SPR (Surface Plasmon Resonance) riveting point position detection method and system based on coordinate recognition

Abstract

The invention provides an SPR riveting point position detection method and system based on coordinate recognition, and the method comprises the steps: selecting an assembly digital-analog file which needs to be subjected to SPR riveting point position detection, carrying out the software recognition of digital-analog files of all SPR riveting points in the assembly digital-analog file, carrying out the highlight display of the SPR riveting point information in the digital-analog file, and carrying out the detection of the SPR riveting point position. Inputting a recognition model to obtain two-dimensional coordinates of highlight information of a front view, a left view and a top view, converting the three two-dimensional coordinates into three-dimensional coordinates of SPR riveting points, recording the three-dimensional coordinates, and detecting parts intersecting with the SPR riveting points; detecting whether the edge distance and the dot distance of each SPR riveting point are qualified or not; and for SPR riveting points with unqualified edge distances and / or point distances, storing the positions of the SPR riveting points in a screenshot manner, and outputting the screenshot on a software interface for a user to click and view. Detection, judgment and recognition of the SPR riveting point position are completed in a software mode, judgment and auditing are more efficient and accurate, and time cost and personnel cost are saved.

Description

Technical Field

[0001] The present invention belongs to the field of automobile automation design, and in particular relates to a method and system for detecting SPR rivet point positions based on coordinate recognition. Background Art

[0002] In the field of automotive automation design, the SPR riveting process (SPR, short for Self-Piercing Riveting) offers excellent joint strength and fatigue resistance, meeting the needs of various industrial applications. The joints formed using this process offer high strength and stable, controllable quality. The location of the SPR rivets in this process has a significant impact on their performance. Proper rivet point placement ensures sufficient strength and rigidity while avoiding excessive stress concentration or damage to surrounding materials. Furthermore, optimizing the layout and number of rivet points can further enhance the strength and stability of the overall structure. With the increasing demand for lightweighting in the automotive industry, SPR technology, with its ability to join lightweight materials (such as aluminum alloys and composite materials), has become a key tool for achieving lightweight design. By optimizing the location and number of rivets, structural weight can be further reduced while maintaining sufficient strength and stability.

[0003] At present, the detection and judgment of the position of SPR rivet points still need to be judged manually. It is necessary to measure the relationship between the margin distance (margin distance refers to the distance between the SPR rivet point and the intersecting part) and the point distance (point distance refers to the distance between the SPR rivet point and the adjacent SPR rivet point) of each point, and perform planing processing. This process is cumbersome and completely relies on the experience of engineers. It is time-consuming and prone to large manual errors. Summary of the Invention

[0004] The present invention proposes a SPR rivet point position detection method and system based on coordinate recognition, which completes the detection, judgment and identification of the SPR rivet point position through software, making the judgment and review more efficient and accurate, saving time and personnel costs.

[0005] To achieve the above object, the technical solution of the present invention is achieved as follows: A method for detecting SPR rivet point positions based on coordinate recognition, comprising: S1. Select the assembly digital model file that needs to be tested for SPR rivet point position and input the model file into the software; S2. The software identifies the digital model files of all SPR rivet points in the assembly digital model file, highlights the SPR rivet point information in the digital model through the input identification mark, and takes position screenshots of the highlighted information through the main view, left view, and top view to obtain the two-dimensional main view, left view, and top view, and inputs the recognition model to obtain the two-dimensional coordinates of the highlighted information of the main view, left view, and top view respectively, converts the three two-dimensional coordinates into the three-dimensional coordinates of the SPR rivet point and records them, and detects the parts intersecting with each SPR rivet point; S3. Check whether the margin between each SPR rivet point and the intersecting part is qualified. The detection method is to generate a cylinder that covers the digital model of the intersecting part and does not interfere with other parts, and determine whether the intersection of the cylinder and the intersecting part completely covers the fitting surface between the SPR rivet point and the intersecting part. If the fitting surface is covered, it is judged as qualified, otherwise it is unqualified. S4. Check whether the distance between each SPR rivet point and the adjacent SPR rivet point is qualified. The detection method is to calculate the distance between each SPR rivet point and the adjacent SPR rivet point according to the centroid coordinates of the SPR rivet point. If the calculated distance value meets the specified value, it is qualified, otherwise it is unqualified; S5. For SPR rivet points with unqualified margins and / or unqualified point distances, their positions are captured and stored, and the screenshots are output to the software interface for users to click and view.

