Tire wear monitoring system, method, and medium based on digital twin comparison

Abstract

The application relates to a tire wear monitoring system, method and medium based on digital twin comparison, and belongs to the technical field of tire state monitoring; the application comprises the following units: a data acquisition unit; a model optimization unit, which is based on deep learning image recognition and segmentation technology, identifies distortion areas in a scanning model, performs data elimination and optimization, and generates a three-dimensional digital twin model capable of representing the real state of the tire; a model alignment unit, which realizes high-precision alignment of a measured model and a pre-stored initial model of a non-worn tire; a wear calculation unit, which calculates the wear depth of different areas of the tire tread by using a 3D deviation analysis principle; and a state management unit, which intelligently identifies the current wear state and form of the tire and displays the tire wear state in a visual manner. The application constructs a tire digital twin model through three-dimensional scanning, performs high-precision comparison with a non-worn initial model, realizes full-area and high-precision tire wear monitoring, and supports full-life-cycle management.

Description

Technical Field

[0001] This invention relates to a tire wear monitoring system, method, and medium based on digital twin comparison, belonging to the field of tire condition monitoring technology. Background Technology

[0002] Engineering radial tires are used in harsh environments. As the tread area that is in direct contact with the road surface, its wear condition directly determines its performance and driving safety. Therefore, regular tread wear monitoring is a key part of the daily maintenance of engineering radial tires. Currently, tire wear monitoring methods typically rely on manual, mechanical multi-point depth measurement and measurement via laser ranging or image recognition technology when the vehicle is driven to a designated monitoring station. Traditional manual measurement can only sample a limited number of tread grooves, failing to capture uneven wear across the entire tread surface. This can easily lead to misjudgments of the overall tire wear degree and pattern. Furthermore, current monitoring methods often use the currently worn tread as a zero-point reference; when uneven wear occurs, the measurement reference plane can drift, resulting in significant errors in the measurement results. Similarly, laser ranging or image recognition technology, when the vehicle is driven to a designated location, struggles to measure the overall tire wear degree. Moreover, 3D model size measurements based on algorithmic optimization and correction also have inherent measurement errors. Additionally, the measured model data is often stored in isolated data within the system, resulting in low digitization and limited support for achieving full tire lifecycle management and data analysis and mining.

[0003] Digital twin technology has shown significant advantages in the field of health monitoring of heavy equipment. For example, the existing patent application publication number CN118862682A discloses a method for predicting the life of the roller surface of a high-pressure roller mill based on digital twins. By constructing a digital twin of the high-pressure roller mill and combining EDEM wear evolution analysis with deep learning algorithms, it has achieved online prediction of the remaining life of the roller surface. However, it has not achieved the construction of a high-precision digital twin of the tire across the entire field based on 3D scanning, and cannot directly quantify the wear amount through geometric comparison. The accuracy of the digital twin model is insufficient.

[0004] In summary, existing tire wear monitoring technologies suffer from common problems such as lack of full-domain monitoring, large reference drift errors, insufficient digital management, and low accuracy in wear quantification, making it difficult to meet the high-precision, full-lifecycle management requirements of engineering radial tires. Therefore, constructing a high-precision tire digital twin using 3D scanning technology, achieving accurate quantification of wear through full-domain geometric comparison, and establishing a full-lifecycle data management system have become core technical issues that urgently need to be addressed in the field of intelligent vehicle operation and maintenance. Summary of the Invention

[0005] The purpose of this invention is to propose a tire wear monitoring system, method, and medium based on digital twin comparison. By constructing a tire digital twin model through three-dimensional scanning, a high-precision comparison is made with the initial unworn model to achieve full-area, high-precision tire wear monitoring and support full life-cycle management, thus overcoming the shortcomings of existing technologies.

[0006] The tire wear monitoring system based on digital twin comparison described in this invention includes:

[0007] The data acquisition unit performs 3D scanning on the tire to be monitored, acquires the original three-dimensional point cloud data of the tire surface, and reconstructs it into a measured three-dimensional model.

[0008] The model optimization unit identifies distorted areas in the scanned model, removes and optimizes the data, and generates a three-dimensional digital twin model that can represent the true state of the tire.

