Soldering tin optimization method and system based on AOI detection

Abstract

The invention relates to the technical field of soldering tin optimization, and discloses a soldering tin optimization method and system based on AOI detection, and the method comprises the steps: obtaining the AOI temperature image information of a preheating stage; the AOI temperature image information is input into a preset defect comparison model, multiple defect comparison images are preset in the defect comparison model, and the defect comparison images refer to the AOI temperature image information when the temperature of the corresponding preheating stage is abnormal; based on the defect comparison model, comparing the AOI temperature image information with various defect comparison images to obtain a comparison result; obtaining a temperature defect area according to the comparison result; based on the temperature defect area and preset standard temperature setting, a current temperature difference value is obtained, and temperature parameter setting in the welding stage is adjusted according to the current temperature difference value; the method has the effect of improving the recognition accuracy of the temperature defect of the welding spot.

Description

Technical Field

[0001] This application relates to the technical field of solder optimization, and in particular to a solder optimization method and system based on AOI detection. Background Technology

[0002] With the rapid development of the electronics manufacturing industry, people's demand for electronic products is becoming increasingly strong. The production of electronic products usually requires the use of a large number of circuit boards, and the soldering process is a key factor in the quality of circuit boards. Precise control of soldering temperature is of great significance to ensure soldering quality. Traditional soldering temperature control methods mainly rely on manual adjustment, which has the problems of low accuracy and low efficiency.

[0003] Currently, many printed circuit board (PCB) manufacturers use AOI (Automated Optical Inspection) equipment and image processing technology to inspect solder temperature and quality, which can greatly improve the detection rate of solder joint quality problems. However, in actual production, due to the high soldering temperature and the significant differences in the types of components on different PCB products, directly adjusting the solder temperature because of quality defects in some solder joints may lead to other unpredictable consequences, such as damage to some components caused by increased solder temperature. On the other hand, some temperature defects in solder joints may affect product quality, while others may not. Blindly addressing temperature defects in solder joints that do not affect product quality may actually create new product quality defects. Therefore, the aforementioned technologies suffer from low accuracy in identifying solder joint temperature defects. Summary of the Invention

[0004] To improve the accuracy of identifying temperature defects in solder joints, this application provides a solder optimization method and system based on AOI detection.

[0005] The first objective of this invention is achieved by the following technical solution: Solder optimization methods based on AOI inspection include: Acquire AOI temperature image information during the preheating stage; The AOI temperature image information is input into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. Based on the defect comparison model, the AOI temperature image information and various defect comparison images are compared to obtain the comparison results; Based on the comparison results, the temperature defect region is obtained; Based on the temperature defect area and the preset standard temperature setting, the current temperature difference is obtained, and the temperature parameter settings for the welding stage are adjusted according to the current temperature difference.

[0006] By employing the above technical solution, AOI temperature image information during the preheating stage is obtained, including image information combining product image and temperature information. Thus, the AOI temperature image information can show both the surface condition and temperature of the product, or the temperature of different parts of the product through heat distribution. Furthermore, by comparing the product's temperature under normal conditions with the current temperature, the internal heat distribution of the product can be determined, for example, whether there are overheated or underheated areas. The defect comparison model has multiple preset defect comparison images, because different temperature defects will lead to different quality problems in the product. Therefore, the defect comparison model can filter out defect comparison images through big data analysis combining quality problems and temperature information. Therefore, temperature defects on product images that could lead to quality problems can be identified by comparing AOI temperature image information with various defect comparison images, improving the accuracy of temperature defect identification. This allows for targeted temperature adjustments to address temperature defects that could cause quality problems, thus improving the accuracy of temperature control. After obtaining the temperature defect area, the temperature of subsequent welding stages needs to be adjusted based on the temperature defect area. The preset standard temperature setting represents the temperature adjustment value for the welding stage to address different temperature defects in the preheating stage. Therefore, by combining the preset standard temperature setting, obtaining the current temperature difference, and then adjusting the temperature parameter settings for the welding stage based on the current temperature difference, the accuracy of temperature adjustment is ensured, thereby improving the accuracy of identifying temperature defects at the weld joint.

[0007] In a preferred embodiment, prior to acquiring the AOI temperature image information during the preheating stage, this application includes: A key defect information extraction interface is used to obtain different product temperature defect information, corresponding quality problem information, and corresponding image information. A preliminary machine learning model is established to analyze the temperature of the preheating stage corresponding to quality problems caused by temperature defects in different products. The key defect information interface is connected to the preliminary machine learning model. Based on the key defect information interface, the different product temperature defects, corresponding quality problem information and corresponding image information are obtained. The preliminary machine learning model is then trained to obtain a defect comparison model.

