Railway ballast rotation and track change test method and system

Abstract

The present application relates to a kind of ballast rotation and trajectory variation test method and system, wherein ballast rotation and trajectory variation test method includes: obtaining the multiple continuous original images of the ballast to be detected under the action of vertical load and horizontal load, each original image includes multiple ballast;Each original image is corrected and corrected coordinates, and the corresponding physical coordinate system image is obtained;Each physical coordinate system image is analyzed, and the characteristic attribute of each ballast to be detected in image is obtained;According to the characteristic attribute of each ballast to be detected, all coordinates physical system images are analyzed using time series image analysis technique, and the dynamic parameter of each ballast to be detected is obtained by adjacent frame difference method.It has beneficial effects that ballast under different stress conditions can be monitored with high precision, real-time rotation and trajectory variation, which greatly improves the test precision and efficiency of experiment.

Description

Technical Field

[0001] This invention relates to the field of experimental equipment technology for traffic engineering, and in particular to a method and system for testing the rotation and trajectory change of ballast. Background Technology

[0002] As the basic material of the track structure, railway ballast's dynamic mechanical properties directly affect the safety and stability of train operation. Under long-term train loads, ballast particles are prone to rotational instability, track deviation, and microcrack propagation. Especially under extreme conditions of high-frequency vibration, traditional experimental methods are insufficient to accurately quantify its dynamic response characteristics.

[0003] Currently, existing technologies have the following drawbacks:

[0004] Limitations of single-parameter detection: Most existing ballast mechanical testing equipment uses a single mechanical sensor or numerical simulation, which cannot simultaneously acquire multi-dimensional dynamic data of ballast rotation angle, trajectory evolution and micro-strain field, resulting in incomplete analysis of material failure mechanism;

[0005] Dynamic interference is difficult to eliminate: natural light environment or mechanical vibration can easily cause visual detection system to misjudge. Existing displacement tracking technology based on monocular camera is not accurate enough and lacks sub-pixel motion compensation algorithm;

[0006] Vibration condition simulation is lacking: Traditional experimental platforms only support static or low-frequency loading, which cannot reproduce the deformation process of ballast material caused by high-frequency vibration of trains, and the experimental results deviate significantly from the actual working conditions.

[0007] Therefore, how to overcome the limitations of traditional methods in terms of accuracy and working conditions, and to monitor the rotation and trajectory changes of ballast under different stress conditions with high precision, so as to provide more reliable technical support for the performance evaluation of railway ballast, has become an urgent technical problem to be solved. Summary of the Invention

[0008] (a) Technical problems to be solved

[0009] In view of the above-mentioned shortcomings and deficiencies of the existing technology, the present invention provides a method and system for testing ballast rotation and trajectory change, which breaks through the accuracy and working condition limitations of traditional methods and provides more reliable technical support for railway ballast performance evaluation.

[0010] (II) Technical Solution

[0011] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0012] In a first aspect, embodiments of the present invention provide a method for testing ballast rotation and trajectory change, comprising:

[0013] S100: Acquire multiple consecutive original images of the ballast to be detected under the action of vertical loading force and horizontal loading force simultaneously, each original image including multiple ballast sections;

[0014] S200. Perform distortion correction and coordinate correction on each of the original images to obtain its corresponding physical coordinate system image;

[0015] S300. Perform feature graph analysis on each of the physical coordinate system images to obtain the feature attributes of each ballast to be detected in the image.

[0016] S400. Based on the characteristic attributes of each ballast to be detected, analyze all coordinate physical system images using time series image analysis technology, and obtain the dynamic parameters of each ballast to be detected through the adjacent frame difference method.

[0017] This invention discloses a method for testing the rotation and trajectory change of ballast. By using computer vision technology, it can monitor the rotation and trajectory change of ballast under different stress conditions with high precision and in real time, which greatly improves the testing accuracy and efficiency of the experiment. It can not only simulate static and dynamic loads, but also accurately record the displacement, rotation angle and strain of ballast during the experiment, which facilitates comprehensive data analysis.

[0018] Optionally, the ballast to be detected has identification information. The original image or physical coordinate system image is input into a trained three-layer neural network to obtain the identification information of each ballast to be detected in the image. The three-layer neural network uses softmax as the activation function and sparse class cross-entropy as the loss function.

[0019] Optionally, the feature attributes include the centroid position, radius, area, and rotation angle of the ballast to be detected.

[0020] Optionally, S300 includes:

[0021] The contour of each ballast to be detected is extracted by feature image analysis. The geometric center of the contour of each ballast to be detected is taken as the centroid position of the ballast to be detected. The radius and area of ​​the ballast to be detected are calculated based on the contour.

[0022] Optionally, the S300 includes:

[0023] The rotation angle of each ballast track to be tested is obtained in the following way.

[0024] S301. For each physical coordinate system image, combine all sampling points in the image into pairs to form sampling point pairs. The combinations of all sampling point pairs constitute a set A.

[0025] A={(p i ,pj )∈R 2 ×R 2 |i <N∧j<i∧i,j∈N}

[0026] Where, p i p is the two-dimensional coordinate of the i-th sampling point in the physical coordinate system image. j R is the two-dimensional coordinate of the j-th sampling point in the physical coordinate system image. 2 Let N be a set of two-dimensional real numbers, where N is the number of sampling points in the physical coordinate system image;

[0027] S302. For each pair of sampling points in set A, obtain its local gradient g(p). i p j ),

[0028]

[0029] Among them, I(p) i , σ i ) represents the i-th sampling point in the physical coordinate system image at scale σ. i The pixel value below; I(p) j , σ j ) represents the j-th sampling point in the physical coordinate system image at scale σ. j The pixel value below;

[0030] S303. Construct a long-distance subset L of point pairs.

[0031]

[0032] Where, δ min =13.67t min , t min It is the minimum value of the scale parameter among all feature points on the ballast profile;

[0033] S304. Obtain the main direction position of the ballast to be detected based on the point pairs in the long-distance point pair subset L and the local gradient.

[0034]

[0035] Where g is the principal direction position vector of the ballast to be detected, g x The component of the principal direction position vector g along the x-axis, g y M is the component of the main direction position vector g along the y-axis; M is the number of elements in the long-distance point pair subset L.

