A good product prediction method and system for a computer industrial control mainboard

Abstract

The application provides a good product prediction method and system for a computer industrial control mainboard. Images of each solder joint of a computer industrial control mainboard to be detected are collected when the computer industrial control mainboard runs in a detection device, and all solder joint images are obtained, and then the pixel contribution of each solder joint image is determined. The global feature vector and the local feature vector of the solder joint are determined based on each pixel contribution, the global feature vector and the local feature vector are fused into a defect description quantity of the solder joint, the defect similarity coefficient of each solder joint image and a historical mainboard solder joint defect sample set is determined, the solder joint damage degree of the computer industrial control mainboard to be detected is determined based on all defect similarity coefficients and defect description quantities, the use quality of the computer industrial control mainboard to be detected is predicted based on the solder joint damage degree, and then good product marking is performed. By adopting the scheme of the application, the hidden trouble fault of the solder joint defect of the computer industrial control mainboard can be detected in advance, and thus the reliable range of good product detection of the computer industrial control mainboard is increased.

Description

Technical Field

[0001] This application relates to the field of computer industrial control motherboard testing technology, and more specifically, to a method and system for predicting the yield of computer industrial control motherboards. Background Technology

[0002] Industrial control computer motherboards, also known as industrial control computer motherboards, are computer motherboards specifically designed for industrial environments. They offer higher reliability, stability, and durability, enabling stable operation in harsh environments. Therefore, industrial control computer motherboards are widely used in industrial automation, transportation, energy, medical equipment, and other fields. Due to their application in mission-critical and harsh environments, the quality requirements for industrial control computer motherboards are extremely high, making product testing particularly important.

[0003] The quality inspection of industrial computer motherboards is a crucial step in ensuring their overall quality. Components on these motherboards are connected via soldering, and soldering issues are inevitable, such as cold solder joints, excessive or insufficient solder, solder bridging, and poor solder joint coplanarity. Therefore, quality inspection of industrial computer motherboards requires examining the internal soldering quality, checking the integrity and consistency of solder joints, and identifying potential manufacturing defects. However, current technology often judges industrial computer motherboards as good simply by their ability to conduct electricity during the inspection. This overlooks the fact that after prolonged use or severe vibration, small amounts of solder may become trapped in the solder joints. Solder joints are prone to breakage, causing some pins to become misaligned or even bridging with other pins, resulting in short circuits or open circuits on the computer industrial control motherboard during use. This leads to potential malfunctions after the motherboard is installed and used. Existing product inspection equipment often relies on manual observation and judgment, which is difficult to detect internal defects and surface defects in solder joints. This is inefficient and prone to misjudgment, allowing defective computer industrial control motherboards with potential solder joint defects to be released, causing malfunctions only after use and creating greater trouble for users. Therefore, how to achieve early detection of potential malfunctions due to solder joint defects on computer industrial control motherboards, thereby increasing the reliability of product inspection, has become a challenge for the industry. Summary of the Invention

[0004] This application provides a method and system for predicting the yield of computer industrial control motherboards, which can detect potential faults in solder joints of computer industrial control motherboards in advance, thereby increasing the reliability of the yield test of computer industrial control motherboards.

[0005] In a first aspect, this application provides a method for predicting the yield of industrial control motherboards for computers, comprising the following steps:

[0006] Images of each solder joint on the computer's industrial control motherboard under test are collected during operation in the testing equipment to obtain all solder joint images, and then the pixel contribution of each solder joint in each image is determined.

[0007] Based on the pixel contribution of each solder joint, feature decomposition is performed on the pixels in all solder joint images to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects. The global feature vector and the local feature vector are then fused into a defect description of the solder joint in the computer industrial control motherboard to be inspected.

[0008] A historical motherboard solder joint defect sample set is obtained. Each solder joint image is matched with the historical motherboard solder joint defect sample set to obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set. The degree of solder joint damage of the computer industrial control motherboard to be tested during operation is determined by all the defect similarity coefficients and the defect description quantity.

[0009] Based on the degree of solder joint damage, the quality of use of the computer industrial control motherboard under test is predicted, and then the computer industrial control motherboard under test is marked as good based on the prediction results.

[0010] In some embodiments, determining the pixel contribution of a solder joint in each solder joint image specifically includes:

[0011] Select an image of a solder joint;

[0012] Based on preset judgment pixel values, each pixel in the weld joint image is divided into defective pixels and normal pixels.

[0013] The pixel contribution of the solder joint in the solder joint image is determined by all defective pixels and all normal pixels;

[0014] Repeat the above steps to obtain the pixel contribution of the solder joints in the remaining solder joint images.

[0015] In some embodiments, feature decomposition is performed on pixels in all solder joint images based on the pixel contribution of each solder joint to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects. Specifically, this includes:

[0016] The global and local feature spaces of solder joint defects in the computer industrial control motherboard to be inspected are determined by the pixels in each solder joint image and the pixel contribution of each solder joint.

[0017] The global feature space is decomposed to obtain the global feature vector of the solder joints on the global defects of the computer industrial control motherboard to be inspected.

[0018] The local feature space is decomposed to obtain the local feature vector of the solder joints of the computer industrial control motherboard to be inspected on the local defects.

