Engine support assembly assembling quality appearance detection method based on machine vision

Through intelligent sampling and model improvement technology based on machine vision, the problem of subjectivity and low efficiency of manual visual inspection in engine bracket assembly assembly quality inspection is solved, high accuracy and high efficiency automated inspection is achieved, and quality traceability is established. mechanism reduces labor costs.

CN119048442AActive Publication Date: 2024-11-29NANJING UNIV OF AERONAUTICS & ASTRONAUTICS

Patent Information

Application Number
CN202411077811.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-07
Publication Date
2024-11-29
Estimated Expiration
2044-08-07

AI Technical Summary

Technical Problem

In the prior art, the appearance inspection of the assembly quality of the engine bracket assembly mainly relies on manual visual inspection, and there are problems such as strong subjectivity, low efficiency, and difficulty in meeting the needs of large-scale production lines.

Method used

Using intelligent sampling based on machine vision, standard template library production, GAN model combined with data enhancement, HAT attention mechanism and Ghost convolution to improve the YOLOv9 model, to realize automated assembly quality evaluation and quality traceability, and improve detection through adaptive algorithms and continuous learning Accuracy and efficiency.

Benefits of technology

It realizes comprehensive and accurate detection of the assembly quality of the engine bracket assembly, improves the detection accuracy and efficiency, reduces labor costs, and establishes an effective traceability and continuous improvement mechanism for quality problems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048442A_ABST
    Figure CN119048442A_ABST
Patent Text Reader

Abstract

An engine support assembly assembling quality appearance detection method based on machine vision is characterized by comprising the following steps that 1, an adaptive sampling algorithm is combined with actual features, and a sampling scheme is formulated; 2, after an engine support assembly image is collected and marked, a GAN model is trained to generate a composite image, and a data set is made; 3, an attention mechanism and lightweight convolution are introduced into the YOLOv9 algorithm, and then the model is trained; 4, using a camera light source adaptive algorithm to automatically adjust camera and light source parameters, and then carrying out image acquisition; 5, comparing and analyzing a detection result with a qualified standard after target detection, and judging the assembly state of the part; 6, associating a detection result with the product to establish a quality tracing mechanism; and 7, regularly triggering a continuous learning automation process, and iteratively optimizing the model. According to the invention, the detection accuracy and efficiency of the assembly state of the parts of the engine support assembly are improved, the labor cost is reduced, and appearance detection of the assembly quality of the engine support assembly, quality problem tracing and continuous learning are realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a computer image recognition technology, in particular to a product assembly quality appearance detection technology, specifically to an engine bracket assembly quality appearance detection method based on machine vision. Background Art

[0002] The engine bracket assembly is one of the most important parts in the automobile engine system, used to support and fix the engine. It is generally composed of the following key parts: bracket body, fasteners such as bolts and nuts, and shock-absorbing elements such as rubber pads and hydraulic shock absorbers. Some engine bracket assemblies may be equipped with adjustment mechanisms, such as adjustable support arms, height adjustment bolts, etc. Appearance inspection is a crucial part of the quality control of the engine bracket assembly. This link can effectively identify assembly defects such as wrong installation, missing installation, and multiple installation of various parts in the engine bracket assembly, as well as other potential problems. If these defects and problems are not discovered and corrected in time, they will directly threaten the stability and reliability of the engine bracket, and then affect the overall safety and driving experience of the vehicle. At present, manufacturers mainly use manual visual inspection methods to conduct appearance inspection of the assembly quality of engine bracket assembly parts. However, the manual visual inspection method is highly subjective, and the inspection results are easily affected by factors such as the experience and vision of the inspector. It is also inefficient and difficult to meet the rapid inspection needs of large-scale production lines. At the same time, the labor cost is high. Summary of the invention

[0003] The purpose of the present invention is to mainly use manual visual inspection methods to detect the appearance of assembly quality of existing engine bracket assemblies. However, the manual visual inspection method is highly subjective, and the inspection results are easily affected by factors such as the inspection personnel's experience and eyesight. In addition, the efficiency is low and it is difficult to meet the rapid inspection needs of large-scale production lines. At the same time, there is the problem of labor costs. A method for appearance inspection of assembly quality of engine bracket assemblies based on machine vision is invented. It realizes comprehensive and accurate inspection of the assembly quality of engine bracket assemblies through intelligent sampling, standard template library production, data set enhancement, model improvement, assembly quality evaluation, quality traceability and continuous learning, thereby improving inspection accuracy and efficiency and reducing labor costs.

