Large special-shaped ring section contour structured light intelligent measurement method and device

Abstract

The invention discloses a large special-shaped ring section contour structured light intelligent measurement method and system, and the method comprises the following steps: S1, collecting the point cloud data of a to-be-measured large special-shaped ring section contour, and the collection modes comprise radial collection and circumferential collection; s2, processing the point cloud data to obtain complete point cloud data of the section contour of the to-be-measured large special-shaped ring piece; s3, inputting the point cloud data into a pre-trained deep learning model, and outputting information, including feature points, size and shape information, of the section contour of the to-be-measured large special-shaped ring piece; the deep learning model is pre-constructed, and a data set is made according to collected point cloud data of various large special-shaped ring part section contours for training and evaluation; and S4, performing comparative analysis on the information, output by the deep learning model, of the section contour of the to-be-measured large special-shaped ring piece and the size in the actual design drawing of the ring piece, and calculating and outputting a machining error in real time. According to the invention, the detection precision of the large special-shaped ring piece can be obviously improved.

Description

Technical Field

[0001] The present invention relates to the field of machine vision technology, and in particular to a structured light intelligent measurement method and device for the cross-sectional profile of a large-sized special-shaped ring, which is suitable for accurately measuring the cross-sectional profile and size of the large-sized special-shaped ring. Background Art

[0002] Large special-shaped rings are key basic components in the fields of wind power, aerospace, industrial machinery, and building bridges. They are widely used in important equipment such as wind turbines, spacecraft launch platforms, port cranes, and elevated bridge support columns. Their quality and performance have attracted much attention. To ensure that these rings meet strict quality standards, high-performance seamless ring forming technology is usually used for rolling, followed by precision machining. Machining, as the core link in the key structural forming of large special-shaped rings, can further accurately adjust the shape and size of the rings to ensure that they meet design requirements and industry standards. The quality of machining is directly related to the final performance of the rings and whether they can meet service conditions.

[0003] However, the current measurement of the cross-sectional profile of large special-shaped rings mainly relies on handheld laser cutting standard profile calipers. This traditional measurement method has many limitations, including large measurement errors, low efficiency, and susceptibility to human factors. In addition, the data obtained by the manual caliper is too localized and cannot fully and accurately reflect the dimensional deviation of the cross-sectional profile of large special-shaped rings, resulting in incomplete information and difficulty in rework, thereby reducing processing efficiency.

[0004] Therefore, there is an urgent need to develop a more accurate, real-time and intelligent measurement method that can comprehensively measure the cross-sectional profile shape and size of large special-shaped rings during the machining process, and display the measurement results and output the machining errors in real time, and promptly feedback the problems in the machining process, thereby improving the machining efficiency and quality of the rings. Summary of the invention

[0005] The main purpose of the present invention is to provide a structured light intelligent measurement method for the cross-section contour line of a large-scale special-shaped ring part, which is used to measure the size and shape of the cross-section of the large-scale special-shaped ring part after fine processing and output the processing error in real time.

[0006] The technical solution adopted by the present invention is:

[0007] A method for intelligently measuring the cross-sectional profile of a large-scale special-shaped ring with structured light is provided, comprising the following steps:

[0008] S1. Collect the cross-sectional contour point cloud data of the large-scale special-shaped ring to be tested, wherein the collection methods include radial collection and circumferential collection. The radial collection is specifically performed by controlling the scanning frequency, data interval and step displacement of the line structured light scanner and the radial collection displacement stage, and the large-scale special-shaped ring to be tested is scanned once at a certain distance; the circumferential collection is specifically performed by controlling the scanning frequency and data interval of the line structured light scanner and the step speed and distance of the circumferential collection rotating platform, and the ring to be tested is scanned once at a certain angle; and the coverage rate between adjacent scans meets the preset value;

[0009] S2. Processing the point cloud data to obtain complete point cloud data of the cross-section profile of the large-sized special-shaped ring to be tested;

[0010] S3, inputting the cross-sectional profile point cloud data of the large special-shaped ring to be tested into a pre-trained deep learning model, and outputting the cross-sectional profile information of the large special-shaped ring to be tested, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected cross-sectional profile point cloud data of various large special-shaped rings for training and evaluation;

[0011] S4. Compare and analyze the information of the cross-sectional profile of the large special-shaped ring to be tested output by the deep learning model with the dimensions in the actual design drawings of the ring, calculate the processing error in real time and output it.

