second laser light source, the second light supplementing light source, the first 2D camera, the first coarse detection 3D camera and the first fine detection 3D camera are all arranged in the fixed clamping groove, and the third laser light source, the third light supplementing light source, the second 2D camera, the second coarse detection 3D camera and the second fine detection 3D camera are all arranged in the fixed clamping groove.

Another embodiment provided herein is: the data processing module comprises an industrial personal computer, and a visual data storage unit and a data analysis unit are arranged in the industrial personal computer.

The application also provides an application of the inspection device, wherein the inspection device is used for identifying surface cracks and void off blocks of the track plates, the structures among the plates and the filling layers, and identifying ballastless track plates, interlayer gaps and quantitative measurement of the size.

Another embodiment provided herein is: collecting images, preprocessing the images, identifying the defects, guiding the images after the defect identification into a U-shaped semantic segmentation model, and combining cracking behaviors at the distribution positions of the images to realize accurate identification and crack size measurement of track plate cracks, plate-to-plate structure cracks, track plate-to-plate structure gaps and track plate-to-filling layer gaps; meanwhile, primarily identifying the different-surface gap between the track plate and the filling layer; the morphological image processing method is utilized to realize the preliminary identification of broken falling blocks and filling layer void falling blocks of the track plate and the structure between the plates; extracting only 3D point cloud information of the position corresponding to the preliminary identification defect to perform three-dimensional reconstruction, and performing precise identification and quantitative size detection of the preliminary identification; if the primary identification is that the top surface of the track slab is damaged and the structure between the track slab is damaged and the filling layer is empty, the 3D camera point cloud roughly detected at the corresponding position is called to carry out three-dimensional reconstruction, and the accurate identification and the volume quantitative calculation of the damaged and empty block are realized; if the preliminary identification is that the filling layer-supporting layer is different in surface off-seams, the 3D camera point cloud accurately detected at the corresponding position is called to carry out three-dimensional reconstruction on the lower part of the filling layer and the upper surface part of the base plate near the different surface off-seams, the abnormal height and the abnormal length of the spatial point position are identified, and the different surface off-seams are accurately identified, and the off-seam height and the length are quantitatively calculated.

3. Advantageous effects

Compared with the prior art, the beneficial effect of the inspection device that this application provided lies in:

the utility model provides a patrol and examine device is a quick accurate work patrol and examine device of ballastless track structure hidden defect based on machine vision technique.

The inspection device fills the blank of technical equipment for quick and accurate identification and quantitative measurement of hidden defects of the current ballastless track structure. The method realizes rapid and flexible acquisition and accurate detection of visual information of the hidden defects of the ballastless track structure in a short skylight period, improves the speed, efficiency and intelligent degree of the service inspection, and has obvious application and research values in the aspects of timely grasping the occurrence and development states of the hidden defects of the ballastless track structure, improving the defect renovation process and the like in a railway operation and maintenance department.

The inspection device provided by the application realizes synchronous improvement of detection speed and precision of hidden defects, improves detection efficiency and intelligent degree, and the detection process can not pollute and destroy track structures and surrounding environment, and has obvious research value for engineering application of the daily quick work inspection device for pushing the hidden defects of ballastless tracks.

Drawings

FIG. 1 is a schematic view of the inspection device of the present application;

FIG. 2 is a schematic perspective view of the inspection device of the present application;

FIG. 3 is a schematic structural view of the gimbal of the present application;

fig. 4 is a schematic diagram of a data processing flow of the data processing structure of the present application.

Detailed Description

Hereinafter, specific embodiments of the present application will be described in detail with reference to the accompanying drawings, and according to these detailed descriptions, those skilled in the art can clearly understand the present application and can practice the present application. Features from various embodiments may be combined to obtain new implementations or to replace certain features from certain embodiments to obtain other preferred implementations without departing from the principles of the present application.