[0008] By adopting the above technical solution, since different temperature defects can lead to different quality problems in products, a key defect information interface is first extracted. This interface can acquire different product temperature defect information and corresponding image information. A preliminary machine learning model is used to analyze the temperature of the preheating stage corresponding to the quality problems caused by different product temperature defects. That is, it analyzes the quality problems caused by different product temperature defects and the specific temperature of the corresponding product during the preheating stage, thereby determining the product quality problems caused by different temperature defects more accurately. Therefore, the key defect information interface is connected to the preliminary machine learning model. Different product temperature defects and corresponding image information are acquired through the key defect information interface, thereby training the preliminary machine learning model. This allows the preliminary machine learning model to filter out defect comparison images, resulting in a defect comparison model. This ensures the accuracy of identifying product temperature defects. The defect comparison model can filter out defect comparison images. Thus, normal temperature defects and temperature defects that do not affect product quality on the product image can be identified by comparing AOI temperature image information and multiple defect comparison images, improving the accuracy of identifying product temperature defects. This allows for targeted temperature adjustments for normal temperature defects, improving the accuracy of temperature control.

[0009] In a preferred embodiment, this application further includes, prior to the interface for extracting critical defect information: The interface for extracting product defect images and the interface for extracting quality scoring information are provided. The product defect image interface is used to obtain various product quality problem information, corresponding product temperature defect information, and corresponding image information. The quality scoring information interface is used to obtain quality scoring information for products with different quality problems. A secondary machine learning model is established to analyze the training availability of image information corresponding to quality problems of different products. The product defect image interface and the quality score information interface are connected to the secondary machine learning model to obtain quality problem information, product temperature defect information and corresponding image information of different products, as well as the corresponding quality scores of products with different quality problems. The secondary machine learning model is then trained in combination with a preset first score threshold to obtain a quality analysis model. The quality analysis model is used to filter temperature defect images used for training the primary machine learning model.

[0010] By adopting the above technical solution, the product defect image interface is used to obtain information on different product quality problems, corresponding product temperature defect information, and corresponding image information. The quality score information interface is used to obtain the quality score information of the product corresponding to the information obtained by the product defect image interface. Since the quality score reflects the degree of impact of the quality problem caused by the temperature defect on the actual use requirements of the product, a high quality score indicates that the corresponding temperature defect will not have a high impact on the use of the product, while a low quality score indicates that the corresponding temperature defect will have a high impact on the use of the product. Therefore, the secondary machine learning model can learn from the product quality score and the quality problem, and based on the requirements of the quality problem represented by the current market demand, determine the defect image that meets the market demand and has the ability to affect the actual use of the product. This generates a key defect information interface, which facilitates the defect comparison model to obtain image information that is more suitable for analysis and learning.

[0011] In a preferred embodiment of this application: obtaining the temperature defect region based on the comparison result includes: If the comparison result indicates that the AOI temperature image information corresponds to the defect comparison image, then the temperature defect region is obtained based on the defect comparison image; If the comparison result indicates that the AOI temperature image information does not correspond to any of the various defect comparison images, then based on the defect comparison model, the temperature defect information of the product is predicted, and the temperature defect area is obtained according to the prediction result.

[0012] By adopting the above technical solution, when comparing AOI temperature image information with defect comparison images, there may be situations where they correspond and situations where they do not. When the AOI temperature image information corresponds to the defect comparison image, it means that the temperature defect of the current product is consistent with the temperature defect shown in the defect comparison image. Therefore, the temperature defect in the defect comparison image can be taken as the temperature defect of the current product. This facilitates understanding the quality problem corresponding to the temperature defect based on the temperature defect in the defect comparison image, thereby enabling temperature control by combining the temperature defect with the corresponding quality problem. This improves the success rate of temperature control in solving the corresponding quality problem and enhances the accuracy of temperature control. However, when the AOI temperature image information and the defect comparison image are not consistent, the temperature defect in the AOI temperature image information may not correspond to the temperature defect in the defect comparison image. When the comparison images do not correspond, it indicates that the temperature defect of the current product may be quite unique. It is unknown whether the quality problem caused by the temperature defect on the current product will affect the actual use of the product. Therefore, the temperature defect of the product is predicted by the defect comparison model. In addition to the preset multiple defect comparison images, the defect comparison model also includes a large number of other defect comparison images. The defect comparison images are obtained by the defect comparison model through analysis of the current quality score. Thus, the prediction results can be obtained by comparing with a large number of other defect comparison images, and the temperature defect area can be obtained. This allows the temperature defect area of ​​the product to be found even if there is no correspondence between the AOI temperature image information and the defect comparison image, which is convenient for subsequent adjustment measures.

[0013] In a preferred embodiment of this application, obtaining the temperature defect region based on the prediction result includes: The product temperature defect information represented by the prediction results is input into the quality analysis model, and the quality score corresponding to the product temperature defect information represented by the prediction results is determined based on the quality analysis model. The quality score corresponding to the product temperature defect information represented by the prediction result is compared with a preset second scoring threshold. If the quality score corresponding to the product temperature defect information represented by the prediction result is greater than the second scoring threshold, the product temperature defect information represented by the prediction result is taken as a temperature defect area. If the quality score corresponding to the product temperature defect information represented by the prediction result is not greater than the second scoring threshold, the product temperature defect information represented by the prediction result is adjusted according to the quality problem information corresponding to the product temperature defect information represented by the prediction result to generate a temperature defect area.