[0036] S305, the rotation angle is obtained based on the principal direction position vector.

[0037] α=arctan2(gyg x )

[0038] Where α is the rotation angle of the ballast to be tested.

[0039] Optionally, S400 includes:

[0040] The dynamic parameters include the movement trajectory and relative displacement;

[0041] Specifically,

[0042] The set T of the movement trajectories of the ballast to be detected.

[0043] T={P(t0),P(t0+Δt),..,P(t k )}

[0044] Where, P(t0) = (x(t0), y(t0)) is the physical coordinate of the centroid position of the ballast to be tested at time t0, and P(t0+Δt) = (x(t0+Δt), y(t0+Δt)) is the physical coordinate of the centroid position of the ballast to be tested at time t0+Δt. k )=(x(t k ), (y(t) k )) is the ballast to be tested at t k The physical coordinates of the centroid position at time t = 100 ms;

[0045] Adjacent displacements include the difference in displacement vectors between adjacent frames and the displacement magnitude.

[0046] The difference in displacement vectors between adjacent frames is calculated using the following formula.

[0047] ΔP(t)=P(t+Δt)-P(t)=(Δx,Δy)

[0048] Where ΔP(t) is the displacement vector difference between adjacent frames of the ballast to be detected at time t, P(t+Δt)=(x(t+Δt), y(t+Δt)) is the physical coordinate of the centroid position of the ballast to be detected at time t+Δt, P(y)=(x(t), y(t)) is the physical coordinate of the centroid position of the ballast to be detected at time t, Δt=100ms, Δx is the component of ΔP(t) in the x-axis direction, and Δy is the component of ΔP(t) in the y-axis direction;

[0049] The displacement modulus ||ΔP|| is calculated using the following formula based on the displacement vector difference between adjacent frames.

[0050]

[0051] Optionally, the dynamic parameters include the change in rotation angle, angular velocity, circumferential deformation, and surface area deformation;

[0052] Specifically,

[0053] The change in rotation angle is calculated using the following formula.

[0054] Δα(t)=α(t+Δt)-α(t)

[0055] Where Δα(t) is the change in rotation angle of the ballast to be tested at time t, α(t+Δt) is the rotation angle of the ballast to be tested at time t+Δt, and α(t) is the rotation angle of the ballast to be tested at time t.

[0056] The angular velocity is obtained from the change in the rotation angle.

[0057]

[0058] Where ω(t) is the angular velocity of the ballast to be tested at time t, in rad / s, and Δα(t) is the change in the rotation angle of the ballast to be tested at time t, Δt = 100 ms;

[0059] Perimeter deformation is calculated using the following formula:

[0060] C 物理 (t)=C 像素 (t)×Pixel Physical Ratio

[0061] ΔC(t)=(C 物理 (t+Δt)-C 物理 (t)

[0062] Among them, C 物理 (t) represents the physical perimeter of the ballast to be tested at time t, and C 像素 (t) represents the total number of pixels in the ballast outline to be detected extracted through feature image analysis, and ΔC(t) represents the perimeter deformation of the ballast to be detected at time t. 物理 (t+Δt) is the physical perimeter of the ballast to be tested at time t+Δt, where Δt = 100ms;

[0063] Surface area deformation is calculated using the following formula.

[0064] ΔC(t)=C 物理 (t+Δt)-C 物理 (t)

[0065] A 表面积 (t)=2πr(t)L 物理 (t)+2πr(t) 2

[0066] ΔA(t)=A 表面积 (t+Δt)-A 表面积 (t)

[0067] Among them, L物理 (t) represents the distance between the feature points at both ends of the ballast to be tested at time t, x1 and y1 are the x-coordinates and y-coordinates of the feature point at one end of the ballast to be tested at time t, respectively, and x2 and y2 are the x-coordinates and y-coordinates of the feature point at the other end of the ballast to be tested at time t, respectively. A 表面积 Let r(t) be the surface area of ​​the ballast to be tested at time t, and r(t) be the radius of the ballast to be tested at time t. 表面积 (t+Δt) is the surface area of ​​the ballast to be tested at time t+Δt, ΔA(t) is the surface area deformation of the ballast to be tested at time t, and Δt=100ms.

[0068] Optionally, S400 includes:

[0069] The dynamic parameters include the deformation of the contact rod;

[0070] Specifically,

[0071] For any two ballast sections in the same physical coordinate system image, they are considered adjacent if they satisfy the following formula.

[0072] (A x -B x ) 2 -(A y -B y ) 2 ≤(A r +B r ) 2

[0073] Among them, A x A y These are the x-axis and y-axis coordinates of ballast A, respectively. r Let B be the radius of ballast A; x B y These are the x-axis and y-axis coordinates of ballast B, respectively. r Let B be the radius of the ballast.

[0074] For a ballast to be detected, the above formula is used to iterate through all other ballasts in the physical coordinate system image to obtain all contact ballasts of the ballast.

[0075] The deformation of the contact bar is calculated based on the physical coordinates of the contact ballast.

[0076] Secondly, embodiments of the present invention provide a ballast rotation and trajectory change testing system, including a loading subsystem and a vision subsystem;

[0077] The loading subsystem includes a placement slot, a vertical loading mechanism, and a horizontal loading mechanism.

[0078] The placement slot is used to place the ballast to be inspected.

[0079] The vertical loading mechanism includes a first ball screw, a first loading motor for driving the first screw of the first ball screw to rotate, a vertical pressure plate connected to the first nut of the first ball screw, and a vertical linear guide rail.

[0080] The first screw of the first ball screw is vertically positioned.

[0081] The vertical pressure plate is positioned directly above the ballast to be inspected.

[0082] The horizontal loading mechanism includes a second ball screw, a second loading motor for driving the second screw of the second ball screw to rotate, a horizontal pressure plate connected to the second nut of the second ball screw, and a horizontal linear guide rail.

[0083] The second screw of the second ball screw is set horizontally.

[0084] The horizontal pressure plate is located on one side of the ballast to be inspected;

[0085] The vision subsystem includes an industrial camera and a host computer.

[0086] The industrial camera is used to capture continuous raw images of the ballast in the placement trough under the combined action of the vertical loading force of the vertical loading mechanism and the horizontal loading force of the horizontal loading mechanism.