[0004] The technical solution of the present invention is:

[0005] A method for appearance inspection of assembly quality of engine bracket assembly based on machine vision is implemented through the following steps: intelligent sampling of adaptive algorithm, preparation of standard template library of engine bracket assembly, preparation of data set by GAN model combined with data enhancement technology, improvement of YOLOv9 model by HAT attention mechanism and Ghost convolution, improved model target detection, assembly quality assessment according to inspection results, quality traceability mechanism and continuous learning automation process, which realizes comprehensive and accurate appearance inspection of assembly quality of engine bracket assembly, improves the accuracy and efficiency of assembly status inspection of parts such as engine bracket, fasteners and shock absorbing elements, reduces labor costs, and realizes effective traceability of quality problems of engine bracket assembly and continuous improvement and learning of appearance inspection accuracy.

[0006] The specific steps are:

[0007] Step 1: Intelligent sampling plan formulation, using an adaptive sampling algorithm based on surface curvature and part density, taking into account camera parameters and combining the engine bracket assembly CAD model or actual scan data, adding sampling points A for key parts assembly areas, parts assembly dense areas and complex surfaces i , formulate an intelligent sampling scheme A(A0,A1,...A m ).

[0008] Step 2: Establish a standardized template library. Establish a standard template library for engine bracket assemblies that includes various models, batches, and production lines. Each template contains detailed information such as the size (W, H), position (U, V), and part type C of the engine bracket assembly. The i-th sampling point A i The jth part parameter is denoted as T ij (C ij ,(W ij ,H ij ),(U ij ,V ij )), when sampling point A i When the total number of parts is n, the sampling point A i The standard assembly template is denoted as N i (T i1 , T i2 , ... T in ). At the same time, the template update tool was introduced to allow production personnel to fine-tune the template according to actual production conditions to ensure synchronization between the template and the actual product.

[0009] Step 3: Dataset creation: collect and annotate images of engine bracket assembly parts under multiple lighting, multiple angles, and multiple backgrounds. Use annotated real images to train the GAN model, and use the GAN model to generate a large number of high-quality synthetic images, merge these images into the dataset, and annotate them.

[0010] Step 4: Data enhancement of the training data set. Data enhancement methods such as image rotation, scaling, cutting, color jittering and Gaussian filtering are used to enhance the data set.

[0011] Step 5: Model improvement. Based on the YOLOv9 algorithm, the HAT attention mechanism module is introduced to extract comprehensive features of parts, and the Ghost convolution is introduced to replace some of the original convolution modules to reduce the computational complexity of the model.

[0012] Step 6: Model training: Continuously adjust the hyperparameters of the target detection model training through the hyperparameter evolution strategy until the model recognition and classification accuracy is optimal.

[0013] Step 7: Image acquisition, using the camera light source adaptive algorithm to automatically adjust the camera and light source parameters according to image features and lighting conditions.

[0014] Step 8: YOLOv9 improved model target detection: Input the collected image into the YOLOv9 improved model for part target detection classification and pixel coordinate positioning. Record the detection result R of each part at each sampling point, where R includes part type RC, part pixel size (RW, RH), part pixel coordinate system positioning (RU, RV) and other part parameters. The i-th sampling point A i The test result is recorded as M i (R i1 , R i2 , ... R in ).

[0015] Step 9: Assembly quality assessment, compare the test result M with the standard template T, compare the part size, position and type, and calculate the relevant error in combination with the custom assembly status formula to determine whether the part assembly is qualified. Introduce data visualization technology to produce a detailed and intuitive appearance quality inspection report.

[0016] Step 10: Quality traceability, link the results of each test with product information and establish a quality traceability mechanism. When quality problems occur, they can be quickly traced back to the specific production links and causes.

[0017] Step 11: Continuous learning, regularly triggering the continuous learning automation process of collecting actual data, evaluating performance, updating parameters and retraining, so that the YOLOv9 improved model can be continuously iterated and optimized to gradually adapt to various complex scenarios and target types.