[0006] Furthermore, in step S2, the recognition model calls the Caffe2 framework for model training. The model training includes: cleaning the two-dimensional image of the highlighted SPR rivet point and retaining the data to be detected; then performing data annotation, annotating the retained data with SPR rivet points, and then performing preprocessing operations including rotation, enhancement, and grayscale processing on the annotated two-dimensional image. The two-dimensional image set obtained after preprocessing is divided into a training set, a test set, and a validation set. The two-dimensional image is used as input and the two-dimensional coordinates of the SPR rivet point are used as the target for training.

[0007] Furthermore, the method of converting into three-dimensional coordinates in step S2 includes: Obtain the X and Z coordinates from the two-dimensional coordinates of the main view, obtain the X and Y coordinates from the two-dimensional coordinates of the top view, and obtain the Y and Z coordinates from the two-dimensional coordinates of the side view; and convert the X, Y, and Z coordinates into three-dimensional coordinates.

[0008] Furthermore, in step S2, the method for detecting parts intersecting with each SPR rivet point includes: S211, detecting the part closest to the center of gravity of the SPR rivet point according to the center of gravity coordinates of the SPR rivet point, and generating a normal line through the center of gravity point and the surface of the part closest to the center of gravity of the SPR rivet point; S212. Call the "Collision" command, use the normal as the detection direction, perform collision detection on the SPR rivet point and each part, obtain the detection result based on the displayed interference result, and the parts that have an interference relationship with the SPR rivet point are intersecting parts.

[0009] Furthermore, the method for generating the cylinder in step S3 includes: S301, extracting the surface where a certain SPR rivet point is located; S302, obtaining the fitting surface between the surface where the SPR rivet point is located and an intersecting part; S303, generating a normal line through the centroid coordinates of the SPR riveting point and the fitting surface; S304 , generating a detection cylinder along the normal line, wherein the cylinder generates the shortest cylinder that meets the condition that the cylinder covers the digital model of the intersecting part and does not interfere with other parts.

[0010] On the other hand, the present invention also proposes an SPR rivet point position detection system based on coordinate recognition, comprising: Input module: Select the assembly digital model file that needs to be tested for SPR rivet point position, and input the model file into the software; Identification and coordinate module: The software identifies the digital model files of all SPR rivet points in the assembly digital model file, highlights the SPR rivet point information in the digital model through the input identification mark, and takes position screenshots of the highlighted information through the main view, left view, and top view to obtain the two-dimensional main view, left view, and top view. The recognition model is input to obtain the two-dimensional coordinates of the highlighted information of the main view, left view, and top view respectively, and the three two-dimensional coordinates are converted into the three-dimensional coordinates of the SPR rivet points and recorded, and the parts intersecting with each SPR rivet point are detected; Margin determination module: Checks the margin between each SPR rivet point and the intersecting part to see if it is qualified. The detection method is to generate a cylinder that covers the digital model of the intersecting part and does not interfere with other parts. It then determines whether the intersection of the cylinder and the intersecting part completely covers the fitting surface between the SPR rivet point and the intersecting part. If so, it is qualified, otherwise it is unqualified. Point distance judgment module: for each SPR rivet point, the distance between it and the adjacent SPR rivet point is detected to see if it is qualified. The detection method is to calculate the distance between each SPR rivet point and the adjacent SPR rivet point through the centroid coordinates of the SPR rivet point. If the calculated distance value meets the specified value, it is judged to be qualified, otherwise it is unqualified. Output module: For SPR rivet points with unqualified margins and / or unqualified point distances, their positions are captured and stored, and the screenshots are output to the software interface for users to click and view.

