A rail fastener intelligent inspection method and intelligent inspection vehicle

Abstract

The present invention provides a track fastener intelligent inspection method and an intelligent inspection trolley, which are applied to the field of track inspection technology, and include the following steps: obtaining an induction pulse of a benchmark, a benchmark picture, and a two-dimensional image and a three-dimensional depth image of the track fastener; determining the geographical location of the benchmark according to the induction pulse count and the pole number character of the benchmark; obtaining a detection result and position information of the track fastener according to the two-dimensional image of the track fastener and the three-dimensional depth image of the track fastener, the position information including the image coordinates and the pulse number of the fastener; calculating the relative position of the benchmark and the track fastener according to the geographical location of the benchmark and the pulse number of the track fastener to infer the geographical location of the track fastener; judging whether the track fastener is qualified, and if unqualified, giving an alarm and outputting the geographical location of the corresponding track fastener; detecting and judging the shape of the fastener and the height of the nut by collecting the fastener information, locating the fastener in combination with the benchmark information, and determining the accurate position of the fastener.

Description

Technical Field

[0001] The present invention belongs to the technical field of track inspection, and in particular relates to an intelligent inspection method for track fasteners and an intelligent inspection trolley. Background Art

[0002] Track inspection is an important means to ensure railway safety and improve transportation efficiency. Through regular inspection of tracks, defects and hidden dangers of the tracks can be discovered in time, and effective maintenance measures can be taken to ensure the smooth progress of railway transportation.

[0003] Rail fasteners are important parts of rails. Usually, rail intelligent inspection carts are used to inspect rails and their accessories. Rail intelligent inspection carts (hereinafter referred to as "inspection carts") can detect and identify rail defects and diseases, locate and locate them, evaluate quality, and predict and monitor trends. Through real-time detection, comprehensive analysis, and data mining, the correlation between various professional detection data can be found, and the development laws and status change trends of rail diseases can be mastered. It has the characteristics of comprehensiveness, efficiency, and real-time, which can reduce cost investment, greatly improve detection efficiency, reduce the frequency of personnel on-track operations, guide track maintenance, extend the service life of track facilities, and ensure operational safety.

[0004] At present, most of the rail fastener installation status inspection instruments on the market are used to inspect rail fasteners. This method is not convenient for personnel to work for a long time, the fatigue strength is too high, and the degree of automation, digitization and intelligence is insufficient. It only stays in the manual operation stage, which is not conducive to the liberation of labor and vigorous promotion.

[0005] At present, with the rapid development of my country's high-speed railways and urban rail transit, their safety issues have become increasingly important. However, my country's technology in rail transit safety detection and assurance is relatively traditional, and many aspects still rely on manual inspections. With the rapid development of the rail transit industry today, mechanized and automated rail transit safety detection technology has become increasingly the focus. Summary of the invention

[0006] In view of the above problems in the prior art, the purpose of the present invention is to provide an intelligent inspection method for rail fasteners, which can detect and judge the fastener shape and nut height by collecting fastener information, and locate the fastener in combination with benchmark information to determine the exact position of the fastener.

[0007] A rail fastener intelligent inspection method is used to inspect the rail fasteners and determine the geographical location of the corresponding rail fasteners according to the benchmarks set beside the track, specifically comprising the following steps:

[0008] Information collection: obtain the induction pulse of the current benchmark, the picture of the benchmark, and the two-dimensional image and three-dimensional depth image of the track fastener;

[0009] Benchmark information processing: Obtain the benchmark's induction pulse count and the benchmark's pole number character based on the collected information to determine the benchmark's geographical location;

[0010] Fastener information processing: obtaining the detection result and position information of the rail fastener according to the rail fastener 2D image and the rail fastener 3D depth image, wherein the position information includes the image coordinates and pulse number of the fastener;

[0011] Locating fasteners: Calculate the relative position of the benchmark and the track fastener based on the geographical location of the benchmark and the pulse number of the track fastener, thereby inferring the geographical location of the track fastener;

[0012] The rail fasteners are inspected based on their two-dimensional images and three-dimensional depth images to determine whether they are qualified. If they are unqualified, an alarm is issued and the geographical location of the corresponding rail fasteners is output. If they are qualified, the inspection vehicle continues to move forward to complete the inspection.