Referring to fig. 1-4, the present application provides a patrol device, including interconnect's defect vision collection module and data processing module 14, defect vision collection module is connected with electric power and drive module 13, defect vision collection module is including the first defect vision collection subassembly, second defect vision collection subassembly and the third defect vision collection subassembly that connect gradually, electric power and drive module 13 can drive the second defect vision collection subassembly removes, because first defect vision collection subassembly, second defect vision collection subassembly and third defect vision collection subassembly connect gradually, so electric power and drive module 13 can make whole defect vision collection module remove through drive second defect vision collection subassembly, electric power and drive module 13 is for defect vision collection module and data processing module 14 power supply.

The second defect vision acquisition component realizes rapid acquisition, precise identification and measurement of gaps, cracks and broken falling blocks on the upper surfaces of the ballastless track plates 16 and the inter-plate structures 18; the first and third defective vision acquisition components enable accurate identification and dimensional measurement of side inter-panel gaps, inter-panel structures 18 and filling layer 15 side cracks, void off blocks.

The electric power and driving module 13 provides electric power and driving force for the defect vision acquisition module, so that the defect vision acquisition module can rapidly move along the railway line direction, and the moving speed is adjustable. Meanwhile, the power and driving module 13 can provide power for each element of the top surface and side surface defect vision acquisition assembly, the data processing module 14 and the speed and distance measuring element of the detection mobile platform 1 in a wired mode.

The data processing module 14 can be borne by a vehicle-mounted computer or a remote server, is responsible for acquisition, storage and analysis processing of the 2D images and the 3D point clouds of the top surface and side surface defect detection modules, and supports visual data transmission of the 2D and 3D cameras in a wired or wireless mode.

Preferably, the data processing module 14 analyzes the 2D image information of the top and side defect detection modules through image preprocessing and a deep learning algorithm to realize accurate identification and quantitative size detection of top cracks and inter-plate gaps of the top track plate and the inter-plate structure; accurately identifying and quantitatively measuring the side surface gap of the track plate-plate structure, the gap of the track plate-filling layer, the side surface of the plate-plate structure and the gap of the filling layer; and carrying out preliminary identification on broken falling blocks of the top surface track plate and the plate-to-plate structure, empty and broken side plate-to-plate structure and filling layer, and different surface gaps between the filling layer and the supporting layer.

Preferably, the data processing module 14 performs secondary identification and quantitative description of the size according to the result of the primary identification after analyzing the 2D image information of the top and side defect detection module, specifically, if the primary identification is that the top track slab and the inter-slab structure are damaged, or the inter-side slab structure and the filling layer are broken by void, the point cloud information of the 3D camera (coarse detection) at the corresponding position is called in combination with the mileage and time information corresponding to the defect, and the three-dimensional reconstruction of the side of the ballastless track structure is performed, so that the broken block and the broken block defect are accurately identified, and the quantitative calculation of the defect volume and the defect form is completed. If the abnormal surface gap between the filling layer and the supporting layer is primarily identified, 3D camera (fine inspection) point cloud information at the corresponding position is called, three-dimensional reconstruction of partial structures of the lower part of the filling layer and the upper surface of the base plate near the abnormal surface gap is carried out, abnormal height and length of the spatial point position are identified, the abnormal surface gap is accurately identified, and meanwhile quantitative calculation of the abnormal surface gap height and length is completed.

