Non-contact measurement method and system for measuring the height of combined steel springs in large bogies

Abstract

This invention discloses a non-contact measurement method and system for the height of combined steel springs in large bogies. Based on this height measurement method, a steel spring height detection system based on a dual-camera 3D intelligent contour sensor is constructed. The system involves fixing the dual-camera 3D intelligent contour sensor and scanning a set of combined steel springs to obtain 3D point cloud data. The 3D point cloud data is processed and fitted to obtain the end faces, reference surfaces, and height data of the large and small steel springs in the combined steel springs. The system then performs a cyclic scanning motion when scanning subsequent sets of combined steel springs. This method enables real-time, high-precision steel spring height detection, significantly improving the efficiency of steel spring height detection, reducing the workload of relevant personnel, increasing measurement efficiency, reducing labor costs, simplifying measurement procedures, and enhancing measurement automation.

Description

Technical Field

[0001] This invention relates to a steel spring for a railway freight car bogie, and more particularly to a non-contact method for measuring the height of a large combined steel spring used on a railway freight car bogie. Background Technology

[0002] To achieve rapid and stable development in the railway freight car transportation market, the railway vehicle sector urgently needs to automate and intelligently inspect bogies, while simultaneously improving inspection efficiency and accuracy. Bogie springs, a crucial component of the bogie, provide support and damping for the suspension system. Their dimensions directly impact the vehicle's damping capacity and characteristics; therefore, height measurement is of paramount importance. The height difference in the bogie springs directly affects the wheel load distribution, thus influencing vehicle performance. Traditional inspection methods are inefficient and labor-intensive. With technological advancements, even a 1mm difference in the mounting surface height of the bogie springs, for railway vehicles weighing tens or even hundreds of tons, can have an impact on wheel load deviation reaching hundreds of millions of tons. Therefore, measuring the height of the bogie springs is crucial.

[0003] Bogie combined steel springs (see) Figure 4 The bogie spring height measurement system consists of a large steel spring 50 with a larger diameter and a smaller steel spring 60 nested inside it. Current methods for detecting bogie spring height require manual removal of both springs from the bogie and placing them on a conveyor belt. When the springs are transported to the testing area, they are manually separated and their heights are measured in batches. The larger springs are then manually sorted and placed on brackets for the smaller springs to be tested. This method is labor-intensive, inefficient, and lacks automation. With the rapid development of computer technology and science, more and more high-precision sensors based on point cloud processing are being used in spring measurement. Using detection technologies such as CCD and CMOS cameras to capture a set of two-dimensional images of the steel spring, and then manually performing stereo matching to obtain a three-dimensional image for measurement, the time cost is relatively high and the accuracy is relatively low; point laser displacement sensors require N measurements to determine the end face of the steel spring, and it is difficult to determine the end face 51 of the large steel spring and the end face 61 of the small steel spring, which is time-consuming and complicated; line laser displacement sensors need to move to find the contour of the end face of the steel spring through the emitted laser, and then determine the height of the steel spring, but line laser sensors have high requirements for the surface of the measured object. For objects with uneven end faces such as steel springs, additional processing may be required, which takes a long time. Summary of the Invention

[0004] This invention addresses the current challenges of high labor costs, low efficiency, low automation, difficulty in determining the end faces of large and small springs, and lengthy and complex procedures in measuring the height of large bogie composite springs on railway freight cars. It provides a non-contact measurement method and system for large bogie composite spring height that improves measurement efficiency, reduces labor costs, simplifies measurement procedures, and enhances measurement automation.

[0005] The specific technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows: a non-contact measurement method for the height of combined steel springs in large bogies, comprising the following measurement method steps:

[0006] A1. Align the combined steel springs, aligning the dual-camera 3D intelligent contour sensor with the single set of steel springs to be measured;

[0007] A2. Image acquisition of combined steel springs: The three-dimensional point cloud data of the single combined steel springs in step A1 above are obtained by scanning with a dual-camera 3D intelligent contour sensor.

[0008] A3. Process the combined steel spring image. Process the three-dimensional point cloud data obtained in step A2 above and fit it to obtain the end face of the large steel spring, the end face of the small steel spring, and the common reference plane of the large and small steel springs.

[0009] A4. Calculate the height of the combined steel springs. The distance between the end face of the large steel spring and the reference plane is the height of the large steel spring. The distance between the end face of the small steel spring and the reference plane is the height of the small steel spring.