[0018] In the step 1: Surface curvature: Curvature reflects the degree of curvature of the surface. The greater the curvature, the more complex the surface. Part density: represents the number of parts in the area near the surface. The adaptive sampling algorithm is mainly divided into the following steps: 1. Data input: The adaptive sampling algorithm first needs to obtain the geometric data of the engine bracket assembly, including the CAD model file. Actual scanning data (such as point cloud data) or other forms of geometric representation. These data should contain the detailed size, position, shape information of the engine bracket assembly and its parts, and whether they are key parts. 2. Sampling point initialization: The algorithm preliminarily determines the number and approximate distribution of sampling points based on the overall size and complexity of the geometric data of the engine bracket assembly. 3. Sampling point adjustment based on the assembly area of ​​key parts: According to the geometric data, add sampling points to the part area where key parts exist. 4. Curvature-based sampling point adjustment: The algorithm defines the grid parameters according to the geometric data of the engine bracket assembly, performs semi-automatic grid division on the geometric data, and after passing the grid quality check, uses numerical analysis or geometric processing algorithms to calculate the curvature of the surface grid points of the engine bracket assembly, and adds sampling points on the surface grid points with larger curvature. 5. Adjustment of sampling points based on part density: Calculate the relative position, overlap or occupied space between parts to achieve part density calculation, and add sampling points in areas with high part density.

[0019] In step 2, the order of part records in the standard assembly template N of each sampling point is arranged in ascending order according to the x-axis coordinates of the part pixels. If the x-axis coordinates are the same, they are arranged in ascending order according to the y-axis coordinates. The part types C are represented by mapping with increasing positive integers. Each positive integer C a Represents a type of part.

[0020] In step 3: at least 100 images are collected for each part classification in the produced data set, and at least 50 images are collected for each sampling point. For color detection requirements, some images of other colors need to be introduced into the data set to ensure that the data set contains all light backgrounds and images of detection conditions. The side images and surface images of the same part are each classified into one category. Different fasteners and shock-absorbing components are divided into multiple categories according to their types and characteristics, and different holes and slots without bolts are classified according to their characteristics.

[0021] In the step 5: the YOLOv9 improved model network structure parameter information table is shown in Table 1, where HAT represents the introduced hybrid attention mechanism HAT (Hybrid Attention Transformer), and GS_Conv represents lightweight Ghost convolution.

[0022] Table 1 YOLOv9 improved model network structure parameter information table

[0023]

[0024]

[0025] In step 6, the hyperparameter evolution strategy mainly adjusts the values ​​of hyperparameters such as the model batch size setting value, the initial learning rate, and the training cycle (epoch) value. The hyperparameter evolution strategy is implemented by a genetic algorithm, which imitates the selection, crossover, and mutation process in biological evolution. The best performing part is selected from multiple hyperparameter combinations, and they are crossed to generate a new hyperparameter combination, and then some hyperparameter values ​​are mutated in the new hyperparameter combination, so as to continuously evolve the hyperparameter combination and explore the optimal solution, and generate the best precision part assembly detection and positioning model with the highest average accuracy of all classifications.

[0026] In step 7: the specific implementation process of the camera light source adaptive algorithm is as follows: 1. According to the sampling point A i The standard template information sets the camera parameters: including exposure time, white balance, focal length, etc. i 1. Set the light source parameters according to the standard template information: including light intensity, direction, color temperature, etc. 2. The current image obtained from the camera. 2. Extract features of the captured image, including key features such as brightness, contrast, and clarity, and use image processing techniques (such as histogram analysis, edge detection, etc.) to quantify these features. 3. Evaluate the current lighting conditions through image analysis, including light intensity, direction, and color temperature. 4. Develop an adjustment strategy for camera and light source parameters based on the extracted image features and lighting conditions. 5. Adjust parameters through the API or control interface of the camera and light source. 6. Evaluate the quality of the adjusted image. Image quality assessment algorithms (such as SSIM, PSNR, etc.) can be used to quantify the image quality to ensure that the image quality meets the requirements. 7. Iterate and optimize. If the image quality does not meet the requirements, re-extract image features and evaluate lighting conditions, and adjust parameters.

[0027] In step 9: Assembly quality assessment, the test result M is compared with the standard template T, and the relevant errors are calculated by comparing the part size, position and type, combined with the custom assembly state formula to determine whether the part assembly is qualified. Data visualization technology is introduced to produce a detailed and intuitive appearance quality inspection report.

[0028] Since the sampling points established in step 1 are fixed, the surface positions taken are the same, so according to the standard assembly template N i , for each part's standard assembly template T ij (C ij ,(W ij ,H ij ),(U ij , V ij )) and test results Mi The actual assembly status of each part in R ij (RC ij ,(RW ij ,RH ij ),(RU ij , RV ij ) for comparison. When the custom assembly state formula (1) is satisfied, the jth part of the i-th sampling point is considered to be assembled properly. ij =1, otherwise the part is considered unqualified DR ij =0.