[0009] The model alignment unit uses the geometric features of the non-wear area outside the tire tread as a spatial constraint reference to achieve high-precision alignment between the measured model and the pre-stored initial model of the unworn tire.

[0010] The wear calculation unit uses the 3D deviation analysis principle to calculate the wear depth of different areas of the tread.

[0011] The status management unit intelligently identifies the current wear status and pattern of the tire and displays the tire wear status in a visual way, realizing intelligent management of the entire tire life cycle.

[0012] Preferably, the model optimization unit identifies distorted regions based on deep learning image recognition and segmentation technology. Specifically, it includes: performing feature recognition and instance segmentation on multi-angle photos using the YOLO_v11-seg algorithm, determining the specific location of the distorted region on the three-dimensional model through 2D-3D spatial coordinate mapping, performing data removal, and smoothing and repairing the removed region by referring to the surrounding normal pattern features.

[0013] Preferably, the 2D-3D spatial coordinate mapping relationship is based on the perspective projection model. By establishing the conversion relationship between image pixel coordinates and 3D point cloud spatial coordinates, the position of the distorted area in the three-dimensional model can be accurately located.

[0014] Preferably, in the model alignment unit, the geometric features of the non-wear area include at least one of the following: tire sidewall symmetrical feature circles, brand logo, specification characters, mold positioning holes, and wheel hub edge contours; the alignment process includes: calculating the model center based on the average coordinates of the center of at least three sets of tire sidewall symmetrical feature circles, aligning the center and axis of the measured model with the center of the initial model, and then using an incremental rotation strategy to step the measured model circumferentially along the central axis until it is aligned with the circumferential features of the initial model.

[0015] Preferably, in the wear calculation unit, the wear depth is obtained by calculating the signed normal distance from the sampling point of the measurement model to the corresponding projection point of the reference model, and positive deviation data of the measurement point located outside the reference surface are regarded as abnormal data and removed.

[0016] Preferably, the wear calculation unit calculates the wear volume by Boolean operation difference set. In the tread area, the geometric entity difference set of the comparison model is obtained by Boolean operation, and then the geometric volume of the difference set area is calculated, which is the wear volume of the tire in three-dimensional geometry.

[0017] Preferably, in the state management unit, the current wear state and form of the tire are intelligently identified based on multi-dimensional wear indicators. The multi-dimensional wear indicators include at least two of the following: maximum wear depth, average value, standard deviation, wear volume, and wear rate. The visualization method includes a panoramic differential cloud map, which intuitively displays the tread wear distribution through color changes.

[0018] Preferably, the data acquisition unit uses a portable handheld scanner to perform multi-angle, multi-region 3D scanning and simultaneously acquire multi-angle photos of the tire; the reconstruction process includes triangulation of the original three-dimensional point cloud data.

[0019] The tire wear monitoring method based on digital twin comparison described in this invention, applied to the tire wear monitoring system based on digital twin comparison, includes the following steps:

[0020] S1: Perform a 3D scan on the tire to be monitored to obtain the original three-dimensional point cloud data and reconstruct it into a measured three-dimensional model;

[0021] S2: Identify distorted areas in the scanned model, remove and optimize the data, and generate a three-dimensional digital twin model that represents the true state of the tire;

[0022] S3: Using the geometric features of the tire's non-wear area as a constraint benchmark, the measured model is spatially aligned with the pre-stored initial model of the unworn tire with high precision;

[0023] S4: The wear depth and volume of different areas of the tire tread are calculated using 3D deviation analysis and Boolean operation principles;

[0024] S5: Based on multi-dimensional wear indicators, identify the tire wear status and form, and display the wear status in a visual way to realize tire life cycle data management.

[0025] The computer-readable storage medium of the present invention stores a computer program thereon, which, when executed by a processor, implements the steps of the tire wear monitoring method based on digital twin comparison.