[0010] A5. Send the calculation results: Send the measured height data of a single combined steel spring obtained from step A4 above to the remote control center;

[0011] A6. The dual-camera 3D intelligent contour sensor automatically aligns with all the individual combined steel springs on the next multi-combined steel spring tray to be measured and repeats the above steps A1 to A5 to perform the measurement task on all the individual combined steel springs on the next multi-combined steel spring tray to be measured.

[0012] A7. Repeat steps A1 to A6 above until all individual combined steel springs on multiple combined steel spring trays on all conveyor belts to be measured have been measured;

[0013] A8. In step A7 above, as the conveyor belt operates, repeat steps A1 to A7 above until the height of each individual combined steel spring on all the multiple combined steel spring trays that need to be measured is completed.

[0014] A method for measuring bogie spring height based on a dual-camera 3D intelligent profile sensor utilizes the sensor to acquire point cloud data, enabling a top-to-bottom scan of the spring even when stationary, resulting in a 3D model. This method achieves high-precision spring height measurement, addressing issues such as high labor costs, low efficiency, low automation, and complex procedures. Point cloud data processing-based detection technology offers higher automation and efficiency, representing a future trend. It improves measurement efficiency, reduces labor costs, simplifies measurement procedures, and enhances measurement automation.

[0015] Preferably, each of the multiple combined steel spring trays has nine single combined steel springs. These nine single combined steel springs are distributed in three rows and four columns on the trays. The first and third rows each have four single combined steel springs, while the middle second row has only one. This single combined steel spring in the middle second row is positioned in the fourth column of the trays on the far side of the moving direction towards the dual-camera 3D intelligent contour sensor. It combines with the single combined steel springs in the fourth column of the first and third rows to form a fourth column with three single combined steel springs. The other three columns each have only two single combined steel springs, thus forming a U-shaped arrangement. The opening of the U-shape faces the front of the trays on the conveyor belt in the transport direction. This improves the efficiency of non-contact measurement of the combined steel spring height.

[0016] Preferably, in step A1 above, a dual-camera 3D intelligent contour sensor is installed above multiple combined steel spring trays containing nine steel springs, enabling scanning of all individual combined steel springs on the multiple combined steel spring trays and acquiring the 3D point cloud data of each individual combined steel spring. This improves the efficiency and convenience of non-contact measurement of the combined steel spring height.

[0017] Preferably, in step A2 above, the dual-camera 3D intelligent contour sensor is controlled by a computer or corresponding host computer device connected to the dual-camera 3D intelligent contour sensor to scan and acquire the point cloud data of the first single combined steel spring. This improves the effectiveness of controlling the scanning acquisition of the point cloud data of the combined steel spring.

[0018] Preferably, step A3 above includes the following processing and fitting method:

[0019] A31. Preprocess the collected point cloud data of a single combined steel spring to obtain the three-dimensional point cloud data of the smooth end faces of the large and small steel springs and their reference surfaces in a single combined steel spring. The preprocessing includes noise reduction and downsampling.

[0020] A32. Remove outlier and mixed noise from individual combined steel spring point cloud data. The point cloud denoising methods used to remove outlier and mixed noise include radius filtering, statistical filtering, and pass-through filtering.

[0021] A33. To improve the algorithm's running speed and ultimately increase the measurement speed of the height of a single combined steel spring, the sampling simplification methods adopted to reduce the number of single combined steel springs include uniform sampling and nearest neighbor sampling.

[0022] A34. When the heights of the large and small springs in a single combined steel spring are different after the noise reduction and simplification processing of the point cloud data of the single combined steel spring in steps A31 to A33, plane fitting processing is performed on the end faces and reference planes of the large and small springs in the single combined steel spring after noise reduction and downsampling processing.

[0023] Improve the accuracy and effectiveness of non-contact measurement processing and fitting of point cloud data for composite steel springs.

[0024] Preferably, in step A4 above, when the heights of the large and small steel springs in a single combined steel spring are different after noise reduction, downsampling, and simplification, planar fitting processing is performed on the end faces and reference planes of the large and small steel springs in the single combined steel spring. The distance between the point cloud planes of the single combined steel spring is calculated, where the distance between the end face of the large steel spring and the reference plane is the height of the large steel spring, and the height between the end face of the small steel spring and the reference plane is the height of the small steel spring. This improves the accuracy and effectiveness of non-contact measurement and fitting of the height of the combined steel spring.

[0025] Preferably, in step A5 above, the calculated height data of a single combined steel spring is correlated with the corresponding point cloud data of the combined steel spring and sent to the remote control center for storage in the host computer software. This allows staff to view the height information of this single combined steel spring and determine whether it needs to be replaced. This improves the accuracy and effectiveness of non-contact measurement of combined steel spring height data and increases the efficiency of combined steel spring screening.