[0011] Furthermore, in the recognition and coordinate module, the recognition model calls the Caffe2 framework for model training. The model training includes: cleaning the two-dimensional image of the highlighted SPR rivet points and retaining the data that needs to be detected; then performing data annotation, annotating the retained data with SPR rivet points, and then performing preprocessing operations including rotation, enhancement, and grayscale processing on the annotated two-dimensional image. The two-dimensional image set obtained after preprocessing is divided into a training set, a test set, and a validation set. The two-dimensional image is used as input and the two-dimensional coordinates of the SPR rivet points are used as targets for training.

[0012] Furthermore, the recognition and coordinate module includes: Obtain the X and Z coordinates from the two-dimensional coordinates of the main view, obtain the X and Y coordinates from the two-dimensional coordinates of the top view, and obtain the Y and Z coordinates from the two-dimensional coordinates of the side view; and convert the X, Y, and Z coordinates into three-dimensional coordinates.

[0013] Furthermore, the recognition and coordinate module includes: According to the centroid coordinates of the SPR rivet point, the part closest to the centroid of the SPR rivet point is detected, and a normal line is generated through the centroid point and the surface of the part closest to the centroid of the SPR rivet point; Call the "Collision" command, use the normal as the detection direction, perform collision detection on the SPR rivet point and each part, and obtain the detection results based on the displayed interference results. The parts that have an interference relationship with the SPR rivet point are intersecting parts.

[0014] Furthermore, the margin determination module includes: Extract the surface where a certain SPR rivet point is located; obtain the fitting surface between the surface where the SPR rivet point is located and a certain intersecting part; generate a normal line through the centroid coordinates of the SPR rivet point and the fitting surface; generate a detection cylinder along the normal line, and generate the shortest cylinder that meets the conditions, wherein the condition is to cover the digital model of the intersecting part and not interfere with other parts.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention provides software logic for SPR rivet point identification and SPR rivet point position detection. Through software judgment and identification, the steps of manual SPR rivet point position detection are simplified to software operations. After the corresponding logical rules are input, the judgment and review are more efficient and accurate, which reduces the professional requirements of vehicle body design and saves time and personnel costs.

[0016] 2. Based on the software method of the present invention, with the continuous optimization and iteration of the algorithm, subsequent detection and judgment will become more and more intelligent, and the overall judgment data can be customized and changed, providing a reference sample for subsequent software development based on deep learning intelligent design. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic diagram of the process of Example 1 of the present invention; Figure 2 This is a diagram of the main interface of the software of Example 1 of the present invention; Figure 3 This is a schematic diagram of the internal logic of the software of Example 1 of the present invention; Figure 4 Schematic diagram of normal generation in Example 1 of the present invention; Figure 5 Schematic diagram of the detection cylinder of Example 1 of the present invention; Figure 6 Schematic diagram of the relationship between the detection cylinder and the fitting surface in Example 1 of the present invention; Figure 7 is a schematic diagram of equally divided normal lines of the screenshot in Example 1 of the present invention; Figure 8 This is an illustration of the definition of the screenshot time line in Example 1 of the present invention; Figure 9 is a schematic diagram of generating a cutting plane for a screenshot according to Example 1 of the present invention; Figure 10 This is a cutting plane effect diagram of the screenshot of Example 1 of the present invention; Figure 11 It is a structural diagram of embodiment 2 of the present invention. DETAILED DESCRIPTION

[0018] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0019] In order to make the purpose and features of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are further described below with reference to the accompanying drawings.

[0020] Example 1: The SPR rivet point position detection method based on coordinate recognition proposed in this embodiment is as follows: Figure 1 Shown, including: S1. Select the assembly digital model file that needs to be tested for SPR rivet point position and input the model file into the software; The SPR rivet point position detection method based on coordinate recognition of the present invention is implemented by software, so it is required to select the digital model file of the vehicle body assembly for input in the software interface, and identify the digital model of the SPR rivet point from the digital model file. The function of this step is to realize the selection and input of the assembly digital model file.