[0026] Compared with existing technologies, the tire wear monitoring system, method, and medium based on digital twin comparison of the present invention exhibit the following beneficial effects in terms of technical performance and practical application:

[0027] 1. High-precision full-field wear quantification:

[0028] Employing 3D scanning and digital twin comparison technology, the wear depth measurement accuracy reaches ±0.01mm, and the volume measurement error is less than 5%, achieving precise quantification of wear across the entire tread area and increasing the abnormal wear identification rate to over 95%. Using an initial model of an unworn tire as a benchmark avoids benchmark drift after wear, reducing measurement errors compared to traditional methods and completely resolving pain points such as "sampling misjudgment" in manual measurement and "lack of full-area coverage" in laser ranging.

[0029] 2. High-efficiency batch testing capability:

[0030] The inspection time for a single tire is shortened, and the inspection efficiency is improved compared to traditional manual measurement, meeting the rapid inspection needs of batch tires in mines, logistics fleets, and other industries. The portable handheld scanner supports flexible on-site inspection, eliminating the need to remove tires to a designated workstation, further enhancing inspection convenience and adaptability to various scenarios.

[0031] 3. Strong anti-interference and data repair capabilities:

[0032] The Yolo_v11-seg algorithm accurately identifies foreign objects such as stones and mud on the tire surface. Combined with 2D-3D coordinate mapping, it achieves precise positioning and data repair of distorted areas, improving data accuracy. It effectively avoids interference from foreign object adhesion on measurement results, solves problems such as "reflection interference" in laser ranging and "lighting effects" in image recognition, and adapts to the complex and harsh operating environment of engineering tires.

[0033] 4. Intelligent management throughout the entire lifecycle:

[0034] By constructing a digital twin model archive for tires, long-term storage and traceability of wear data can be achieved, supporting wear trend analysis and remaining life prediction, thereby extending the average tire lifespan. A panoramic differential cloud map visually displays the tread wear distribution, and combined with a big data model, intelligently identifies wear patterns, providing a scientific basis for tire maintenance and replacement decisions. Attached Figure Description

[0035] Figure 1 This is a flowchart of the tire wear monitoring method based on digital twin comparison according to the present invention;

[0036] Figure 2 The following is a schematic diagram of tire twin models at different wear stages in an embodiment of the present invention: In the figure, (a) represents the initial twin model; (b) represents the initial wear twin model; (c) represents the middle wear twin model; and (d) represents the late wear twin model.

[0037] Figure 3 This is an image distortion feature recognition and segmentation diagram based on the Yolo_v11-seg algorithm in an embodiment of the present invention; in the diagram, (a) represents distortion feature recognition; (b) represents distortion feature segmentation;

[0038] Figure 4 This is a diagram showing the precise alignment of coordinates in an embodiment of the present invention; in the diagram, (a) represents model import; (b) represents precise axial alignment of the model based on the dual tire side feature circles; and (c) represents precise axial alignment of the model based on the dual tire side identifiers.

[0039] Figure 5 This is a visualization of the three-dimensional color deviation cloud map in an embodiment of the present invention; in the figure, (a) represents the visualization of the overall three-dimensional wear cloud map; (b) represents the visualization of the wear cloud map of a local pattern block.

[0040] Figure 6 This is a visualization of the two-dimensional wear depth cloud map of the cross section in an embodiment of the present invention;

[0041] Figure 7 The figures are visualizations of three-dimensional wear cloud maps of tires at different wear stages in embodiments of the present invention; in the figures, (a) represents the initial baseline twin model; (b) represents the cloud map visualization of the early wear stage; (c) represents the cloud map visualization of the middle wear stage; and (d) represents the cloud map visualization of the late wear stage. Detailed Implementation

[0042] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0043] Example 1:

[0044] This embodiment discloses a tire wear monitoring system based on digital twin comparison, including:

[0045] The data acquisition unit performs 3D scanning on the tire to be monitored, acquires the original three-dimensional point cloud data of the tire surface, and reconstructs it into a measured three-dimensional model.

[0046] The model optimization unit identifies distorted areas in the scanned model, removes and optimizes the data, and generates a three-dimensional digital twin model that can represent the true state of the tire.

[0047] The model alignment unit uses the geometric features of the non-wear area outside the tire tread as a spatial constraint reference to achieve high-precision alignment between the measured model and the pre-stored initial model of the unworn tire.

