Railway vehicle wheel set full circumference dimension detection method and system

Abstract

The invention relates to a rail vehicle wheel set full circumference dimension detection method and system. The detection method comprises the steps that multiple sets of three-dimensional tread modules and wheel set dimension measurement modules are installed on the two sides of a rail; when a wheel passes through the three-dimensional tread module, multiple groups of original data are automatically collected; the size of the wheel set is calculated based on the multiple sets of original data; when a wheel passes through the wheel set size measurement module, original laser line data of a tread and a side image of the wheel are automatically collected, and curve fitting is carried out by using a neural network fitting algorithm based on the collected data and image to obtain the diameter of the wheel. According to the invention, the measurement module is installed beside the rail, so that the measurement of the size of the wheel set and the fault detection of the surface of the wheel set can be automatically completed when a train passes, the measurement speed is high, the real-time performance is good, and the automation degree is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of railway vehicles, and particularly relates to a method and system for detecting the full-circumference dimensions of railway vehicle wheelsets. Background Art

[0002] In the field of railway transportation, the dimensions of the wheelsets of train vehicles are one of the key factors ensuring the safe operation of trains. As the direct contact components between the train and the track, the dimensional accuracy of the wheelsets directly affects the running stability and safety of the trains. Accurate measurement of the wheelset dimensions and corresponding fault detection play an important role in preventing train operation failures and ensuring railway transportation safety.

[0003] In the prior art, hand-held or fixed-size measuring instruments or two-dimensional measuring devices are usually used to measure the wheelset dimensions. Among them, the measurement methods using hand-held or fixed-size measuring instruments cannot achieve real-time online monitoring of the wheelset dimensions, resulting in lagging data and difficult unified management. When two-dimensional measuring devices measure three-dimensional wheelsets, they are easily affected by changes in the external environment, and it is difficult to guarantee the accuracy of data acquisition. Subsequent algorithm processing is also difficult to completely make up for this defect.

[0004] Although some designs in the prior art have realized using three-dimensional sensors to obtain the surface data of the wheelset online and obtaining the dimensional parameters through fitting, the data obtained by the existing three-dimensional measurement technologies are mostly limited to local parts of the wheelset, with incomplete coverage. Moreover, most methods only have the function of dimensional measurement and cannot identify faults on the surface of the wheelset, making it difficult to meet the requirements of comprehensive detection of wheelsets in practical applications. Summary of the Invention

[0005] The purpose of the present invention is to solve one of the above technical problems, and provide a method and system for detecting the full-circumference dimensions of railway vehicle wheelsets. By using multiple groups of measurement and detection modules to perform non-contact dynamic measurement on the wheelset dimensions and the inner distance between wheel diameters, it is possible to automatically complete the measurement of the wheelset dimensions and the inner distance between wheel diameters when the train passes by.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is:

[0007] A method for detecting the full-circumference dimensions of railway vehicle wheelsets includes the following steps:

[0008] Install multiple groups of three-dimensional tread modules and wheelset dimension measurement modules on both sides of the track; the three-dimensional tread module includes a wheelset camera and a first laser component; the wheelset dimension measurement module includes a wheel diameter camera and a second laser component;

[0009] When the wheel passes through the three-dimensional tread module, use the wheelset camera and the first laser component to collect multiple groups of original data;

[0010] Obtain the three-dimensional point cloud data corresponding to each group of original data based on the optical plane parameters of the first laser component, and perform curve registration on the point cloud curve of the three-dimensional point cloud data;

[0011] Calculate the size of the wheel set based on the registered point cloud curve;

[0012] When the wheel passes through the wheel set size measurement module, use the wheel diameter camera to collect multiple groups of original tread laser line data and wheel side images in combination with the laser line emitted by the second laser component;

[0013] Obtain the tread three-dimensional curve and the wheel set capture position based on the original tread laser line data and the wheel side images;

[0014] Based on the curve coordinates and the wheel set capture position, obtain multiple coordinate points on the rolling circle of the wheel set, mark the corresponding relationship between the curve coordinates and the coordinate points on the rolling circle, and use the neural network fitting algorithm to perform curve fitting to obtain the wheel diameter.

[0015] In some embodiments of the present invention, the following steps are further included:

[0016] Perform internal parameter calibration on the wheel set camera to obtain the internal parameters and distortion parameters of the wheel set camera;

[0017] Perform three-dimensional calibration on the wheel set camera and the first laser component to obtain the three-dimensional calibration parameters of the three-dimensional tread module. The three-dimensional calibration parameters include the optical plane parameters of the first laser component;

[0018] Perform joint calibration on multiple groups of three-dimensional tread modules on the same side of the track to obtain the external parameters between the three-dimensional tread modules on the same side.