[0013] Preferably, a picture of the current benchmark is obtained by taking pictures continuously with a benchmark camera, and the induction pulses emitted by the sensor on the benchmark are received and counted by a pulse receiver, where the sensor on the benchmark is a photoelectric switch; each benchmark is equipped with a corresponding unique pole number, and image recognition is performed on the benchmark picture to obtain the pole number characters; the geographical location of the current benchmark is inferred by the pole number characters and the induction pulse count.

[0014] Preferably, a two-dimensional image of the current fastener is obtained through a fastener camera. In addition, when the fastener camera takes a two-dimensional image of the track fastener, the pulse numbers of the track fasteners are accumulated in sequence; a three-dimensional depth image of the current fastener is obtained through a constructed structured light system; the track fastener includes a left fastener and a right fastener, and image processing is performed on the two-dimensional image of the current fastener to achieve fastener classification and contour extraction, so as to determine whether the current fastener is qualified; the nut height and spatial position of the current fastener are obtained through the structured light system, and whether the current fastener is qualified is determined according to the nut height, and the corresponding image coordinates are determined according to the spatial position of the current fastener.

[0015] Preferably, the structured light system includes a structured light projector and a system camera. The system camera and the structured light projector are installed at a set angle. The structured light projector emits line structured light to the surface of the object so that the line structured light plane is tangent to the surface of the object, forming a light strip on the surface of the object.

[0016] Preferably, the unqualified fastener conditions include that the left fastener and the right fastener are installed in reverse positions, the fastener is damaged, the nut is not installed, and the nut height is not within a threshold range.

[0017] Another object of the present invention is to provide a track fastener intelligent inspection trolley, a track fastener intelligent inspection trolley comprising:

[0018] The benchmark detection module is used to obtain and count the induction pulses output by the benchmark, obtain the benchmark image and perform image recognition on the benchmark image to obtain the pole number characters, and infer the current geographic location of the benchmark through the pole number characters and the induction pulse count;

[0019] The fastener detection module is used to obtain a two-dimensional image of the rail fastener and perform image processing on the two-dimensional image of the current fastener to achieve fastener classification and contour extraction; at the same time, when the fastener camera is set to shoot the two-dimensional image of the rail fastener, the pulse numbers of the rail fastener are accumulated in sequence; the three-dimensional depth image of the rail fastener is obtained to obtain the nut height and spatial position of the current fastener, and the spatial position reflects the image coordinates of the current fastener;

[0020] The central control module is used to determine whether the current rail fastener is unqualified based on the processing results of the two-dimensional image and the three-dimensional depth image of the rail fastener; calculate the relative position of the benchmark and the rail fastener based on the geographical location of the benchmark and the pulse number of the rail fastener, so as to infer the geographical location of the rail fastener;

[0021] When it is detected that the rail fastener is unqualified, the central control module drives the alarm module to sound an alarm and output the geographical location of the corresponding rail fastener.

[0022] Preferably, the inspection trolley also includes a driving module and a power supply module. The driving module is used to drive the trolley to run along the track so as to inspect the track fasteners in sequence. The driving module and the chassis are positioned by a latch and locked by a toggle clamp mechanism. The power supply module is used to provide corresponding power supply to the inspection trolley and is placed vertically downward on the chassis through an external shape limiter.

[0023] Preferably, the benchmark detection module includes a benchmark camera and a pulse receiver, the benchmark camera is used to obtain a benchmark image, and a photoelectric switch for outputting an induction pulse is installed on the benchmark; the fastener detection module includes a fastener camera and a structured light system, the fastener camera is used to obtain a two-dimensional image of the track fastener, and the structured light system is used to obtain a three-dimensional image of the track fastener, and the structured light system includes a structured light projector and a system camera, and the system camera and the structured light projector are installed at a set angle, and the structured light projector emits line structured light to the surface of the object, so that the line structured light plane is tangent to the surface of the object, forming a light strip on the surface of the object.