[0048] The wear calculation unit uses 3D deviation analysis and Boolean operation principles to calculate the wear depth and volume of different areas of the tread.

[0049] The status management unit intelligently identifies the current wear status and form of the tire based on multi-dimensional wear indicators, and displays the tire wear status in a visual way, realizing intelligent management of the entire tire life cycle.

[0050] In this embodiment, the data acquisition module uses a portable handheld scanner paired with a multi-angle photography device. During the scanning process, multi-angle photos of the tire are simultaneously acquired and uploaded to the terminal system, providing image data support for model optimization. A FARO FocusS 70 3D laser scanner paired with a Canon EOS R5 camera is used to perform 3D scanning of the tire at four angles: 0°, 90°, 180°, and 270°. Each angle is scanned for 10 seconds, simultaneously acquiring multi-angle photos of the tire. The original 3D point cloud data acquired by the scan has a resolution of 0.1mm. It is then processed into a triangular mesh using MeshLab software and reconstructed into a measured 3D model.

[0051] In this embodiment, image recognition and segmentation technology based on deep learning is used to identify distorted areas (such as stones, mud, and tread damage) in the scanned model, remove and optimize the data, and generate a three-dimensional digital twin model that can represent the true state of the tire. The specific implementation includes:

[0052] The model optimization unit uses the YOLOv11-seg algorithm to perform image feature recognition and instance segmentation. The final segmentation mask is generated by a linear combination of the prototype set and the mask coefficient vector, followed by the Sigmoid activation function.

[0053] The system uses the Yolo_v11-seg algorithm to identify and segment distorted regions in multi-view photo models. Then, through a 2D-3D spatial coordinate mapping method, it establishes a conversion relationship between image pixel coordinates and 3D point cloud spatial coordinates, determining the specific coordinate positions of the distorted regions on the 3D model before performing data removal. For the removed "holes," the system uses the geometric curvature features of the surrounding normal tread pattern for smoothing repair, ensuring the integrity of the digital twin model. In this embodiment, the generated tire twin models at different wear stages are as follows: Figure 2 As shown, Figure 2 (a) is the initial twin model; Figure 2 (b) shows the twin model in the early stage of wear; Figure 2 (c) represents the mid-wear twin model; Figure 2 (d) represents the twin model in the later stage of wear.

[0054] The specific algorithmic theories involved in the above process are as follows:

[0055] (1) YOLO_v11-seg deep learning algorithm and image recognition:

[0056] The core of the Yolo_v11-seg algorithm lies in its decoupling and parallel processing of detection and segmentation tasks. Based on a convolutional neural network (CNN), it automatically learns the features of distorted regions in the image (abnormal patterns, damaged areas, and features of stones, mud, and impurities) through pre-training, which helps the model achieve localization and ensures that the algorithm can simultaneously identify patterned damage, punctured stones, and large mud-covered areas. On this basis, a set of prototype masks is generated through the segmentation head branch, and coefficients are generated for each detected target. Pixel segmentation is achieved through linear combination. The final segmentation mask M is obtained by linear combination of the prototype set S and the mask coefficient vector C and then processed by an activation function.

[0057] ;

[0058] in, The basic geometric prototype representing the surface of a tire. It is a specific weighting coefficient for areas containing foreign objects such as rocks or mud. The sigmoid activation function is used to constrain the output to... The interval is used to determine the boundary of the foreign object. To ensure the accuracy of the model's recognition, its total loss is... It consists of classification loss, localization loss, and segmentation loss:

[0059] ;

[0060] in, (Classification loss) can solve the problem of imbalance between positive and negative samples, ensuring that the system will not mistakenly identify the tread pattern as a stone and cause over-rejection; (Location loss) can quickly lock the interference area in the complex 3D scanning field of view, ensuring that the model can quickly locate the rectangular area where the foreign object is located, narrowing the search range of the segmentation algorithm, and significantly improving the real-time performance and spatial positioning accuracy of 2D image processing; (Segmentation loss) uses binary cross-entropy (BCE) to measure the consistency between the predicted mask and the actual foreign object region. Image distortion feature recognition and segmentation based on the Yolo_v11-seg algorithm is as follows: Figure 3 (a) and Figure 3 As shown in (b). Figure 3 (a) is the original photograph of the tire (including stones and damage). Figure 3 (b) shows the segmented image, separating the stone from the damaged area.