[0019] In some embodiments, performing similarity matching between each solder joint image and the historical motherboard solder joint defect sample set to obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set specifically includes:

[0020] Select an image of a solder joint;

[0021] Determine the matching value between the current solder joint image and each defective solder joint sample image in the historical motherboard solder joint defect sample set;

[0022] The defect similarity coefficient between the solder joint image and the historical motherboard solder joint defect sample set is determined by all matching values;

[0023] Repeat the above steps to obtain the defect similarity coefficient between the remaining solder joint images and the historical motherboard solder joint defect sample set.

[0024] In some embodiments, determining the solder joint damage degree of the computer industrial control motherboard under test during operation by using all defect similarity coefficients and the defect descriptive quantity specifically includes:

[0025] The overall matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set is determined based on all defect similarity coefficients.

[0026] The degree of solder joint damage during operation of the computer industrial control motherboard under test is determined by the comprehensive compatibility and the defect description quantity.

[0027] In some embodiments, predicting the usability of the industrial control computer motherboard under test based on the solder joint damage degree, and then marking the industrial control computer motherboard under test as good based on the prediction result, specifically includes:

[0028] Preset the good product judgment threshold for the computer industrial control motherboard to be tested;

[0029] When the solder joint damage exceeds the good product determination threshold, it is predicted that the quality of the computer industrial control motherboard under test is poor, and the computer industrial control motherboard under test is marked as a defective product.

[0030] When the solder joint damage is less than or equal to the good product determination threshold, the computer industrial control motherboard under test is predicted to be of good quality and is marked as a good product.

[0031] In some embodiments, the image acquisition device in the computer industrial control motherboard testing equipment acquires images of each solder joint corresponding to the computer industrial control motherboard under test during operation in the testing equipment.

[0032] Secondly, this application provides a yield prediction system for computer industrial control motherboards, comprising:

[0033] The acquisition module is used to acquire images of each solder joint of the computer industrial control motherboard under test during operation in the testing equipment, obtain all solder joint images, and then determine the pixel contribution of each solder joint in each solder joint image;

[0034] The processing module is further configured to perform feature decomposition on the pixels in all solder joint images based on the pixel contribution of each solder joint, to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects, and to fuse the global feature vector and the local feature vector into a defect description quantity of the solder joint in the computer industrial control motherboard to be inspected.

[0035] The processing module is also used to acquire a historical motherboard solder joint defect sample set, perform similarity matching between each solder joint image and the historical motherboard solder joint defect sample set, obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set, and determine the degree of solder joint damage of the computer industrial control motherboard to be tested during operation by all the defect similarity coefficients and the defect description quantity.

[0036] The execution module is used to predict the usability of the computer industrial control motherboard under test based on the solder joint damage degree, and then mark the computer industrial control motherboard under test as good based on the prediction result.

[0037] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device executes the above-described yield prediction method for computer industrial control motherboards.

[0038] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned method for predicting the yield of a computer industrial control motherboard.

[0039] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0040] In this application, images of each solder joint on the industrial control motherboard of the computer under test are collected during operation in the testing equipment to obtain all solder joint images, and then the pixel contribution of each solder joint in each image is determined. Based on the pixel contribution of each solder joint, feature decomposition is performed on the pixels in all solder joint images to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects. The global feature vector and the local feature vector are fused into the defect description of the solder joint in the industrial control motherboard of the computer under test. A historical motherboard solder joint defect sample set is obtained, and each solder joint image is similarly matched with the historical motherboard solder joint defect sample set to obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set. The solder joint damage degree of the industrial control motherboard of the computer under test during operation is determined by all the defect similarity coefficients and the defect description. Based on the solder joint damage degree, the service quality of the industrial control motherboard of the computer under test is predicted, and then the industrial control motherboard of the computer under test is marked as good according to the prediction result.

[0041] Therefore, in this application, the solder joint damage degree of the computer industrial control motherboard under test during operation is determined by all defect similarity coefficients and the defect descriptive quantity. Since the global feature vector reflects the global defect situation of the solder joints on the computer industrial control motherboard under test, and the local feature vector reflects the local defect situation of the solder joints on the computer industrial control motherboard under test, the global feature vector and the local feature vector can be fused to obtain the defect degree (i.e., defect descriptive quantity) of the solder joints in the computer industrial control motherboard under test. Furthermore, the matching degree (i.e., defect similarity) between each solder joint image and the historical motherboard solder joint defect sample set is determined by the similarity between the solder joint image and the historical motherboard solder joint defect sample set. The similarity coefficients and the defect description coefficients are used to determine the solder joint damage degree of the computer industrial control motherboard under test during operation. This determines the severity of the solder joint defects on the motherboard during operation, thus effectively making a preliminary judgment on potential faults caused by solder joint defects. Finally, based on the solder joint damage degree, the usability of the computer industrial control motherboard under test is predicted, and then a good product mark is made according to the prediction results, thereby completing the good product inspection of the computer industrial control motherboard under test. In summary, this scheme realizes the preliminary detection of potential faults caused by solder joint defects on computer industrial control motherboards, thereby increasing the reliability of good product inspection of computer industrial control motherboards. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is an exemplary flowchart of a yield prediction method for computer industrial control motherboards according to some embodiments of this application;

[0044] Figure 2 This is an exemplary flowchart illustrating the determination of pixel contribution according to some embodiments of this application;