[0012] Following the above technical solution, in step S2, the point cloud data is processed, specifically including noise identification and removal, outlier processing, point cloud resampling and optimization.

[0013] Following the above technical solution, in step S2, processing the point cloud data includes using a point cloud coarse registration algorithm to align the multiple sets of point cloud data collected in preliminary position and posture; and using a point cloud fine registration algorithm to perform fine registration on the basis of the coarse registration, so that the multiple sets of point cloud data are highly aligned, and a complete large-scale special-shaped ring cross-section contour point cloud is output.

[0014] Following the above technical solution, step S2 also includes the following steps: calibrating the line structured light scanner using a checkerboard calibration plate to correct the distortion of the laser and the lens.

[0015] Following the above technical solution, in step S3, when making the data set of the deep learning model, a sufficient number of complete point clouds of cross-sectional contours of different special-shaped ring parts acquired in advance are enhanced by data enhancement methods, including mirror flipping, point cloud rotation, and changing color channels, to produce a point cloud feature point recognition data set.

[0016] Following the above technical solution, PointNet or PointNet++ is selected as the deep learning model.

[0017] Following the above technical solution, when collecting radial data, the coverage rate must satisfy the following formula:

[0018]

[0019] Where H is the height of the ring to be measured, L is the range of the line structured light scanner, and v r is the stepping speed of the radial acquisition stage, t r is the step time, and N is the number of radial acquisition steps.

[0020] Following the above technical solution, during circumferential acquisition, the scanning frequency f of the line structured light scanner satisfies the following formula:

[0021]

[0022] Where y is the moving direction accuracy of the scanner, α is the fixed angle that the ring rotates at each step, D is the outer diameter of the raceway of the large special-shaped ring to be measured, and t c Acquire the step time of the rotating platform in the circumferential direction.

[0023] The present invention also provides a large-scale special-shaped ring cross-section profile structured light intelligent measurement system, comprising:

[0024] The point cloud data acquisition module is used to acquire the point cloud data of the cross-sectional profile of the large-scale special-shaped ring to be tested, wherein the acquisition methods include radial acquisition and circumferential acquisition. The radial acquisition is specifically performed by controlling the scanning frequency, data interval and step displacement of the line structured light scanner and the radial acquisition displacement of the displacement stage, and the large-scale special-shaped ring to be tested is scanned once at a certain distance; the circumferential acquisition is specifically performed by controlling the scanning frequency and data interval of the line structured light scanner and the step speed and distance of the circumferential acquisition rotating platform, and the ring to be tested is scanned once at a certain angle; and the coverage rate between adjacent scans meets the preset value;

[0025] A data processing module is used to process the point cloud data to obtain complete point cloud data of the cross-section profile of the large-sized special-shaped ring to be tested;

[0026] The ring recognition module is used to input the cross-sectional profile point cloud data of the large special-shaped ring to be tested into a pre-trained deep learning model, and output the cross-sectional profile information of the large special-shaped ring to be tested, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected cross-sectional profile point cloud data of various large special-shaped rings for training and evaluation;

[0027] The error calculation module is used to compare and analyze the information of the cross-sectional profile of the large special-shaped ring to be tested output by the deep learning model with the dimensions in the actual design drawings of the ring, calculate the processing error in real time and output it.