[0087] The host computer is communicatively connected to the industrial camera and is used to execute any of the methods described in the first aspect.

[0088] Optionally, the placement slot is disposed on the loading frame, and the loading frame is equipped with a fuma wheel;

[0089] The vertical loading mechanism further includes a first reducer connected to the first loading motor, and the first reducer is connected to the first screw of the first ball screw via a first coupling.

[0090] The first nut and the vertical pressure plate of the first ball screw are connected by the first nut seat and the first sensor mounting plate.

[0091] The first nut is connected to the first sensor mounting plate via a first nut seat, and the first sensor mounting plate is connected to the vertical pressure plate.

[0092] The vertical linear guide rails are located on both sides of the first sensor mounting plate.

[0093] A vertical force sensor that communicates with the host computer is installed on the first sensor mounting plate;

[0094] The second screw of the second ball screw is connected to the second loading motor via a second reducer.

[0095] The second nut of the second ball screw is connected to the horizontal pressure plate via a second sensor mounting plate and a thrust transmission plate.

[0096] The second nut is connected to the second sensor mounting plate, which is connected to the thrust transmission plate via a ball joint. The thrust transmission plate is connected to the horizontal pressure plate.

[0097] The second sensor mounting plate is equipped with a horizontal force sensor that communicates with the host computer.

[0098] (III) Beneficial Effects

[0099] The beneficial effects of the present invention are as follows: The method and system for testing the rotation and trajectory change of ballast of the present invention, through computer vision technology, can monitor the rotation and trajectory change of ballast under different stress conditions with high precision and in real time, which greatly improves the testing accuracy and efficiency of the experiment. It can not only simulate static and dynamic loads, but also accurately record the displacement, rotation angle and strain of ballast during the experiment, which facilitates comprehensive data analysis.

[0100] The system of this invention employs automated operation, reducing manual intervention and improving the stability and reliability of experiments. Simultaneously, its integrated sensors and vision subsystem work in concert, enabling more accurate acquisition of data such as loading force and displacement, providing strong data support for subsequent engineering research. The entire system has a compact structure, is easy to operate, and possesses high application value and practicality.

[0101] This invention integrates the ORB feature matching algorithm with Digital Image Correlation (DIC) technology to eliminate dynamic background interference and achieve sub-pixel level displacement tracking. By integrating high-precision visual inspection, servo control and real-time data processing technologies, it breaks through the accuracy and working condition limitations of traditional methods, providing more reliable technical support for railway ballast performance evaluation. Attached Figure Description

[0102] Figure 1 This is a schematic diagram of the loading subsystem in a ballast rotation and trajectory change testing system according to this embodiment;

[0103] Figure 2 This is a schematic diagram of the vertical loading mechanism of the loading subsystem in a ballast rotation and trajectory change testing system according to this embodiment;

[0104] Figure 3 This is a schematic diagram of the loading frame of the loading subsystem in a ballast rotation and trajectory change testing system according to this embodiment;

[0105] Figure 4 This is a schematic diagram of the loading subsystem placement slot in a ballast rotation and trajectory change testing system according to this embodiment.

[0106] Figure 5 This is a schematic diagram of the horizontal loading mechanism of the loading subsystem in a ballast rotation and trajectory change test system according to this embodiment.

[0107] Figure 6 This is a flowchart illustrating a method for testing ballast rotation and trajectory change in this embodiment.

[0108] [Explanation of Labels in the Attached Image]

[0109] 1: Ballast; 2: Loading frame; 3: Fuma wheel; 4: Vertical load-bearing plate; 5: Horizontal load-bearing plate; 6: Vertical pressure plate; 7: Horizontal pressure plate; 8: First screw; 9: First loading motor; 10: First nut; 11: Vertical linear guide; 12: First reducer; 13: First coupling; 14: First nut seat; 15: First sensor mounting plate; 16: First front bearing seat; 17: Second rear bearing seat; 18: Second screw; 19: Second loading motor; 20: Second nut; 21: Joint ball joint; 22: Horizontal linear guide; 23: Second reducer; 24: Second sensor mounting plate; 25: Thrust transmission plate; 26: Second front bearing seat; 27: Second rear bearing seat; 28: Second coupling; 29: Reducer mounting plate; 30: Horizontal pressure head; 31: Balance connecting rod. Detailed Implementation

[0110] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0111] In real railway environments, ballast stress is influenced by a combination of factors, including train speed, load, track geometry, temperature, and humidity, making it difficult to isolate single variables for mechanistic analysis. Existing dynamic mechanical testing technologies for ballast suffer from several technical problems, including a lack of coordinated measurement of multi-dimensional parameters, insufficient dynamic interference suppression, inadequate accuracy in vibration condition simulation, and low efficiency in data correlation analysis. Therefore, this embodiment provides a computer vision-based ballast rotation and trajectory change testing system, particularly suitable for the synchronous monitoring and analysis of the rotational behavior, trajectory evolution, and micro-strain characteristics of railway ballast under complex loads. This system integrates servo control, high-precision visual inspection, and dynamic data processing technologies to simulate the mechanical response of ballast under train vibration conditions. It achieves synchronous acquisition of multi-dimensional parameters, uses ORB feature matching algorithms and checkerboard calibration technology to eliminate dynamic interference, configures high-frequency vibration loading to improve vibration condition reproduction capabilities, and achieves rapid data response based on data interaction and dynamic spectrum analysis algorithms. This systematically solves the aforementioned technical problems and provides experimental basis for railway track structure stability assessment and ballast material optimization design.

[0112] The ballast rotation and trajectory change testing system in this embodiment is mainly used for experimental simulation, that is, to reproduce the mechanical conditions of actual railway ballast working conditions (such as train vibration and bidirectional load) in a controlled laboratory environment. By applying vertical / horizontal loads and high-frequency vibrations through the loading subsystem, the dynamic response of ballast in real railways is simulated. This embodiment can achieve accurate reproduction of actual ballast working conditions and accurate measurement of experimental data.

[0113] Example

[0114] See Figures 1-5 This embodiment provides a ballast rotation and trajectory change testing system, including a loading subsystem and a vision subsystem.