[0045] Figure 3 This is an exemplary flowchart illustrating the determination of a defect description quantity according to some embodiments of this application;

[0046] Figure 4 This is a schematic diagram of the structure of a yield prediction system for computer industrial control motherboards according to some embodiments of this application;

[0047] Figure 5 This is a schematic diagram of the structure of a computer device that implements a yield prediction method for a computer industrial control motherboard, according to some embodiments of this application. Detailed Implementation

[0048] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0049] This application provides a method and system for predicting the yield of industrial control computer motherboards. The core of the method involves starting a testing device for the industrial control computer motherboard and then connecting the motherboard to be tested to the testing device; acquiring images of each solder joint on the motherboard during operation in the testing device to obtain all solder joint images; determining the pixel contribution of each solder joint in each image; performing feature decomposition on the pixels in all solder joint images based on the pixel contribution of each solder joint to obtain a global feature vector of the solder joint on global defects and a local feature vector of the solder joint on local defects; and fusing the global feature vector and the local feature vector to form the yield prediction result for the industrial control computer motherboard under test. The system employs a method to describe the defects in solder joints on the motherboard. It acquires a historical set of motherboard solder joint defects, performs similarity matching between each solder joint image and the historical set, and obtains the defect similarity coefficient between each image and the historical set. The solder joint damage degree of the motherboard under test during operation is determined using all the similarity coefficients and the defect description. Based on the solder joint damage degree, the system predicts the usability of the motherboard and marks it as a good product. This method enables early detection of potential faults caused by solder joint defects on the motherboard, thereby increasing the reliability of good product testing of the motherboard.

[0050] To better understand the above technical solutions, a detailed description of the solutions will be provided below in conjunction with the accompanying drawings and specific implementation methods. (Reference) Figure 1 The figure is an exemplary flowchart of a yield prediction method for a computer industrial control motherboard according to some embodiments of this application.

[0051] In practice, the testing equipment for the industrial control motherboard needs to be started first. Then, the industrial control motherboard to be tested is connected to the testing equipment. Specifically: First, the testing equipment is started and performs self-tests and initialization, including hardware self-tests, sensor calibration, and image acquisition device adjustments, to confirm that each subsystem of the testing equipment (such as the camera, sensor, and image processing unit) is functioning correctly. Next, the interfaces on the industrial control motherboard to be tested (such as power interfaces and data interfaces) are matched and connected to the connection interface of the testing equipment. That is, an appropriate connecting cable is used to connect the industrial control motherboard to the testing equipment, ensuring that the motherboard can be powered on normally and that the connection is secure and without looseness, so that the testing equipment can receive the operating signals from the motherboard. Finally, the control software of the testing equipment confirms that the industrial control motherboard to be tested has been correctly connected and recognized, thus completing the connection of the industrial control motherboard to the testing equipment. Other methods may be used in other embodiments, which are not limited here.

[0052] In step 101, images of each solder joint of the computer industrial control motherboard under test are acquired when it is running in the testing equipment to obtain all solder joint images, and then the pixel contribution of each solder joint in each solder joint image is determined.

[0053] In a specific implementation, the image acquisition device in the testing equipment of the computer industrial control motherboard can automatically scan the entire running computer industrial control motherboard under test and acquire images of each solder joint one by one, thereby obtaining images of all solder joints of the computer industrial control motherboard under test during operation. The image acquisition device can be, for example, a high-resolution optical microscope, a scanning electron microscope, an automated optical inspection (AOI) device, or a machine vision system. Other methods can also be used for acquisition in other embodiments, and no specific limitation is made here.

[0054] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart of determining pixel contribution in some embodiments of this application. In this embodiment, determining the pixel contribution of a solder joint in each solder joint image can be achieved by the following steps:

[0055] First, in step 1011, select an image of a solder joint;

[0056] Secondly, in step 1012, each pixel in the weld point image is divided into defective pixels and normal pixels based on the preset judgment pixel value.

[0057] Then, in step 1013, the pixel contribution of the solder joint in the solder joint image is determined by all defective pixels and all normal pixels;

[0058] Finally, in step 1014, the above steps are repeated to obtain the pixel contribution of the solder joints in the remaining solder joint image.

[0059] It should be noted that the defective pixels in this application refer to the pixels corresponding to defects (e.g., poor soldering, cold soldering, oxidation, etc.) at the solder joints of the computer industrial control motherboard to be tested. In addition, the normal pixels refer to the pixels that appear under normal conditions at the solder joints of the computer industrial control motherboard to be tested.

[0060] In specific implementation, the judgment pixel value in this application is a pixel grayscale threshold used to distinguish defective pixels from normal pixels in a solder joint image. The division of each pixel in the solder joint image into defective pixels and normal pixels based on the preset judgment pixel value can be achieved in the following way: First, the judgment pixel value can be set by relevant experts based on the pixel value of solder joint defects (e.g., cold solder joint, oxidation, etc.) in historical computer industrial control motherboards. Then, pixels in the solder joint image with grayscale pixel values ​​less than or equal to the judgment pixel value are judged as defective pixels, and pixels in the solder joint image with grayscale pixel values ​​greater than the judgment pixel value are judged as normal pixels. Other methods can also be used for determination in other embodiments, which are not limited here.