[0028] The present invention also provides a large-scale special-shaped ring cross-section profile structured light intelligent measurement device, comprising:

[0029] The radial acquisition displacement stage is used to move along the radial direction of the large special-shaped ring to be measured according to the control instructions;

[0030] The circumferential acquisition rotating platform is used to rotate along the circumference of the large special-shaped ring to be tested according to the control instructions;

[0031] A line structured light scanner is used to be arranged on a radial acquisition displacement stage or a line structured light scanner, and scans according to a preset scanning frequency and data interval;

[0032] An image acquisition card is connected to a data output port of the line structured light scanner to obtain scanned point cloud data;

[0033] The controller is connected to the radial acquisition displacement stage, the circumferential acquisition rotating platform, and the image acquisition card. During radial acquisition, the scanning frequency and data interval of the line structured light scanner and the stepping displacement of the radial acquisition displacement stage are controlled to scan the large special-shaped ring to be measured once at a certain distance; during circumferential acquisition, the scanning frequency and data interval of the line structured light scanner and the stepping speed and distance of the circumferential acquisition rotating platform are controlled to scan the ring to be measured once at a certain angle; and the coverage rate between adjacent scans meets the preset value;

[0034] The processor is connected to the image acquisition card and is used to process the point cloud data collected by the line structured light scanner to obtain complete point cloud data of the cross-sectional contour of the large special-shaped ring to be tested; the point cloud data of the cross-sectional contour of the large special-shaped ring to be tested is input into a pre-trained deep learning model, and the information of the cross-sectional contour of the large special-shaped ring to be tested is output, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected point cloud data of the cross-sectional contour of various large special-shaped rings for training and evaluation; the information of the cross-sectional contour of the large special-shaped ring to be tested output by the deep learning model is then compared and analyzed with the dimensions in the actual design drawings of the ring, and the processing error is calculated and output in real time.

[0035] The beneficial effects of the present invention are as follows: the present invention uses special radial acquisition and circumferential acquisition to obtain the point cloud data of the cross-sectional profile of the ring according to the characteristics of large special-shaped rings, and processes the acquired point cloud data to obtain complete and accurate cross-sectional profile data. In addition, the present invention pre-trains and evaluates the deep learning model comprehensively to ensure that the model can accurately identify and segment the cross-sectional profile point cloud. With the help of this intelligent measurement method, the shape and size parameters of the cross-sectional profile of the ring can be output efficiently and accurately. At the same time, the method can also timely discover the processing errors of the ring during rolling and machining, thereby improving the overall efficiency and economic benefits of the ring rolling process. It can be seen that the present invention not only significantly improves the detection accuracy of large special-shaped rings and ensures high-quality control of each link in the production process, but also reduces labor costs through automation and intelligent means, optimizes the production process, and provides strong technical guarantees and economic advantages for enterprises in a highly competitive market environment.

[0036] Furthermore, the present invention first performs a preliminary alignment of the segmented acquired point cloud data through an effective coarse alignment method. This step not only speeds up the subsequent processing, but also significantly reduces the computational burden during fine alignment, and improves the robustness of the overall processing; then, the fine alignment technology is applied to further optimize the preliminary aligned point cloud to ensure highly accurate alignment of each segment of the point cloud, and finally output a complete and high-precision cross-sectional contour point cloud; the application of this method solves the problem that traditional measurement methods are difficult to obtain a complete cross-sectional point cloud of large special-shaped rings at one time, and significantly improves the measurement accuracy; through effective point cloud processing technology, the present invention not only reduces the hardware requirements and costs for the image acquisition system, but also finds an ideal balance between accuracy and efficiency, providing a reliable solution for high-precision measurement of large special-shaped workpieces.

[0037] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1A Flow chart 1 of the structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to an embodiment of the present invention;

[0040] Figure 1B The process of the structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to the embodiment of the present invention is as follows Figure 2 ;

[0041] Figure 2 This is a layout diagram of a structured light intelligent measurement device for the cross-sectional profile of a large-scale special-shaped ring according to an embodiment of the present invention;

[0042] Figure 3 is a schematic diagram of a radial scanning method according to an embodiment of the present invention;

[0043] Figure 4 is a schematic diagram of a circumferential scanning method according to an embodiment of the present invention;

[0044] Figure 5 Schematic diagram of measurement results of an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0046] It should be noted that the illustrations provided in the embodiments of the present invention are only used to illustrate the basic concept of the present invention in a schematic manner. Therefore, the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout type may also be more complicated.