[0026] Preferably, in steps A6 to A7 above, after all individual combined springs in the same multiple combined spring trays have been scanned and processed, the conveyor belt continues to rotate and transport the springs. When the next multiple combined spring tray reaches the position of the dual-camera 3D intelligent contour sensor, steps A1 to A5 above are repeated to perform the height measurement task for the next group of individual combined springs; this continues until the height measurement task for all the combined springs to be measured in the multiple combined spring trays on the conveyor belt has been completed. This improves the efficiency of non-contact automatic measurement of combined springs in multiple combined spring trays.

[0027] Preferably, when the nine single-unit combined steel springs are scanned using a dual-camera 3D intelligent contour sensor in step A2 to acquire the 3D point cloud data of each single-unit combined steel spring, the sampling scanning order is as follows: every three single-unit combined steel springs in the U-shape formed by the nine single-unit combined steel springs are used as a scanning unit, and the scanning order is executed clockwise or counterclockwise, scanning one scanning unit at a time to obtain the 3D point cloud data of each single-unit combined steel spring. This improves the efficiency of non-contact automatic measurement of combined steel springs in multiple combined steel spring trays.

[0028] Another objective of this invention application is to provide a non-contact measurement system for the height of combined steel springs in large bogies, employing one of the aforementioned technical solutions. This system includes a point cloud data acquisition module, a data receiving module, a data storage module, a data processing module, and a display module, connected sequentially in a hierarchical manner. The point cloud data acquisition module includes a dual-camera 3D intelligent contour sensor and multiple combined steel spring tray transmission lines. The data receiving module uses an STM32F407 microcontroller. The data storage module uses a Macronix MX25L25635F memory chip. The data processing module uses the CUDA data processing platform. CUDA is a general-purpose parallel computing platform and programming model from NVIDIA, used for data processing.

[0029] The beneficial effects of this invention are as follows: The bogie spring height measurement method based on a dual-camera 3D intelligent contour sensor utilizes a dual-camera 3D intelligent contour sensor to acquire point cloud data, enabling a top-to-bottom scan of the spring even when stationary, thus obtaining a three-dimensional model of the spring. Using a dual-camera 3D intelligent contour sensor for spring height measurement can achieve high-precision measurement of the spring height, solving problems such as high labor costs, low efficiency, low automation, and complex processes. Point cloud data processing-based detection technology offers higher automation and efficiency, representing a development trend. It can improve measurement and detection efficiency, reduce labor costs, simplify measurement procedures, and enhance measurement automation.

[0030] This application designs a method for measuring the height of large bogie springs based on a dual-camera 3D intelligent contour sensor, and constructs a spring height detection system based on this method. The fixing method of the dual-camera 3D intelligent contour sensor and its movement method when scanning subsequent groups of springs after scanning one group are designed. Software for point cloud image acquisition, image processing, and data storage is also designed. This system can achieve real-time, high-precision spring height detection, significantly improving the efficiency of spring height detection and reducing the workload of relevant personnel. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the height measurement process of the non-contact measurement method and measurement system for the combined steel springs of large bogies according to the present invention.

[0032] Figure 2 This is a schematic diagram of the system block diagram of the non-contact measurement method and measurement system for the combined steel spring height of a large bogie according to the present invention.

[0033] Figure 3 This is a schematic diagram of the measurement transmission line structure used in the non-contact measurement method and measurement system for the combined steel spring height of large bogies of the present invention.

[0034] Figure 4 This is a schematic diagram of a single steel spring assembly structure that needs to be measured by the non-contact measurement method and measurement system for the height of the combined steel springs of the large bogie of the present invention. Detailed Implementation

[0035] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0036] Example 1:

[0037] Figure 1 , Figure 3 , Figure 4 In the illustrated embodiment, a non-contact method for measuring the height of a large bogie's combined steel springs includes the following measurement steps:

[0038] A1. Align the combined steel spring 01, and align the dual-camera 3D intelligent contour sensor with the single set of steel springs to be measured;

[0039] A2. Image acquisition of combined steel springs 02: The three-dimensional point cloud data of the single combined steel springs in step A1 above are obtained by scanning with a dual-camera 3D intelligent contour sensor.

[0040] A3. Process the combined steel spring image 03. Process the three-dimensional point cloud data obtained in step A2 above and fit it to obtain the end face of the large steel spring, the end face of the small steel spring, and the common reference plane of the large and small steel springs.