[0061] The model optimization unit determines the location of the distorted region in the 3D model through 2D-3D spatial coordinate mapping. The mapping relationship is based on the perspective projection model.

[0062] Principle of 2D-3D spatial coordinate mapping method:

[0063] To achieve accurate alignment between the 2D foreign object region identified by YOLOv11-seg and the 3D point cloud data, a 2D-to-3D coordinate transformation was performed based on the camera's linear model, and the 3D point cloud model was located. The coordinate system transformations involved in the mapping process are as follows:

[0064] Pixel coordinate system: Points on an image, measured in pixels;

[0065] Image physical coordinate system Physical coordinates of the sensor plane, in millimeters;

[0066] Camera coordinate system The origin is the optical center of the camera.

[0067] World coordinate system (3D point cloud coordinate system) The global coordinate system of the tire digital twin model; coordinate transformation and positioning based on the perspective projection model, where the perspective projection transformation model is as follows:

[0068] ;

[0069] in: (Scale factor) represents the depth information from the 3D point to the camera's optical center. The (intrinsic parameter matrix) represents the camera's performance parameters. The extrinsic parameter matrix describes the camera's pose during the scanning process and is a rotation matrix. Let be the translation vector, where the intrinsic parameter matrix is ​​. The definition is as follows:

[0070] ;

[0071] in: and Focal length The coordinates of the main point.

[0072] Model alignment unit:

[0073] By utilizing multiple geometric features of the non-wear area on the tire sidewall as spatial constraint benchmarks, high-precision alignment between the measured model and the pre-stored initial model of the unworn tire is achieved. Specifically, this includes:

[0074] When the optimized tire twin model enters the comparison stage, it is not aligned geometrically with the initial twin model being compared. To improve the alignment accuracy, this embodiment uses multiple geometric features of non-wear areas as spatial constraint benchmarks. Non-wear areas such as the tire sidewall do not directly contact the ground during actual use, and most geometric features are almost undeformed compared to the initial tire state. Using these as alignment geometric features enables accurate positioning of the comparison model. The method is explained below with reference to the following embodiments:

[0075] Engineering radial finished tires typically have symmetrical feature circles on the sidewalls, and several pairs of these symmetrical feature circles usually exist at different heights along the radial direction, such as... Figure 4 (a) indicates model import; Figure 4 (b) indicates that the model's axial alignment was achieved based on the feature circles on the twin tire sides; Figure 4 (c) indicates that the model axial alignment was completed based on the dual tire side identifier; Figure 4 (b) The red dashed line indicates a certain feature circle. A similar feature circle exists on the symmetrical plane of the tire sidewall. In this embodiment, geometric circle 1 and symmetrical geometric circle 2 are constructed using the feature circle of the twin model. A line segment 1 is then drawn through the centers of the two geometric circles; the center of line segment 1 is the center 1 of the current twin tire model. To prevent significant errors in the tire center found based on a single symmetrical feature circle at different times, the method described in this embodiment typically uses the coordinate averaging of at least three sets of symmetrical circle centers as the final model center coordinates. The specific calculation method is shown below. This indicates the number of sets of symmetrical feature circles selected:

[0076] ;

[0077] After aligning the center and axis of the initial twin model of the tire and the twin model of the current state to be monitored, the two models have locked five degrees of freedom in space on the system. For the final circumferential phase difference, the system establishes a one-dimensional search space based on the geometric topology of the marker region. Using an incremental rotation strategy, the measured model is circumferentially stepped along the central axis until alignment is complete. Tire information geometric features such as brand logos and specification characters on the tire sidewall area typically appear at specific positions in the circumferential direction and are asymmetrical, serving as markers for the aforementioned incremental rotation alignment. If tire information geometric feature matching fails during the alignment process, it is usually because the tire information geometric features of the two models do not appear in the same direction. In this case, the model is rotated 180° radially once, and then the tire information geometric feature alignment is performed again until the circumferential alignment is complete. The objective function set during the circumferential alignment process is as follows:

[0078] ;

[0079] exist Traversal During the process, find the measured feature points With reference feature points The one with the smallest sum of squared Euclidean distances between them .in For the axis Rotation The angle rotation matrix has the following specific form:

[0080] .