[0047] In the present invention, it is also necessary to explain that, if the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application. In addition, if the terms "first" and "second" appear, they are only used for description and distinction purposes, and cannot be understood as indicating or implying relative importance.

[0048] In addition, it should be noted that the features of the various embodiments of the present invention may be combined or combined in part or in whole, and may interact and operate in different ways as will be appreciated by those skilled in the art. Each embodiment may be implemented independently of one another, or in an associated relationship.

[0049] like Figure 1A As shown, the structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to an embodiment of the present invention comprises the following steps:

[0050] S1. Collect the cross-sectional contour point cloud data of the large-scale special-shaped ring to be tested, wherein the collection methods include radial collection and circumferential collection. The radial collection is specifically performed by controlling the scanning frequency, data interval and step displacement of the line structured light scanner and the radial collection displacement stage, and the large-scale special-shaped ring to be tested is scanned once at a certain distance; the circumferential collection is specifically performed by controlling the scanning frequency and data interval of the line structured light scanner and the step speed and distance of the circumferential collection rotating platform, and the ring to be tested is scanned once at a certain angle; and the coverage rate between adjacent scans meets the preset value;

[0051] S2. Processing the point cloud data to obtain complete point cloud data of the cross-section profile of the large-sized special-shaped ring to be tested;

[0052] S3, inputting the cross-sectional profile point cloud data of the large special-shaped ring to be tested into a pre-trained deep learning model, and outputting the cross-sectional profile information of the large special-shaped ring to be tested, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected cross-sectional profile point cloud data of various large special-shaped rings for training and evaluation;

[0053] S4. Compare and analyze the information of the cross-sectional profile of the large special-shaped ring to be tested output by the deep learning model with the dimensions in the actual design drawings of the ring, calculate the processing error in real time and output it.

[0054] Specifically, the process of making the dataset is as follows:

[0055] (1) The process design for collecting the cross-sectional profile of large special-shaped rings is carried out. The collection process involves multiple, multi-angle, and multi-directional collection of the cross-sectional profile of large special-shaped rings, which is divided into radial collection and circumferential collection.

[0056] Radial acquisition is: for large special-shaped rings whose cross-sectional profile shape and size need to be acquired, the number of radial sampling and step distance are designed according to the ring height, the measuring range of the linear structured light scanner, and the radial acquisition stage combined with the requirements of the point cloud registration algorithm;

[0057] Circumferential acquisition: For large special-shaped rings that need to acquire cross-sectional contour shapes and sizes, the sampling interval time and sampling interval angle are designed according to the processing technology of the rings, the scanning frame rate and data interval of the line structured light scanner, and the stepping speed and rotation angle of the circumferential acquisition rotating platform.

[0058] Among them, whether radial or circumferential acquisition, there needs to be a reasonable overlapping scanning part between adjacent scans of the line structured light scanner, and the general overlap rate is 30%-40%, so that the point cloud data points are sufficient to cover the entire ring contour.

[0059] (2) After obtaining a sufficient number of complete point clouds of the cross-sectional profiles of different special-shaped ring parts, data enhancement methods are used, including mirror flipping, point cloud rotation, and changing color channels, to produce a point cloud feature point recognition data set to provide data support for model training.

[0060] In step S2, the acquired point cloud data needs to be preprocessed, including noise identification and removal, outlier processing, point cloud resampling and optimization, etc.

[0061] Furthermore, in step S2, processing the point cloud data includes using a point cloud coarse registration algorithm to align the multiple sets of collected point cloud data in preliminary position and posture; and using a point cloud fine registration algorithm to perform fine registration on the basis of the coarse registration, so that the multiple sets of point cloud data are highly aligned, and a complete large-scale special-shaped ring cross-section contour point cloud is output.