[0021] like Figure 2 Shown is a software interface diagram for inputting the model file of this embodiment.

[0022] The input location is the "Select" button in the "Part Number" box. After clicking "Select," the software uses an API function to call the 3D design software to read the digital model file. In this embodiment, the 3D design software uses CATIA. The software reads the model under the CATIA SM node and then allows the user to click and select. After selection, the model is displayed in the "Part Number" box.

[0023] S2. The software identifies the digital model files of all SPR rivet points in the assembly digital model file, highlights the SPR rivet point information in the digital model through the input identification mark, and takes position screenshots of the highlighted information through the main view, left view, and top view to obtain the two-dimensional main view, left view, and top view, and inputs the recognition model to obtain the two-dimensional coordinates of the highlighted information of the main view, left view, and top view respectively, convert the three two-dimensional coordinates into the three-dimensional coordinates of the SPR rivet points and record them, and detect the parts intersecting with each SPR rivet point.

[0024] exist Figure 2 In the "Identifier Selection" box, select the identifier to be tested from the drop-down menu to complete the identifier selection. If there is no corresponding identifier in the drop-down menu, click the "+" at the back to manually add it. This identifier is a required option. The digital model of the SPR rivet point is identified through the identifier.

[0025] After identifying the digital model of the SPR rivet point, the center of gravity coordinates of the SPR rivet point are usually obtained and recorded by calling the "Measure Inertia" command in CATIA. However, this method also has certain defects: 1. Depends on the accuracy of material properties. Incorrect material properties will lead to deviations in the calculation results of the center of gravity coordinates.

[0026] 2. Accuracy is affected by the details of the digital model. Some tiny features may be simplified and ignored, but this will affect the actual center of gravity position.

[0027] 3. Poor adaptability to dynamic scenes. If the plate undergoes plastic deformation, it cannot respond to the deformation in real time.

[0028] Based on the above defects, this embodiment adopts another method of determining the center of gravity coordinates, the logic is as follows Figure 3 Said, including: The software will highlight the SPR rivet point information in the digital model through the input identification mark, and at the same time display the connection information (highlight information) through the main view, left view, top view Figure 3The operator takes a screenshot of the position, converts the three-dimensional data into two-dimensional data, and performs data cleaning (retaining the data that needs to be tested) and annotation on the two-dimensional image (annotating the retained test data, mainly the coordinates and SPR rivet point number information and view information, etc., to facilitate later machine learning). The data cleaning mainly focuses on the quality of the screenshot (whether it is blurred), the position (whether it is in the correct direction), and the number of screenshots (retaining the same style at the same position), etc., and retains the data that meets the conditions.

[0029] And confirm that the marked information is the correct SPR rivet point. Then the data is pre-processed (image data is rotated, enhanced, and grayscale processed).

[0030] Integrate its algorithm into the computational model. Call the Caffe2 model framework to perform model training (adjusting model parameters to meet desired performance by validating evaluation metrics on the dataset). Model training involves data processing and image recognition within the framework. Input is an image or multi-dimensional sample (batch_size, channels, height, width); output is a sequence of positions, i.e., coordinate values.

[0031] First, the algorithm generates the 2D coordinates of the required SPR rivet points. These 2D coordinates are converted to 3D coordinates and output. (Positioning is performed using three views. The positioning process includes: obtaining the X and Z coordinates from the 2D coordinates of the main view, the X and Y coordinates from the 2D coordinates of the top view, and the Y and Z coordinates from the 2D coordinates of the side view; and converting the X, Y, and Z coordinates into 3D coordinates.) The coordinates of all SPR rivet points in the model are recorded, and margin and point distance detection is performed.

[0032] Methods for detecting parts intersecting with each SPR riveting point include: According to the coordinates of the center of gravity of the SPR rivet point, detect the part closest to the center of gravity of the SPR rivet point.