[0014] By adopting the above technical solution, since it is unknown whether the quality problems caused by temperature defects in the current product will affect the actual use of the product, the product defect image represented by the prediction result is input into the quality analysis model. The quality analysis model determines the quality score corresponding to the product defect image represented by the prediction result, and then compares the quality score corresponding to the product defect image represented by the prediction result with a preset second scoring threshold. The second scoring threshold is higher than the first scoring threshold because the first scoring threshold is used by the quality analysis model to select defect images for training the initial machine learning model. Model training requires a large amount of data. If the first scoring threshold is set too high, fewer defect images may be selected, and the initial machine learning model will further select defect comparison images. Therefore, the quality analysis model can set a lower first scoring threshold to obtain defect images for training. A second screening is then performed by a primary machine learning model, which is equivalent to raising the first scoring threshold. Therefore, the first scoring threshold is relatively low, possibly lower than the current normal quality score, while the second scoring threshold is relatively high, possibly equal to the normal quality score. If the quality score corresponding to the product defect image represented by the prediction result is greater than the second scoring threshold, the product defect image represented by the prediction result can be directly regarded as a temperature defect area. If the quality score corresponding to the product defect image represented by the prediction result is not greater than the second scoring threshold, the specific temperature defect situation can be determined based on the quality problem corresponding to the product temperature defect image represented by the prediction result, combined with other factors other than the quality score that may cause quality problems to affect the actual use of the product. This allows for adjustment of the product temperature defect represented by the prediction result, thereby obtaining more accurate product temperature defect information.

[0015] The second objective of this application is achieved by the following technical solution: The AOI-based solder optimization system, applied to any of the above-described AOI-based solder optimization methods, includes: The image acquisition module is used to acquire AOI temperature image information during the preheating stage; The model input module is used to input the AOI temperature image information into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. The image comparison module is used to compare the AOI temperature image information and multiple defect comparison images based on the defect comparison model to obtain a comparison result; The defect determination module is used to obtain the temperature defect area based on the comparison results; The temperature adjustment module is used to obtain the current temperature difference based on the temperature defect area and the preset standard temperature setting, and adjust the temperature parameter settings for the welding stage according to the current temperature difference.

[0016] In a preferred embodiment of this application, the image acquisition module further includes: The information interface extraction submodule is used to extract the key defect information interface, which is used to obtain different product temperature defect information, corresponding quality problem information and corresponding image information. The first-order model building submodule is used to build a preliminary machine learning model. The preliminary machine learning model is used to analyze the temperature of the preheating stage corresponding to the quality problems caused by temperature defects in different products. The key defect information interface is connected to the preliminary machine learning model. Based on the key defect information interface, the different product temperature defects, the corresponding quality problem information and the corresponding image information are obtained. The preliminary machine learning model is trained to obtain a defect comparison model.

[0017] In a preferred embodiment of this application, the defect determination module further includes: The defect comparison image channel determination submodule is used to obtain the temperature defect region based on the defect comparison image if the comparison result indicates that the AOI temperature image information corresponds to the defect comparison image. The defect comparison model channel determination submodule is used to predict the temperature defect information of the product based on the defect comparison model if the comparison result indicates that the AOI temperature image information does not correspond to any of the various defect comparison images, and to obtain the temperature defect area based on the prediction result.

[0018] The third objective of this invention is achieved by the following technical solution: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described solder optimization method based on AOI detection.

[0019] The fourth objective of this invention is achieved by the following technical solution: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described solder optimization method based on AOI detection.

[0020] In summary, this application includes at least one of the following beneficial technical effects: 1. Acquire AOI temperature image information during the preheating stage, including image information combining product image and temperature information. This AOI temperature image information can show both the surface condition and temperature of the product, or the temperature of different parts of the product through heat distribution. Furthermore, by comparing the product's temperature under normal conditions with the current temperature, the internal heat distribution of the product can be determined, for example, whether there are overheated or underheated areas. The defect comparison model has multiple preset defect comparison images. Because different temperature defects can lead to different quality problems in the product, the defect comparison model can filter defect comparison images through big data analysis combining quality problems and temperature information. Thus, temperature defects on the product image that may cause quality problems can be identified by comparing AOI temperature image information with multiple defect comparison images, improving the accuracy of product temperature defect identification. This allows for targeted temperature adjustments to address temperature defects that may cause quality problems, improving the accuracy of temperature control and ultimately enhancing the accuracy of identifying temperature defects at solder joints.