[0061] In a specific implementation, the pixel contribution of a solder joint in a solder joint image can be determined by combining all defective pixels and all normal pixels in the following manner: First, obtain the total number of pixels in the solder joint image. Then, multiply the absolute value of the difference between the number of all defective pixels and the number of all normal pixels by the total number of defective pixels. Finally, use the ratio of the resulting product to the total number of pixels as the pixel contribution of the solder joint in the solder joint image. Other methods can also be used in other embodiments, which are not limited here.

[0062] It should be noted that the pixel contribution of the solder joint in this application reflects the proportion of defective pixels in the corresponding solder joint image. The higher the pixel contribution of the solder joint, the higher the proportion of defective pixels in the corresponding solder joint image. Conversely, the lower the pixel contribution of the solder joint, the lower the proportion of defective pixels in the corresponding solder joint image.

[0063] In step 102, feature decomposition is performed on the pixels in all solder joint images based on the pixel contribution of each solder joint to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects. The global feature vector and the local feature vector are then fused into a defect description of the solder joint in the computer industrial control motherboard to be inspected.

[0064] In some embodiments, the feature decomposition of pixels in all solder joint images based on the pixel contribution of each solder joint to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects can be achieved by the following steps:

[0065] The global and local feature spaces of solder joint defects in the computer industrial control motherboard to be inspected are determined by the pixels in each solder joint image and the pixel contribution of each solder joint.

[0066] The global feature space is decomposed to obtain the global feature vector of the solder joints on the global defects of the computer industrial control motherboard to be inspected.

[0067] The local feature space is decomposed to obtain the local feature vector of the solder joints of the computer industrial control motherboard to be inspected on the local defects.

[0068] In specific implementation, the global feature space in this application represents a high-dimensional space obtained by mapping the correlation between each solder joint image in the computer industrial control motherboard to be inspected, and the local feature space represents a high-dimensional space obtained by mapping the correlation between the pixels of the solder joints in the computer industrial control motherboard to be inspected. The global and local feature spaces for determining solder joint defects in the computer industrial control motherboard to be inspected, based on the pixels in each solder joint image and the pixel contribution of each solder joint, can be implemented in the following way: The correlation between any two solder joint images can be evaluated using deep learning or machine learning algorithms combined with the pixel contribution of each solder joint, thereby... The kernel method is used to construct the global feature space of solder joint defects in the computer industrial control motherboard to be inspected by using all the evaluation values ​​obtained. Similarly, deep learning or machine learning algorithms are used in conjunction with the determination results of pixels (i.e., whether a pixel is a defective pixel or a normal pixel) to evaluate the correlation between every two pixels in each solder joint image. Then, the kernel method is used to construct the local feature space of solder joint defects in the computer industrial control motherboard to be inspected by using all the evaluation values ​​obtained. The kernel method can be, for example, a linear kernel, a Gaussian kernel, a Laplacian kernel, and a Bayesian kernel. Other methods can also be used in other embodiments, which are not limited here.

[0069] In specific implementation, the global feature space is decomposed to obtain the global feature vector of the solder joints of the computer industrial control motherboard under test on global defects. This can be achieved in the following way: multiple global principal components can be decomposed from the global feature space using Principal Component Analysis (PCA). Each global principal component corresponds to a global defect of a solder joint on the computer industrial control motherboard under test. Global defects of solder joints may include: uneven distribution of solder joints, poor solder joints throughout the board, overall oxidation and corrosion, overall solder joint size defects, solder joint misalignment, and solder joint coplanarity problems, etc. Then, all the global principal components obtained from the decomposition in the global feature space are used to form the global feature vector of the solder joints of the computer industrial control motherboard under test on global defects. The local feature space is decomposed to obtain the local feature vector of the solder joints of the computer industrial control motherboard under test on local defects. The following method is used to achieve this: multiple local principal components can be decomposed from the local feature space using principal component analysis. Each local principal component corresponds to a local defect of a solder joint on the computer industrial control motherboard to be inspected. Local defects of solder joints may include: cold solder joints, solder joint cracks, solder joint holes, too many or too few solder joints, cold solder joints, solder joint contamination, and solder joint bridging, etc. Then, all the local principal components obtained from the decomposition in the local feature space are used to form the local feature vector of the solder joints of the computer industrial control motherboard to be inspected on the local defects, that is, feature decomposition is achieved for the pixels in all solder joint images. Other methods can also be used to determine this in other embodiments, which are not limited here.

[0070] It should be noted that the global feature vector in this application is an array of feature vectors representing the overall solder joint defects of the computer industrial control motherboard to be tested. The global feature vector reflects the defect situation of the solder joints of the computer industrial control motherboard to be tested in terms of global defects. In addition, the local feature vector is an array of feature vectors representing the local solder joint defects of the computer industrial control motherboard to be tested. The local feature vector reflects the defect situation of the solder joints of the computer industrial control motherboard to be tested in terms of local defects.