[0115] It should be noted that the ballast can be made of aluminum or rubber rods, and its surface is marked with random textures or monochrome coatings for identification and differentiation.

[0116] The loading subsystem is used to apply bidirectional loads to the ballast in both vertical and horizontal directions, while the vision subsystem is used to collect the displacement field, rotation angle, and micro-strain of the ballast bars in real time and analyze them to obtain their dynamic parameters.

[0117] The loading subsystem includes a placement slot, a vertical loading mechanism, and a horizontal loading mechanism.

[0118] Both of the above loading mechanisms are driven by motors to achieve vertical and horizontal loading of ballast bars.

[0119] The placement slot is used to place the ballast 1 to be inspected. Specifically, the placement slot is located on the loading frame 2, and the loading frame 2 is equipped with casters 3. The loading frame 2 is an integral welded frame, with casters 3 at the bottom for movement and fixation. The loading frame 2 integrates a vertical vibration loading mechanism and a horizontal loading mechanism to achieve coordinated control of dual-axis loading. A balance link 31 is provided at the top of the loading frame to enhance the structural stability of the loading frame and prevent deformation of the frame under high-frequency vibration conditions.

[0120] The placement slot includes a vertical support plate 4 and a horizontal support plate 5. The ballast 1 to be tested is placed in the placement slot and thus fixed between the vertical support plate 4 and the horizontal support plate 5. The upper surface of the ballast to be tested contacts the vertical pressure plate 6 (described below), and the side contacts the horizontal pressure plate 7 (described below). The placement slot is a cuboid with an open front side. The vertical support plate and the horizontal support plate are the bottom and left side, respectively, and the vertical pressure plate and the horizontal pressure plate are the top and right side, respectively. The rear baffle is the rear side. The horizontal support plate and the rear baffle are rigidly connected by bolts to form an L-shaped fixed support structure. The vertical pressure plate and the horizontal pressure plate are independently installed on the guide rail. There is a horizontal baffle on the left side of the placement slot.

[0121] The vertical loading mechanism includes a first ball screw, a first loading motor 9 for driving the first screw 8 of the first ball screw to rotate, a vertical pressure plate 6 connected to the first nut 10 of the first ball screw, and a vertical linear guide rail 11. The first screw 8 of the first ball screw is vertically arranged, and the vertical pressure plate 6 is located directly above the ballast to be tested.

[0122] Specifically, the vertical loading mechanism further includes a first reducer 12 connected to the first loading motor 9. The first reducer 12 is connected to the first screw 8 of the first ball screw via a first coupling 13. The first nut 10 of the first ball screw and the vertical pressure plate 6 are connected to the first sensor mounting plate 15 via a first nut seat 14. The first nut 10 is connected to the first sensor mounting plate 15 via the first nut seat 14, and the first sensor mounting plate 15 is connected to the vertical pressure plate 6.

[0123] The vertical linear guide rail 11 is disposed on both sides of the first sensor mounting plate 15, and a vertical force sensor that is connected to the host computer described below is mounted on the first sensor mounting plate 15.

[0124] In the vertical loading mechanism, the first loading motor can be a high-frequency servo motor. After being reduced by the first reducer 13, the first loading motor drives the first screw of the first ball screw to rotate through the first coupling. The first screw is fixed by the first front bearing seat 16 and the second rear bearing seat 17. The first nut 10 on the first screw 8 is connected to the first sensor mounting plate through the first nut seat 14, which converts the rotational motion into the vertical linear motion of the first nut, and finally drives the vertical pressure plate 6 to press down or vibrate at high frequency. Vertical linear guide rails 11 are connected to both sides of the first sensor mounting plate, and the vertical linear guide rails 11 are used for guidance.

[0125] The horizontal loading mechanism includes a second ball screw, a second loading motor 19 for driving the second screw 18 of the second ball screw to rotate, a horizontal pressure plate connected to the second nut 20 of the second ball screw, and a horizontal linear guide rail 22. The second screw 18 of the second ball screw is horizontally arranged, and the horizontal pressure plate is located on one side of the ballast to be tested.

[0126] Specifically, the second screw 18 of the second ball screw is connected to the second loading motor 19 via the second reducer 23. The second screw is fixed via the second front bearing seat 26 and the second rear bearing seat 27, and is connected to the second reducer via the second coupling 28. The second reducer is fixed on the reducer mounting plate 29. The second nut 20 of the second ball screw is connected to the horizontal pressure plate via the second sensor mounting plate 24 and the thrust transmission plate 25.

[0127] The second nut 20 is connected to the second sensor mounting plate 24, the second sensor mounting plate 24 is connected to the thrust transmission plate 25 through the joint ball joint 21, the thrust transmission plate 25 is connected to the horizontal pressure plate, and a horizontal force sensor that communicates with the host computer is installed on the second sensor mounting plate 24.

[0128] The horizontal loading mechanism also includes a horizontal pressure head 30, which is connected to the horizontal pressure plate and directly contacts the ballast to be tested to transfer the horizontal load. The spherical design avoids stress concentration.

[0129] In the horizontal loading mechanism, the second loading motor can be a stepper motor. The second loading motor 19 drives the second screw 18 of the second ball screw to rotate through the second reducer 23. The horizontal force sensor is mounted on the second sensor mounting plate 24 to provide real-time feedback of the horizontal loading force. The second sensor mounting plate 24 is flexibly connected to the thrust transmission plate 25 through the articulated ball joint 21 to eliminate horizontal assembly gaps. The horizontal linear guide rail 22 is used to limit lateral displacement and provide guidance.

[0130] The first loading motor drives the first screw to rotate, pressing down the vertical pressure plate and thus loading the ballast bars vertically. Similarly, the second loading motor drives the second screw to rotate, bringing the horizontal pressure plate closer to and pressing the ballast bars, thus loading them horizontally. When the first loading motor drives the first screw to rotate in both directions, the vertical pressure plate reciprocates up and down, mimicking the stress on the ballast under vibration. Therefore, the loading subsystem of this embodiment supports low-frequency vibration loading and can simulate the dynamic loading of ballast. Furthermore, the loading subsystem of this embodiment includes a multi-axis collaborative control unit to achieve dual closed-loop feedback of force / displacement for both vertical and horizontal loading; and it has an emergency stop protection mechanism that directly cuts off the servo driver power supply to respond to abnormal operating conditions.