[0024] Preferably, unqualified fastener conditions include the left fastener and the right fastener being installed in reverse positions, the fastener being damaged, the nut not being installed, and the nut height not being within a threshold range.

[0025] The beneficial effects of the present invention are as follows: the intelligent inspection method for rail fasteners collects information about the rail fasteners, uses image recognition technology to identify and analyze the two-dimensional images of the rail fasteners, and determines whether the rail fasteners have problems such as reverse installation or external damage; at the same time, a constructed structured light system is used to collect three-dimensional depth images of the rail fasteners, and measures whether the nuts on the fasteners are lost or loose, thereby realizing intelligent inspection of the rail fasteners and improving detection efficiency.

[0026] Furthermore, the fastener is positioned in combination with the benchmark information. The current position of the benchmark is determined by collecting the benchmark's induction pulses and pole number characters. The relative position between the current benchmark and the current track fastener is calculated to determine the exact position of the track fastener, which facilitates fastener positioning.

[0027] The rail fastener intelligent inspection trolley adopts a lightweight integrated design, combining the benchmark detection module, the fastener detection module and the central control module, and utilizing the fusion of two-dimensional images and three-dimensional depth images to achieve target recognition through two-dimensional images and precise measurement through three-dimensional images. The benchmark detection module is combined with the current rail fastener to locate the position, which can effectively eliminate the accumulated error of the trolley's travel and achieve precise positioning of fasteners, interfaces and other accessories. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0029] Figure 1 It is a flowchart of the present invention;

[0030] Figure 2 It is an image output after the benchmark information of the present invention is processed;

[0031] Figure 3 is a two-dimensional image of the rail fastener collected by the present invention;

[0032] Figure 4 is a three-dimensional depth image of the rail fastener collected by the present invention;

[0033] Figure 5 It is an image output after the track fastener information of the present invention is processed;

[0034] Figure 6 It is a schematic diagram of an ideal linear structured light model in the prior art;

[0035] Figure 7 is a schematic diagram of a structured light system model of the present invention;

[0036] Figure 8 It is a structural schematic diagram of the intelligent inspection vehicle of the present invention.

[0037] The markings in the marking diagram are: 1. Benchmark detection module; 2. Fastener detection module; 3. Central control module; 4. Drive module; 5. Power supply module; 6. Chassis. DETAILED DESCRIPTION

[0038] Embodiment 1

[0039] like Figure 1 As shown, a rail fastener intelligent inspection method is used to inspect the rail fasteners and determine the geographical location of the corresponding rail fasteners according to the benchmarks set beside the track, so as to facilitate the positioning of the rail fasteners. The inspection method includes the following steps:

[0040] Collecting information: During the operation of the inspection vehicle, the sensing pulse of the current benchmark, the image of the benchmark, the two-dimensional image of the track fastener, and the three-dimensional depth image of the track fastener are obtained. Among them, the benchmark information is collected through the benchmark detection module 1, which includes a benchmark camera and a pulse receiver. The benchmark camera continuously takes pictures to obtain the image of the current benchmark, and the pulse receiver receives and counts the sensing pulses emitted by the sensor on the benchmark. In this embodiment, the sensor on the benchmark is set as a photoelectric switch. In addition, the fastener information is collected through the fastener detection module 2, which includes a fastener camera and a structured light system. The structured light system includes a structured light projector and a system camera. The fastener camera is used to obtain a two-dimensional image of the current fastener, and the structured light system is used to obtain a three-dimensional depth image of the current fastener.