[0019] In some embodiments of the present invention, the following steps are further included:

[0020] Perform internal parameter calibration on the wheel diameter camera to obtain the internal parameters and distortion parameters of the wheel diameter camera;

[0021] Perform three-dimensional calibration on the wheel diameter camera and the second laser component to obtain the three-dimensional calibration parameters of the wheel set size measurement module; the three-dimensional calibration parameters include the optical plane parameters of the second laser component;

[0022] Mark the position points of the line laser in the wheel diameter camera coordinate system and the corresponding world coordinate positions of each position point, and use the deep learning algorithm to fit the external parameters between the wheel set size measurement modules on both sides of the track.

[0023] In some embodiments of the present invention, the following steps are further included:

[0024] Fuse the three-dimensional point cloud data of each group of three-dimensional tread modules on both sides respectively to obtain the complete three-dimensional point cloud data of the wheels on both sides;

[0025] Use a two-dimensional curve registration algorithm and a standard wheel pair curve to perform curve registration on the point cloud curves of the complete point cloud data.

[0026] In some embodiments of the present invention, the method for performing curve registration on the point cloud curves of three-dimensional point cloud data specifically includes the following steps:

[0027] For each point cloud curve, use a two-dimensional curve registration algorithm and a standard wheel pair curve to determine the corresponding points between the current curve and the standard curve based on the nearest point principle. The expression of the correspondence relationship between the corresponding points of the current curve and the standard curve is:

[0028]

[0029] where θ, b1, and b2 are all undetermined coefficients;

[0030] Solve the undetermined coefficients. The principle for solving the undetermined coefficients is that the distance from the transformed target point to the reference point is the shortest;

[0031] Establish an objective function based on the solution principle:

[0032]

[0033] Construct a new target point set based on the obtained undetermined coefficients and the objective function;

[0034] Repeat the steps of determining the corresponding points between the current curve and the standard curve, solving the undetermined coefficients, and constructing a new target point set until a predetermined termination condition is met.

[0035] In some embodiments of the present invention, it further includes the following steps:

[0036] Perform difference detection on the point cloud data of the calculated wheel pair size and the point cloud data of the standard wheel diameter to obtain a difference image;

[0037] Perform deep learning defect segmentation on the difference image to obtain a defect area to achieve wheel pair defect detection.

[0038] In some embodiments of the present invention, a wheel positioning sensor is provided on the track. When the wheel positioning sensor is triggered to sense the arrival of the wheel, the wheel pair camera and the wheel diameter camera are activated to capture images.

[0039] In some embodiments of the present invention, multiple groups of three-dimensional tread modules and wheel pair size measurement modules on both sides of the track are symmetrically installed with respect to the track.

[0040] In some embodiments of the present invention, the wheel pair camera and the first laser component in the same three-dimensional tread module and between each three-dimensional tread module are connected by mechanical connectors;

[0041] The wheel diameter camera and the second laser assembly of the same wheel set dimension measurement module, as well as between each wheel set dimension measurement module, are connected through mechanical connectors.

[0042] Some embodiments of the present invention further provide a track vehicle wheel set full circumference dimension detection system, including:

[0043] At least one processor;

[0044] At least one memory for storing at least one program;

[0045] When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned track vehicle wheel set full circumference dimension detection method.

[0046] The beneficial effects of the present invention are as follows:

[0047] 1. By installing multiple groups of three-dimensional tread modules and wheel set dimension measurement modules on both sides of the track, and using the cooperation of three-dimensional cameras and lasers to collect original data, the present invention realizes non-contact accurate measurement of the full circumference dimensions of the wheel set and accurate calculation of the wheel diameter, ensuring the comprehensiveness and accuracy of the measurement data;

[0048] 2. By performing internal parameter calibration and three-dimensional calibration on the three-dimensional tread modules and wheel set dimension measurement modules, accurate camera parameters and light plane parameters are obtained, and joint calibration is performed on multiple groups of three-dimensional tread modules and multiple groups of wheel set dimension measurement modules to obtain the external parameters between the modules on the same side, thereby improving the measurement accuracy and data consistency and effectively reducing measurement errors;

[0049] 3. In the present invention, the fitting process of calibration is constructed by building a deep learning neural network, which can adapt to different complexity requirements by adjusting the number of network nodes and layer depth to achieve higher calibration accuracy.