[0071] It should be noted that the feature decomposition in this application refers to the process of mapping the correlation between each solder joint image in the computer industrial control motherboard to be inspected to obtain a high-dimensional space (i.e., global feature space) and mapping the correlation between solder joint pixels in the computer industrial control motherboard to be inspected to obtain a high-dimensional space (i.e., local feature space). Then, the global feature vector of the solder joints of the computer industrial control motherboard to be inspected on global defects is obtained from the global feature space, and the local feature vector of the solder joints of the computer industrial control motherboard to be inspected on local defects is obtained from the local feature space. Through the principal components decomposed from the global feature space and the local feature space, the global features and changes of solder joint defects, as well as the local features and changes, can be revealed in detail. This allows for the classification and identification of different types of global and local defects in the computer industrial control motherboard. At the same time, it can filter noise, reduce dimensionality, and facilitate visualization, enabling more accurate solder joint defect detection and improving the quality control effect of the computer industrial control motherboard.

[0072] In some embodiments, reference Figure 3 As shown, this figure is an exemplary flowchart for determining the defect descriptor quantity in some embodiments of this application. In this embodiment, the fusion of the global feature vector and the local feature vector into the defect descriptor quantity of the solder joint in the computer industrial control motherboard to be inspected can be achieved by the following steps:

[0073] First, in step 1021, the global fusion influence factor of the global feature vector is determined;

[0074] Secondly, in step 1022, the local fusion influence factor of the local feature vector is determined;

[0075] Finally, in step 1023, the global feature vector and the local feature vector are linearly fused using the global fusion influence factor and the local fusion influence factor to obtain the defect description quantity of the solder joints in the computer industrial control motherboard to be inspected.

[0076] In specific implementation, the global fusion influence factor in this application reflects the degree of influence of external environmental factors when fusing global defects of solder joints in the computer industrial control motherboard to be tested. The larger the global fusion influence factor, the greater the degree of influence of external environmental factors when fusing global defects of solder joints in the computer industrial control motherboard to be tested. Conversely, the smaller the global fusion influence factor, the smaller the degree of influence of external environmental factors when fusing global defects of solder joints in the computer industrial control motherboard to be tested. The global fusion influence factor of the global feature vector can be determined in the following way: the degree of influence of external environmental factors (such as temperature, humidity, vibration, dust and electromagnetic interference, etc.) on global defects of solder joints in the computer industrial control motherboard to be tested can be comprehensively evaluated by combining historical experimental data with an evaluation algorithm. The obtained evaluation value is used as the global fusion influence factor of the global feature vector. The value of the global fusion influence factor is usually between 0 and 1. The evaluation algorithm is, for example, cross-validation, genetic algorithm, ensemble learning and reinforcement learning. Other methods can also be used to determine it in other embodiments, which are not limited here.

[0077] In specific implementation, the local fusion influence factor reflects the degree of influence of internal factors when fusing local defects of solder joints in the computer industrial control motherboard to be inspected. The larger the local fusion influence factor, the greater the degree of influence of internal factors when fusing local defects of solder joints in the computer industrial control motherboard to be inspected; conversely, the smaller the local fusion influence factor, the smaller the degree of influence of internal factors when fusing local defects of solder joints in the computer industrial control motherboard to be inspected. The local fusion influence factor of the local feature vector can be determined in the following way: the degree of influence of internal factors (soldering temperature, solder composition, welding speed, welding pressure, and PCB material of the computer industrial control motherboard) on local defects of solder joints in the computer industrial control motherboard to be inspected can be comprehensively evaluated by combining historical experimental data with an evaluation algorithm, and the obtained evaluation value is used as the local fusion influence factor of the local feature vector. The value of the local fusion influence factor is usually between 0 and 1. The evaluation algorithm is, for example, cross-validation, genetic algorithm, ensemble learning, and reinforcement learning. Other methods can also be used to determine it in other embodiments, which are not limited here.

[0078] In specific implementation, the global feature vector and the local feature vector are linearly fused using the global fusion influence factor and the local fusion influence factor to obtain the defect description of the solder joints in the computer industrial control motherboard to be inspected. This can be achieved in the following way: First, the global feature vector and the local feature vector can be standardized using the StandardScaler function in the sklearn library of Python to obtain standardized global feature vectors and standardized local feature vectors. Second, the global fusion influence factor is multiplied by the standardized global feature vector to obtain an array. The local fusion influence factor is multiplied by the standardized local feature vector to obtain another array. Then, the two arrays obtained above are added together to obtain a set of data. Finally, the average value of this set of data is used as the defect description of the solder joints in the computer industrial control motherboard to be inspected. Other methods can also be used to determine this in other embodiments, which are not limited here.

[0079] It should be noted that the defect description quantity in this application reflects the degree of defect of the solder joints in the computer industrial control motherboard to be tested. The larger the defect description quantity, the higher the degree of defect of the solder joints in the computer industrial control motherboard to be tested. Conversely, the smaller the defect description quantity, the lower the degree of defect of the solder joints in the computer industrial control motherboard to be tested.

[0080] In step 103, a historical motherboard solder joint defect sample set is obtained, and each solder joint image is matched with the historical motherboard solder joint defect sample set to obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set. The degree of solder joint damage of the computer industrial control motherboard to be tested during operation is determined by all the defect similarity coefficients and the defect description quantity.

[0081] It should be noted that the historical motherboard solder joint defect sample set is a collection of defective solder joint sample images from historical computer industrial control motherboards. In specific implementation, defective solder joint sample images from historical computer industrial control motherboards can be obtained from the computer industrial control motherboard database, and then all defective solder joint sample images can be combined into a historical motherboard solder joint defect sample set for computer industrial control motherboards. Other methods can also be used to obtain the samples in other embodiments, and no specific limitation is made here.