[0021] 2. Because different temperature defects can lead to different quality problems in products, a key defect information interface is first extracted. This interface can acquire different product temperature defect information and corresponding image information. A preliminary machine learning model is used to analyze the preheating temperature corresponding to the quality problems caused by different product temperature defects. That is, it analyzes the quality problems caused by different product temperature defects and the specific temperature of the corresponding product during the preheating stage, thereby determining the product quality problems caused by different temperature defects more accurately. Therefore, the key defect information interface is connected to the preliminary machine learning model. Different product temperature defects and corresponding image information are acquired through the key defect information interface, thereby training the preliminary machine learning model. This allows the preliminary machine learning model to filter out defect comparison images, resulting in a defect comparison model. This ensures the accuracy of identifying product temperature defects. The defect comparison model can filter out defect comparison images. Thus, normal temperature defects and temperature defects that do not affect product quality on the product image can be identified by comparing AOI temperature image information and multiple defect comparison images, improving the accuracy of identifying product temperature defects. This allows for targeted temperature adjustments for normal temperature defects, improving the accuracy of temperature control.

[0022] 3. The product defect image interface is used to obtain information on different product quality problems, corresponding product temperature defect information, and corresponding image information. The quality score information interface is used to obtain the quality score information of the product obtained from the product defect image interface. This is because the quality score reflects the degree of impact of the quality problem caused by the temperature defect on the actual use requirements of the product. A high quality score indicates that the corresponding temperature defect will not have a high impact on the use of the product, while a low quality score indicates that the corresponding temperature defect will have a high impact on the use of the product. Therefore, the secondary machine learning model can learn from the product quality score and the quality problems, and based on the requirements of the quality problems represented by the current market demand, determine the defect images that meet the market demand and have the ability to affect the actual use of the product. This generates a key defect information interface, which facilitates the defect comparison model to obtain image information that is more suitable for analysis and learning. Attached Figure Description

[0023] Figure 1 This is a flowchart of the solder optimization method based on AOI detection in Embodiment 1 of this application.

[0024] Figure 2 This is a principle block diagram of the solder optimization system based on AOI detection in Embodiment 2 of this application.

[0025] Figure 3 This is a schematic diagram of the device in Embodiment 3 of this application. Detailed Implementation

[0026] The following is in conjunction with the appendix Figures 1 to 3 This application will be described in further detail. Example 1

[0027] Reference Figure 1 This application discloses a solder optimization method based on AOI inspection, which specifically includes the following steps: S10: Obtain AOI temperature image information during the preheating stage.

[0028] In this embodiment, AOI temperature image information refers to the visual image and temperature value information of the product being soldered. AOI temperature image information is obtained by capturing images of the product being soldered using an AOI industrial vision camera, combining this with thermal imaging sensors to detect the temperature of different areas of the product, and then fusing the image and temperature data. The product being soldered is a printed circuit board, and the AOI temperature image information can be images acquired at different times during the preheating stage.

[0029] S20: Input the AOI temperature image information into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage.

[0030] In this embodiment, the defect comparison model has multiple preset defect comparison images. These images are used to compare with AOI temperature image information. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. The defect comparison image can be obtained by the defect comparison model analyzing various quality problems that may occur in the current product, and then analyzing the temperature defects corresponding to these quality problems during the preheating stage. The temperature defect is then combined with the standard image of the product to obtain the defect comparison image. The temperature defect during the preheating stage will cause the product to have quality problems due to incorrect temperature.

[0031] Specifically, the AOI temperature image information is input into a preset defect comparison model. The defect comparison model can combine the AOI temperature image information with various defect comparison images to form multiple corresponding image comparison groups, so as to facilitate the comparison of image information in the future.

[0032] S30: Based on the defect comparison model, compare the AOI temperature image information with various defect comparison images to obtain the comparison result.

[0033] Specifically, the defect comparison model compares the AOI temperature image information with various defect comparison images one by one. This can be done by comparing the AOI temperature image information and the defect comparison images in each image comparison group to obtain a comparison result. This comparison result indicates whether the AOI temperature image information and the defect comparison image are consistent. It can be a consistency judgment between the AOI temperature image information and each type of defect comparison image, or a comparison similarity information between the AOI temperature image information and each type of defect comparison image, or a combination of the consistency judgment between the AOI temperature image information and each type of defect comparison image, and the comparison similarity information between the AOI temperature image information and each type of defect comparison image.

[0034] S40: Based on the comparison results, the temperature defect area is obtained.

[0035] Specifically, based on the consistency between the AOI temperature image information and the defect comparison image, if the AOI temperature image information and the defect comparison image are consistent, it indicates that the temperature of the temperature defect area corresponding to the defect comparison image on the product is abnormal. Therefore, the temperature defect area is obtained according to the specific comparison situation. If the AOI temperature image information and the defect comparison image are inconsistent, it indicates that the temperature of each area on the product is accurate. At this time, the temperature of each area on the product can be compared with the corresponding temperature in the product welding standard according to the preset product welding standard. If the comparison result indicates that the temperature of each area on the product is different from the corresponding temperature in the product welding standard, the temperature defect area is obtained according to the comparison result. If the comparison result indicates that the temperature of each area on the product is the same as the corresponding temperature in the product welding standard, there is no need to adjust the temperature in the subsequent welding stage.