[0131] Vertical and horizontal force sensors are used to measure the applied force and its changes. The loading subsystem in this embodiment also includes displacement sensors (which may include vertical and horizontal displacement sensors) for measuring displacement changes. All sensors are communicatively connected to the host computer described below for data transmission and real-time feedback of the experimental status.

[0132] The vision subsystem includes an industrial camera and a host computer.

[0133] The industrial camera is used to capture continuous raw images of the ballast placed in the trough under the combined action of the vertical loading force of the vertical loading mechanism and the horizontal loading force of the horizontal loading mechanism.

[0134] It should be noted that industrial cameras can be selected that can automatically adjust the shooting distance and are equipped with a checkerboard calibration plate distortion correction function.

[0135] The host computer is connected to the industrial camera for communication.

[0136] The host computer in this embodiment is equipped with vision processing software, which is used to analyze physical parameters such as rotation angle, displacement and strain of ballast bars, and supports the export of experimental data to CSV / Excel format. It provides frame-by-frame playback and historical record retrieval functions, and the retrieval conditions include time range or bar number.

[0137] This embodiment presents a computer vision-based experimental testing system for simulating the rotation and trajectory changes of ballast. The system applies vertical and horizontal loads to the ballast through a loading subsystem, simulating the stress state of the ballast material. Simultaneously, an industrial camera in the vision subsystem acquires real-time image data of the ballast under different loading conditions, and the images are analyzed by a host computer to accurately monitor the physical properties of the ballast, such as rotation, displacement, and strain. This system integrates multiple sensors to measure changes in loading force and displacement in real time, ensuring data accuracy. It can simulate not only static loads but also accurately simulate dynamic loads, making it suitable for research on the mechanical properties of ballast materials. It boasts advantages such as high precision, high efficiency, and high automation.

[0138] In this embodiment, the host computer is used to execute the methods described below. It should be noted that before implementing the following methods, an industrial camera is used to aim at the 400mm x 400mm test area of ​​the test bench. If an industrial camera with a checkerboard calibration plate distortion correction function is selected, the camera needs to be calibrated to eliminate the influence of camera distortion and establish physical correlation. Test objects such as aluminum or rubber ballast are placed in the test area, i.e., the placement slot. When the equipment starts running, pressure is applied to the test area, generating controllable vertical loading force, controllable vertical loading speed, controllable horizontal loading force, and controllable horizontal acceleration. These data signals are acquired at a 100ms acquisition rate, and simultaneously, the industrial camera is triggered to take pictures of the test area to achieve image acquisition.

[0139] See Figure 6 A method for testing the rotation and trajectory change of ballast, comprising:

[0140] S100: Acquire multiple consecutive original images of the ballast to be detected under both vertical and horizontal loading forces, each original image including multiple ballast sections.

[0141] S200. Perform distortion correction and coordinate correction on each of the original images to obtain its corresponding physical coordinate system image.

[0142] During the acquisition of raw images of ballast using industrial cameras, geometric distortions can occur in the acquired images due to the inherent characteristics of the camera lens (such as radial and tangential distortion). Distortion correction corrects these distortions, ensuring that the geometric information of the ballast, such as its shape and size, accurately reflects its actual physical state. For example, a normally circular ballast might appear elliptical in a distorted image; distortion correction restores its circular shape, ensuring that subsequent analysis is based on accurate geometric morphology. Accurate image morphology is fundamental for the feature image analysis performed in the subsequent S300 step. After distortion correction, the features of each ballast section in the image (such as texture and edges) can be extracted and identified more accurately. If image distortion exists, it can cause deviations in feature positions and shapes, leading to errors in subsequent feature matching and displacement calculations, affecting the accurate analysis of parameters such as ballast rotation and trajectory changes.

[0143] Furthermore, the pixel coordinates of the original image differ from the actual physical coordinates, and the coordinate systems of images captured by different cameras may be inconsistent. Coordinate correction converts the pixel coordinates in the image into a unified physical coordinate system, enabling images captured at different times and angles to be analyzed under the same standard. For example, accurately mapping the pixel position of the ballast in the image to its actual spatial position facilitates accurate calculation of dynamic parameters such as displacement and rotation angle of the ballast. Moreover, in the subsequent time series analysis in step S400, the unified coordinate system ensures accurate alignment and fusion between different images. Through coordinate correction, the ballast in different images can be mapped to the same position in actual physical space. Based on this, analysis using techniques such as adjacent frame difference can more accurately obtain the dynamic changes of the ballast, such as displacement trajectory and rotation trend. Therefore, coordinate correction ensures the accuracy of the movement trajectory of dynamic parameters obtained in subsequent steps.

[0144] S300. Perform feature graph analysis on each of the physical coordinate system images to obtain the feature attributes of each ballast to be detected in the image.

[0145] Specifically, the characteristic attributes include the centroid position, radius, area, and rotation angle of the ballast to be detected.

[0146] More specifically, in step S300, the contour of each ballast to be detected is extracted through feature graphic analysis, the geometric center of the contour of each ballast to be detected is taken as the centroid position of the ballast to be detected, and the radius and area of ​​the ballast to be detected are calculated based on the contour.

[0147] The radius is obtained by calculating the minimum circumscribed circle radius or fitting ellipse parameters from the bar profile extracted through feature graphic analysis, and the area is obtained by calculating the radius.

[0148] In step S300, the rotation angle of each ballast track to be inspected is obtained in the following manner:

[0149] S301. For each physical coordinate system image, combine all sampling points in the image into pairs to form sampling point pairs. The combinations of all sampling point pairs constitute a set A.

[0150] A={(p i ,p j )∈R 2 ×R 2 |i <N∧j<i∧i,j∈N}

[0151] Where, p i p is the two-dimensional coordinate of the i-th sampling point in the physical coordinate system image. j R is the two-dimensional coordinate of the j-th sampling point in the physical coordinate system image. 2 Let p be a set of two-dimensional real numbers. i p j All belong to the two-dimensional real number space set R 2 N is the number of sampling points in the physical coordinate system image, and there are N(N-1) / 2 combinations in set A.