[0041] A4. Calculate the height of the combined steel spring 04. The distance between the end face of the large steel spring and the reference plane is the height of the large steel spring. The distance between the end face of the small steel spring and the reference plane is the height of the small steel spring.

[0042] A5. Send calculation result 05, send the measured height data of a single combined steel spring obtained from step A4 above to the remote control center;

[0043] A6. Repeat the measurement of the combined steel spring 06. The dual-camera 3D intelligent contour sensor automatically aligns with all the individual combined steel springs on the next tray of multiple combined steel springs to be measured and repeats the above steps A1 to A5 to perform the measurement task of all the individual combined steel springs on the next tray of multiple combined steel springs to be measured.

[0044] A7. Repeat steps A1 to A6 above until all individual combined steel springs on multiple combined steel spring trays on all conveyor belts to be measured have been measured;

[0045] A8. In step A7 above, as the conveyor belt operates, repeat steps A1 to A7 above until the height of each individual combined steel spring on all the multiple combined steel spring trays to be measured is completed, and the measurement ends after the required measurement task is completed.

[0046] The measurement data included in this combined steel spring measurement method mainly consists of the end face height information and the reference plane height information of the combined steel spring. The common reference plane for the large and small steel springs is the unloading plane of multiple combined steel spring trays.

[0047] A series of combined steel spring trays 20 are placed on the combined steel spring conveyor belt 10. A measuring frame 30 is provided above the combined steel spring conveyor belt 10, and a dual-camera 3D intelligent contour sensor 40 is installed on the measuring frame 30. Each of the multiple combined steel spring trays 20 is fixed with nine single-set combined steel spring positioning seats 21. The nine single-set combined steel springs are respectively placed on the nine single-set combined steel spring positioning seats 21. Each of the nine single-set combined steel springs has a combined set structure in which large and small steel springs are nested together. The nine single combined steel springs are sampled in three rows and four columns on the multiple combined steel spring trays, with the first and third rows evenly distributed. Four individual combined steel springs are arranged, with only one individual combined steel spring in the second row. This single combined steel spring in the second row is positioned in the fourth column on the far side of the multiple combined steel spring trays in the direction of movement towards the dual-camera 3D intelligent contour sensor. It combines with the single combined steel springs in the fourth column of the first and third rows to form a fourth column with three individual combined steel springs. The other three columns each have only two individual combined steel springs. This overall arrangement forms a U-shaped distribution structure, with the U-shaped opening facing the front end of the multiple combined steel spring trays in the transport direction on the conveyor belt (i.e., see...). Figure 3 As shown, the U-shaped opening faces the measuring frame 30, which is equipped with a dual-camera 3D intelligent contour sensor 40. The dual-camera 3D intelligent contour sensor 40 can automatically detect and measure the three-dimensional point cloud data of the combined steel spring during the forward transport of the conveyor belt.

[0048] In step A1 above, the dual-camera 3D intelligent contour sensor 40 is installed on the measuring frame 30 above the multiple combined steel spring trays 20 containing nine steel springs, so as to scan all the individual combined steel springs on the multiple combined steel spring trays 20 and acquire the three-dimensional point cloud data of the individual combined steel springs.

[0049] In step A2 above, the dual-camera 3D intelligent contour sensor is controlled by a computer or corresponding host computer device connected to the dual-camera 3D intelligent contour sensor to scan and acquire the point cloud data of the first single combined steel spring.

[0050] The above A3 step includes the following processing and fitting methods:

[0051] A31. Preprocess the collected point cloud data of a single combined steel spring to obtain the three-dimensional point cloud data of the smooth end faces of the large and small steel springs and their reference surfaces in a single combined steel spring. The preprocessing includes noise reduction and downsampling.

[0052] A32. Remove outlier and mixed noise from individual combined steel spring point cloud data. The point cloud denoising methods used to remove outlier and mixed noise include radius filtering, statistical filtering, and pass-through filtering.

[0053] A33. To improve the algorithm's running speed and ultimately increase the measurement speed of the height of a single combined steel spring, the sampling simplification methods adopted to reduce the number of single combined steel springs include uniform sampling and nearest neighbor sampling.