[0082] In some embodiments, the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set can be obtained by performing similarity matching between each solder joint image and the historical motherboard solder joint defect sample set using the following steps:

[0083] Select an image of a solder joint;

[0084] Determine the matching value between the current solder joint image and each defective solder joint sample image in the historical motherboard solder joint defect sample set;

[0085] The defect similarity coefficient between the solder joint image and the historical motherboard solder joint defect sample set is determined by all matching values;

[0086] Repeat the above steps to obtain the defect similarity coefficient between the remaining solder joint images and the historical motherboard solder joint defect sample set.

[0087] In specific implementation, the matching value reflects the degree of matching between the corresponding solder joint image and the defective solder joint sample images in the historical motherboard solder joint defect sample set. A larger matching value indicates a greater degree of matching between the corresponding solder joint image and the defective solder joint sample images in the historical motherboard solder joint defect sample set, while a smaller matching value indicates a smaller degree of matching. Determining the matching value between the solder joint image and each defective solder joint sample image in the historical motherboard solder joint defect sample set can be achieved in the following way: selecting the historical motherboard solder joint image... For a defective solder joint sample image in the defect sample set, the Euclidean distance between the defective solder joint sample image and the corresponding pixel in the solder joint image can be calculated using the numpy.linalg.norm() function in Python. The average of all obtained Euclidean distances is used as the matching value between the solder joint image and the defective solder joint sample image. The above steps are repeated to obtain the matching value between the solder joint image and the remaining defective solder joint sample images in the historical motherboard solder joint defect sample set. Other methods can also be used to determine this in other embodiments, which are not limited here.

[0088] It should be noted that the defect similarity coefficient in this application reflects the degree of similarity between the corresponding solder joint image and the historical motherboard solder joint defect sample set. The larger the defect similarity coefficient, the higher the similarity between the corresponding solder joint image and the historical motherboard solder joint defect sample set; conversely, the smaller the defect similarity coefficient, the lower the similarity between the corresponding solder joint image and the historical motherboard solder joint defect sample set. This will not be elaborated further here. As a preferred embodiment, the defect similarity coefficient between the solder joint image and the historical motherboard solder joint defect sample set can be determined by the following method: the ratio of the largest matching value among all matching values ​​to the average value of all matching values ​​can be used as the defect similarity coefficient between the solder joint image and the historical motherboard solder joint defect sample set. Other methods can also be used to determine this in other embodiments, which are not limited here.

[0089] In some embodiments, determining the solder joint damage degree of the computer industrial control motherboard under test during operation by using all defect similarity coefficients and the defect descriptive quantity can be achieved by the following steps:

[0090] The overall matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set is determined based on all defect similarity coefficients.

[0091] The degree of solder joint damage during operation of the computer industrial control motherboard under test is determined by the comprehensive compatibility and the defect description quantity.

[0092] In specific implementation, the comprehensive matching degree reflects the overall matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set. The larger the comprehensive matching degree, the greater the overall matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set; conversely, the smaller the comprehensive matching degree, the smaller the overall matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set. Determining the comprehensive matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set based on all defect similarity coefficients can be achieved in the following way: all... The average value of the defect similarity coefficients is used as the comprehensive matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set. Other methods can also be used to determine this in other embodiments, and are not limited here. The solder joint damage degree of the computer industrial control motherboard to be tested during operation can be determined by the comprehensive matching degree and the defect description quantity. That is, the comprehensive matching degree and the defect description quantity can be multiplied, and the result of the multiplication can be used as the solder joint damage degree of the computer industrial control motherboard to be tested during operation. Other methods can also be used to determine this in other embodiments, and are not limited here.

[0093] It should be noted that the solder joint damage degree in this application reflects the severity of solder joint defects of the computer industrial control motherboard under test during operation. The higher the solder joint damage degree, the greater the severity of solder joint defects of the computer industrial control motherboard under test during operation. Conversely, the lower the solder joint damage degree, the less severe the solder joint defects of the computer industrial control motherboard under test during operation.

[0094] In step 104, the quality of use of the computer industrial control motherboard to be tested is predicted based on the solder joint damage degree, and then the computer industrial control motherboard to be tested is marked as good based on the prediction result.

[0095] In some embodiments, predicting the usability of the industrial control computer motherboard under test based on the solder joint damage degree, and then marking the industrial control computer motherboard under test as a good product based on the prediction result, can be achieved by the following steps:

[0096] Preset the good product judgment threshold for the computer industrial control motherboard to be tested;

[0097] When the solder joint damage exceeds the good product determination threshold, it is predicted that the quality of the computer industrial control motherboard under test is poor, and the computer industrial control motherboard under test is marked as a defective product.

[0098] When the solder joint damage is less than or equal to the good product determination threshold, the computer industrial control motherboard under test is predicted to be of good quality and is marked as a good product.