[0050] Other features and advantages of the present invention will be described in the following specification, and part of them will become obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures pointed out in the specification, claims, and drawings. Description of the Drawings

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will describe the specific embodiments of the present invention in detail with reference to the drawings. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0052] Figure 1 It is a flowchart of a track vehicle wheel set full circumference dimension detection method;

[0053] Figure 2 This is a schematic structural diagram of a three-dimensional tread module provided by an embodiment of the present invention;

[0054] Figure 3 This is a schematic structural diagram of a wheel set dimension measurement module provided by an embodiment of the present invention;

[0055] Figure 4 This is a schematic arrangement structure diagram of a three-dimensional tread module and a wheel set dimension measurement module;

[0056] Figure 5 This is a schematic diagram of a three-dimensional tread curve provided by an embodiment of the present invention;

[0057] Figure 6 This is a schematic diagram of a wheel set capture position provided by an embodiment of the present invention;

[0058] Among them, the reference numerals are:

[0059] 1. Rail;

[0060] 2. Three-dimensional tread module; 21. Wheel set camera; 22. First laser component;

[0061] 3. Wheel set dimension measurement module; 31. Wheel diameter camera; 32. Second laser component;

[0062] 4. Wheel positioning sensor. Detailed implementation manners

[0063] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be described and explained below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application. Based on the embodiments provided by the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0064] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary implementation manners according to the present application. As used herein, unless otherwise clearly specified in the context, the singular form is also intended to include the plural form. In addition, it should also be understood that the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or devices.

[0065] Without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0066] The technical solution of the present invention will be described in detail below in conjunction with specific embodiments and the accompanying drawings of the specification.

[0067] As shown in the Figure 1 - attached Figure 6 figures, in a schematic embodiment of a method for detecting the full circumferential dimensions of a wheel set of a rail vehicle according to the present invention, the detection method includes the following steps.

[0068] S1: Install multiple groups of three-dimensional tread modules and wheel set dimension measurement modules on both sides of the track.

[0069] Among them, the three-dimensional tread module is mainly used for detecting tread abrasion and obtaining the information of the contour curve. A total of ten are set, and ten are arranged opposite to each other on both sides of the track, with 5 on one side, and each three-dimensional tread module photographs 1 / 5 of the wheel tread. As shown in the Figure 2 figures, the three-dimensional tread module includes a wheel set camera and a first laser component. The wheel set camera and the first laser component in each module are connected by a mechanical connector. In this embodiment, the first laser component is specifically a laser grating. The interval of the three-dimensional tread module is set to 800 mm, and a total of five groups of three-dimensional cameras completely cover a wheel set of 4000 mm. In the actual installation process, 3D line scan cameras and lenses with appropriate depths of field can be selected based on the camera installation interval distance and the different wheel set dimensions.

[0070] The wheel set dimension measurement modules are arranged on the outer sides of the rails on both sides of the track, and a total of six are set, with 3 on one side. As shown in the Figure 3 figures, the wheel set dimension measurement module includes a wheel diameter camera and a second laser component.

[0071] Both the wheel set camera and the wheel diameter camera are high-speed three-dimensional cameras, which can continuously image the wheel set when the wheel set passes by to collect the three-dimensional image of the surface of the locomotive wheel set. By the relay shooting of the three-dimensional cameras in multiple modules and collecting the wheel set data in sub-regions, the three-dimensional data collection of the full circumference of the wheel set can be realized, ensuring the integrity of the wheel set data, so as to realize the measurement of the full circumferential dimensions of the wheel set and the detection of defects. The depths of field and working distances corresponding to the wheel set camera and the wheel diameter camera are different and can be adjusted according to the actual working conditions. The present invention does not make further restrictions on this.

[0072] The arrangement of the three-dimensional tread module and the wheel set dimension measurement module on both sides of the track is as shown in the Figure 4 figures.

[0073] The three-dimensional tread module and the wheel set dimension measurement module are centrally installed in terms of position. The installation space in the track direction is about 6m, but they do not interfere with each other to avoid mutual interference between lasers and reduce the installation space at the same time. The technical principle is the hyperspectral high-speed non-contact laser triangulation three-dimensional imaging technology, which uses a hyperspectral grating light source detection to achieve three-dimensional imaging of the high-density cross-section of the locomotive wheel set. The dynamic detection device for the wheel set of the locomotive running gear is usually installed on both sides of the running line track within the normal speed limit of 10 kilometers per hour of the vehicle. Multiple groups of 808nm lasers are used in combination with a three-dimensional camera to dynamically collect the tread images of the wheel set by the laser triangulation method. The three-dimensional camera can generate two-dimensional image data and 3D point cloud data of the tread profile.