[0036] S50: Based on the temperature defect area and the preset standard temperature setting, obtain the current temperature difference value, and adjust the temperature parameter setting of the welding stage according to the current temperature difference value.

[0037] Specifically, after obtaining the temperature defect area, the temperature parameters can be adjusted according to the temperature defect represented by the temperature defect area and the current time of obtaining the AOI temperature image information. If the current time of obtaining the AOI temperature image information is an earlier stage in the preheating stage, the temperature parameters can be adjusted accordingly based on the preset product welding standards. If the current time of obtaining the AOI temperature image information is a later stage in the preheating stage, the temperature parameters of the subsequent welding stage can be adaptively adjusted according to the area and condition of the temperature anomaly represented by the temperature defect area.

[0038] Furthermore, the preset standard temperature setting corresponds to the temperature adjustment value of the welding stage for different temperature defects in the preheating stage. This preset standard temperature setting can be set by the user in advance. Based on the temperature defect area and the preset standard temperature setting, the current temperature difference is obtained. The current temperature difference refers to the temperature difference that needs to be adjusted in the welding stage.

[0039] Based on the current temperature difference, the temperature parameters for subsequent welding stages are adjusted accordingly. This ensures that heating can proceed according to the adjusted temperature parameters when the welding process reaches the next stage.

[0040] In another embodiment of this application, prior to step S10, the following steps are included: S11: Key Defect Information Extraction Interface, which is used to obtain different product temperature defect information, corresponding quality problem information and corresponding image information.

[0041] Specifically, a key defect information extraction interface is used to obtain specific product temperature defect information, corresponding quality problem information, and corresponding image information that may lead to quality issues. This is because during the preheating stage of product production, some temperature defects can cause quality problems, while others may not. Furthermore, the specific temperature distribution on the product represented by the temperature defect is often strongly correlated with the product quality problem. That is, when certain product quality problems occur, it is because the corresponding product exhibits related temperature defects during production. Therefore, the key defect information interface is specifically designed to obtain different product temperature defect information, corresponding image information, and corresponding quality problem information that may lead to quality problems. This facilitates targeted learning by subsequent machine learning models, improving the model's recognition capabilities.

[0042] S12: Establish a preliminary machine learning model. The preliminary machine learning model is used to analyze the temperature of the preheating stage corresponding to the quality problems caused by temperature defects of different products. The key defect information interface is connected to the preliminary machine learning model. Based on the key defect information interface, the different product temperature defects, the corresponding quality problem information and the corresponding image information are obtained. The preliminary machine learning model is trained to obtain a defect comparison model.

[0043] Specifically, a preliminary machine learning model is established. This model is defined as a tool for analyzing the preheating temperature corresponding to quality problems caused by different product temperature defects. In other words, by analyzing the product temperature defects and corresponding image information, as well as the corresponding quality problem information, obtained from the key defect information interface, the correlation between the product temperature defects, image information, and quality problem information is accurately determined through the analysis of the correlation between the three different parameters. Therefore, the product defect image interface is connected to the preliminary machine learning model. Based on the product defect image interface, different product temperature defects and corresponding image information, as well as the corresponding quality problem information, are obtained. The preliminary machine learning model is then trained to obtain a defect comparison model. At this point, the defect comparison model pre-stores multiple defect comparison images, that is, multiple combinations of product temperature defects, image information, and quality problem information.

[0044] In another embodiment of this application, before step S11: extracting the critical defect information interface, the method further includes: S111: Extract product defect image interface and quality score information interface. The product defect image interface is used to obtain various product quality problem information, corresponding product temperature defect information and corresponding image information. The quality score information interface is used to obtain quality score information of products with different quality problems.

[0045] Specifically, the system extracts product defect image interfaces and quality scoring information interfaces. The product defect image interface is used to acquire various product quality problem information, corresponding product temperature defect information, and corresponding image information. This product defect image interface differs from the critical defect information interface in that the critical defect information interface acquires specific product temperature defect information, corresponding quality problem information, and corresponding image information that will affect the actual use of the product due to quality issues. The product defect image interface, however, acquires various product quality problem information, corresponding product temperature defect information, and corresponding image information whose impact on the actual use of the product is unclear. Therefore, the subsequent secondary machine learning model is used to extract information from the various product quality problem information, corresponding product temperature defect information, and corresponding image information regarding their potential impact on the actual use of the product. The information is filtered; in addition, the quality rating information interface is used to obtain quality rating information for products with different quality problems. The quality rating information refers to the quality rating given by the downstream processing unit or market of the product. Specifically, it can be determined by the ratio of the product's quality rating to the standard product's quality rating. The products in the quality rating information of different quality problems obtained by the quality rating information interface correspond to the product quality problem information, the corresponding product temperature defect information, and the corresponding image information obtained by the product defect image interface. That is, for the same quality problem information, the product defect image interface obtains the quality problem information, the corresponding product temperature defect information, and the corresponding image information, and the quality rating information corresponding to the quality problem information is obtained by the quality rating information interface.