[0099] It should be noted that the good product judgment threshold is a threshold used to determine whether the tested computer industrial control motherboard is a good product. In specific implementation, the good product judgment threshold can be set by relevant experts using prior experience or historical experimental data. For example, the minimum solder joint damage degree among all computer industrial control motherboards that have been judged as defective products in historical experimental data can be used as the good product judgment threshold of the computer industrial control motherboard to be tested. Alternatively, the average value of all good product judgment thresholds in historical experimental data can be used as the good product judgment threshold of the computer industrial control motherboard to be tested. Other methods can also be used to set the good product judgment threshold in other embodiments, which are not specifically limited here.

[0100] In another aspect, in some embodiments, this application provides a yield prediction system for computer industrial control motherboards, with reference to... Figure 4 The figure is a schematic diagram of a yield prediction system for computer industrial control motherboards according to some embodiments of this application. The yield prediction system 400 for computer industrial control motherboards includes: a data acquisition module 401, a processing module 402, and an execution module 403, which are described below:

[0101] The acquisition module 401 in this application is mainly used to acquire images of each solder joint corresponding to the computer industrial control motherboard under test when it is running in the testing equipment, to obtain all solder joint images, and then to determine the pixel contribution of each solder joint in each solder joint image.

[0102] Processing module 402, in this application, is mainly used to perform feature decomposition on the pixels in all solder joint images based on the pixel contribution of each solder joint, to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects, and to fuse the global feature vector and the local feature vector into a defect description quantity of the solder joint in the computer industrial control motherboard to be detected.

[0103] The processing module 402 described in this application is also used to obtain a historical motherboard solder joint defect sample set, perform similarity matching between each solder joint image and the historical motherboard solder joint defect sample set, obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set, and determine the degree of solder joint damage of the computer industrial control motherboard to be tested during operation by all the defect similarity coefficients and the defect description quantity.

[0104] The execution module 403 in this application is mainly used to predict the quality of the computer industrial control motherboard to be tested based on the solder joint damage degree, and then mark the computer industrial control motherboard to be tested as good based on the prediction result.

[0105] The foregoing has detailed examples of a yield prediction method and system for computer industrial control motherboards provided in this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0106] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described yield prediction method for computer industrial control motherboards.

[0107] In some embodiments, reference Figure 5 The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the structure of a computer device implementing the yield prediction method for computer industrial control motherboards according to this application. The yield prediction method for computer industrial control motherboards in the above embodiments can be achieved through… Figure 5 The computer device 500 shown is used to implement this, and the computer device 500 includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device 500 may be a terminal device, a server or a chip.

[0108] The processor 501 can be a general-purpose processor or a special-purpose processor. For example, the processor 501 can be a central processing unit (CPU). The CPU can be used to control the computer device 500, execute software programs, and process data from the software programs. The computer device 500 may also include a communication unit 505 for inputting (receiving) and outputting (transmitting) signals.

[0109] For example, computer device 500 may be a chip, communication unit 505 may be the input and / or output circuit of the chip, or communication unit 505 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.

[0110] For example, computer device 500 may be a terminal device or a server, and communication unit 505 may be a transceiver of the terminal device or the server, or communication unit 505 may be a transceiver circuit of the terminal device or the server.

[0111] The computer device 500 may include one or more memories 502 storing a program 504. The program 504 can be executed by a processor 501 to generate instructions 503, causing the processor 501 to perform the methods described in the above method embodiments according to the instructions 503. Optionally, the memory 502 may also store data (such as a target audit model). Optionally, the processor 501 may also read data stored in the memory 502, which may be stored at the same storage address as the program 504, or the data may be stored at a different storage address than the program 504.

[0112] The processor 501 and memory 502 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.

[0113] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 501. The processor 501 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gate, transistor logic devices, or discrete hardware components.

[0114] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0115] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described method for predicting the yield of industrial control motherboards.

[0116] In summary, the yield prediction method and system for computer industrial control motherboards disclosed in this application acquires images of each solder joint of the computer industrial control motherboard under test during operation in the testing equipment, obtaining all solder joint images, and then determining the pixel contribution of each solder joint in each image; based on the pixel contribution of each solder joint, feature decomposition is performed on the pixels in all solder joint images to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects; the global feature vector and the local feature vector are fused into a defect description quantity of the solder joint in the computer industrial control motherboard under test; and historical motherboard solder joint data are obtained. A defect sample set is generated by performing similarity matching between each solder joint image and the historical motherboard solder joint defect sample set to obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set. The solder joint damage degree of the computer industrial control motherboard under test during operation is determined by all the defect similarity coefficients and the defect descriptive quantity. Based on the solder joint damage degree, the service quality of the computer industrial control motherboard under test is predicted, and then the computer industrial control motherboard under test is marked as good according to the prediction results. This can realize the early detection of potential faults of solder joint defects of computer industrial control motherboards, thereby increasing the reliability of good product detection of computer industrial control motherboards.