[0029]

[0030] Where i represents the sampling point number in step 1, and j represents the number of the part record in N in step 2. i∈[1,m],j∈[1,n](i,j∈N). α, β, γ, δ are all initial preset constants, representing the allowable error constants, where α represents the allowable error of the part width, β represents the allowable error of the part height (length), γ represents the allowable error of the pixel coordinate of the center of the part on the x-axis, and δ represents the allowable error of the pixel coordinate of the center of the part on the y-axis.

[0031] After the comparison is completed, proceed to the next sampling point A i+1 The image acquisition and target detection are performed until all sampling points are detected, and the assembly information of the parts detected at all sampling points is summarized. Through visualization techniques such as image overlay annotation, thermal map and error map, data table and statistical chart, an interactive report on the appearance quality inspection of the engine bracket assembly is produced.

[0032] In step 10: the quality traceability mechanism, first of all, the appearance inspection is used to store a database for product information, test results and evaluation results, wherein the product information includes unique identifiers such as batch number, serial number, production date, production line number, etc., which are used to uniquely identify each engine bracket assembly. After each inspection is completed, the test results and related product information are automatically entered into the database. The product information, test records and result evaluation table are associated through unique identifiers to form a complete quality traceability chain. Parts or assemblies that fail the inspection are specially marked so that they can be focused on in subsequent production or quality control processes. A data query interface is provided to allow quality management personnel to quickly retrieve relevant quality traceability information based on product information or test records.

[0033] In step 11: the specific steps of the continuous learning automation process include: regularly collecting and labeling actual engine bracket assembly image data on the production line, deploying the model to test and analyze on the actual collected data set, adjusting parameters and retraining the model based on the analysis, testing and verifying the performance of the model after learning optimization on the validation set, and deploying it to the production environment if the model performance is better, logging the entire process, executing the above steps in a timed loop, and regularly supervising and assisting manually.

[0034] The beneficial effects of the present invention are:

[0035] The present invention ensures that key assembly areas and complex surfaces are fully sampled by formulating an intelligent sampling scheme, thereby improving the accuracy and reliability of detection. The establishment of a standardized template library ensures that the detection standard is synchronized with the actual product, and improves the versatility and flexibility of detection. The introduction of the GAN model and data enhancement technology when making the data set can effectively solve the problem of insufficient image data diversity caused by lighting conditions, shooting angles and background differences. On the basis of the YOLOv9 algorithm, the HAT attention mechanism module is introduced and lightweight Ghost convolution is used. The HAT attention mechanism combines the advantages of self-attention, channel attention and spatial attention, and can more comprehensively capture and utilize key information in the image, so that the model can more accurately locate and identify the assembly state of parts, and improve the accuracy and reliability of detection. Ghost convolution is an efficient convolution method. It reduces the complexity and calculation amount of the model by reducing redundant calculations in the convolution operation. The application of Ghost convolution in YOLOv9 can enable the model to achieve faster reasoning speed while maintaining high-precision detection, meeting real-time requirements. The image acquisition camera light source adaptive algorithm is used to effectively solve the problem of poor image quality caused by changes in lighting conditions and improper camera parameter settings. The quality traceability mechanism can effectively solve the problem of difficulty in tracing quality issues. Continuous learning improves the adaptability, accuracy and generalization ability of the engine bracket assembly appearance inspection model, reduces manual intervention, and supports continuous optimization and improvement of the model. In summary, the present invention comprehensively improves the inspection efficiency and accuracy of the assembly quality of the engine bracket assembly through a series of innovative technical means, reduces the inspection cost, and establishes a complete quality traceability and continuous optimization mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is the initial configuration flow chart of the present invention.

[0037] Figure 2 The invention discloses an engine bracket assembly product detection workflow diagram. DETAILED DESCRIPTION

[0038] The present invention is further described in detail below with reference to the accompanying drawings and implementation examples.

[0039] like Figure 1 and Figure 2 shown.