[0081] Wear calculation unit:

[0082] The wear depth and volume of different areas of the tire tread are calculated using 3D deviation analysis and Boolean operation principles.

[0083] After the twin model is precisely aligned, there will be a thickness deviation (normal) between the twin model reflecting the tire wear state and the initial twin model in the tread area. The difference in normal thickness in different areas calculated by 3D deviation analysis is the tire wear value. The core logic of 3D deviation analysis lies in calculating the signed normal distance from the measuring point to the reference surface. This is a sampling point on the measurement model (a twin model reflecting tire wear conditions). Distance on the surface of the reference model (initial twin model of the tire) The nearest projection point. Definition The unit normal vector at point is Then the deviation value at that point The calculation formula is:

[0084] ;

[0085] when When the measurement point is located outside the reference surface, it is defined as a positive deviation (increment / bulge).

[0086] when When the measurement point is located inside the reference surface, it is defined as a negative deviation (reduction / concavity).

[0087] when When the measurement point is completely coincident with the reference surface, it indicates that the measurement point is completely coincident with the reference surface.

[0088] In tire wear monitoring, the deviation values ​​of most areas of the 3D model A negative deviation indicates the wear value of the tire due to friction with the road surface during actual use; in some cases, wear may occur in very few locations on the tire tread. or The error in this positional data is mostly due to permanent deformation of the tread rubber at this location, resulting in outward bulging. Such measurement data will be removed during actual wear statistics (this situation has an extremely low probability of occurring and accounts for no more than 1% of the overall data, so it will not affect the measurement accuracy). In this embodiment, points with d>0.1mm are regarded as abnormal data and removed, and only valid wear data are retained, with the wear depth measurement accuracy reaching ±0.01mm.

[0089] Boolean operations to calculate wear volume:

[0090] By performing Boolean difference operations (initial model - measured model) on the tread area using the OpenCASCADE library, the geometric volume of the difference region is calculated, which is the tire wear volume, with a volume measurement error of less than 5%.

[0091] Boolean operations are the mathematical theory for handling intersection, union, and difference relationships between three-dimensional geometric entities. Its core lies in the operation of two entities... and Logically combine the occupied spatial areas:

[0092] Intersection ( ):reserve and A shared space area;

[0093] Union ( ):merge and All the space occupied;

[0094] Difference set( ): From entity Subtract from the middle Overlapping spatial regions.

[0095] After completing the wear numerical calculation, the geometric difference set of the comparison model is obtained by Boolean operation in the tread area. Then, the geometric volume of the difference set area is calculated, which is the wear volume of the tire in three-dimensional geometry.

[0096] Status Management Unit:

[0097] By using a built-in big data model to intelligently identify the current wear state and pattern of tires based on multi-dimensional wear indicators, and visually displaying this information through a panoramic differential cloud map, intelligent management of the entire tire lifecycle is achieved.

[0098] Wear condition identification: Based on multi-dimensional indicators such as maximum, average, and standard deviation of wear depth, wear volume, and wear rate, the wear condition (normal / abnormal) and form (uniform wear / uneven wear / spot wear) are identified through a random forest machine learning model.

[0099] Visualization: A panoramic gradient cloud map is generated using Matplotlib. Colors from blue to red represent wear depths from -2mm to 8mm. Negative values ​​indicate that the surface geometry is higher than the original reference plane due to tire deformation, typically appearing in the sidewall area. Figure 5 (a) Overall wear contour map Figure 5 As shown in (b) (partial wear cloud map of patterned blocks). Figure 6 A visual representation of the wear depth cloud map of a tire on a certain cross section, where the red area on the tread represents the wear value; Figure 7 A 3D cloud map visualization of tire wear at different wear stages; Figure 7 (a) is the initial baseline twin model; Figure 7 (b) shows the cloud map visualization of the initial wear stage; Figure 7 (c) shows the mid-stage wear cloud map visualization; Figure 7 The middle (d) is a visualization of the wear phase cloud map. As can be seen from the cloud map, the longer the wear cycle, the darker the red wear area in the cloud map, indicating greater tread wear.