Further, the second defect vision collection assembly comprises a mobile platform 1, the mobile platform 1 comprises a chassis, a plurality of first laser light sources 2, namely top surface laser light sources, a plurality of first 2D camera arrays 3, namely top surface 2D camera arrays and a plurality of first light supplementing light sources 4, namely top surface light supplementing light sources, are arranged on the chassis, the chassis is connected with the power and driving module 13, the first defect vision collection assembly comprises a first universal frame 11, the first universal frame 11 is connected with the chassis, a second laser light source 5, namely side laser light sources, a second light supplementing light source 6, namely side light supplementing light sources, a first 2D camera 7, namely side surface 2D camera, a first coarse inspection 3D camera 8, namely 3D camera (coarse inspection) and a first fine inspection 3D camera 9, namely 3D camera (fine inspection), the third defect vision collection assembly comprises a second universal frame, the second universal frame is connected with the chassis, and a third light source, a third light supplementing camera 2D camera, a second fine inspection 3D camera and a second coarse inspection 3D camera are arranged on the second universal frame 11; a first vibration reduction cradle head 10 is arranged between the first universal frame 11 and the chassis, and a second vibration reduction cradle head is arranged between the second universal frame and the chassis; the ground is provided with a plurality of third vibration reduction cloud platforms, a plurality of first laser light 2 sources, a plurality of first 2D camera arrays 3 and a plurality of first light supplementing light sources 4 are arranged on the third vibration reduction cloud platforms. In the moving inspection process, vibration of the plurality of first laser light sources 2, the plurality of first 2D camera arrays 3 and the plurality of first light supplementing light sources 4 is reduced.

The electric power and driving module 13 provides electric power and driving force for the mobile platform 1, so that the mobile platform 1 can rapidly move along the railway line direction, and the moving speed can be adjusted. Meanwhile, the power and driving module 13 can provide power for each element of the top surface and side surface defect vision acquisition assembly, the data processing module 14 and the speed and distance measuring element of the detection mobile platform 1 in a wired mode.

The data processing module 14 is fixed on the mobile platform 1 through a rigid support, the data processing module 14 can be borne by a vehicle-mounted computer or a remote server, is responsible for acquisition, storage and analysis processing of 2D images and 3D point clouds of the top surface and side surface defect detection modules, and supports visual data transmission of the 2D and 3D cameras in a wired or wireless mode.

The vibration reduction cradle head transversely distributed on the lower surface of the chassis of the mobile platform 1 is provided with a top surface laser light source, a top surface 2D camera array and a top surface light supplementing light source. The system comprises a first defect vision acquisition assembly, a second defect vision acquisition assembly and a third defect vision acquisition assembly, wherein the first defect vision acquisition assembly consists of a top laser light source, a top 2D camera array and a top light supplementing light source; the two sides of the platform are symmetrically provided with side defect vision acquisition components (a first defect vision acquisition component or a third defect vision acquisition component) consisting of a side laser light source, a side light supplementing light source, a side 2D camera, a 3D camera (coarse inspection) and a 3D camera (fine inspection) through universal frames; the full coverage from the edge line on the side surface of the track plate to the boundary line between the different surfaces of the filling layer and the base plate is vertically realized in the acquisition range of the side surface 2D camera and the 3D camera (rough detection); the field of view of the 3D camera (fine inspection) is vertically and fully covered by the boundary line between the filling layer and the base plate layer. The lateral laser light source, the lateral light supplementing light source, the lateral 2D camera, the 3D camera (coarse inspection) and the 3D camera (fine inspection) are not invaded into the transverse and vertical dimensions of the lateral stop block from the structural distance of the ballastless track; the field of view of the 3D camera (fine inspection) is vertically and fully covered by the boundary line between the filling layer and the base plate layer.

The image range of the top surface 2D camera array, i.e. the first 2D camera array 3, transversely covers the width of the track plate, the image resolution must meet the recognition accuracy of the cracks of the track plate 16, the inter-plate structure 18 and the inter-plate gap, the transverse overlapping duty ratio of the image edges of the adjacent 2D cameras meets the image splicing requirement, and the number of the 2D cameras is determined by the conditions. The number of laser sources on the top surface, the fixed position and the angle of the upper surface of the laser incidence track plate are required to acquire the image information of the 2D camera array. The number and positions of the top surface light supplementing light sources are determined according to the acquisition requirement of the 2D camera array.