[0040] A machine vision-based appearance inspection method for the assembly quality of an engine bracket assembly is mainly composed of two parts. The first part is the initial configuration of the method, and the second part is the inspection of the engine bracket assembly product. First, an intelligent sampling scheme is formulated based on the geometric data of the engine bracket assembly using an adaptive sampling algorithm based on surface curvature and part density, and a standard template library of engine bracket assemblies containing multiple models, batches, and production lines is established. Then, images of engine bracket assembly parts under multiple illuminations, multiple angles, and multiple backgrounds are collected and annotated, a GAN model is trained, and synthetic images are produced using the GAN model. The training data set is merged and other data enhancement operations are performed on the data set. Then, the YOLOv9 model improved by the HAT attention mechanism and lightweight convolution is trained on this data set. After completing the above initial configuration work, the actual inspection process of the engine bracket assembly product can be entered. The second part of the engine bracket assembly product inspection work first uses the camera light source adaptive algorithm to automatically adjust the camera and light source parameters to perform image acquisition, and then uses the YOLOv9 improved model for all images to detect and identify parts and locate pixel coordinates. Then, the inspection results are compared with the standard template parameters, and the customized assembly status formula is used to determine whether the parts assembly is qualified. Finally, the data visualization technology is introduced to produce a detailed and intuitive appearance quality inspection report. Then, based on the quality traceability mechanism, the results of each inspection are associated with the product information and stored in the database. At the same time, the continuous learning automation process of collecting actual data, evaluating performance, updating parameters, and retraining is triggered regularly.

[0041] Figure 1 The flowchart of the initial configuration work of the method of the present invention specifically includes steps 1-6:

[0042] Step 1: Formulate an intelligent sampling scheme, using an adaptive sampling algorithm based on surface curvature and part density. According to the overall size and complexity of the actual geometric data of a certain type of engine bracket assembly, preliminarily determine the number and approximate distribution of sampling points, and add sampling points to the key parts assembly area based on the part information in the geometric data. Then, define the grid parameters based on the geometric data of the engine bracket assembly, and perform semi-automatic grid division. After the grid quality check, use numerical analysis or geometric processing algorithms to calculate the curvature of the surface grid points of the engine bracket assembly, and add sampling points to the surface grid points with larger curvature. Then, calculate the relative position, overlap or occupied space between parts to achieve part density acquisition, and add sampling points in areas with high part density. Finally, formulate an intelligent sampling scheme A (A0, A1, ... A m ).

[0043] Step 2: Establish a standardized template library. Establish a standard template library for engine bracket assemblies that includes various models, batches, and production lines. Each template contains detailed dimensions (W, H), positions (U, V), and part types C of the engine bracket assembly at a sampling point. i The jth part parameter is denoted as T ij (C ij ,(W ij ,H ij ),(U ij ,V ij )), when sampling point A i When the total number of parts is n, the sampling point A i The standard assembly template is denoted as N i (T i1 , T i2 , ... T in ). The order of part records in the standard assembly template N of each sampling point is arranged in ascending order according to the x-axis coordinates of the part pixels. If the x-axis coordinates are the same, they are arranged in ascending order according to the y-axis coordinates. The type of part C is represented by an increasing positive integer mapping. Each positive integer C a Represents a type of part. At the same time, the template update tool is introduced to allow production personnel to fine-tune the template according to actual production conditions to ensure that the template information is synchronized with the qualified assembly standards of the sampling point products.

[0044] Step 3: Dataset creation, collect images of engine bracket assembly parts under multiple lighting, multiple angles and multiple backgrounds, and annotate them. Use annotated real images to train the GAN model, and use the GAN model to generate a large number of high-quality synthetic images, merge these images into the dataset and annotate them. In the created dataset, at least 100 images are required for each part classification, and at least 50 images are collected for each sampling point. For those with color detection requirements, some images of other colors need to be introduced into the dataset to ensure that the dataset contains images of all light backgrounds and detection conditions. The side images and surface images of the same part are classified into one category each. Different fasteners and shock-absorbing elements are divided into multiple categories according to their types and characteristics, and are classified accordingly according to the characteristics of different holes and slots without bolts.

[0045] Step 4: Data enhancement of the training data set. Data enhancement methods such as image rotation, scaling, cutting, color jittering and Gaussian filtering are used to enhance the data set.

[0046] Step 5: Model improvement. Based on the YOLOv9 algorithm, the HAT attention mechanism module is introduced to extract comprehensive features of parts, and the Ghost convolution is introduced to replace some of the original convolution modules to reduce the amount of model calculation. The detailed YOLOv9 improved model network structure parameter information table is shown in Table 1.