[0046] The product defect image interface and quality rating information interface can be accessed through an SQL or NoSQL query interface to access a database storing historical product defect images and temperature data, or by obtaining data through a RESTful API or web service in cooperation with third-party services from other enterprises or research institutions.

[0047] S112: Establish a secondary machine learning model, which is used to analyze the training availability of image information corresponding to quality problems of different products.

[0048] Specifically, a secondary machine learning model is established. This model is used to analyze the training availability of image information corresponding to quality problems of different products. The training availability refers to the extent to which image information can be used to improve the recognition ability of the primary machine learning model.

[0049] S113: Connect the product defect image interface and the quality score information interface to the secondary machine learning model to obtain quality problem information, product temperature defect information and corresponding image information of different products, as well as the corresponding quality scores of products with different quality problems, and train the secondary machine learning model in combination with a preset first score threshold to obtain a quality analysis model. The quality analysis model is used to filter temperature defect images used for training the primary machine learning model.

[0050] Specifically, the product defect image interface and quality score information interface are connected to a secondary machine learning model to obtain quality problem information, corresponding product temperature defect information, corresponding image information, and corresponding quality score information for different products. This information is then combined with a preset first score threshold to train the secondary machine learning model. It's important to note that the secondary machine learning model is trained using the quality problem information, corresponding quality score information, and the first score threshold for different products, but not using product temperature defect information and corresponding image information. This is because the secondary machine learning model is used to learn about quality problems by combining product quality scores with quality problem information. Based on the requirements for quality problems represented by current market demand, it determines the quality scores of products with different quality problems that can affect the actual use of the product. Then, combined with the first score threshold used for filtering based on product quality scores, it filters out defect images and corresponding related information that meet market demand and have quality problems that can affect the actual use of the product, thereby generating a key defect information interface.

[0051] In another embodiment of this application, step S40 includes: S41: If the comparison result indicates that the AOI temperature image information corresponds to the defect comparison image, then the temperature defect region is obtained based on the defect comparison image.

[0052] Specifically, if the comparison result indicates that the AOI temperature image information corresponds to the defect comparison image, it means that the temperature defect of the current product is consistent with the temperature defect represented by the defect comparison image. Therefore, the defect comparison image and the corresponding temperature defect information can be used as the temperature defect of the current product to generate a temperature defect area, which is a combination of image information and temperature distribution information.

[0053] S42: If the comparison result indicates that the AOI temperature image information does not correspond to any of the various defect comparison images, then based on the defect comparison model, the temperature defect information of the product is predicted, and the temperature defect area is obtained according to the prediction result.

[0054] Specifically, if the AOI temperature image information does not correspond to the defect comparison image, it indicates that the temperature defect of the current product may be quite special. It is unknown whether the quality problem caused by the temperature defect on the current product will affect the actual use of the product. Therefore, the temperature defect of the product is predicted by the defect comparison model. In addition to the preset multiple defect comparison images, the defect comparison model also includes a large number of other defect comparison images. Thus, the prediction result can be obtained by comparing with a large number of other defect comparison images, and the temperature defect area can be obtained.

[0055] In another embodiment of this application, step S42: obtaining the temperature defect region based on the prediction result includes: S421: Input the product temperature defect information represented by the prediction result into the quality analysis model, and determine the quality score corresponding to the product temperature defect information represented by the prediction result based on the quality analysis model.

[0056] Specifically, the defect comparison model includes not only the defect comparison image corresponding to the current quality score, but also a large number of other defect comparison images. These defect comparison images correspond to quality problems that, based on the current quality score, will affect the actual use of the product. However, since there may be other quality problems affecting the actual use of the product besides those in the defect comparison images due to incomplete quality score investigations or other factors, a large number of other defect comparison images may not correspond to the quality problems that, based on the current quality score, will affect the actual use of the product. Therefore, after predicting the quality problems corresponding to the product's temperature defects through the defect comparison model, the quality analysis model is used to determine whether these quality problems will affect the actual use of the product, thereby identifying the corresponding temperature defect areas. Based on this, the product temperature defect information represented by the prediction results is input into the quality analysis model, and the quality score corresponding to the product temperature defect information represented by the prediction results is determined based on the quality analysis model.

[0057] S422: Compare the quality score corresponding to the product temperature defect information represented by the prediction result with a preset second scoring threshold. If the quality score corresponding to the product temperature defect information represented by the prediction result is greater than the second scoring threshold, then the product temperature defect information represented by the prediction result is taken as a temperature defect area. If the quality score corresponding to the product temperature defect information represented by the prediction result is not greater than the second scoring threshold, then adjust the product temperature defect information represented by the prediction result according to the quality problem information corresponding to the product temperature defect information represented by the prediction result to generate a temperature defect area.