And the space positions of the sensing elements of the lateral defect detection module are changed by adjusting the angles of the rotary universal joint, the telescopic rigid arm and the rotary clamping groove, so that the lateral defect detection module bypasses a space region of the convex baffle table, and the visual angle of the side 2D camera is vertical to the side of the track plate. The vertical acquisition range of the incident laser of the side laser light source, the 2D camera image information and the point cloud information of the 3D camera (coarse inspection) is required to cover the edge line on the side surface of the track plate and the boundary line between the filling layer and the supporting layer. The vertical acquisition range of the 3D camera (fine inspection) point cloud information is concentrated on the boundary line between the filling layer and the supporting layer. The side light supplementing light source is used for adjusting illumination intensity required by 2D image information and 3D point cloud acquisition.

Further, the mobile platform 1 comprises a plurality of wheel sets, the wheel sets are coupled with steel rails, and the wheel sets are connected with the chassis through damping springs 12. The mobile platform 1 is a mobile platform capable of realizing electric drive, the mobile speed is adjustable, the chassis rigidity is high, the deformation of the platform itself in the mobile inspection process is negligible, the power mobile platform is coupled with the steel rail through a wheel pair, the rapid movement on a track is met, the mobile platform has the rapid disassembly and assembly function on a railway line in a short time, and the mobile platform has the manned mobile capability.

Through wheel pair and rail coupling contact, install speed sensor, meter axle and be used for gathering record platform travel speed, mileage, time. During movement, the motion of the chassis may be considered rigid motion. The chassis and wheel sets of the mobile platform 1 are vertically provided with one-stage or multi-stage vibration reduction structures for vibration reduction of the mobile platform 1 (specifically, the 12 position of fig. 1 can be referred to as 12 position, and here, springs or other vibration reduction combinations can be used for realizing one-stage or multi-stage vibration reduction), the center of the upper surface of the two sides of the chassis of the mobile platform 1 is longitudinally, and a vibration reduction holder capable of installing sensing elements and universal frames is fixed at a plurality of points on the lower surface transversely.

And the third vibration reduction holders for installing a plurality of first laser light sources 2, namely top surface laser light sources, a plurality of first 2D camera arrays 3, namely top surface 2D camera arrays and a plurality of first light supplementing light sources 4, namely top surface light supplementing light sources are transversely arranged at multiple positions on the two sides and the lower surface of the upper surface of the chassis, and the chassis of the mobile platform 1 is connected with the wheel set through a damping structure of a vibration reduction spring 12. In the quick inspection process of the device, a plurality of first laser light sources 2, namely top surface laser light sources, irradiate on the upper surfaces of ballastless track plates and structures between plates, a plurality of first 2D camera arrays 3, namely top surface 2D camera arrays, transversely cover the width of the track plates, receive emitted light of the laser light sources on the upper surfaces of the track plates and structures between plates, acquire two-dimensional image data of the upper surfaces of the ballastless tracks, and a first light supplementing light source 4, namely a top surface light supplementing light source, is used for adjusting illumination intensity of a camera view field. And then the data processing module 14 is used for accurately identifying and measuring the gap between the plates, the track plate and the cracking of the plate-to-plate structure, and simultaneously, the damaged falling blocks of the top surface track plate and the plate-to-plate structure are primarily identified.