[0152] S302. For each pair of sampling points in set A, obtain its local gradient g(p). i p j ),

[0153]

[0154] Among them, I(p) i , σ i ) represents the i-th sampling point in the physical coordinate system image at scale σ. i The pixel value below; I(p) j , σ j ) represents the j-th sampling point in the physical coordinate system image at scale σ. j The pixel value below; σ i and σ j It is a scale-space parameter in feature graph analysis, used to control the range of the local neighborhood of feature points, thereby determining the computation time and accuracy.

[0155] S303. Construct a long-distance subset L of point pairs.

[0156]

[0157] Where, δ min =13.67t min , t min It is the minimum value of the scale parameter among all feature points on the ballast profile.

[0158] Simultaneously, a subset S of short-distance point pairs is constructed to filter effective feature points, thereby eliminating local noise or duplicate features.

[0159]

[0160] Where, δ max =9.57t max , t max It is the maximum value of the scale parameter among all feature points on the ballast profile.

[0161] δ min and δ max All thresholds are calculated based on the feature point scale parameter and are linearly related to it. The feature point scale parameter is generated by the feature detection algorithm used for feature image detection. Each feature point has an independent scale parameter value, which is used to dynamically adjust the screening threshold for short-distance and long-distance point pairs.

[0162] S304. Obtain the main direction position of the ballast to be detected based on the point pairs in the long-distance point pair subset L and the local gradient.

[0163]

[0164] Where g is the principal direction position vector of the ballast to be detected, used to correct the centroid position, g x The component of the principal direction position vector g along the x-axis, g y The y-component of the main direction position vector g; M is the number of elements in the long-distance point pair subset L, g x and g y It is calculated by gradient-weighted average of long-distance point pairs.

[0165] S305, the rotation angle is obtained based on the principal direction position vector.

[0166] α=arctan2(g y ,g x )

[0167] Where α is the rotation angle of the ballast to be tested.

[0168] In addition, the ballast to be detected has identification information. In step S300, the original image or physical coordinate system image can be input into a trained three-layer neural network to obtain the identification information of each ballast to be detected in the image. The three-layer neural network uses softmax as the activation function and sparse class cross-entropy as the loss function. The identification information can be a character number used to identify and distinguish the bars. After obtaining the feature attributes of each ballast to be detected in the image, the bars can be labeled according to the identified character number of the ballast for subsequent tracking.

[0169] The algorithm for recognizing ballast marking information is trained using deep learning on a 128x128 pixel grayscale image. It can recognize character numbers of any combination of 0-9 and A-Z, and can recognize test scenarios composed of up to a thousand bars, meeting the maximum needs.

[0170] S400. Based on the characteristic attributes of each ballast to be detected, analyze all coordinate physical system images using time series image analysis technology, and obtain the dynamic parameters of each ballast to be detected through the adjacent frame difference method.

[0171] Specifically, the dynamic parameters include movement trajectory, relative displacement, change in rotation angle, angular velocity, circumferential deformation, and surface area deformation. The system acquires image frame sequences at 100ms intervals and uses the following core algorithm to achieve real-time calculation of the bar's kinematic parameters.

[0172] More specifically:

[0173] The set T of movement trajectories of the ballast to be detected consists of a sequence of positions from consecutive frames.

[0174] T = {P(t0), P(t0+Δt), ..., P(t k )}

[0175] Where, P(t0) = (x(t0), y(t0)) is the physical coordinate of the centroid position of the ballast to be tested at time t0, and P(t0+Δt) = (x(t0+Δt), y(t0+Δt)) is the physical coordinate of the centroid position of the ballast to be tested at time t0+Δt. k )=(x(t k ), y(t) k )) is the ballast to be tested at t k The physical coordinates of the centroid position at time t = 100 ms.

[0176] Adjacent displacements include the difference in displacement vectors between adjacent frames and the displacement magnitude. The difference in displacement vectors between adjacent frames is calculated using the following formula.

[0177] ΔP(t)=P(t+Δt)-P(t)=(Δx,Δy) (1)

[0178] Where ΔP(t) is the displacement vector difference between adjacent frames of the ballast to be detected at time t, P(t+Δt)=(x(t+Δt), y(t+Δt)) is the physical coordinate of the centroid position of the ballast to be detected at time t+Δt, P(t)=(x(t), y(t)) is the physical coordinate of the centroid position of the ballast to be detected at time t, Δt=100ms, Δx is the component of ΔP(t) in the x-axis direction, and Δy is the component of ΔP(t) in the y-axis direction.

[0179] The displacement modulus ‖ΔP‖ is calculated using the following formula based on the displacement vector difference between adjacent frames.

[0180]

[0181] The pixel physical ratio is the actual length corresponding to a unit pixel after calibration. It is a preset value and is determined based on the ratio of the actual object length to the pixel value.

[0182] The change in rotation angle is calculated using the following formula.

[0183] Δα(t)=α(t+Δt)-α(t)

[0184] Where Δα(t) is the change in rotation angle of the ballast to be tested at time t, α(t+Δt) is the rotation angle of the ballast to be tested at time t+Δt, and α(t) is the rotation angle of the ballast to be tested at time t.

[0185] The angular velocity is obtained from the change in the rotation angle.

[0186]

[0187] Where ω(t) is the angular velocity of the ballast to be tested at time t, in rad / s, and Δα(t) is the change in the rotation angle of the ballast to be tested at time t, Δt = 100 ms.

[0188] Perimeter deformation is calculated using the following formula:

[0189] C 物理 (t)=C 像素 (t)×Pixel Physical Ratio

[0190] ΔC(t)=C 物理 (t+Δt)-C 物理 (t)

[0191] Among them, C 物理 (t) represents the physical perimeter of the ballast to be tested at time t, and C 像素 (t) represents the total number of pixels in the ballast outline to be detected extracted through feature image analysis, and ΔC(t) represents the perimeter deformation of the ballast to be detected at time t. 物量 (t+Δt) is the physical perimeter of the ballast to be tested at time t+Δt, where Δt = 100 ms.

[0192] Surface area deformation is calculated using the following formula.