[0117] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0118] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for predicting the yield of industrial control motherboards for computers, characterized in that, Includes the following steps: Images of each solder joint on the computer's industrial control motherboard under test are collected during operation in the testing equipment to obtain all solder joint images, and then the pixel contribution of each solder joint in each image is determined. Based on the pixel contribution of each solder joint, feature decomposition is performed on the pixels in all solder joint images to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects. The global feature vector reflects the defect status of the solder joints of the computer industrial control motherboard under test on global defects, and the local feature vector reflects the defect status of the solder joints of the computer industrial control motherboard under test on local defects. The global feature vector and the local feature vector are fused into a defect description quantity of the solder joints in the computer industrial control motherboard under test, wherein the defect description quantity reflects the degree of defect of the solder joints in the computer industrial control motherboard under test. A historical motherboard solder joint defect sample set is obtained. Each solder joint image is matched with the historical motherboard solder joint defect sample set to obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set. The degree of solder joint damage of the computer industrial control motherboard to be tested during operation is determined by all the defect similarity coefficients and the defect description quantity. Based on the degree of solder joint damage, the quality of use of the computer industrial control motherboard under test is predicted, and then the good product mark is made on the computer industrial control motherboard under test according to the prediction result. Specifically, feature decomposition is performed on the pixels in all solder joint images based on the pixel contribution of each solder joint, resulting in the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects. The global and local feature spaces of solder joint defects in the computer industrial control motherboard to be inspected are determined by the pixels in each solder joint image and the pixel contribution of each solder joint. The global feature space is decomposed to obtain the global feature vector of the solder joints on the global defects of the computer industrial control motherboard to be inspected. The local feature space is decomposed to obtain the local feature vector of the solder joints on the local defects of the computer industrial control motherboard to be inspected. The following steps are used to fuse the global feature vector and the local feature vector into a defect description of the solder joints in the computer industrial control motherboard to be inspected: Determine the global fusion influence factor of the global feature vector; Determine the local fusion influence factor of the local feature vector; By linearly fusing the global feature vector and the local feature vector using the global fusion influence factor and the local fusion influence factor, the defect description quantity of the solder joint in the computer industrial control motherboard to be inspected is obtained.

2. The method as described in claim 1, characterized in that, Determining the pixel contribution of each solder joint in each solder joint image specifically includes: Select an image of a solder joint; Based on preset judgment pixel values, each pixel in the weld joint image is divided into defective pixels and normal pixels; The pixel contribution of the solder joint in the solder joint image is determined by all defective pixels and all normal pixels; Repeat the above steps to obtain the pixel contribution of the solder joints in the remaining solder joint images.

3. The method as described in claim 1, characterized in that, The process of performing similarity matching between each solder joint image and the historical motherboard solder joint defect sample set to obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set specifically includes: Select an image of a solder joint; Determine the matching value between the current solder joint image and each defective solder joint sample image in the historical motherboard solder joint defect sample set; The defect similarity coefficient between the solder joint image and the historical motherboard solder joint defect sample set is determined by all matching values; Repeat the above steps to obtain the defect similarity coefficient between the remaining solder joint images and the historical motherboard solder joint defect sample set.

4. The method as described in claim 1, characterized in that, The degree of solder joint damage on the computer industrial control motherboard under test during operation is determined by all defect similarity coefficients and the defect descriptive quantity, specifically including: The overall matching degree between the computer industrial control motherboard to be tested and the historical motherboard solder joint defect sample set is determined based on all defect similarity coefficients. The degree of solder joint damage during operation of the computer industrial control motherboard under test is determined by the comprehensive compatibility and the defect description quantity.

5. The method as described in claim 1, characterized in that, Based on the solder joint damage degree, the usability of the computer industrial control motherboard under test is predicted, and then a good product mark is made on the computer industrial control motherboard under test according to the prediction result. Specifically, this includes: Preset the good product judgment threshold for the computer industrial control motherboard to be tested; When the solder joint damage exceeds the good product determination threshold, it is predicted that the quality of the computer industrial control motherboard under test is poor, and the computer industrial control motherboard under test is marked as a defective product. When the solder joint damage is less than or equal to the good product determination threshold, the computer industrial control motherboard under test is predicted to be of good quality and is marked as a good product.

6. The method as described in claim 1, characterized in that, The computer industrial control motherboard is an automotive industrial control motherboard.

7. A yield prediction system for computer industrial control motherboards, comprising using the method described in any one of claims 1 to 6 to predict the yield of computer industrial control motherboards, characterized in that, This yield prediction system for industrial control motherboards includes: The acquisition module is used to acquire images of each solder joint of the computer industrial control motherboard under test during operation in the testing equipment, obtain all solder joint images, and then determine the pixel contribution of each solder joint in each solder joint image; The processing module is used to perform feature decomposition on the pixels in all solder joint images based on the pixel contribution of each solder joint, to obtain the global feature vector of the solder joint on global defects and the local feature vector of the solder joint on local defects, and to fuse the global feature vector and the local feature vector into a defect description quantity of the solder joint in the computer industrial control motherboard to be inspected. The processing module is also used to acquire a historical motherboard solder joint defect sample set, perform similarity matching between each solder joint image and the historical motherboard solder joint defect sample set, obtain the defect similarity coefficient between each solder joint image and the historical motherboard solder joint defect sample set, and determine the degree of solder joint damage of the computer industrial control motherboard to be tested during operation by all the defect similarity coefficients and the defect description quantity. The execution module is used to predict the usability of the computer industrial control motherboard under test based on the solder joint damage degree, and then mark the computer industrial control motherboard under test as good based on the prediction result.

8. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, causing the computer device to perform the yield prediction method for computer industrial control motherboards as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the yield prediction method for a computer industrial control motherboard as described in any one of claims 1 to 6.

Images