Further, the first gimbal 11 includes a first gimbal arm, a first gimbal, a second gimbal and a third gimbal that are sequentially connected, where the first gimbal arm is connected to the chassis through a first vibration-reduction cradle head 10; the second universal frame comprises a fourth universal arm, a third universal joint, a fifth universal arm, a fourth universal joint and a sixth universal arm which are sequentially connected, and the fourth universal arm is connected with the chassis. The universal frame has enough rigidity, and the universal frame tip can be installed on the damping cloud platform of mobile platform 1 upper surface both sides, provides secondary damping for side defect vision acquisition assembly. The universal frame is provided with at least two universal joints (three universal arms), each universal arm is telescopic, and rotatable sensor fixing clamping grooves are formed in the universal frame. The clamping groove can be used for clamping and fixing any sensor with proper size. The sensors to be fixed in this application are the second laser light source 5 in fig. 1, i.e. the side laser light source, the second light supplementing light source 6 in fig. 1, i.e. the side light supplementing light source, the first 2D camera 7 in fig. 2D camera, the first coarse inspection 3D camera 8 in fig. 3D camera (coarse inspection) and the first fine inspection 3D camera 9 in fig. 3D camera (fine inspection). In the quick inspection process of the device, a second laser light source 5, namely a side laser light source irradiates the side surfaces of the track plate and the filling layer, the laser completely covers the side surfaces of the track plate and the filling layer (as shown in figure 1) in a range, reflected light is collected by a first 2D camera 7, a first coarse inspection 3D camera 8 and a first fine inspection 3D camera 9, and the side light supplementing light source supplements light. The first 2D camera 7 and the first coarse inspection 3D camera 8 respectively acquire two-dimensional images and three-dimensional point cloud data of the side surfaces of the track plate and the filling layer. The data analysis of the first 2D camera 7 realizes the accurate identification and the quantitative measurement of the side surface gap of the track plate-plate structure, the gap of the track plate-filling layer, the side surface of the plate-plate structure and the gap of the filling layer, and can only primarily identify the gap of the different surfaces of the filling layer-supporting layer; the data analysis of the first coarse inspection 3D camera 8 realizes the accurate identification and volume measurement of the side panel structure and the broken and empty block of the filling layer. The first fine inspection 3D camera 9 focuses on the field of view and covers the intersection line position of the different surfaces of the filling layer and the supporting layer. When the out-of-plane seam of the filling layer-supporting layer is initially identified, the three-dimensional point cloud acquired by the first fine inspection 3D camera corresponding to the mileage is called, and the data processing module 14 performs point cloud reconstruction and realizes accurate identification and size measurement of the out-of-plane seam of the filling layer-supporting layer through the abnormal relative height and abnormal transverse depth of the point cloud space.

Further, the first universal arm, the second universal arm, the third universal arm, the fourth universal arm, the fifth universal arm and the sixth universal arm are all telescopic, and fixing clamping grooves are formed in the first universal arm, the second universal arm, the third universal arm, the fourth universal arm, the fifth universal arm and the sixth universal arm. All components and parts in the side defect vision collection subassembly are installed on the rotatable sensor fixed slot of universal frame, through adjusting the universal joint, side 2D, 3D camera gather the reflection laser signal that side laser source shines at track board, inter-plate structure, filling layer side, and side light filling source adjusts ambient brightness.

Further, the second laser light source 5, the second light supplementing light source 6, the first 2D camera 7, the first coarse inspection 3D camera 8 and the first fine inspection 3D camera 9 are all arranged in the fixed clamping groove, and the third laser light source, the third light supplementing light source, the second 2D camera, the second coarse inspection 3D camera and the second fine inspection 3D camera are all arranged in the fixed clamping groove.

Further, the data processing module 14 includes an industrial personal computer, and a visual data storage unit and a data analysis unit are disposed in the industrial personal computer. The industrial personal computer drives the camera to rapidly acquire and store all 2D images and 3D point cloud data in the top surface defect visual acquisition assembly and the side surface defect visual acquisition assembly under the rapid movement of the platform, the data transmission supports a wireless or wired mode, and the data analysis program can selectively perform hidden defect identification detection on stored data in an online or offline mode.

According to typical hidden defect sizes and distribution characteristics of a typical ballastless track, 2D images and 3D point cloud visual sensing equipment are carried on a track moving platform, and quick acquisition and storage of visual information of the hidden defects under quick movement are realized by utilizing a machine visual processing technology, and accurate identification of defect types and quantitative analysis of defect morphology sizes in various modes on line or off line can be realized. The detection technology and the detection equipment are suitable for ballastless tracks in various forms, fill in the blank of the current ballastless track structure recessive defect quick and accurate service inspection technology and the detection equipment, and provide the quick service inspection operation requirement of early discovery and early diagnosis of railway infrastructure defects in fourteen-five stages.