[0193]

[0194] A 表面积 (t)=2πr(t)L 物理 (t)+2πr(t)2

[0195] ΔA(t)=A 表面积 (t+Δt)-A 表面积 (t)

[0196] Among them, L 物理 (t) represents the distance between the feature points at both ends of the ballast to be tested at time t, x1 and y1 are the x-coordinates and y-coordinates of the feature point at one end of the ballast to be tested at time t, respectively, and x2 and y2 are the x-coordinates and y-coordinates of the feature point at the other end of the ballast to be tested at time t, respectively. A 表面积 Let r(t) be the surface area of ​​the ballast to be tested at time t, and r(t) be the radius of the ballast to be tested at time t. 表面积 (t+Δt) is the surface area of ​​the ballast to be tested at time t+Δt, ΔA(t) is the surface area deformation of the ballast to be tested at time t, and Δt=100ms.

[0197] The dynamic parameters also include the deformation of the contact rod; these are calculated through the following steps:

[0198] S401. For any two ballast sections in the same physical coordinate system image, they are considered adjacent if they satisfy the following formula.

[0199] (A x -B x ) 2 -(A y -B y ) 2 ≤(A r +B r ) 2

[0200] Among them, A x A y These are the x-axis and y-axis coordinates of ballast A, respectively. r Let B be the radius of ballast A; x B y These are the x-axis and y-axis coordinates of ballast B, respectively. r Let B be the radius of the ballast.

[0201] S402. For a ballast to be detected, use the above formula to iterate through all other ballasts in the physical coordinate system image to obtain all contact ballasts of the ballast.

[0202] S403. The deformation of the contact bar is calculated based on the physical coordinates of the contact ballast.

[0203] Specifically, the displacement modulus can be used to reflect the deformation of the contact bar. The calculation method is the same as the above calculation method for the displacement modulus of the ballast to be tested. For details, refer to formulas (1) to (2).

[0204] It should be noted that when selecting an industrial camera with a checkerboard calibration plate distortion correction function, since the industrial camera itself has distortion correction and coordinate correction functions, the S200 step can be omitted, and the feature image analysis can be performed directly on the image transmitted by the industrial camera to the host computer.

[0205] During the experiment, data sequences can be generated sequentially according to the identification information or number of the ballast bars to provide users with basic experimental data.

[0206] This embodiment is based on a collaborative architecture of the loading subsystem and the vision subsystem. It applies vertical / horizontal loads in real time, and simultaneously collects the displacement field and rotation angle of the ballast bar material using an industrial camera. The sensor and vision features are synchronized at the millisecond level via Ethernet protocol. The ORB feature matching algorithm is combined with checkerboard calibration technology. The motion area is separated by background modeling through a sliding window, and the gray-scale centroid method is used to optimize the positioning of the marker points. The dynamic fatigue and motion process is reproduced using a high-frequency vibration loading module. It supports CSV / Excel format data backtracking and frame-by-frame reproduction of abnormal working conditions.

[0207] Compared with the prior art, this embodiment has the following technical advantages:

[0208] 1. In terms of parameter measurement, existing technologies can only acquire single mechanical parameters (such as pressure and displacement) and cannot simultaneously analyze rotation angle, trajectory evolution and micro-strain field distribution. However, this embodiment adopts multi-dimensional parameter collaborative measurement, that is, simultaneously acquires rotation angle, trajectory path and micro-strain, and supports millisecond-level time synchronization between mechanical sensors and visual features.

[0209] 2. Regarding interference suppression, existing technologies suffer from the drawback of visual detection failure due to natural light reflection and mechanical vibration. However, this embodiment uses the ORB feature matching algorithm and checkerboard calibration technology to achieve sub-pixel level displacement tracking, and uses Fourier transform to filter out high-frequency vibration noise, which can achieve a very good interference suppression effect.

[0210] 3. Regarding working condition simulation, existing technologies lack this function. This embodiment can realize high-frequency vibration loading, thereby achieving working condition simulation.

[0211] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0212] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0213] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0214] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0215] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0216] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A ballast rotation and track change test method, characterized by, The method comprises the following steps: S100, acquiring a plurality of continuous original images of ballast under the action of vertical and horizontal loading forces, each of the original images comprising a plurality of ballasts; S200, performing distortion correction and coordinate correction on each of the original images to obtain a corresponding physical coordinate system image; S300, performing feature pattern analysis on each of the physical coordinate system images to obtain feature attributes of each of the ballasts to be detected in the images; The feature attributes comprise the center of mass position, radius, area and rotation angle of the ballast to be detected; The contour of each of the ballasts to be detected is extracted through feature pattern analysis, the geometric center of the contour of each of the ballasts to be detected is taken as the center of mass position of the ballast, and the radius and area of the ballast are calculated based on the contour; The rotation angle of each of the ballasts to be detected is obtained by the following method, S301, for each of the physical coordinate system images, all the sampling points in the image are combined in pairs to form sampling point pairs, and the combination mode of all the sampling point pairs forms a set A, ； wherein p i is a two-dimensional coordinate of the i-th sampling point in the physical coordinate system image, p j is a two-dimensional coordinate of the j-th sampling point in the physical coordinate system image, R 2 is a set of two-dimensional real numbers, and N is the number of sampling points in the physical coordinate system image; S302、for each sample point pair in set A, get its local gradient , ； wherein, is the pixel value of the i-th sample point in the physical coordinate system image at scale i is the pixel value of the j-th sample point in the physical coordinate system image at scale j .​ S303, a long-distance point pair subset L is constructed, ； wherein δ min = 13.67t min , t min is the minimum value among the scale parameter values of all feature points on the ballast profile. S304, the main direction position of the ballast to be detected is obtained according to the point pairs in the long-distance point pair subset L and the local gradient, ； where g is a main direction position vector of the ballast to be detected, g x is a component of the main direction position vector g in the x-axis direction, g y is a component of the main direction position vector g in the y-axis direction; M is the number of elements in the long distance point pair subset L. S305, the rotation angle is obtained according to the main direction position vector, ； Wherein, α is the rotation angle of the ballast to be detected; S400, according to the feature attributes of each of the ballasts to be detected, all the physical coordinate system images are analyzed by using time sequence image analysis technology, and the dynamic parameters of each of the ballasts to be detected are obtained by using adjacent frame difference method.