[0054] A34. When the heights of the large and small springs in a single combined steel spring are different after denoising and simplification processing in steps A31 to A33, plane fitting processing is performed on the spring end faces and reference planes of the large and small springs in the single combined steel spring after denoising and downsampling. The plane fitting process uses the random sample consensus algorithm, which is an iterative method to calculate mathematical model parameters from a series of data containing outliers. The processing steps are as follows:

[0055] A331. Three points are randomly selected from the point cloud. These three points form a plane. The equation of the plane is in the format: A·x + B·y + C·z + D = 0; A, B, and C are the x, y, and z components of the plane's normal vector. The normal vector points in a direction perpendicular to the plane, and its components represent the degree of inclination of the plane in the x, y, and z axes; D is the constant term of the plane equation, representing the distance between the plane and the origin.

[0056] A332. Calculate the distance from all other points to the plane. If the distance is less than the threshold T, consider them to be points on the same plane.

[0057] A333. If there are more than n points in the same plane, save the plane and mark all the points in the plane as matched;

[0058] A334. The termination condition is that after N iterations, the plane found has fewer than n points, or no three marked points can be found.

[0059] In step A4 above, when the heights of the large and small springs in a single combined spring system differ after noise reduction, downsampling, and simplification, plane fitting is performed on the spring end faces and reference planes of the large and small springs respectively. The distance between the point cloud planes of the single combined spring system is calculated, where the distance between the end face of the large spring and the reference plane is the height of the large spring, and the height between the end face of the small spring and the reference plane is the height of the small spring. Note: The reference plane is the common reference plane for both the large and small springs, i.e., the unloading plane of multiple combined spring trays is used as the common reference plane for both the large and small springs; the point cloud plane is the common reference plane for the end faces of the large and small springs after plane fitting; and the 3D point cloud data is the raw point cloud data obtained from a dual-camera 3D intelligent contour sensor without noise reduction or downsampling; the distance between the end face of the large spring and the reference plane after plane fitting of the 3D point cloud data is the height of the large spring, and the distance between the end face of the small spring and the reference plane after fitting is the height of the small spring.

[0060] In step A5 above, the calculated height data of a single combined steel spring is correlated with the corresponding point cloud data of the combined steel spring, and sent to the remote control center for storage in the host computer software. This allows staff to view the height information of the single combined steel spring and determine whether replacement is necessary. The order in which the dual-camera 3D intelligent contour sensor used in this solution scans the point cloud data of a single combined steel spring corresponds to the height data of each group of steel springs. For example, nine groups of steel springs are scanned in a U-shaped sequence: downward scan, horizontal scan, and upward scan, resulting in point cloud data D1, D2, D3, D4, D5, D6, D7, D8, and D9. The final height data is also D1, D2, D3, D4, D5, D6, D7, D8, and D9. The height data of each group of combined steel springs corresponds one-to-one with its point cloud data. When using a dual-camera 3D intelligent contour sensor to scan the same combined steel spring tray 20 task, it is necessary to scan nine times. However, the point cloud data obtained by scanning three times in sequence is a group. If scanning is performed in a U-shape, the bottom, horizontal and top are three groups, but the bottom, horizontal and top each contain three groups of combined steel springs, so it is necessary to scan three times for a total of nine times.

[0061] In steps A6 to A7 above, after all individual combined steel springs in the same multiple combined steel spring trays have been scanned and processed, the conveyor belt continues to rotate and transport the next multiple combined steel spring trays to the position of the dual-camera 3D intelligent contour sensor. When the next multiple combined steel spring trays reach the position of the dual-camera 3D intelligent contour sensor, steps A1 to A5 above are repeated to perform the task of measuring the height of the next group of single combined steel springs; until the task of measuring the height of the combined steel springs to be measured in all multiple combined steel spring trays on the conveyor belt is completed.

[0062] When the nine single-group combined steel springs are scanned using a dual-camera 3D intelligent contour sensor in step A2 to acquire the 3D point cloud data of the single-group combined steel springs, the sampling scanning order is as follows: every three single-group combined steel springs in the U-shape formed by the nine single-group combined steel springs constitute one scanning unit 22 (see...). Figure 3 The scanning unit (defined by the dashed line) is scanned in a clockwise or counterclockwise order to obtain the three-dimensional point cloud data of a single steel spring.

[0063] The core algorithms for processing 3D point cloud data in the aforementioned measurement methods, which are based on existing technologies, include the following processing methods:

[0064] (1) Point cloud noise reduction method

[0065] Radius filtering: The main principle is that in point cloud data, there must be enough points within a given radius around each point. Points that meet the condition are retained, otherwise they are deleted.