Examples

The application discloses a quick and accurate work inspection device for hidden defects of a ballastless track structure based on a machine vision technology; fig. 1 is a cross-sectional view of a quick and accurate work inspection device for hidden defects of a ballastless track structure based on a machine vision technology. The mobile platform has the functions of quick disassembly and assembly in a short time on a railway line and the manned capacity, the chassis and the wheel sets are made of rigid materials, and the length, the width and the thickness of the chassis are 2.5m multiplied by 1.6m multiplied by 0.1m. The power and driving module 13 provides power, track advancing power and speed adjusting functions for the mobile platform, and simultaneously provides electric energy for each sensing element of the top surface and side surface defect vision acquisition module in a wired mode, so that the cruising ability of at least 5 hours is met.

The chassis of the mobile platform shown in fig. 1 and 2 is longitudinally 1.25m on the upper surface and is 10cm away from the left edge and the right edge, and 7 vibration-damping holders provided with sensing elements are longitudinally 1.25m on the lower surface and are transversely and approximately equidistantly fixed, and the mobile platform 1 realizes the connection between the chassis and the wheel pair and the vibration damping of the platform through a vibration-damping spring damping structure;

the top surface defect vision acquisition module shown in fig. 1 and 2 consists of 2 laser light sources, 8 2D cameras and 2 light supplementing light sources, and is connected with a chassis through a vibration reduction cradle head, wherein the image acquisition range of the image acquisition array formed by the 2D cameras is fully covered on the surface of the plate, the transverse vision boundary parts of the adjacent 2D cameras are overlapped, and the acquisition of the upper surface image information of the track plate and the plate-to-plate structure is met;

the universal frame shown in fig. 3 is composed of three telescopic rigid arms and two universal joints, each rigid arm can realize three-dimensional rotation, a rotatable clamping groove for installing a sensing element is formed in each rigid arm, and the end part of the universal frame is fixed to the vibration reduction cradle head on the upper surface of the chassis of the mobile platform, so that secondary vibration reduction of the side defect vision acquisition module is realized. Because the two sides of the ballastless track plate on the bridge are provided with the convex baffle tables, the universal frame is wound around the convex baffle tables through the expansion and contraction of the rigid arms and the rotation of the three-dimensional rotation and clamping grooves in order to ensure the universality of quick and accurate detection of hidden defects of the ballastless track structure on the roadbed and the bridge by the detection device. Meanwhile, the full coverage from the upper edge line of the side surface of the track plate to the boundary line between the different surfaces of the filling layer and the base plate is ensured to be vertically realized in the acquisition range of the side surface 2D camera (7) and the 3D camera (rough inspection); the field of view of the 3D camera (fine inspection) is vertically and fully covered by the boundary line between the filling layer and the base plate layer.

The top and side defect vision collection modules shown in fig. 1 and 2 are designed as a light supplementing light source, a laser light source, a 2D camera and a 3D camera. The light supplementing light source is used for adjusting the ambient illumination intensity required by top surface and side surface hidden defect image and point cloud information acquisition, the color temperature range is 5000-7000k, the laser light source irradiates on the surfaces of ballastless track plates, plate-to-plate structures and filling layers, reflected laser is provided for acquiring image and three-dimensional point cloud information by the 2D and 3D cameras, the laser wavelength is 638nm/405nm and the power is more than 10W, the resolution of the 2D camera meets 4000 x 3000, the exposure time is 2-14 mu s, the number of vertical space points of the 3D camera is not less than 500 points, the line direction is not less than 2000 points, 2500profile/s, the resolution of the 2D image and the 3D point cloud meet hidden defect detection requirements, and the light supplementing light source is applicable to visual information acquisition under the moving condition of not lower than 40km/h at most.