[0062] Specifically, point cloud rough registration algorithms such as RANSAC (Random Sampling Consistency) and FPFH (Point Cloud Feature Descriptor) can be used to align the collected multiple sets of point cloud data in preliminary position and posture, providing a good starting point for subsequent precise registration. Point cloud fine registration algorithms such as ICP (Iterative Closest Point) and GMM-Registration (Gaussian Mixture Model Registration) can be used to further improve the accuracy of point cloud registration on the basis of rough registration, so that multiple sets of point cloud data are highly aligned and a complete point cloud of the cross-section profile of a large special-shaped ring is output.

[0063] Furthermore, step S2 also includes the following steps: adopting Zhang Zhengyou calibration method and using a checkerboard calibration plate to calibrate the line structured light scanner, correcting the distortion of the laser and lens, and ensuring the integrity and accuracy of the point cloud data.

[0064] Specifically, in step S3, a point cloud deep learning model, such as PointNet, PointNet++, etc., can be used as the deep learning model of this method to complete the training and evaluation of the deep learning model and meet the evaluation indicators such as accuracy required by the task.

[0065] In step S4, the measurement results output by the deep learning model can be imported into the large-scale special-shaped ring cross-section profile structured light intelligent measurement software, compared with its design drawings, and the finishing error is calculated and output in real time.

[0066] In another embodiment of the present invention, a more specific acquisition method is provided for radial acquisition and circumferential acquisition:

[0067] like Figure 3As shown in the figure, the specific process of radial acquisition design for large special-shaped rings is as follows: by controlling the scanning frequency and data interval of the line structured light scanner and the step displacement of the radial acquisition stage, the ring to be tested is scanned once at a certain interval to obtain a portion of point cloud data. The height of the ring to be tested is H, the range of the line structured light scanner is L, and the step speed of the radial acquisition stage is v r With step time t r , the number of radial acquisition steps N, to ensure the coverage rate, must satisfy the following formula (1):

[0068]

[0069] like Figure 4 As shown in the figure, the specific process of the circumferential acquisition design of large special-shaped rings is as follows: by controlling the scanning frequency and data interval of the line structured light scanner and the stepping speed and distance of the circumferential acquisition rotating platform, the ring to be tested is scanned once at a certain angle interval to obtain a part of the point cloud data. The scanner's motion direction accuracy is y, the outer diameter of the ring to be tested at the raceway is D, and by setting the stepping time t of the circumferential acquisition rotating platform c , ensuring that each step causes the ring to rotate a fixed angle α, so that the scanning frequency f should satisfy the following formula (2):

[0070]

[0071] Since circumferential acquisition requires the collection of cross-sectional profile data at many different angles in the circumferential direction, the acquisition can be divided into multiple times, and each acquisition is numbered according to the rotation angle.

[0072] After scanning the cross-sectional contour of a large special-shaped ring, the point cloud data of each part is processed to obtain a complete cross-sectional contour point cloud. The recognition rate and accuracy of the deep learning model can be improved by cross-validation or other training and testing methods to evaluate the performance of the model, such as the retention method, K-fold cross-validation, etc., which effectively avoids information confusion between different groups and improves the generalization ability of the network model, thereby improving the accuracy and robustness of the model. A trained deep learning model is used to identify the cross-sectional contour point cloud of a complete large special-shaped ring to be tested, and the size and shape of the complete contour point cloud of the large special-shaped ring are detected, and compared and analyzed with the size in the actual design drawing of the ring to obtain the final calculation result, and the cross-sectional contour shape, size and processing error of the large special-shaped ring are output. Through this systematic measurement method, the present invention realizes high-precision and intelligent measurement of the cross-sectional contour of large special-shaped rings, which not only improves the detection efficiency, but also ensures the accuracy and reliability of the measurement, and provides strong support for the precision manufacturing of large special-shaped rings.