[0066] Implementation steps:

[0067] 1. Construct a KD tree for the point cloud data and establish topological relationships;

[0068] 2. Traverse any point in the point cloud and calculate the number of points within a given radius around it;

[0069] 3. Given a threshold, if the number of points within a given radius is less than the threshold, they are judged as noise and removed; points greater than the threshold are retained; traverse all points until processing is complete.

[0070] Adjust the given radius and decision threshold parameters according to the actual point cloud scanning situation to achieve the best noise reduction effect.

[0071] Statistical filtering: This algorithm is mainly used to remove obvious outliers. These noise points are primarily measurement errors and are sparsely distributed in space. Each point in point cloud data contains a certain amount of information; the higher the density of the point cloud within a certain region, the more information it contains. Outliers have low density and contain relatively little information, which can be ignored. The principle of statistical filtering is to define the density of the point cloud at a certain point. If the density at that point is less than a given threshold, the point cloud is considered invalid and will be removed.

[0072] Processing steps:

[0073] 1. Calculate the distance d from each point to its neighborhood K;

[0074] 2. Calculate the mean μ and standard deviation σ of the Gaussian distribution, and then set the judgment threshold as D = μ + nσ by setting the standard deviation;

[0075] 3. Compare d with the given judgment threshold D. If d is less than D, it is determined to be noise removal; otherwise, it is retained. Iterate through all points in the point cloud data until the processing is complete.

[0076] The calculation process is as follows:

[0077] Statistical analysis is performed on the neighborhood of each point in the point cloud data. It is assumed that the distances between all points in the point cloud follow a Gaussian distribution, the shape of which is determined by the mean μ and standard deviation σ. Let the coordinates of the nth point in the point cloud be Pn(Xn,Yn,Zn), and the distances from this point to any point P... m (X m ,Y m Z m The distance is:

[0078]

[0079] The formula for calculating the average distance from each point to any other point is:

[0080]

[0081] Standard deviation is

[0082]

[0083] Let std be the standard deviation factor. In the process of algorithm implementation, only two thresholds, k and std, need to be input. When the average distance of a point to its k nearest neighbors is within the standard range (μ-σ·std, μ+σ·std), the point is retained. Points outside this range are defined as outliers and deleted.

[0084] Pass-through filtering: The function of pass-through filtering is to filter out points whose values ​​in a specified dimension (x,y,z) are not within the given range. This algorithm is simple and efficient and is suitable for eliminating background noise over a large area.

[0085] Processing steps:

[0086] 1. Specify a dimension and the range of values ​​within that dimension;

[0087] 2. Determine whether the value of a point in a specified dimension is within the range, and delete points whose values ​​are within the range;

[0088] 3. Iterate through all points until all points within the specified value range have been deleted.

[0089] (2) Point cloud downsampling algorithm

[0090] Voxel grid method: A voxel grid is a three-dimensional cube that surrounds a point. Voxel filtering divides the entire point cloud space into multiple fixed voxel grids of a given size, and replaces other points in the grid with the center or centroid of all points in the voxel. The filtered point cloud is thus a set of center or centroid points. This filtering method does not destroy the geometry of the point cloud itself, so it will not damage the end face point cloud of the steel spring point cloud and affect the measurement results.

[0091] The calculation process for voxel downsampling is as follows:

[0092] 1. For all coordinate sets in the point cloud data, find the maximum value X in the X, Y, and Z directions respectively. max Y max Z max and minimum value X min Y min Z min ;

[0093] 2. Set the side length r of the voxel grid;

[0094] 3. Calculate the side length l of the minimum bounding box of the point cloud based on the maximum and minimum values ​​on the X, Y, and Z axes respectively. x l y l z .

[0095]

[0096] 4. Calculate the size of the voxel mesh

[0097]

[0098] In the formula, Indicates rounding down

[0099] 5. Calculate the index h of each voxel grid in the point cloud.

[0100]

[0101] 6. Sort the elements in h in ascending order, calculate the centroid of each voxel grid, and replace all points in the grid with the centroid.

[0102] (3) Point cloud smoothing algorithm

[0103] The purpose of point cloud smoothing is to reduce the density of point cloud data, fill in missing data, and obtain a smoother and more continuous surface. Smoothing helps improve the quality of point cloud data and provides a more accurate and stable foundation for subsequent analysis and processing.

[0104] Mean filtering: Mean filtering replaces a given point with the point at the mean distance between that point and its neighbors. The algorithm involves iterating through all points in the point cloud model, finding the k nearest neighbors of each point, calculating the location of the mean point using these k neighbors, and replacing the given point with the mean point.