The data processing modules shown in fig. 1 and 2 are borne by a vehicle-mounted computer fixed on a support of the mobile platform 1, support a wired or wireless mode to realize real-time transmission, storage and processing of 2D images and 3D point cloud detection data in the visual acquisition modules of the defects of the top surface and the side surface, and support the wired or wireless mode to carry out data transmission.

Fig. 4 is a schematic flow diagram of hierarchical analysis and utilization of information collected by the 2D and 3D cameras of the data processing module shown in fig. 1, wherein the hierarchical utilization of 2D images with larger data volume and 3D point cloud information satisfies high-speed collection of visual data, and intelligent and accurate identification and measurement of hidden defect characteristics of ballastless tracks are also ensured. The method comprises the following steps:

firstly, image preprocessing including distortion correction, bilateral filtering denoising, thresholding and binarization is carried out on 2D image information in the top surface and side surface defect vision acquisition modules.

And then, introducing the preprocessed data into a semantic segmentation model oriented to an attention mechanism, and combining the distribution positions of the cracking behaviors in the images to realize accurate identification and crack size measurement on the top surface cracks of the track plates, the top surface and side surface cracks of the inter-plate structures, the surface cracks of the filling layers, the separation seams of the track plates and the inter-plate structures and the separation seams of the track plates and the filling layers. Meanwhile, the different surface separation between the track plate and the filling layer is primarily identified. Meanwhile, the morphological image processing method is utilized to realize the preliminary identification of broken blocks on the top surface and the side surface of the track plate and the plate-to-plate structure and the empty blocks of the filling layer.

And finally, only extracting 3D point cloud information of the position corresponding to the primary identification defect to perform three-dimensional reconstruction, and performing accurate identification and size quantitative detection of the primary identification. If the primary identification is that the top surface of the track slab is damaged and the top surface and the side surface of the inter-slab structure are damaged and the filling layer is emptied, the 3D camera (rough detection) point cloud at the corresponding position is called to carry out three-dimensional reconstruction, and accurate identification and volume quantitative calculation of the damaged and emptied blocks are realized. If the preliminary identification is that the filling layer-supporting layer is different in surface off-seams, the 3D camera (fine inspection) point cloud at the corresponding position is called to reconstruct the lower part of the filling layer and the upper surface part of the base plate near the different surface off-seams in a three-dimensional manner, the abnormal height and the length of the spatial point position are identified, and the different surface off-seams are accurately identified, and the off-seam height and the length are calculated quantitatively.

Although the present application has been described with reference to particular embodiments, those skilled in the art will appreciate that many modifications are possible in the principles and scope of the disclosure. The scope of the application is to be determined by the appended claims, and it is intended that the claims cover all modifications that are within the literal meaning or range of equivalents of the technical features of the claims.

Claims (9)

1. The utility model provides a patrol and examine device which characterized in that: the defect vision acquisition module is connected with the electric power and driving module, the defect vision acquisition module comprises a first defect vision acquisition assembly, a second defect vision acquisition assembly and a third defect vision acquisition assembly which are sequentially connected, the electric power and driving module can drive the second defect vision acquisition assembly to move, and the electric power and driving module supplies power for the defect vision acquisition module and the data processing module.