[0041] Benchmark information processing: Obtain the benchmark's induction pulse count and the benchmark's pole number character based on the collected information to determine the benchmark's geographical location. Specifically, the benchmark's induction pulse is set to correspond to the pulse number and image coordinates of the track fastener, and since different benchmarks are configured with corresponding unique pole numbers, image recognition is performed on the acquired benchmark image and the pole number data is further inferred through the received benchmark induction pulse count, so that the current benchmark's geographical location can be determined. Figure 2 As shown, the pole number of the current benchmark is determined to be L009 through image recognition.

[0042] Fastener information processing: The detection results and position information of the rail fasteners are obtained based on the two-dimensional image and three-dimensional depth image of the rail fasteners; the position information includes the image coordinates and pulse number of the fasteners. The two-dimensional image and the three-dimensional depth image are fused through image fusion technology to achieve complete alignment, target recognition and detection are achieved through two-dimensional images, and accurate measurement is achieved through three-dimensional depth images.

[0043] First, the process of obtaining the rail fastener detection result based on the rail fastener 2D image and the rail fastener 3D depth image is as follows:

[0044] On the one hand, since the track fasteners include left fasteners and right fasteners, which are installed on the left and right sides of the track respectively, the two-dimensional image can realize fastener classification and detect whether the left and right fasteners are installed reversely; on the other hand, the two-dimensional image can realize contour extraction and determine whether the fasteners are damaged and whether the nuts are installed. In addition, the nut height can be obtained through the three-dimensional depth image, and whether the nut is loose can be determined based on the nut height. Figures 3 to 5 As shown, Figure 3 Shown is a 2D image of a rail fastener. Figure 4 Shown is a 3D depth image of a rail fastener. Figure 5 Shown is a two-dimensional image of a rail fastener marked with nut height values ​​obtained from a three-dimensional depth image measured by a structured light system.

[0045] Therefore, the detection result of the rail fastener can be obtained based on the two-dimensional image of the rail fastener and the three-dimensional depth image of the rail fastener. If the left and right fasteners are installed in reverse, the outline of the rail fastener is damaged, the nut of the rail fastener is not installed, and the height of the nut of the rail fastener is not within the set value range, the detection result of the rail fastener is judged to be unqualified.

[0046] Secondly, the specific process of obtaining the position information of the rail fastener according to the two-dimensional image of the rail fastener and the three-dimensional depth image of the rail fastener is as follows:

[0047] When the fastener camera takes a two-dimensional image of the track fastener, the pulse numbers of the track fasteners are set through the corresponding software system and accumulated in sequence; the spatial position of the current fastener is obtained through the three-dimensional depth image of the track fastener. The spatial position can reflect the image coordinates of the corresponding fastener, and the corresponding relationship between the pulse number and the image coordinates is determined.

[0048] Positioning fasteners: The relative position of the benchmark and the track fastener is calculated based on the geographical location of the benchmark and the pulse number of the track fastener, so as to infer the geographical location of the track fastener and accurately position the track fastener. Based on the pole number recognition and positioning, the cumulative error of the inspection trolley travel can be eliminated, and the accurate positioning of the track fasteners, interfaces, etc. can be achieved.

[0049] The rail fasteners are inspected based on their two-dimensional images and three-dimensional depth images to determine whether they are qualified. If they are unqualified, an alarm is output corresponding to the geographical location of the rail fasteners. If they are qualified, the inspection vehicle continues to move forward to complete the inspection.

[0050] Furthermore, since there are inevitably process differences in the rail fasteners themselves and operational errors are prone to occur during the detection process, quantitative data can be determined through relative detection through repeated measurements to reflect the relative change trend. If the data difference of a single measurement is within a preset range, it is considered that the current track is still in a normal state. If the data difference of a single measurement exceeds the preset range, an alarm signal is issued to avoid frequent false alarms due to process differences.

[0051] Embodiment 2

[0052] A method for intelligent inspection of rail fasteners includes the steps as described in Example 1. The difference between this embodiment and Example 1 is that in order to improve the offset problem caused by lens distortion and improve measurement accuracy, lens distortion is considered when establishing a structured light system.