[0105] Processing steps:

[0106] Select a template composed of its neighboring data points, and replace the value of the current sampling point with the mean of the template. For a 3D point cloud, the template is a sphere D, meaning the value at a sampling point (x0, y0, z0) in the 3D point cloud is replaced with the mean of its neighbors.

[0107] D(x0,y0,z0)=(x,y,z)|(x-x0) 2 +(y-y0) 2 +(z-z0) 2 <d 2 Replace with the average of the point set:

[0108]

[0109] In the formula,

[0110] d is the neighborhood radius

[0111] *D(x0,y0,z0) is the number of points in D(x0,y0,z0). The larger the neighborhood radius value, the smoother the point cloud after denoising, but the more severe the loss of detail information will be. Therefore, it is necessary to set the radius value reasonably.

[0112] Example 2:

[0113] Figure 2 , Figure 3In the illustrated embodiment, a non-contact measurement system for the height of combined steel springs in a large bogie, to better meet the needs of automated measurement, employs the non-contact measurement method for the height of combined steel springs in a large bogie as described in Embodiment 1. The system includes a point cloud data acquisition module B1, a data receiving module B2, a data storage module B3, a data processing module B4, and a display module B5, connected sequentially. The point cloud data acquisition module includes a dual-camera 3D intelligent contour sensor 40 and multiple combined steel spring tray transmission lines 10. The data receiving module uses an STM32F407 microcontroller, the data storage module uses a Macronix MX25L25635F memory chip, and the data processing module uses a CUDA data processing platform. During each measurement, the sensor promptly transmits the data to the STM32F407 microcontroller data receiving module. The Macronix MX25L25635F memory chip is used for data storage, and then the data is processed to obtain the height of the steel springs. The measurement results are displayed on a terminal device for further processing.

[0114] This invention proposes a method for measuring the height of large steel springs in bogies based on three-dimensional sensors. According to the maintenance procedures and processes for bogie steel springs, the system underwent point cloud image processing and height measurement analysis, achieving the required accuracy for steel spring height measurement and overcoming the shortcomings of low efficiency and high labor costs associated with traditional measurement techniques. With the development of science and technology, the measurement technology for combined steel springs in bogies should further evolve towards automation, higher efficiency, and reduced labor costs.

[0115] The above content and structure describe the basic principles, main features, and advantages of the product of this invention, which should be understood by those skilled in the art. The examples and descriptions above are merely illustrative of the principles of this invention. Various changes and modifications can be made to this invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A non-contact method for measuring the height of combined steel springs in a large bogie, characterized in that: The measurement methods and steps include the following: A1. Align the combined steel springs, aligning the dual-camera 3D intelligent contour sensor with the single set of steel springs to be measured; A2. Image acquisition of combined steel springs: The three-dimensional point cloud data of the single group of steel springs in step A1 above are obtained by scanning with a dual-camera 3D intelligent contour sensor for all individual combined steel springs on the same tray of multiple combined steel springs. A3. Process the combined steel spring image. Process the three-dimensional point cloud data obtained in step A2 above and fit it to obtain the end face of the large steel spring, the end face of the small steel spring, and the common reference plane of the large and small steel springs. A4. Calculate the height of the combined steel springs. The distance between the end face of the large steel spring and the reference plane is the height of the large steel spring. The distance between the end face of the small steel spring and the reference plane is the height of the small steel spring. A5. Send the calculation results: Send the measured height data of a single combined steel spring obtained from step A4 above to the remote control center; A6. Repeat the measurement of the combined steel springs. The dual-camera 3D intelligent contour sensor automatically aligns with all the individual combined steel springs on the next tray of multiple combined steel springs to be measured and repeats the above steps A1 to A5 to perform the measurement task of all the individual combined steel springs on the next tray of multiple combined steel springs to be measured. A7. Repeat steps A1 to A6 above until all individual combined steel springs on multiple combined steel spring trays on all conveyor belts to be measured have been measured; A8. In step A7 above, as the conveyor belt operates, steps A1 to A7 are repeated until the height of each individual combined steel spring on all the multiple combined steel spring trays to be measured is completed; each combined steel spring tray is provided with nine single combined steel springs, and the nine individual combined steel springs are distributed in three rows and four columns on the multiple combined steel spring trays. The first and third rows are each provided with four individual combined steel springs, and the middle second row is provided with only one individual combined steel spring. The one individual combined steel spring in the middle second row is located at the fourth column position on the far end side of the multiple combined steel spring trays in the direction of movement of the dual-camera 3D intelligent contour sensor, and together with the one individual combined steel spring at the fourth column position of the first and third rows, it forms a fourth column with three individual combined steel springs, while the other three columns have only two individual combined steel springs. Thus, the whole arrangement forms a U-shaped distribution structure, with the opening of the U-shape facing the front end of the multiple combined steel spring trays in the transportation direction on the conveyor belt; The above A3 step includes the following processing and fitting methods: A31. Preprocess the collected point cloud data of a single combined steel spring to obtain the three-dimensional point cloud data of the smooth end faces of the large and small steel springs and their reference surfaces in a single combined steel spring. The preprocessing includes noise reduction and downsampling. A32. Remove outlier and mixed noise from individual combined steel spring point cloud data. The point cloud denoising methods used to remove outlier and mixed noise include radius filtering, statistical filtering, and pass-through filtering. A33. To improve the algorithm's running speed and ultimately increase the measurement speed of the height of a single combined steel spring, the sampling simplification methods adopted to reduce the number of individual combined steel springs include uniform sampling and nearest neighbor sampling. A34. When the heights of the large and small springs in a single combined steel spring are different after denoising and simplification processing in steps A31 to A33, perform plane fitting processing on the end faces and reference planes of the large and small springs in the single combined steel spring after denoising and downsampling processing; In step A4 above, when the heights of the large and small steel springs in a single combined steel spring are different after noise reduction, downsampling, and simplification, plane fitting is performed on the end faces and reference planes of the large and small steel springs in the single combined steel spring. The distance between the point cloud planes of the single combined steel spring is calculated after the fitting process. The distance between the end face of the large steel spring and the reference plane is the height of the large steel spring, and the height between the end face of the small steel spring and the reference plane is the height of the small steel spring.