[0033] Generate a normal line for the SPR rivet point based on the coordinates of the center of gravity and its distance to the nearest part plane. The length of the normal line is 5 mm from each SPR rivet point.

[0034] A preliminary judgment is made on the relationship between the SPR rivet point and the part. The "Collision" command of CATIA is called, and the normal line is used as the detection direction to perform collision detection on the SPR rivet point and each part. If there is no interference between the part and the SPR rivet point digital model, the position judgment process will not be performed on this point. If there is an interference relationship, the part will be recorded as the intersection part of the SPR rivet point.

[0035] S3. Check whether the margin between each SPR rivet point and the intersecting part is qualified. The detection method is to generate a cylinder that covers the digital model of the intersecting part and does not interfere with other parts, and determine whether the intersection of the cylinder and the intersecting part completely covers the fitting surface between the SPR rivet point and the intersecting part. If the fitting surface is covered, it is judged as qualified, otherwise it is unqualified. Specifically include: First, the coordinates of the center of gravity of all SPR rivet points are sorted. Next, the software detects the margins and distances between adjacent points for each SPR rivet point. Margins and distances that meet the requirements are not marked with a color; those that do not meet the requirements are marked with a red font. A screenshot is taken of the locations where the margins and distances do not meet the requirements and stored in the "Screenshot" directory. The screenshot is then displayed in a table and can be viewed by clicking on it. The screenshot process involves marking the coordinates of the locations where the requirements do not meet, centering each location, setting the screenshot ratio, and then taking the screenshot.

[0036] Margin detection methods include: Extract the surface where the SPR rivet point is located. The surface where the SPR rivet point is located is the surface where the SPR rivet point head is located. The position of the SPR rivet point head is identified by confirming the intersecting part plate or thickness. The intersecting part surface closest to the SPR rivet point head is the surface where the rivet point is located. Call the "extraction function" to perform extraction and identification operations.

[0037] The SPR rivet point and the fitting surface generate normals. The fitting surface is obtained by the fitting operation between the two parts to obtain the common surface, that is, the fitting surface; the fitting surface refers to the fitting intersection surface between the upper part (the intersecting part closest to the SPR rivet point head) and the lower part (the rivet mold side plate) when the SPR rivets two layers; when the SPR rivets three layers or more, the fitting intersection surface between the upper part and the middle part. The method of generating normals is to call the line generation command, and generate normals through the center of gravity of the SPR rivet point and the fitting surface. The normal is as follows Figure 4 shown.

[0038] Then generate a detection cylinder along this normal, such as Figure 5 The generated cylinder is the shortest cylinder that meets the conditions. The shortest cylinder that meets the conditions refers to a cylinder that covers the SPR rivet point and does not interfere with other parts (determined by interference detection). It can be understood that the length of the cylinder just meets the SPR rivet point length and can just cover the SPR rivet point model.

[0039] The process of generating a test cylinder along the normal line includes: identifying the SPR rivet point digital module size, where the size value in the normal direction is the cylinder length value, and the size value in the direction perpendicular to the normal line is the cylinder diameter value, and then generating the corresponding cylinder based on the length value and the cylinder diameter.

[0040] Determine the relationship between the detection cylinder and the fitting surface, such as Figure 6 If the intersection line of the cylinders completely covers the fitting surface, the test is considered qualified; if not, the test is considered unqualified. The method for determining whether the intersection line of the cylinders completely covers the fitting surface is to determine whether the line generated by the intersection of the cylinders and the fitting surface forms a complete circle.

[0041] S4. Check whether the distance between each SPR rivet point and the adjacent SPR rivet point is qualified. The detection method is to calculate the distance between each SPR rivet point and the adjacent SPR rivet point through the centroid coordinates of the SPR rivet point. If the calculated distance value meets the specified value, it is qualified, otherwise it is unqualified.

[0042] S5. For SPR rivet points with unqualified margins and / or unqualified point distances, their positions are captured and stored, and the screenshots are output to the software interface for users to click and view.