[0074] S2: When the wheel passes through the three-dimensional tread module, use the wheel set camera and the first laser component to collect multiple groups of original data.

[0075] In some embodiments of the present invention, in step S2, before using the wheel set camera and the first laser component to collect the original data, the following steps are further included.

[0076] Calibrate the internal parameters of the wheel set camera to obtain the internal parameters and distortion parameters of the wheel set camera.

[0077] Perform three-dimensional calibration on the wheel set camera and the first laser component to obtain the three-dimensional calibration parameters of the three-dimensional tread module. The three-dimensional calibration parameters include the light plane parameters of the first laser component.

[0078] Perform joint calibration on multiple groups of three-dimensional tread modules on the same side of the track to obtain the external parameters between the three-dimensional tread modules on the same side.

[0079] S3: Obtain the three-dimensional point cloud data corresponding to each group of original data based on the light plane parameters of the first laser component, and perform curve registration on the point cloud curves of the three-dimensional point cloud data.

[0080] In some embodiments of the present invention, the method for performing curve registration further includes the following steps.

[0081] Fuse the three-dimensional point cloud data of the three-dimensional tread modules on both sides respectively to obtain the complete three-dimensional point cloud data of each side of the wheel.

[0082] Use a two-dimensional curve registration algorithm and a standard wheel set curve to perform curve registration on the point cloud curves of the complete point cloud data.

[0083] In some embodiments of the present invention, the method for performing curve registration specifically includes the following steps.

[0084] For the point cloud curve of each wheel, use a two-dimensional curve registration algorithm and a standard wheel set curve to determine the corresponding points between the current curve and the standard curve based on the nearest point principle. The corresponding relationship expression between the corresponding points of the current curve and the standard curve is:

[0085]

[0086] Among them, θ, b1, and b2 are all undetermined coefficients;

[0087] Solve for the undetermined coefficients. The principle for solving the undetermined coefficients is that the distance from the transformed target point to the reference point is the shortest.

[0088] Establish an objective function based on the solution principle:

[0089]

[0090] Construct a new set of target points based on the obtained undetermined coefficients and the objective function;

[0091] Repeat the steps of determining the corresponding points between the current curve and the standard curve, solving for the undetermined coefficients, and constructing a new set of target points until a predetermined termination condition is met.

[0092] In some embodiments of the present invention, the predetermined termination condition is that the error is less than a predetermined threshold or the number of iterations is greater than a predetermined number of iterations.

[0093] S4: Calculate the dimensions of the wheel set based on the registered point cloud curve.

[0094] In some embodiments of the present invention, in step S4, after obtaining the dimensions of the wheel set, the following steps are further included.

[0095] In some embodiments of the present invention, the following steps are further included:

[0096] Perform a difference detection on the point cloud data of the calculated wheel set dimensions and the point cloud data of the standard wheel diameter to obtain a difference image;

[0097] Perform deep learning defect segmentation on the difference image to obtain the defect area, so as to realize wheel set defect detection.

[0098] S5: When the wheel passes through the wheel set dimension measurement module, use the wheel diameter camera and the laser line emitted by the second laser component to collect multiple groups of original laser line data of the tread surface and the side image of the wheel.

[0099] In some embodiments of the present invention, in step S5, before using the wheel diameter camera and the second laser component to collect data, the following steps are further included.

[0100] S6: Obtain four groups of three-dimensional tread curves and the wheel set capture positions based on the original laser line data of the tread surface and the side image of the wheel. Among them, the three-dimensional tread curve is as shown in the appendix Figure 5 shown, and the wheel set capture position is as shown in the appendix Figure 6 shown.

[0101] In some embodiments of the present invention, the following steps are further included:

[0102] Perform internal parameter calibration on the wheel diameter camera to obtain the internal parameters and distortion parameters of the wheel diameter camera;

[0103] Perform three-dimensional calibration on the wheel diameter camera and the second laser assembly to obtain the three-dimensional calibration parameters of the wheel set size measurement module; the three-dimensional calibration parameters include the optical plane parameters of the second laser assembly;

[0104] Mark the position points of the line laser in the wheel diameter camera coordinate system and the corresponding world coordinate positions of each position point, and use a deep learning algorithm to fit the external parameters between the wheel set size measurement modules on both sides of the track.