[0053] Structured light technology: Use a pre-designed pattern with a special structure (such as discrete light spots, stripe light, coded structured light, etc.), then project the pattern onto the surface of a three-dimensional object, and use another camera to observe the distortion of the image on the three-dimensional physical surface. If the structured light pattern is projected on a plane on the surface of the object, then the structured light pattern in the observed imaging is similar to the projected pattern, without deformation, but with a certain scale change depending on the distance. However, if the surface of the object is not a plane, the observed structured light pattern will produce different distortions due to the different geometric shapes of the surface of the object, and it will be different depending on the distance. Based on the known structured light pattern and the observed deformation, the three-dimensional shape and depth information of the object under test can be calculated according to the algorithm.

[0054] Specifically, the structured light system includes a structured light projector and a system camera. The system camera and the structured light projector are installed at a set angle. The structured light projector emits line structured light to the surface of the object, so that the line structured light plane is tangent to the surface of the object, forming a light strip on the surface of the object. The system camera captures the laser image from another angle and realizes measurement through optical triangulation.

[0055] The schematic diagram of ideal linear structured light model imaging is as follows Figure 6 As shown, point P (X, Y) is the object, point is the imaging point of the object. However, in the actual environment, the perspective imaging of the lens is not ideal and always has different degrees of distortion. These lens distortions make the image of the actual object point not at the ideal imaging point position described by the linear structured light model, but there is a certain offset, such as Figure 7 As shown, the coordinates of the actual imaging point in the image coordinate system are , the corresponding pixel coordinates in the pixel coordinate system are , the coordinates of the actual object point in the camera coordinate system are , the coordinates of the actual object point in the world coordinate system are .

[0056] The distortion of the lens includes tangential distortion and radial distortion. Due to the characteristics of the linear structured light model, the influence of tangential distortion on imaging is small and can be ignored. Therefore, without considering the tangential distortion, the relationship between the ideal point and the actual imaging point can be expressed as:

[0057] (1)

[0058] In the formula, is the first-order radial distortion coefficient; .

[0059] From the measurement model of the linear structured light model, we can get:

[0060] (2)

[0061] Combining the above formula with the camera measurement model under ideal environment, we can get:

[0062] (3)

[0063] Simplifying the above formula, we can get:

[0064] (4)

[0065] The above formula describes the measured three-dimensional surface point Its image point In formula (4), Through image processing, we can get is the pixel coordinate position of the physical coordinate origin of the image; is the radial distortion coefficient of the camera; is the focal length of the lens; is the internal parameter of the camera; ,as well as The relationship between the camera coordinate system and the world coordinate system is called the camera external parameters.

[0066] Since the measured point is located on the light plane, so, , put it into formula (4), when the internal and external parameters of the camera are known, we can get and Relationship:

[0067] (5)

[0068] The above model is the mathematical model of the line structured light vision sensor taking distortion into account.

[0069] According to the line structured light 3D scanning model, the points on the light strip on the surface of the object The coordinates are obtained from the above formula. When the one-dimensional displacement platform moves, the point on the light strip The coordinates are determined by the platform displacement:

[0070] (6)

[0071] Among them, Represents the reading of the grating ruler or encoder. The coordinates of the point on the light strip can be obtained by equations (5) and (6): .

[0072] Embodiment 3

[0073] like Figure 8 As shown, a patrol vehicle for intelligent patrol inspection of rail fasteners includes a benchmark detection module 1, a fastener detection module 2, a central control module 3, a drive module 4, a power supply module 5, and a chassis 6.

[0074] The benchmark detection module 1 is used to obtain and count the induction pulses output by the benchmark, obtain a benchmark image and perform image recognition on the benchmark image to obtain the pole number character, and infer the current geographic location of the benchmark through the pole number character and the induction pulse count. Specifically, the benchmark detection module 1 includes a benchmark camera and a pulse receiver, the benchmark camera is used to obtain the benchmark image, and a photoelectric switch for outputting the induction pulse is installed on the benchmark.