2. The method of claim 1, wherein, The ballast to be detected has identification information, the original image or the physical coordinate system image is input into a trained three-layer neuron network to obtain the identification information of each of the ballasts to be detected in the image, and the three-layer neuron network takes softmax as an activation function and takes sparse class cross entropy as a loss function.

3. The method of claim 1, wherein, The S400 comprises: The dynamic parameters comprise a moving track and an adjacent displacement; Specifically, A moving track set T of the ballast to be detected, ； wherein, is the physical coordinate of the mass center position of the ballast to be detected at the time t0, is the physical coordinate of the mass center position of the ballast to be detected at the time t0+ t, is the physical coordinate of the mass center position of the ballast to be detected at the time t0+ 2t, is the physical coordinate of the mass center position of the ballast to be detected at the time t0+ 3t, k is the physical coordinate of the mass center position of the ballast to be detected at the time t0+ 4t, t = 100 ms; The adjacent displacement comprises an adjacent frame displacement vector difference and a displacement module length, The adjacent frame displacement vector difference is calculated by the following formula, ； wherein, is the adjacent frame displacement vector difference of the ballast to be detected at time t, is the adjacent frame displacement vector difference of the ballast to be detected at time t, is the physical coordinates of the centroid position of the ballast to be detected at time t, is the physical coordinates of the centroid position of the ballast to be detected at time t, t = 100 ms, x is the component in the x-axis direction, y is the component in the y-axis direction; The displacement module length is calculated according to the adjacent frame displacement vector difference through the following formula , 。 4. The method of claim 1, wherein, The S400 comprises: The dynamic parameters comprise a rotation angle change amount, an angular velocity, a circumference deformation and an area deformation; Specifically, the rotation angle change amount is calculated by the following formula, ； wherein, is a change in a rotation angle of the ballast to be detected at time t, is a rotation angle of the ballast to be detected at time t+ t, is a rotation angle of the ballast to be detected at time t; The angular velocity is obtained according to the rotation angle change amount, ； wherein, is the angular velocity of the ballast to be detected at time t, in units of , t = 100 ms; The circumference deformation is calculated by the following formula, ； ； C(t) = C(t) + C(t) 物理 (t) is the physical circumference of the ballast to be detected at time t, C 像素 (t) is the total number of pixels of the profile of the ballast to be detected extracted by the analysis of the characteristic pattern, C(t) is the circumference deformation of the ballast to be detected at time t, C 物理 (t+ t) is the physical circumference of the ballast to be detected at time t+ t, t = 100 ms; The area deformation is calculated by the following formula, ； ； ； wherein L 物理 (t) is the distance between the two end feature points of the ballast to be detected at time t, x1 and y1 are respectively the x-axis coordinate and y-axis coordinate of one end feature point of the ballast to be detected at time t, x2 and y2 are respectively the x-axis coordinate and y-axis coordinate of the other end feature point of the ballast to be detected at time t, A 表面积 (t) is the surface area of the ballast to be detected at time t, r(t) is the radius of the ballast to be detected at time t, A 表面积 (t) is the surface area of the ballast to be detected at time t, r(t) is the radius of the ballast to be detected at time t, A t) is the surface area of the ballast to be detected at time t t, A(t) is the surface area deformation of the ballast to be detected at time t, t = 100 ms.

5. The method of claim 1, wherein, The S400 comprises: The dynamic parameters comprise a deformation of a contact rod; Specifically, For any two ballasts in the same physical coordinate system image, if the following formula is satisfied, it is judged that the two ballasts are adjacent, ； wherein A x , A y are the x-axis coordinate and the y-axis coordinate of ballast A, respectively, A r is the radius of ballast A; B x , B y are the x-axis coordinate and the y-axis coordinate of ballast B, respectively, B r is the radius of ballast B; For one ballast to be detected, all the other ballasts in the physical coordinate system image are sequentially cycled by using the above formula to obtain all the contact ballasts of the ballast; The deformation of the contact rod is calculated according to the physical coordinates of the contact ballasts.

6. A ballast rotation and trajectory change testing system, characterized by, The system comprises a loading subsystem and a vision subsystem; The loading subsystem comprises a placing groove, a vertical loading mechanism and a horizontal loading mechanism, The placing groove is used for placing the ballast to be detected, The vertical loading mechanism comprises a first ball screw, a first loading motor for driving the first ball screw to rotate, a vertical pressing plate connected with a first nut of the first ball screw, and a vertical linear guide rail, The first screw rod of the first ball screw is vertically arranged, The vertical pressing plate is arranged directly above the ballast to be detected, The horizontal loading mechanism comprises a second ball screw, a second loading motor for driving the second ball screw to rotate, a horizontal pressing plate connected with a second nut of the second ball screw, and a horizontal linear guide rail, The second screw rod of the second ball screw is horizontally arranged, The horizontal pressing plate is arranged on one side of the ballast to be detected; The visual subsystem comprises an industrial camera and an upper computer, The industrial camera is used to collect continuous original images of the ballast in the placing groove under the action of the vertical loading force of the vertical loading mechanism and the horizontal loading force of the horizontal loading mechanism, The upper computer is in communication connection with the industrial camera and is used to execute the method of any one of claims 1-5.

7. The system of claim 6, wherein, The placing groove is arranged on a loading frame, and a Foma wheel is installed on the loading frame; The vertical loading mechanism further comprises a first speed reducer connected with the first loading motor, and the first speed reducer is connected with the first screw rod of the first ball screw through a first coupling, The first nut of the first ball screw and the vertical pressing plate are connected through a first nut seat and a first sensor mounting plate, The first nut is connected with the first sensor mounting plate through the first nut seat, and the first sensor mounting plate is connected with the vertical pressing plate, The vertical linear guide rail is arranged on both sides of the first sensor mounting plate, A vertical force sensor in communication connection with the upper computer is installed on the first sensor mounting plate; The second screw rod of the second ball screw is connected with the second loading motor through a second speed reducer, The second nut of the second ball screw is connected with the horizontal pressing plate through a second sensor mounting plate and a thrust transmission plate, The second nut is connected with the second sensor mounting plate, the second sensor mounting plate is connected with the thrust transmission plate through a joint ball head, and the thrust transmission plate is connected with the horizontal pressing plate, A horizontal force sensor in communication connection with the upper computer is installed on the second sensor mounting plate.

Images