[0073] When the measuring range of the scanner is smaller than the height of the ring to be measured, a segmented measurement strategy is used for radial acquisition, and it is ensured that the data of adjacent segments have an appropriate overlapping area to facilitate subsequent data fusion.

[0074] In order to implement the above method embodiment, the present invention also provides a large-scale special-shaped ring cross-section profile structured light intelligent measurement system, the system comprising:

[0075] The point cloud data acquisition module is used to acquire the point cloud data of the cross-sectional profile of the large-scale special-shaped ring to be tested, wherein the acquisition methods include radial acquisition and circumferential acquisition. The radial acquisition is specifically performed by controlling the scanning frequency, data interval and step displacement of the line structured light scanner and the radial acquisition displacement of the displacement stage, and the large-scale special-shaped ring to be tested is scanned once at a certain distance; the circumferential acquisition is specifically performed by controlling the scanning frequency and data interval of the line structured light scanner and the step speed and distance of the circumferential acquisition rotating platform, and the ring to be tested is scanned once at a certain angle; and the coverage rate between adjacent scans meets the preset value;

[0076] A data processing module is used to process the point cloud data to obtain complete point cloud data of the cross-section profile of the large-sized special-shaped ring to be tested;

[0077] The ring recognition module is used to input the cross-sectional profile point cloud data of the large special-shaped ring to be tested into a pre-trained deep learning model, and output the cross-sectional profile information of the large special-shaped ring to be tested, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected cross-sectional profile point cloud data of various large special-shaped rings for training and evaluation;

[0078] The error calculation module is used to compare and analyze the information of the cross-sectional profile of the large special-shaped ring to be tested output by the deep learning model with the dimensions in the actual design drawings of the ring, calculate the processing error in real time and output it.

[0079] The above modules are mainly used to implement the steps of the method embodiment, which will not be described one by one here.

[0080] The present application also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a disk, an optical disk, a server, an App application store, etc., on which a computer program is stored, and when the program is executed by a processor, a corresponding function is realized. When the computer-readable storage medium of this embodiment is executed by a processor, the structured light intelligent measurement method of the cross-sectional profile of a large-scale special-shaped ring part of the method embodiment is realized.

[0081] like Figure 5 , which is a display interface of a structured light intelligent measurement system for the cross-sectional profile of a large-scale special-shaped ring according to an embodiment of the present invention.

[0082] like Figure 2 As shown, the present invention also provides a large-scale special-shaped ring cross-section profile structured light intelligent measurement device, comprising:

[0083] The radial acquisition displacement stage is used to move along the radial direction of the large special-shaped ring to be measured according to the control instructions;

[0084] The circumferential acquisition rotating platform is used to rotate along the circumference of the large special-shaped ring to be tested according to the control instructions;

[0085] A line structured light scanner is used to be arranged on a radial acquisition displacement stage or a line structured light scanner, and scans according to a preset scanning frequency and data interval;

[0086] An image acquisition card is connected to a data output port of the line structured light scanner to obtain scanned point cloud data;

[0087] The controller is connected to the radial acquisition displacement stage, the circumferential acquisition rotating platform, and the image acquisition card. During radial acquisition, the scanning frequency and data interval of the line structured light scanner and the stepping displacement of the radial acquisition displacement stage are controlled to scan the large special-shaped ring to be measured once at a certain distance; during circumferential acquisition, the scanning frequency and data interval of the line structured light scanner and the stepping speed and distance of the circumferential acquisition rotating platform are controlled to scan the ring to be measured once at a certain angle; and the coverage rate between adjacent scans meets the preset value;

[0088] The processor is connected to the image acquisition card and is used to process the point cloud data collected by the line structured light scanner to obtain complete point cloud data of the cross-sectional contour of the large special-shaped ring to be tested; the point cloud data of the cross-sectional contour of the large special-shaped ring to be tested is input into a pre-trained deep learning model, and the information of the cross-sectional contour of the large special-shaped ring to be tested is output, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected point cloud data of the cross-sectional contour of various large special-shaped rings for training and evaluation; the information of the cross-sectional contour of the large special-shaped ring to be tested output by the deep learning model is then compared and analyzed with the dimensions in the actual design drawings of the ring, and the processing error is calculated and output in real time.