2. The inspection device of claim 1, wherein: the second defect vision acquisition assembly comprises a mobile platform, the mobile platform comprises a chassis, a plurality of first laser light sources, a plurality of first 2D camera arrays and a plurality of first light supplementing light sources are arranged on the chassis, the chassis is connected with the power and driving module, the first defect vision acquisition assembly comprises a first universal frame, the first universal frame is connected with the chassis, a second laser light source, a second light supplementing light source, a first 2D camera, a first coarse inspection 3D camera and a first fine inspection 3D camera are arranged on the first universal frame, the third defect vision acquisition assembly comprises a second universal frame, the second universal frame is connected with the chassis, and a third laser light source, a third light supplementing light source, a second 2D camera, a second coarse inspection 3D camera and a second fine inspection 3D camera are arranged on the second universal frame; the first universal frame with be provided with first damping cloud platform between the chassis, the second universal frame with be provided with the second damping cloud platform between the chassis, be provided with a plurality of third damping cloud platform on the chassis, a plurality of first laser source, a plurality of first 2D camera array and a plurality of first light filling light source sets up in a plurality of on the third damping cloud platform.

3. The inspection device of claim 2, wherein: the mobile platform comprises a plurality of wheel pairs, the wheel pairs are coupled with steel rails, and the wheel pairs are connected with the chassis through damping springs.

4. A patrol apparatus according to claim 3, wherein: the first universal frame comprises a first universal arm, a first universal joint, a second universal arm, a second universal joint and a third universal arm which are sequentially connected, and the first universal arm is connected with the chassis; the second universal frame comprises a fourth universal arm, a third universal joint, a fifth universal arm, a fourth universal joint and a sixth universal arm which are sequentially connected, and the fourth universal arm is connected with the chassis.

5. The inspection device of claim 4, wherein: the first universal arm, the second universal arm, the third universal arm, the fourth universal arm, the fifth universal arm and the sixth universal arm are all telescopic, and fixing clamping grooves are formed in the first universal arm, the second universal arm, the third universal arm, the fourth universal arm, the fifth universal arm and the sixth universal arm.

6. The inspection device of claim 5, wherein: the second laser light source, the second light supplementing light source, the first 2D camera, the first coarse detection 3D camera and the first fine detection 3D camera are all arranged in the fixed clamping groove, and the third laser light source, the third light supplementing light source, the second 2D camera, the second coarse detection 3D camera and the second fine detection 3D camera are all arranged in the fixed clamping groove.

7. The inspection device of claim 6, wherein: the data processing module comprises an industrial personal computer, and a visual data storage unit and a data analysis unit are arranged in the industrial personal computer.

8. Use of the inspection device according to any one of claims 1 to 7, characterized in that: the inspection device is used for identifying cracks, inter-plate structures and surface cracks of the filling layer, removing empty blocks, and quantitatively measuring the sizes of gaps and cracks among ballastless track plates and between layers.

9. Use of a patrol apparatus according to claim 8, wherein: collecting images, preprocessing the images, identifying the defects, guiding the images after the defect identification into a U-shaped semantic segmentation model, and combining cracking behaviors at the distribution positions of the images to realize accurate identification and crack size measurement of track plate cracks, plate-to-plate structure cracks, track plate-to-plate structure gaps and track plate-to-filling layer gaps; meanwhile, primarily identifying the different-surface gap between the track plate and the filling layer; the morphological image processing method is utilized to realize the preliminary identification of broken falling blocks and filling layer void falling blocks of the track plate and the structure between the plates; extracting only 3D point cloud information of the position corresponding to the preliminary identification defect to perform three-dimensional reconstruction, and performing precise identification and quantitative size detection of the preliminary identification; if the primary identification is that the top surface of the track slab is damaged and the structure between the track slab is damaged and the filling layer is empty, the 3D camera point cloud roughly detected at the corresponding position is called to carry out three-dimensional reconstruction, and the accurate identification and the volume quantitative calculation of the damaged and empty block are realized; if the preliminary identification is that the filling layer-supporting layer is different in surface off-seams, the 3D camera point cloud accurately detected at the corresponding position is called to carry out three-dimensional reconstruction on the lower part of the filling layer and the upper surface part of the base plate near the different surface off-seams, the abnormal height and the abnormal length of the spatial point position are identified, and the different surface off-seams are accurately identified, and the off-seam height and the length are quantitatively calculated.

Images