[0100] Full lifecycle management: Wear data and digital twin models are stored in a MySQL database to enable full lifecycle data traceability of tires from manufacturing, use to scrapping, supporting wear trend analysis and remaining life prediction, with a remaining life prediction error of less than 10%.

[0101] Example 2:

[0102] like Figure 1 As shown, the tire wear monitoring method based on digital twin comparison of the present invention, based on the tire wear monitoring system based on digital twin comparison described in Example 1, includes the following steps:

[0103] S1: Perform a 3D scan on the tire to be monitored to obtain the original three-dimensional point cloud data and reconstruct it into a measured three-dimensional model;

[0104] S2: Identify distorted areas in the scanned model, remove and optimize the data, and generate a three-dimensional digital twin model that represents the true state of the tire;

[0105] S3: Using the geometric features of the tire's non-wear area as a constraint benchmark, the measured model is spatially aligned with the pre-stored initial model of the unworn tire with high precision;

[0106] S4: The wear depth and volume of different areas of the tire tread are calculated using 3D deviation analysis and Boolean operation principles;

[0107] S5: Based on multi-dimensional wear indicators, identify the tire wear status and form, and display the wear status in a visual way to realize tire life cycle data management.

[0108] A FARO FocusS 70 3D laser scanner was used in conjunction with a Canon EOS R5 camera to perform 3D scanning of the radial tires of the project under monitoring from four angles: 0°, 90°, 180°, and 270°. The raw 3D point cloud data (resolution 0.1mm) and multi-angle photos were obtained and then reconstructed into a measured 3D model using MeshLab software.

[0109] Based on the trained Yolo_v11-seg model, distortion regions are identified and segmented from multi-angle photos (with an accuracy of 98%). The position of the distortion region on the 3D model is determined by 2D-3D spatial coordinate mapping, and data removal is performed. The B-spline interpolation algorithm is used to smooth and repair the model with reference to the surrounding normal tread features, generating a 3D digital twin model that represents the true state of the tire.

[0110] Three sets of symmetrical feature circles and brand logos were extracted from the tire sidewall. The center of the model was determined based on the average coordinates of the center of the three sets of symmetrical circles. The center and axis of the measured model were aligned with those of the initial model (axial error <0.1mm). The circumferential phase difference was aligned using an incremental rotation strategy (step size 1°) (circumferential error <0.5°) to achieve high-precision spatial alignment.

[0111] 3D deviation analysis was performed using CloudCompare software (sampling density 100 points / square centimeter) to calculate the wear depth in different areas of the tread and remove outlier data with d>0.1mm; Boolean difference operations were performed using the OpenCASCADE library to calculate the tread wear volume.

[0112] Based on wear depth and volume multidimensional indicators, the tire wear state and form are identified; a panoramic differential cloud map is generated to visualize the wear state; and wear data and twin models are stored in a MySQL database to realize tire life cycle management and remaining life prediction.

[0113] Example 3:

[0114] This embodiment discloses a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements all the steps of the tire wear monitoring method based on digital twin comparison described in Embodiment 2.

[0115] The computer-readable storage medium may be, but is not limited to, a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, an optical disk, or a Blu-ray disc; the computer program may run on operating systems such as Windows 10, Linux Ubuntu 22.04, Android 13, and iOS 17, and the processor may be a general-purpose or special-purpose processor such as an Intel Core i7, AMD Ryzen 7, or ARM Cortex-A76. When the program executes, it can automatically complete the entire tire wear monitoring process, outputting results such as wear depth, volume, and wear status, and generating a visual cloud map. The detection time for a single tire is less than 5 minutes.