[0075] The fastener detection module 2 is used to obtain a two-dimensional image of the rail fastener and perform image processing on the two-dimensional image of the current fastener to achieve fastener classification and contour extraction; at the same time, when the fastener camera is set to shoot the two-dimensional image of the rail fastener, the pulse numbers of the rail fastener are accumulated in sequence; the three-dimensional depth image of the rail fastener is obtained to obtain the nut height and spatial position of the current fastener, and the spatial position reflects the image coordinates of the current fastener.

[0076] The fastener detection module 2 includes a fastener camera and a structured light system. The fastener camera is used to obtain a two-dimensional image of the track fastener. The structured light system is used to obtain a three-dimensional image of the track fastener. The structured light system includes a structured light projector and a system camera. The system camera and the structured light projector are installed at a set angle. The structured light projector emits line structured light to the surface of the object so that the line structured light plane is tangent to the surface of the object, forming a light strip on the surface of the object. The construction process of the structured light system in this embodiment is the same as the process described in the second embodiment, and will not be repeated here.

[0077] The central control module 3 is used to determine whether the current fastener is unqualified based on the processing results of the two-dimensional image and the three-dimensional depth image of the track fastener; calculate the relative position of the benchmark and the track fastener based on the geographical location of the benchmark and the pulse number of the track fastener, so as to infer the geographical location of the track fastener.

[0078] The rail fasteners are inspected through the cooperation between the benchmark detection module 1, the fastener detection module 2, and the central control module 3, and the geographical location of the corresponding rail fastener is determined according to the benchmark set beside the track, which is convenient for positioning the rail fastener. When the rail fastener is detected to be unqualified, the central control module 3 drives the alarm module to sound an alarm and output the geographical location of the corresponding rail fastener. Among them, the unqualified rail fasteners include the reverse installation of the left and right fasteners, the fasteners are damaged, the nuts are not installed, and the nut height is not within the threshold range.

[0079] In addition, the driving module 4 is used to drive the trolley to run along the track and inspect the track fasteners in turn, and the power supply module 5 is used to provide corresponding power supply to each part of the trolley. The driving module 4 and the chassis 6 are locked by a latch and a toggle clamp mechanism, and the central control module 3 and the chassis 6 are also locked by a latch and a toggle clamp mechanism. The power supply module 5 is placed vertically downward on the chassis 6 through the outer shape limit.

[0080] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention is described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A rail fastener intelligent inspection method, characterized in that: It is used to inspect the rail fasteners and determine the geographical location of the corresponding rail fasteners according to the benchmarks set beside the track, specifically including the following steps: Information collection: obtain the induction pulse of the current benchmark, the picture of the benchmark, and the two-dimensional image and three-dimensional depth image of the track fastener; Benchmark information processing: Obtain the induction pulse count and pole number character of the benchmark according to the collected information to determine the geographic location of the benchmark; continuously take pictures with the benchmark camera to obtain the image of the current benchmark, receive and count the induction pulses emitted by the sensor on the benchmark through the pulse receiver, and the sensor on the benchmark is a photoelectric switch; each benchmark is equipped with a corresponding unique pole number, and image recognition is performed on the benchmark picture to obtain the pole number character; the geographic location of the current benchmark is inferred from the pole number character and the induction pulse count; Fastener information processing: obtaining the detection result and position information of the rail fastener according to the rail fastener 2D image and the rail fastener 3D depth image, wherein the position information includes the image coordinates and pulse number of the fastener; Locating fasteners: Calculate the relative position of the benchmark and the track fastener based on the geographical location of the benchmark and the pulse number of the track fastener, thereby inferring the geographical location of the track fastener; The rail fasteners are inspected based on their two-dimensional images and three-dimensional depth images to determine whether they are qualified. If they are unqualified, an alarm is issued and the geographical location of the corresponding rail fasteners is output. If they are qualified, the inspection vehicle continues to move forward to complete the inspection.