[0043] The screenshot process includes: like Figure 7 As shown, four equally divided normal lines are generated within the cylinder outline (the curve distribution is 0 / 0.25 / 0.5 / 0.75), and the direction is the same as the center line.

[0044] Take a screenshot of the SPR rivet point that does not meet the result, and the screenshot shows that the installation point position is sliced, such as Figure 8 、 9 As shown, the planing operation is performed along the surface formed by the normal line that does not meet the conditions and its diagonal normal line and the point centroid. Figure 10 This is the effect diagram of the cutting plane.

[0045] The present invention provides software logic for SPR rivet point recognition and SPR rivet point position detection. Through software judgment and recognition, the steps of manual SPR rivet point position detection are simplified to software operations. After the corresponding logical rules are input, the judgment and review are more efficient and accurate, which reduces the professional requirements of vehicle body design and saves time and personnel costs.

[0046] Example 2: This embodiment proposes an SPR rivet point position detection system based on coordinate recognition, such as Figure 11 Shown, including: Input module: Select the assembly digital model file that needs to be tested for SPR rivet point position, and input the model file into the software; Identification and coordinate module: The software identifies the digital model files of all SPR rivet points in the assembly digital model file, highlights the SPR rivet point information in the digital model through the input identification mark, and takes position screenshots of the highlighted information through the main view, left view, and top view to obtain the two-dimensional main view, left view, and top view. The recognition model is input to obtain the two-dimensional coordinates of the highlighted information of the main view, left view, and top view respectively, and the three two-dimensional coordinates are converted into the three-dimensional coordinates of the SPR rivet points and recorded, and the parts intersecting with each SPR rivet point are detected; Margin determination module: Checks the margin between each SPR rivet point and the intersecting part to see if it is qualified. The detection method is to generate a cylinder that covers the digital model of the intersecting part and does not interfere with other parts. It then determines whether the intersection of the cylinder and the intersecting part completely covers the fitting surface between the SPR rivet point and the intersecting part. If so, it is qualified, otherwise it is unqualified. Point distance judgment module: for each SPR rivet point, the distance between it and the adjacent SPR rivet point is detected to see if it is qualified. The detection method is to calculate the distance between each SPR rivet point and the adjacent SPR rivet point through the centroid coordinates of the SPR rivet point. If the calculated distance value meets the specified value, it is judged to be qualified, otherwise it is unqualified. Output module: For SPR rivet points with unqualified margins and / or unqualified point distances, their positions are captured and stored, and the screenshots are output to the software interface for users to click and view.

[0047] In the recognition and coordinate module, the recognition model calls the Caffe2 framework for model training. The model training includes: cleaning the two-dimensional image of the highlighted SPR rivet point and retaining the data to be detected; then performing data annotation, annotating the retained data with SPR rivet points, and then performing preprocessing operations including rotation, enhancement, and grayscale processing on the annotated two-dimensional image. The two-dimensional image set obtained after preprocessing is divided into a training set, a test set, and a validation set. The two-dimensional image is used as input and the two-dimensional coordinates of the SPR rivet point are used as the target for training.

[0048] The recognition and coordinate modules include: Obtain the X and Z coordinates from the two-dimensional coordinates of the main view, obtain the X and Y coordinates from the two-dimensional coordinates of the top view, and obtain the Y and Z coordinates from the two-dimensional coordinates of the side view; and convert the X, Y, and Z coordinates into three-dimensional coordinates.

[0049] The recognition and coordinate module also includes: According to the centroid coordinates of the SPR rivet point, the part closest to the centroid of the SPR rivet point is detected, and a normal line is generated through the centroid point and the surface of the part closest to the centroid of the SPR rivet point; Call the "Collision" command, use the normal as the detection direction, perform collision detection on the SPR rivet point and each part, and obtain the detection results based on the displayed interference results. The parts that have an interference relationship with the SPR rivet point are intersecting parts.