[0116] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A tire wear monitoring system based on digital twin comparison, characterized in that, include: The data acquisition unit performs 3D scanning on the tire to be monitored, acquires the original three-dimensional point cloud data of the tire surface, and reconstructs it into a measured three-dimensional model. The model optimization unit identifies distorted areas in the scanned model, removes and optimizes the data, and generates a three-dimensional digital twin model that can represent the true state of the tire. The model alignment unit uses the geometric features of the non-wear area outside the tire tread as a spatial constraint reference to achieve high-precision alignment between the measured model and the pre-stored initial model of the unworn tire. The wear calculation unit uses the 3D deviation analysis principle to calculate the wear depth of different areas of the tread. The status management unit intelligently identifies the current wear status and form of the tire and displays the tire wear status in a visual way, realizing intelligent management of the entire tire life cycle; The model optimization unit identifies distorted regions based on deep learning image recognition and segmentation technology. Specifically, it performs feature recognition and instance segmentation on multi-angle photos using the YOLO_v11-seg algorithm, determines the specific location of the distorted region on the three-dimensional model through 2D-3D spatial coordinate mapping, performs data removal, and smooths and repairs the removed region by referring to the surrounding normal pattern features.

2. The tire wear monitoring system based on digital twin comparison according to claim 1, characterized in that, The 2D-3D spatial coordinate mapping relationship is based on the perspective projection model. By establishing the conversion relationship between image pixel coordinates and 3D point cloud spatial coordinates, the position of the distorted area in the three-dimensional model can be accurately located.

3. The tire wear monitoring system based on digital twin comparison according to claim 1, characterized in that, In the model alignment unit, the geometric features of the non-wear area include at least one of the following: tire sidewall symmetrical feature circles, brand logo, specification characters, mold positioning holes, and wheel hub edge contours; the alignment process includes: calculating the model center based on the average coordinates of the center of at least three sets of tire sidewall symmetrical feature circles, aligning the center and axis of the measured model with the center of the initial model, and then using an incremental rotation strategy to step the measured model circumferentially along the central axis until it is aligned with the circumferential features of the initial model.

4. The tire wear monitoring system based on digital twin comparison according to claim 1, characterized in that, In the wear calculation unit, the wear depth is obtained by calculating the signed normal distance from the sampling point of the measurement model to the corresponding projection point of the reference model, and positive deviation data of the measurement point located outside the reference surface are regarded as abnormal data and removed.

5. The tire wear monitoring system based on digital twin comparison according to claim 4, characterized in that, In the wear calculation unit, the wear volume is calculated by Boolean operation difference set. In the tread area, the geometric entity difference set of the comparison model is obtained by Boolean operation. After calculating the geometric volume of the difference set area, the wear volume of the tire in three-dimensional geometry is obtained.

6. The tire wear monitoring system based on digital twin comparison according to claim 1, characterized in that, The status management unit intelligently identifies the current wear status and form of the tire based on multi-dimensional wear indicators. The multi-dimensional wear indicators include at least two of the following: maximum wear depth, average value, standard deviation, wear volume, and wear rate. The visualization method includes a panoramic differential cloud map, which intuitively displays the tread wear distribution through color changes.

7. The tire wear monitoring system based on digital twin comparison according to claim 1, characterized in that, The data acquisition unit uses a portable handheld scanner to perform multi-angle, multi-region 3D scanning and simultaneously acquire multi-angle photos of the tire; the reconstruction process includes triangulation of the original 3D point cloud data.

8. A tire wear monitoring method based on digital twin comparison, applied to the tire wear monitoring system based on digital twin comparison as described in any one of claims 1-7, characterized in that, The method includes the following steps: S1: Perform a 3D scan on the tire to be monitored to obtain the original three-dimensional point cloud data and reconstruct it into a measured three-dimensional model; S2: Identify distorted areas in the scanned model, remove and optimize the data, and generate a three-dimensional digital twin model that represents the true state of the tire; S3: Using the geometric features of the tire's non-wear area as a constraint benchmark, the measured model is spatially aligned with the pre-stored initial model of the unworn tire with high precision; S4: The wear depth and volume of different areas of the tire tread are calculated using 3D deviation analysis and Boolean operation principles; S5: Based on multi-dimensional wear indicators, identify the tire wear status and form, and display the wear status in a visual way to realize tire life cycle data management.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the tire wear monitoring method based on digital twin comparison as described in claim 8.

Images