2. The intelligent inspection method for rail fasteners according to claim 1, characterized in that: A two-dimensional image of the current fastener is obtained through the fastener camera. In addition, when the fastener camera takes a two-dimensional image of the track fastener, the pulse numbers of the track fastener are accumulated in sequence; a three-dimensional depth image of the current fastener is obtained through the constructed structured light system; The track fastener includes a left fastener and a right fastener, and image processing is performed on the two-dimensional image of the current fastener to achieve fastener classification and contour extraction, so as to determine whether the current fastener is qualified; The nut height and spatial position of the current fastener are obtained through the structured light system, and whether the current fastener is qualified is judged according to the nut height, and the corresponding image coordinates are determined according to the spatial position of the current fastener.

3. The intelligent inspection method for rail fasteners according to claim 2, characterized in that: The structured light system includes a structured light projector and a system camera. The system camera and the structured light projector are installed at a set angle. The structured light projector emits line structured light to the surface of the object, so that the line structured light plane is tangent to the surface of the object, forming a light strip on the surface of the object.

4. The intelligent inspection method for rail fasteners according to claim 1, characterized in that: The unqualified conditions of the rail fasteners include the left fastener and the right fastener being installed in reverse positions, the fastener being damaged, the nut not being installed, and the nut height not being within the threshold range.

5. A rail fastener intelligent inspection vehicle, characterized in that: include: The benchmark detection module is used to obtain and count the induction pulses output by the benchmark, obtain the benchmark image and perform image recognition on the benchmark image to obtain the pole number characters, and infer the current geographic location of the benchmark through the pole number characters and the induction pulse count; The fastener detection module is used to obtain a two-dimensional image of the rail fastener and perform image processing on the two-dimensional image of the current fastener to achieve fastener classification and contour extraction; at the same time, when the fastener camera is set to shoot the two-dimensional image of the rail fastener, the pulse numbers of the rail fastener are accumulated in sequence; the three-dimensional depth image of the rail fastener is obtained to obtain the nut height and spatial position of the current fastener, and the spatial position reflects the image coordinates of the current fastener; The central control module is used to determine whether the current rail fastener is unqualified based on the processing results of the two-dimensional image and the three-dimensional depth image of the rail fastener; calculate the relative position of the benchmark and the rail fastener based on the geographical location of the benchmark and the pulse number of the rail fastener, so as to infer the geographical location of the rail fastener; When it is detected that the rail fastener is unqualified, the central control module drives the alarm module to sound an alarm and output the geographical location of the corresponding rail fastener.

6. The intelligent rail fastener inspection vehicle according to claim 5, characterized in that: The inspection trolley also includes a driving module and a power supply module. The driving module is used to drive the trolley to run along the track so as to inspect the track fasteners in sequence. The driving module and the chassis are positioned by a latch and locked by a toggle clamp mechanism. The power supply module is used to provide the inspection trolley with corresponding power supply and is placed vertically downward on the chassis through an external limiter.

7. The intelligent rail fastener inspection vehicle according to claim 5, characterized in that: The benchmark detection module includes a benchmark camera and a pulse receiver, the benchmark camera is used to obtain a benchmark image, and a photoelectric switch for outputting an induction pulse is installed on the benchmark; The fastener detection module includes a fastener camera and a structured light system. The fastener camera is used to obtain a two-dimensional image of the track fastener. The structured light system is used to obtain a three-dimensional image of the track fastener. The structured light system includes a structured light projector and a system camera. The system camera and the structured light projector are installed at a set angle. The structured light projector emits line structured light to the surface of the object, so that the line structured light plane is tangent to the surface of the object, forming a light strip on the surface of the object.

8. The intelligent rail fastener inspection vehicle according to claim 5, characterized in that: The unqualified conditions of the rail fasteners include the left fastener and the right fastener being installed in reverse positions, the fastener being damaged, the nut not being installed, and the nut height not being within the threshold range.

Images