[0050] The margin determination module includes: Extract the surface where a certain SPR rivet point is located; obtain the fitting surface between the surface where the SPR rivet point is located and a certain intersecting part; generate a normal line through the centroid coordinates of the SPR rivet point and the fitting surface; generate a detection cylinder along the normal line, and generate the shortest cylinder that meets the conditions, wherein the condition is to cover the digital model of the intersecting part and not interfere with other parts.

[0051] The SPR rivet point position detection system based on coordinate recognition proposed in this embodiment can implement the SPR rivet point position detection method based on coordinate recognition described in Example 1, and has the same technical effect as Example 1.

[0052] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A SPR rivet point position detection method based on coordinate recognition, characterized in that: include: S1. Select the assembly digital model file that needs to be tested for SPR rivet point position and input the model file into the software; S2. The software identifies the digital model files of all SPR rivet points in the assembly digital model file, highlights the SPR rivet point information in the digital model through the input identification mark, and takes position screenshots of the highlighted information through the main view, left view, and top view to obtain the two-dimensional main view, left view, and top view, and inputs the recognition model to obtain the two-dimensional coordinates of the highlighted information of the main view, left view, and top view respectively, converts the three two-dimensional coordinates into the three-dimensional coordinates of the SPR rivet point and records them, and detects the parts intersecting with each SPR rivet point; S3. Check whether the margin between each SPR rivet point and the intersecting part is qualified. The detection method is to generate a cylinder that covers the digital model of the intersecting part and does not interfere with other parts, and determine whether the intersection of the cylinder and the intersecting part completely covers the fitting surface between the SPR rivet point and the intersecting part. If the fitting surface is covered, it is judged as qualified, otherwise it is unqualified. S4. Check whether the distance between each SPR rivet point and the adjacent SPR rivet point is qualified. The detection method is to calculate the distance between each SPR rivet point and the adjacent SPR rivet point according to the centroid coordinates of the SPR rivet point. If the calculated distance value meets the specified value, it is qualified, otherwise it is unqualified; S5. For SPR rivet points with unqualified margins and / or unqualified point distances, their positions are captured and stored, and the screenshots are output to the software interface for users to click and view.

2. The SPR rivet point position detection method based on coordinate recognition according to claim 1, characterized in that: In step S2, the recognition model calls the Caffe2 framework to perform model training. The model training includes: cleaning the two-dimensional image of the highlighted SPR rivet point and retaining the data to be detected; then performing data annotation, annotating the retained data with SPR rivet points, and then performing preprocessing operations including rotation, enhancement, and grayscale processing on the annotated two-dimensional image. The two-dimensional image set obtained after preprocessing is divided into a training set, a test set, and a validation set. The two-dimensional image is used as input and the two-dimensional coordinates of the SPR rivet point are used as the target for training.

3. The SPR rivet point position detection method based on coordinate recognition according to claim 1, characterized in that: The method of converting into three-dimensional coordinates in step S2 includes: Obtain the X and Z coordinates from the two-dimensional coordinates of the main view, obtain the X and Y coordinates from the two-dimensional coordinates of the top view, and obtain the Y and Z coordinates from the two-dimensional coordinates of the side view; and convert the X, Y, and Z coordinates into three-dimensional coordinates.

4. The SPR rivet point position detection method based on coordinate recognition according to claim 1, characterized in that: In step S2, the method for detecting parts intersecting with each SPR rivet point includes: S211, detecting the part closest to the center of gravity of the SPR rivet point according to the center of gravity coordinates of the SPR rivet point, and generating a normal line through the center of gravity point and the surface of the part closest to the center of gravity of the SPR rivet point; S212. Call the "Collision" command, use the normal as the detection direction, perform collision detection on the SPR rivet point and each part, obtain the detection result based on the displayed interference result, and the parts that have an interference relationship with the SPR rivet point are intersecting parts.