[0105] In some embodiments of the present invention, the method of using a deep learning algorithm to fit the external parameters between the wheel set size measurement modules on both sides of the track specifically includes the following steps.

[0106] Select a predetermined propagation route, which is the complete calculation path when the neural network of the deep learning algorithm processes a single input data, and perform forward propagation, that is, the data flows from the input layer to the output layer to calculate the predicted value. The forward propagation calculation process is as follows:

[0107]

[0108] Among them, x is the coordinate of the laser point in the camera coordinate system, y is the corresponding world coordinate, w ij is the weight, v i is the intermediate variable.

[0109] Let

[0110] Then The expression of is deformed into:

[0111]

[0112] Obtain the final forward propagation result:

[0113]

[0114] Obtain the final error expressions for multiple lines as:

[0115]

[0116] Let

[0117] Determine the end of forward propagation:

[0118]

[0119] Connect the input and output of the deep learning algorithm through the hidden layer, allocate the error to each layer node, and use the gradient descent algorithm to make the error of each node decrease along the gradient direction. After multiple iterations, when the error is less than the predetermined threshold, it is determined that the fitting training stops. The expression is as follows:

[0120]

[0121] Among them, taking as an example, substituting gives:

[0122]

[0123] Similarly, it can be obtained that:

[0124]

[0125] Use the gradient descent algorithm to update the weights,

[0126]

[0127] where η is the learning rate, and the matrix form of the final result is:

[0128]

[0129] Use the above matrix to jointly calibrate the six modules on both sides to obtain the external parameters between the modules.

[0130] S7: Based on the curve coordinates and the wheel pair capture position, obtain three coordinate points on the rolling circle of the wheel pair, mark the corresponding relationship between the curve coordinates and the coordinate points on the rolling circle, and use the neural network fitting algorithm of the same calibrated deep learning algorithm to perform curve fitting to obtain the wheel diameter.

[0131] In the above-mentioned schematic embodiments, the measurement of the wheel pair size and the fault detection of the wheel pair surface can be automatically completed when the train passes by. The measurement speed is fast, the real-time performance is good, and the degree of automation is high. At the same time, the measurement module has strong environmental adaptability, can effectively reduce the influence of vehicle vibration and ground settlement on the measurement accuracy, and can detect the faults on the wheel pair surface on the basis of realizing the measurement of the wheel pair size data.

[0132] In some embodiments of the present invention, wheel position sensors are arranged on the track. The wheel position sensors are used to detect whether the wheels are in place. When it detects that the wheels have arrived, the wheel pair cameras and wheel diameter cameras on both sides are activated to capture images, ensuring that the modules on both sides can exactly capture the complete wheels, and the images on both sides are on a straight line, which is convenient for subsequent size calculation.

[0133] In the above-mentioned schematic embodiments, the order between steps S2-S4 and steps S5-S7 can be interchanged.

[0134] In some embodiments of the present invention, multiple groups of three-dimensional tread modules and wheel set size measurement modules on both sides of the track are symmetrically installed with respect to the track.

[0135] In some embodiments of the present invention, the wheel set camera and the first laser component of the same three-dimensional tread module are connected by a mechanical connection member, and between each three-dimensional tread module is also connected by a mechanical connection member;

[0136] The wheel diameter camera and the second laser component of the same wheel set size measurement module are connected by a mechanical connection member, and between each wheel set size measurement module is also connected by a mechanical connection member.

[0137] Some embodiments of the present invention further provide a wheel set full circumference size detection system for a rail vehicle, including:

[0138] At least one processor;

[0139] At least one memory for storing at least one program;

[0140] When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned wheel set full circumference size detection method for a rail vehicle.

[0141] Finally, it should be noted that: the embodiments in this specification are described in a progressive manner, and the key points of each embodiment are the differences from other embodiments. For the same and similar parts between each embodiment, reference can be made to each other.

[0142] The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that: still can modify the specific implementation manners of the present invention or perform equivalent substitution on some technical features; without departing from the spirit of the technical solutions of the present invention, they should all be covered within the scope of the technical solutions claimed by the present invention.