[0047] Step 6: Model training, through the hyperparameter evolution strategy, continuously adjust the values ​​of hyperparameters such as the model batch size setting value, the initial learning rate, and the training cycle (epoch) value until the model generates an engine bracket assembly detection model with the highest average accuracy for all classifications.

[0048] Figure 2 This is a flow chart of the engine bracket assembly product detection process of the present invention, which specifically includes steps 7-11:

[0049] Step 7: Image acquisition, using the camera light source adaptive algorithm, first according to the sampling point A i The standard template information sets the camera parameters: including exposure time, white balance, focal length, etc., and then according to the sampling point A i The standard template information sets the light source parameters: including light intensity, direction, color temperature, etc. Then use the camera to capture images. Then extract features from the captured image, including key features such as brightness, contrast, and clarity, and use image processing techniques (such as histogram analysis, edge detection, etc.) to quantify these features. Next, use image analysis to evaluate the current lighting conditions, including light intensity, direction, and color temperature. According to the extracted image features and lighting conditions, formulate an adjustment strategy for camera and light source parameters. Then use the API or control interface of the camera and light source to adjust the parameters, and evaluate the quality of the adjusted image. Image quality evaluation algorithms (such as SSIM, PSNR, etc.) can be used to quantify the image quality to ensure that the image quality meets the requirements. Finally, if the image quality does not meet the requirements, re-extract the image features and evaluate the lighting conditions, adjust the parameters, and iterate the optimization.

[0050] Step 8: YOLOv9 improved model target detection: Input the preprocessed image into the YOLOv9 improved model for part target detection classification and pixel coordinate positioning. Record the detection result R of each part at each sampling point, where R includes part type RC, part pixel size (RW, RH), part pixel coordinate system positioning (RU, RV) and other part parameters. The i-th sampling point A i The test result is recorded as M i (R i1 , R i2 , ... R in ). The recording order of R in M ​​is the same as the template recording standard.

[0051] Step 9: Since the sampling points established in step 1 are fixed, the surface positions taken are the same, so according to the standard assembly template N i , for each part's standard assembly template T ij (C ij ,(W ij ,H ij ),(U ij , V ij )) and test results M i The actual assembly status of each part in R ij (RC ij ,(RW ij ,RH ij ),(RU ij , RV ij ) for comparison. When the custom assembly state formula (1) is satisfied, the jth part of the i-th sampling point is considered to be assembled properly. ij =1, otherwise the part is considered unqualified DR ij =0.

[0052]

[0053] After the comparison is completed, proceed to the next sampling point A i+1 The image acquisition and target detection are performed until all sampling points are detected, and the assembly information of the parts detected at all sampling points is summarized. Through the visualization technology such as image overlay annotation, thermal map and error map, data table and statistical chart, an interactive report on the appearance quality inspection of the engine bracket assembly is produced. The step jump is as follows:

[0054] if(i <m):

[0055] Save sampling point A i Parts inspection information, enter sampling point A i+1 The image acquisition step is skipped to step 7.

[0056] else:

[0057] After making the interactive report on the appearance quality inspection of the engine bracket assembly, the step jumps to step 10.

[0058] Step 10: Quality traceability. First, create a database for storing product information, test results, and evaluation results. The product information includes unique identifiers such as batch number, serial number, production date, and production line number, which are used to uniquely identify each engine bracket assembly. After each test is completed, the test results and related product information are automatically entered into the database. The product information, test records, and result evaluation table are associated through unique identifiers to form a complete quality traceability chain. Parts or assemblies that fail the test are specially marked so that they can be focused on in subsequent production or quality control processes. Provide a data query interface to allow quality management personnel to quickly retrieve relevant quality traceability information based on product information or test records and establish a quality traceability mechanism.

[0059] Step 11: Continuous learning: Regularly collect and annotate actual engine bracket assembly image data on the production line, then test and analyze the deployed model on the actual collected data set, adjust parameters and retrain the model based on the analysis, test the performance of the model after learning and optimization on the validation set, and deploy it to the production environment if the model performance is better. Keep a log of the entire process, execute the above steps in a timed loop, regularly supervise and assist manually, and iterate and optimize the model performance, so that the YOLOv9 improved model can gradually adapt to various complex scenarios and target types.

[0060] The present invention does not involve parts that are the same as the prior art or can be implemented by using the prior art.