5. The SPR rivet point position detection method based on coordinate recognition according to claim 1, characterized in that: The method for generating the cylinder in step S3 includes: S301, extracting the surface where a certain SPR rivet point is located; S302, obtaining the fitting surface between the surface where the SPR rivet point is located and an intersecting part; S303, generating a normal line through the centroid coordinates of the SPR riveting point and the fitting surface; S304 , generating a detection cylinder along the normal line, wherein the cylinder generates the shortest cylinder that meets the condition that the cylinder covers the digital model of the intersecting part and does not interfere with other parts.

6. An SPR rivet point position detection system based on coordinate recognition, characterized in that: include: Input module: Select the assembly digital model file that needs to be tested for SPR rivet point position, and input the model file into the software; Identification and coordinate module: The software identifies the digital model files of all SPR rivet points in the assembly digital model file, highlights the SPR rivet point information in the digital model through the input identification mark, and takes position screenshots of the highlighted information through the main view, left view, and top view to obtain the two-dimensional main view, left view, and top view. The recognition model is input to obtain the two-dimensional coordinates of the highlighted information of the main view, left view, and top view respectively, and the three two-dimensional coordinates are converted into the three-dimensional coordinates of the SPR rivet points and recorded, and the parts intersecting with each SPR rivet point are detected; Margin determination module: Checks the margin between each SPR rivet point and the intersecting part to see if it is qualified. The detection method is to generate a cylinder that covers the digital model of the intersecting part and does not interfere with other parts. It then determines whether the intersection of the cylinder and the intersecting part completely covers the fitting surface between the SPR rivet point and the intersecting part. If so, it is qualified, otherwise it is unqualified. Point distance judgment module: for each SPR rivet point, the distance between it and the adjacent SPR rivet point is detected to see if it is qualified. The detection method is to calculate the distance between each SPR rivet point and the adjacent SPR rivet point through the centroid coordinates of the SPR rivet point. If the calculated distance value meets the specified value, it is judged to be qualified, otherwise it is unqualified. Output module: For SPR rivet points with unqualified margins and / or unqualified point distances, their positions are captured and stored, and the screenshots are output on the software interface for users to click and view.

7. The SPR rivet point position detection system based on coordinate recognition according to claim 6, characterized in that: In the recognition and coordinate module, the recognition model calls the Caffe2 framework for model training. The model training includes: cleaning the two-dimensional image of the highlighted SPR rivet point and retaining the data to be detected; then performing data annotation, annotating the retained data with SPR rivet points, and then performing preprocessing operations including rotation, enhancement, and grayscale processing on the annotated two-dimensional image. The two-dimensional image set obtained after preprocessing is divided into a training set, a test set, and a validation set. The two-dimensional image is used as input and the two-dimensional coordinates of the SPR rivet point are used as the target for training.

8. The SPR rivet point position detection system based on coordinate recognition according to claim 6, characterized in that: The recognition and coordinate modules include: Obtain the X and Z coordinates from the two-dimensional coordinates of the main view, obtain the X and Y coordinates from the two-dimensional coordinates of the top view, and obtain the Y and Z coordinates from the two-dimensional coordinates of the side view; and convert the X, Y, and Z coordinates into three-dimensional coordinates.

9. The SPR rivet point position detection system based on coordinate recognition according to claim 6, characterized in that: The recognition and coordinate modules include: According to the centroid coordinates of the SPR rivet point, the part closest to the centroid of the SPR rivet point is detected, and a normal line is generated through the centroid point and the surface of the part closest to the centroid of the SPR rivet point; Call the "Collision" command, use the normal as the detection direction, perform collision detection on the SPR rivet point and each part, and obtain the detection results based on the displayed interference results. The parts that have an interference relationship with the SPR rivet point are intersecting parts.

10. The SPR rivet point position detection system based on coordinate recognition according to claim 6, characterized in that: The margin determination module includes: Extract the surface where a certain SPR rivet point is located; obtain the fitting surface between the surface where the SPR rivet point is located and a certain intersecting part; generate a normal line through the centroid coordinates of the SPR rivet point and the fitting surface; generate a detection cylinder along the normal line, and generate the shortest cylinder that meets the conditions, wherein the condition is to cover the digital model of the intersecting part and not interfere with other parts.

Images