[0058] Specifically, after obtaining the quality score corresponding to the product temperature defect information represented by the prediction result, the quality score corresponding to the product temperature defect information represented by the prediction result is compared with a preset second scoring threshold. The second scoring threshold will be higher than the first scoring threshold because the first scoring threshold is used by the quality analysis model to filter defect images for training the primary machine learning model. Model training requires a large amount of data. If the first scoring threshold is set too high, fewer defect images may be filtered out. In addition, the primary machine learning model will filter defect comparison images again. Therefore, the quality analysis model can set a lower first scoring threshold to obtain defect images for training, and then the primary machine learning model will perform a second filtering. This is equivalent to the primary machine learning model raising the first scoring threshold. The second scoring threshold can be directly the value threshold set by the user, for example, 95% of the product's selling value, while the first scoring threshold is lower than 95%.

[0059] Based on this, if the quality score corresponding to the product temperature defect information represented by the prediction result is greater than the second scoring threshold, then the product temperature defect information represented by the prediction result is taken as the temperature defect area. If the quality score corresponding to the product temperature defect information represented by the prediction result is not greater than the second scoring threshold, then based on the quality problem corresponding to the product temperature defect image represented by the prediction result, combined with other factors other than the quality score that may cause the quality problem to affect the actual use of the product, the specific temperature defect situation corresponding to other factors can be determined, thereby adjusting the product temperature defect represented by the prediction result and generating the temperature defect area.

[0060] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. Example 2

[0061] An AOI-based solder optimization system is provided, which corresponds to the AOI-based solder optimization method described in the above embodiments.

[0062] like Figure 2 As shown, the solder optimization system based on AOI inspection includes an image acquisition module, a model input module, an image comparison module, a defect determination module, and a temperature adjustment module. Detailed descriptions of each functional module are as follows: The image acquisition module is used to acquire AOI temperature image information during the preheating stage; The model input module is used to input the AOI temperature image information into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. The image comparison module is used to compare the AOI temperature image information and multiple defect comparison images based on the defect comparison model to obtain a comparison result; The defect determination module is used to obtain the temperature defect area based on the comparison results; The temperature adjustment module is used to obtain the current temperature difference based on the temperature defect area and the preset standard temperature setting, and adjust the temperature parameter settings for the welding stage according to the current temperature difference.

[0063] The image acquisition module also includes: The information interface extraction submodule is used to extract the key defect information interface, which is used to obtain different product temperature defect information, corresponding quality problem information and corresponding image information. The first-order model building submodule is used to build a preliminary machine learning model. The preliminary machine learning model is used to analyze the temperature of the preheating stage corresponding to the quality problems caused by temperature defects in different products. The key defect information interface is connected to the preliminary machine learning model. Based on the key defect information interface, the different product temperature defects, the corresponding quality problem information and the corresponding image information are obtained. The preliminary machine learning model is trained to obtain a defect comparison model.

[0064] The defect determination module also includes: The defect comparison image channel determination submodule is used to obtain the temperature defect region based on the defect comparison image if the comparison result indicates that the AOI temperature image information corresponds to the defect comparison image. The defect comparison model channel determination submodule is used to predict the temperature defect information of the product based on the defect comparison model if the comparison result indicates that the AOI temperature image information does not correspond to any of the various defect comparison images, and to obtain the temperature defect area based on the prediction result.

[0065] For specific limitations regarding the AOI-based solder optimization system, please refer to the limitations of the AOI-based solder optimization method mentioned above, which will not be repeated here. Each module in the AOI-based solder optimization system can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in the processor of the computer device in hardware form or independent of it, or it can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module. Example 3

[0066] A computer device, which may be a server, has an internal structure diagram as shown below. Figure 3As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data such as AOI temperature image information, defect comparison models, defect comparison images, AOI temperature image information, comparison results, temperature defect areas, standard temperature settings, current temperature differences, and temperature parameter settings. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a solder optimization method based on AOI inspection.

[0067] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: S10: Obtain AOI temperature image information during the preheating stage; S20: Input the AOI temperature image information into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. S30: Based on the defect comparison model, compare the AOI temperature image information with multiple defect comparison images to obtain a comparison result; S40: Based on the comparison results, the temperature defect area is obtained; S50: Based on the temperature defect area and the preset standard temperature setting, obtain the current temperature difference value, and adjust the temperature parameter setting of the welding stage according to the current temperature difference value.

[0068] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: S10: Obtain AOI temperature image information during the preheating stage; S20: Input the AOI temperature image information into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. S30: Based on the defect comparison model, compare the AOI temperature image information with multiple defect comparison images to obtain a comparison result; S40: Based on the comparison results, the temperature defect area is obtained; S50: Based on the temperature defect area and the preset standard temperature setting, obtain the current temperature difference value, and adjust the temperature parameter setting of the welding stage according to the current temperature difference value.

[0069] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchlink, DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0070] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0071] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A solder optimization method based on AOI inspection, characterized in that, include: Acquire AOI temperature image information during the preheating stage; The AOI temperature image information is input into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. Based on the defect comparison model, the AOI temperature image information and various defect comparison images are compared to obtain the comparison results; Based on the comparison results, the temperature defect region is obtained; Based on the temperature defect area and the preset standard temperature setting, the current temperature difference is obtained, and the temperature parameter settings for the welding stage are adjusted according to the current temperature difference.