Claims

1. A method for appearance inspection of engine bracket assembly quality based on machine vision, characterized by: It includes the following steps: Step 1: Develop an intelligent sampling plan: Adopt an adaptive sampling algorithm, combine the camera hardware parameter analysis and the engine bracket assembly CAD model or actual scan data, determine the sampling point location, and develop an intelligent sampling plan; Step 2: Establish a standardized template library: Establish a standard template library for engine bracket assemblies that includes multiple models, batches, and production lines, and introduce a template update tool to ensure that the template is synchronized with the actual product; Step 3: Training dataset preparation: Collect and annotate images of the engine bracket assembly under multiple lighting, multiple angles and backgrounds, train the GAN model to generate synthetic images, and then merge them into a dataset and re-annotate them; Step 4: Data enhancement of training data set: Use image rotation, scaling, cutting, color jittering and Gaussian filtering data enhancement methods to enhance the data set; Step 5: Model improvement: Introduce attention mechanism and lightweight convolution to improve the model based on YOLOv9; Step 6: Model training: The hyperparameters of the target detection model training are continuously adjusted through the hyperparameter evolution strategy until the model recognition and classification accuracy is optimal, and the optimal accuracy part assembly detection and positioning model is generated; Step 7: Image acquisition: Use the camera light source adaptive algorithm to automatically adjust the camera and light source parameters; Step 8: Target detection: Input the preprocessed image into the YOLOv9 improved model for part target detection classification and pixel coordinate positioning; Step 9: Assembly quality assessment: Compare the test results with the standard template, determine whether the assembly is qualified based on the custom assembly status formula, and introduce data visualization technology to produce a detailed and intuitive appearance quality inspection report; Step 10: Continuous learning: Regularly trigger the continuous learning automation process of collecting actual data, evaluating performance, updating parameters and retraining; Step 11: Quality traceability: Link the results of each test with product information and establish a quality traceability mechanism.

2. The method according to claim 1, characterized in that The adaptive sampling algorithm described in step 1 is specifically an adaptive sampling algorithm based on surface curvature and part density. It needs to be combined with the engine bracket assembly CAD model or actual scanning data, and then consider the camera's field of view, resolution, focal length, and core hardware parameters. The location of sampling points is optimized and added for key parts assembly areas, parts assembly dense areas, and complex surfaces, thereby formulating an intelligent sampling plan.

3. The method according to claim 1, characterized in that The templates described in step 2 are specifically that each template corresponds to standard qualified assembly information such as detailed size, position and assembly type of parts in the sampling points of the engine bracket assembly.

4. The method according to claim 1, characterized in that: The introduction of the attention mechanism module described in step 5 is specifically to introduce the HAT attention mechanism (Hybrid Attention Transformer) that combines the advantages of multiple attention mechanisms. The attention mechanism includes self-attention, channel attention and spatial attention.

5. The method according to claim 1, characterized in that The lightweight convolution described in step 5 is specifically to replace part of the convolution in YOLOv9 with lightweight Ghost convolution.

6. The method according to claim 1, characterized in that Step 9 describes comparing the test results with the standard template, specifically, by comparing the part size, position and type, and calculating the relevant errors in combination with the custom assembly status formula to determine whether the part assembly is qualified.

7. The method according to claim 1, characterized in that The quality traceability mechanism in step 10 is specifically to use a database to store product information, test records, and result evaluation; the product information should include a unique identifier of batch number, serial number, production date, and production line number; After each test is completed, the test results and related product information are automatically entered into the database; at the same time, the parts or assemblies that fail the test are specially marked, and the specific reasons and manifestations of the failure are recorded; the test results include images, coordinate positions, and test classifications.

8. The method according to claim 1, characterized in that The continuous learning automation process described in step 11 specifically includes regular actual image collection and annotation, model deployment on actual data sets for testing and analysis, parameter update and model retraining, model verification, and log recording. The above cycle is triggered regularly with regular manual supervision and assistance.

Citation Information

Patent Citations

  • Lightweight multi-scale aluminum profile surface defect detection method

    CN117911399A

  • Power transmission line construction machinery inspection method based on MDALF-Net

    CN117911862A

  • Method for detecting assembly state of aero-engine

    CN117994237A

  • General target detection method for adaptive attention guidance mechanism

    WO2021139069A1

  • Real-time traffic flow detection system and method based on ghost convolutional feature fusion neural network

    WO2021238019A1

Cited By

  • Copper wire quality online detection method and system

    CN120369742A

  • A method and system for online detection of copper wire quality

    CN120369742B