Claims

1. A method for detecting the full circumferential size of a rail vehicle wheelset, characterized in that: The following steps are involved: A plurality of sets of three-dimensional tread modules and wheelset size measurement modules are installed on both sides of the track; the three-dimensional tread module includes a wheelset camera and a first laser assembly; the wheelset size measurement module includes a wheel diameter camera and a second laser assembly; When the wheel passes through the three-dimensional tread module, using the wheel set camera and the first laser assembly to collect multiple sets of raw data; Obtaining three-dimensional point cloud data corresponding to each group of original data based on the light plane parameters of the first laser assembly, and performing curve registration on point cloud curves of the three-dimensional point cloud data; The size of the wheelset is calculated based on the registered point cloud curve; When the wheel passes through the wheelset size measurement module, the wheel diameter camera is used in combination with the laser line emitted by the second laser assembly to collect multiple sets of tread original laser line data and wheel side images; Obtaining a tread three-dimensional curve and a wheelset capture position based on the tread original laser line data and the wheel side image; Based on the curve coordinates and the captured position of the wheelset, multiple coordinate points on the wheelset rolling circle are obtained, the corresponding relationship between the curve coordinates and the coordinate points on the rolling circle are marked, and a neural network fitting algorithm is used to perform curve fitting to obtain the wheel diameter.

2. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1, characterized in that: Further comprising the steps of: Performing internal parameter calibration on the wheelset camera to obtain internal parameters and distortion parameters of the wheelset camera; Performing three-dimensional calibration on the wheelset camera and the first laser assembly to obtain three-dimensional calibration parameters of the three-dimensional tread module, wherein the three-dimensional calibration parameters include light plane parameters of the first laser assembly; Multiple groups of three-dimensional tread modules on the same side of the track are jointly calibrated to obtain the external parameters between the three-dimensional tread modules on the same side.

3. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1, characterized in that: Further comprising the steps of: Performing internal parameter calibration on the wheel diameter camera to obtain internal parameters and distortion parameters of the wheel diameter camera; Performing three-dimensional calibration on the wheel diameter camera and the second laser assembly to obtain three-dimensional calibration parameters of the wheelset size measurement module; the three-dimensional calibration parameters include light plane parameters of the second laser assembly; The position points of the marking line laser in the wheel diameter camera coordinate system and the world coordinate positions corresponding to each position point are marked, and the external parameters between the wheel pair size measurement modules on both sides of the track are fitted using a deep learning algorithm.

4. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1, characterized in that: Further comprising the steps of: The 3D point cloud data of each group of 3D tread modules on both sides are fused separately to obtain the complete 3D point cloud data of the wheels on both sides; The point cloud curves of the complete point cloud data are aligned using a two-dimensional curve alignment algorithm and a standard wheelset curve.

5. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1 or 4, characterized in that: The method for performing curve registration on the point cloud curve of the three-dimensional point cloud data specifically comprises the following steps: For each point cloud curve, the corresponding points of the current curve and the standard curve are determined based on the nearest point principle using the two-dimensional curve registration algorithm and the standard wheel set curve. The corresponding relationship between the corresponding points of the current curve and the standard curve is expressed as: Among them, θ, b1, and b2 are all unknown coefficients; Solving the undetermined coefficients, the principle for solving the undetermined coefficients is that the distance from the transformed target point to the reference point is the shortest; The objective function is established based on the solution principle: Constructing a new target point set based on the obtained undetermined coefficients and the target function; The steps of determining corresponding points between the current curve and the standard curve, solving undetermined coefficients, and constructing a new target point set are repeated until a predetermined termination condition is met.

6. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1, characterized in that: Further comprising the steps of: Perform difference detection on the point cloud data of the calculated wheelset size and the point cloud data of the standard wheel diameter to obtain a difference image; Deep learning defect segmentation is performed on the difference image to obtain the defect area to achieve wheelset defect detection.

7. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1, characterized in that: A wheel positioning sensor is arranged on the track. When the wheel positioning sensor is triggered to sense the arrival of a wheel, the wheelset camera and the wheel diameter camera start to capture.

8. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1, characterized in that: A plurality of groups of three-dimensional tread modules and wheelset size measuring modules on both sides of the track are symmetrically installed relative to the track.

9. The method for detecting the full circumferential dimension of a rail vehicle wheelset according to claim 1 or 8, characterized in that: The wheelset camera and the first laser assembly of the same three-dimensional tread module and the three-dimensional tread modules are connected via mechanical connectors; The wheel diameter camera and the second laser assembly of the same wheelset dimension measurement module and each wheelset dimension measurement module are connected via a mechanical connector.

10. A rail vehicle wheel set full circumferential dimension detection system, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method for detecting the full circumferential size of a rail vehicle wheelset as described in any one of claims 1 to 9.

Images