2. The non-contact measurement method for the combined steel spring height of a large bogie according to claim 1, characterized in that: In step A1 above, a dual-camera 3D intelligent contour sensor is installed above multiple combined steel spring trays containing nine steel springs, enabling scanning of all individual combined steel springs on the multiple combined steel spring trays and acquiring the three-dimensional point cloud data of each individual combined steel spring.

3. The non-contact measurement method for the combined steel spring height of a large bogie according to claim 1, characterized in that: In step A2 above, the dual-camera 3D intelligent contour sensor is controlled by a computer or corresponding host computer device connected to the dual-camera 3D intelligent contour sensor to scan and acquire the point cloud data of the first single combined steel spring.

4. The non-contact measurement method for the combined steel spring height of a large bogie according to claim 1, characterized in that: In step A5 above, the calculated height data of a single combined steel spring is matched with the corresponding point cloud data of the combined steel spring and sent to the remote control center to be stored in the host computer software. This allows staff to view the height information of the single combined steel spring and determine whether it needs to be replaced.

5. The non-contact measurement method for the combined steel spring height of a large bogie according to claim 1, characterized in that: In steps A6 to A7 above, after all individual combined steel springs in the same multiple combined steel spring trays have been scanned and processed, the conveyor belt continues to rotate and transport the next multiple combined steel spring trays to the position of the dual-camera 3D intelligent contour sensor. When the next multiple combined steel spring trays reach the position of the dual-camera 3D intelligent contour sensor, steps A1 to A5 above are repeated to perform the task of measuring the height of the next group of single combined steel springs; until the task of measuring the height of the combined steel springs to be measured in all multiple combined steel spring trays on the conveyor belt is completed.

6. The non-contact measurement method for the combined steel spring height of a large bogie according to claim 1, characterized in that: When the nine single-group combined steel springs are used to perform step A2 and scan using a dual-camera 3D intelligent contour sensor to obtain the three-dimensional point cloud data of the single-group combined steel springs, the sampling scanning order is as follows: in the U-shape formed by the arrangement of the nine single-group combined steel springs, every three single-group combined steel springs are taken as a scanning unit, and the scanning order is performed clockwise or counterclockwise in turn to obtain the three-dimensional point cloud data of the single-group combined steel springs.

7. A non-contact measurement system for the height of combined steel springs in a large bogie, employing the non-contact measurement method for the height of combined steel springs in a large bogie as described in any one of claims 1 to 6, characterized in that: It includes a point cloud data acquisition module, a data receiving module, a data storage module, a data processing module, and a display module that are connected sequentially. The point cloud data acquisition module includes a dual-camera 3D intelligent contour sensor and multiple combined steel spring tray transmission lines. The data receiving module uses an STM32F407 microcontroller. The data storage module uses a Macronix MX25L25635F memory chip. The data processing module uses the CUDA data processing platform.

Images