2. The solder optimization method based on AOI inspection according to claim 1, characterized in that: Before acquiring the AOI temperature image information during the preheating stage, the following steps are included: A key defect information extraction interface is used to obtain different product temperature defect information, corresponding quality problem information, and corresponding image information. A preliminary machine learning model is established to analyze the temperature of the preheating stage corresponding to quality problems caused by temperature defects in different products. The key defect information interface is connected to the preliminary machine learning model. Based on the key defect information interface, the different product temperature defects, corresponding quality problem information and corresponding image information are obtained. The preliminary machine learning model is then trained to obtain a defect comparison model.

3. The solder optimization method based on AOI inspection according to claim 2, characterized in that: Before the interface for extracting key defect information, the following is also included: The interface for extracting product defect images and the interface for extracting quality scoring information are provided. The product defect image interface is used to obtain various product quality problem information, corresponding product temperature defect information, and corresponding image information. The quality scoring information interface is used to obtain quality scoring information for products with different quality problems. A secondary machine learning model is established to analyze the training availability of image information corresponding to quality problems of different products. The product defect image interface and the quality score information interface are connected to the secondary machine learning model to obtain quality problem information, product temperature defect information and corresponding image information of different products, as well as the corresponding quality scores of products with different quality problems. The secondary machine learning model is then trained in combination with a preset first score threshold to obtain a quality analysis model. The quality analysis model is used to filter temperature defect images used for training the primary machine learning model.

4. The solder optimization method based on AOI inspection according to claim 3, characterized in that: The step of obtaining the temperature defect region based on the comparison results includes: If the comparison result indicates that the AOI temperature image information corresponds to the defect comparison image, then the temperature defect region is obtained based on the defect comparison image; If the comparison result indicates that the AOI temperature image information does not correspond to any of the various defect comparison images, then based on the defect comparison model, the temperature defect information of the product is predicted, and the temperature defect area is obtained according to the prediction result.

5. The solder optimization method based on AOI inspection according to claim 4, characterized in that: The step of obtaining the temperature defect region based on the prediction results includes: The product temperature defect information represented by the prediction results is input into the quality analysis model, and the quality score corresponding to the product temperature defect information represented by the prediction results is determined based on the quality analysis model. The quality score corresponding to the product temperature defect information represented by the prediction result is compared with a preset second scoring threshold. If the quality score corresponding to the product temperature defect information represented by the prediction result is greater than the second scoring threshold, the product temperature defect information represented by the prediction result is taken as a temperature defect area. If the quality score corresponding to the product temperature defect information represented by the prediction result is not greater than the second scoring threshold, the product temperature defect information represented by the prediction result is adjusted according to the quality problem information corresponding to the product temperature defect information represented by the prediction result to generate a temperature defect area.

6. A solder optimization system based on AOI inspection, characterized in that, The solder optimization method based on AOI inspection as described in any one of claims 1-5 includes: The image acquisition module is used to acquire AOI temperature image information during the preheating stage; The model input module is used to input the AOI temperature image information into a preset defect comparison model. The defect comparison model has multiple preset defect comparison images. The defect comparison image refers to the AOI temperature image information when the temperature is abnormal during the preheating stage. The image comparison module is used to compare the AOI temperature image information and multiple defect comparison images based on the defect comparison model to obtain a comparison result; The defect determination module is used to obtain the temperature defect area based on the comparison results; The temperature adjustment module is used to obtain the current temperature difference based on the temperature defect area and the preset standard temperature setting, and adjust the temperature parameter settings of the welding stage according to the current temperature difference.

7. The solder optimization system based on AOI inspection according to claim 6, characterized in that: The image acquisition module further includes: The information interface extraction module is used to extract key defect information interfaces, which are used to obtain different product temperature defect information, corresponding quality problem information and corresponding image information. A first-order model building module is used to build a preliminary machine learning model. The preliminary machine learning model is used to analyze the temperature of the preheating stage corresponding to the quality problems caused by temperature defects in different products. The key defect information interface is connected to the preliminary machine learning model. Based on the key defect information interface, the different product temperature defects, the corresponding quality problem information and the corresponding image information are obtained. The preliminary machine learning model is trained to obtain a defect comparison model.

8. The solder optimization system based on AOI inspection according to claim 7, characterized in that: The defect determination module also includes: The defect comparison image channel determination submodule is used to obtain the temperature defect region based on the defect comparison image if the comparison result indicates that the AOI temperature image information corresponds to the defect comparison image. The defect comparison model channel determination submodule is used to predict the temperature defect information of the product based on the defect comparison model if the comparison result indicates that the AOI temperature image information does not correspond to any of the various defect comparison images, and to obtain the temperature defect area based on the prediction result.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the solder optimization method based on AOI detection as described in any one of claims 1 to 5.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the solder optimization method based on AOI detection as described in any one of claims 1 to 5.

Images