[0089] In summary, the present invention has comprehensively optimized the many problems existing in the contact measurement of manual handheld calipers, and completely solved a series of difficult problems such as the time-consuming and low efficiency of traditional methods in measuring large special-shaped rings, as well as potential safety hazards during operation. Through the present invention, the detection efficiency has been significantly improved, especially in the detection of complex cross-sectional profile shapes and sizes. This method not only supports full inspection of large special-shaped rings after processing to ensure that the quality and size of each ring meet the requirements, but also greatly improves the speed and accuracy of the detection process by introducing intelligent measurement technology based on deep learning. The application of intelligent measurement methods makes the detection process more automated and intelligent, thereby reducing the errors caused by human operation, while significantly improving the overall measurement effect and accuracy.

[0090] In addition, the present invention can meet the measurement requirements of special-shaped rings of different complexities, and can provide reliable and consistent measurement results in scenarios with high accuracy requirements or in environments that require fast and efficient detection. This breakthrough method provides strong technical support for the quality control and production efficiency improvement of large special-shaped rings.

[0091] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, and two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0092] The order of execution of each step in the above embodiment does not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0093] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. A method for intelligent measurement of the cross-sectional profile of a large-scale special-shaped ring by structured light, characterized in that: The following steps are involved: S1. Collect the cross-sectional contour point cloud data of the large-scale special-shaped ring to be tested, wherein the collection methods include radial collection and circumferential collection. The radial collection is specifically performed by controlling the scanning frequency, data interval and step displacement of the line structured light scanner and the radial collection displacement stage, and the large-scale special-shaped ring to be tested is scanned once at a certain distance; the circumferential collection is specifically performed by controlling the scanning frequency and data interval of the line structured light scanner and the step speed and distance of the circumferential collection rotating platform, and the ring to be tested is scanned once at a certain angle; and the coverage rate between adjacent scans meets the preset value; S2. Processing the point cloud data to obtain complete point cloud data of the cross-section profile of the large-sized special-shaped ring to be tested; S3, inputting the cross-sectional profile point cloud data of the large special-shaped ring to be tested into a pre-trained deep learning model, and outputting the cross-sectional profile information of the large special-shaped ring to be tested, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected cross-sectional profile point cloud data of various large special-shaped rings for training and evaluation; S4. Compare and analyze the information of the cross-sectional profile of the large special-shaped ring to be tested output by the deep learning model with the dimensions in the actual design drawings of the ring, calculate the processing error in real time and output it.

2. The structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to claim 1 is characterized in that: In step S2, the point cloud data is processed, specifically including noise identification and removal, outlier processing, point cloud resampling and optimization.

3. The structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to claim 1 is characterized in that: In step S2, processing the point cloud data includes using a point cloud coarse registration algorithm to align the multiple sets of collected point cloud data in preliminary position and posture; and using a point cloud fine registration algorithm to perform fine registration on the basis of the coarse registration, so that the multiple sets of point cloud data are highly aligned, and a complete large-scale special-shaped ring cross-section contour point cloud is output.

4. The structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to claim 1 is characterized in that: Step S2 also includes the following steps: calibrating the line structured light scanner using a checkerboard calibration plate to correct the distortion of the laser and the lens.

5. The structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to claim 1 is characterized in that: In step S3, when making a data set for the deep learning model, a sufficient number of pre-acquired complete point clouds of cross-sectional profiles of different special-shaped ring parts are enhanced by data enhancement methods, including mirror flipping, point cloud rotation, and changing color channels, to produce a point cloud feature point recognition data set.

6. The structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to claim 1 is characterized in that: The deep learning model uses PointNet or PointNet++.

7. The structured light intelligent measurement method for the cross-sectional profile of a large-scale special-shaped ring according to claim 1 is characterized in that: When collecting radial data, the coverage rate must satisfy the following formula: Where H is the height of the ring to be measured, L is the range of the line structured light scanner, and v r is the stepping speed of the radial acquisition stage, t r is the step time, and N is the number of radial acquisition steps.

8. The structured light intelligent measurement method for the cross-section profile of a large-scale special-shaped ring according to claim 1 is characterized in that: When collecting data in the circumferential direction, the scanning frequency f of the line structured light scanner satisfies the following formula: Where y is the moving direction accuracy of the scanner, α is the fixed angle that the ring rotates at each step, D is the outer diameter of the raceway of the large special-shaped ring to be measured, and t c Acquire the step time of the rotating platform in the circumferential direction.

9. A large-scale special-shaped ring cross-section profile structured light intelligent measurement system, characterized in that: include: The point cloud data acquisition module is used to acquire the point cloud data of the cross-sectional profile of the large-scale special-shaped ring to be tested, wherein the acquisition methods include radial acquisition and circumferential acquisition. The radial acquisition is specifically performed by controlling the scanning frequency, data interval and step displacement of the line structured light scanner and the radial acquisition displacement of the displacement stage, and the large-scale special-shaped ring to be tested is scanned once at a certain distance; the circumferential acquisition is specifically performed by controlling the scanning frequency and data interval of the line structured light scanner and the step speed and distance of the circumferential acquisition rotating platform, and the ring to be tested is scanned once at a certain angle; and the coverage rate between adjacent scans meets the preset value; A data processing module is used to process the point cloud data to obtain complete point cloud data of the cross-section profile of the large-sized special-shaped ring to be tested; The ring recognition module is used to input the cross-sectional profile point cloud data of the large special-shaped ring to be tested into a pre-trained deep learning model, and output the cross-sectional profile information of the large special-shaped ring to be tested, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected cross-sectional profile point cloud data of various large special-shaped rings for training and evaluation; The error calculation module is used to compare and analyze the information of the cross-sectional profile of the large special-shaped ring to be tested output by the deep learning model with the dimensions in the actual design drawings of the ring, calculate the processing error in real time and output it.

10. A large-scale special-shaped ring cross-section profile structured light intelligent measurement device, characterized in that: include: The radial acquisition displacement stage is used to move along the radial direction of the large special-shaped ring to be measured according to the control instructions; The circumferential acquisition rotating platform is used to rotate along the circumference of the large special-shaped ring to be tested according to the control instructions; A line structured light scanner is used to be arranged on a radial acquisition displacement stage or a line structured light scanner, and scans according to a preset scanning frequency and data interval; An image acquisition card is connected to a data output port of the line structured light scanner to obtain scanned point cloud data; The controller is connected to the radial acquisition displacement stage, the circumferential acquisition rotating platform, and the image acquisition card. During radial acquisition, the scanning frequency and data interval of the line structured light scanner and the stepping displacement of the radial acquisition displacement stage are controlled to scan the large special-shaped ring to be measured once at a certain distance; during circumferential acquisition, the scanning frequency and data interval of the line structured light scanner and the stepping speed and distance of the circumferential acquisition rotating platform are controlled to scan the ring to be measured once at a certain angle; and the coverage rate between adjacent scans meets the preset value; A processor is connected to the image acquisition card and is used to process the point cloud data collected by the line structured light scanner to obtain complete point cloud data of the cross-sectional profile of the large special-shaped ring to be tested; the point cloud data of the cross-sectional profile of the large special-shaped ring to be tested is input into a pre-trained deep learning model, and the information of the cross-sectional profile of the large special-shaped ring to be tested is output, including feature points, size and shape information; the deep learning model is pre-built, and a data set is prepared based on the collected point cloud data of the cross-sectional profile of various large special-shaped rings for training and evaluation; The information on the cross-sectional profile of the large special-shaped ring to be tested output by the deep learning model is then compared and analyzed with the dimensions in the actual design drawings of the ring, and the processing error is calculated and output in real time.

Images