Method for Identifying Incidental Faults of BGA Welding Defects Based on Clustering Algorithm

Abstract

This application relates to the technical field of data processing, specifically to a method for identifying occasional faults of BGA soldering defects based on a clustering algorithm. The method includes: collecting X-ray data of BGA soldering fault samples, extracting edge information, and determining the contour; analyzing various degrees of defects based on the distribution characteristics between contours; determining the correlation between each fault and various degrees of defects based on the distribution differences of various degrees of defects in the historical fault sample data of each fault after threshold segmentation; clustering all the historical fault samples of each fault type and all the degrees of defects of the fault samples to be identified, and combining the correlation between each fault and various degrees of defects to correct the membership degree of the fault samples to be identified belonging to the clustering clusters of each fault type, and determining the fault type of the fault samples to be identified. This application aims to improve the accuracy of identifying BGA soldering faults.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and specifically to a method for identifying occasional faults of BGA soldering defects based on a clustering algorithm. Background Art

[0002] In the field of electronic manufacturing, the Ball Grid Array (BGA) packaging technology is widely used due to its advantages such as high density and high performance. However, the BGA soldering quality is directly related to the reliability and stability of electronic products. Occasional faults of soldering defects not only affect the performance of products but may also pose safety hazards during product use, bringing huge economic losses and reputation damage to enterprises. Therefore, improving the BGA soldering quality and accurately identifying occasional faults of soldering defects in a timely manner is of crucial significance to electronic manufacturing enterprises.

[0003] As a non-destructive detection method, X-ray detection technology can clearly present the internal structure of BGA solder joints, effectively detect soldering defects, and provide strong support for soldering quality assessment. In the actual production process, enterprises usually extract some samples from each batch of produced boards and use X-ray equipment for detection to evaluate the overall soldering quality. However, when using a clustering algorithm for soldering quality assessment, the relationship between data points and a certain cluster is usually represented by membership degree. This membership degree is fuzzy. Especially for those samples close to the cluster boundary, the membership degree may be assigned to multiple clusters, thus affecting the accurate identification of faults. Particularly in the case of BGA soldering defects, the manifestations of soldering defects may vary due to various factors, which results in a relatively fuzzy membership degree between samples and the cluster center, thus affecting the accuracy of fault identification. Summary of the Invention

[0004] In view of the above, it is necessary to provide a method for identifying occasional faults of BGA soldering defects based on a clustering algorithm to solve the above problems.

[0005] An embodiment of this application provides a method for identifying occasional faults of BGA soldering defects based on a clustering algorithm, and the method includes:

[0006] S1: Collect X-ray data of BGA soldering fault samples;

[0007] S2: Extract the edges in the X-ray data, filter the closed edges, analyze the similarity between each closed edge and other closed edges to obtain the target contours and irregular contours; determine the solder ball contours and bubble contours based on the positional relationship between the target contours; obtain the first defect degree based on the number of bubble contours included in each solder ball contour and the quantity distribution of the target contours; perform clustering on all the closed edges, conduct anomaly detection on the lengths of the closed edges in the clustering result to determine the second defect degree; analyze the distance distribution between the pixel points on all the irregular contours and the pixel points on their corresponding maximum inscribed circles, and combine with the quantity characteristics of the irregular contours to obtain the third defect degree.

[0008] S3: Determine the correlation between each type of fault and various defect degrees based on the distribution differences of various defect degrees of the historical fault sample data of each type of fault after threshold segmentation; form a triple with all the defect degrees of each fault sample, perform clustering on all the historical fault samples of each fault type and the triple of the fault sample to be recognized, and combine with the correlation between each type of fault and various defect degrees to correct the membership degrees of the fault sample to be recognized in the clustering clusters of each fault type, and determine the fault type of the fault sample to be recognized.

[0009] Among them, the specific process of analyzing the similarity between each closed edge and other closed edges to obtain the target contours and irregular contours is as follows:

[0010] Use the shape context algorithm to obtain the similarity between all pairwise combinations of closed edges in each fault sample.

[0011] For each closed edge, calculate the mean similarity between it and all other edges, set a similarity threshold, and regard the closed edges with similarity means greater than or equal to the similarity threshold as target contours, and record the remaining closed edges as irregular contours.

[0012] Among them, the process of determining the solder ball contours and bubble contours based on the positional relationship between the target contours is specifically as follows:

[0013] Extract the center of each target contour, that is, the coordinate mean of all pixel points in each target contour; use the fitting circle algorithm to fit each target contour and extract the edge radius of each target contour.

[0014] When the distance between the centers of two target contours is less than the maximum value of the edge radii of the two target contours, the two target contours are in an inclusion positional relationship.

[0015] Regard the target contour with the largest edge radius among the two target contours with an inclusion relationship as the solder ball contour, and the target contour with the smallest edge radius as the bubble contour.

[0016] Among them, the specific formula for obtaining the first degree of defect based on the number of bubble contours included in each solder ball contour and the number distribution of the target contours is as follows: ; where represents the first degree of defect; represents the number of target contours; L represents the number of solder ball contours containing bubble contours; represents the number of bubble contours contained in the i-th solder ball contour containing bubble contours; represents the length of the j-th bubble contour in the i-th solder ball contour containing bubble contours, and norm() represents the normalization function.

[0017] Among them, the specific process of clustering all closed edges, performing anomaly detection on the lengths of the closed edges in the clustering result, and determining the second degree of defect is as follows:

[0018] Denote all the closed edges in the clustering cluster with the largest number of elements as normal contours, and take the average value of the lengths of all normal contours as the standard solder ball contour length; denote all the remaining contours as abnormal contours;

[0019] Perform anomaly detection on the lengths of all closed edges and the length of the standard solder ball contour, and take the mean value of the anomaly values of all abnormal contours as the second degree of defect.

[0020] Among them, the specific method of analyzing the distance distribution between the pixel points on all irregular contours and the pixel points on their corresponding maximum inscribed circles, and combining the quantity characteristics of the irregular contours to obtain the third degree of defect is as follows:

[0021] Obtain the minimum distance between each pixel point on each irregular contour and the edge pixel points on the maximum inscribed circle of the same irregular contour; calculate the ratio of all irregular contours to all target contours, denoted as the irregular ratio; take the normalized result of the positive fusion of the cumulative sum of the minimum distances of all irregular contours and the irregular ratio as the third degree of defect.

[0022] Among them, the specific process of determining the correlation between each type of fault and various degrees of defect based on the distribution differences of various degrees of defect of the historical fault sample data of each type of fault after threshold segmentation is as follows:

[0023] Perform threshold segmentation on any degree of defect data of all historical fault samples of each type of fault using the Otsu threshold algorithm to obtain the first segmentation threshold;

[0024] Denote the mean value of all the above-mentioned any degree of defect in the category with the largest number of elements after segmentation as the first mean value; denote the mean value of all the above-mentioned any degree of defect in the other category as the second mean value; calculate the absolute value of the difference between the first mean value and the second mean value, denoted as the first absolute difference;

[0025] Calculate the absolute value of the difference between the numbers of the two types of data elements after segmentation, which is denoted as the second absolute value of the difference;

[0026] The correlation is negatively correlated with both the first absolute value of the difference and the second absolute value of the difference.

[0027] Among them, the process of correcting the membership degree of the fault sample to be recognized belonging to each type of clustering cluster is as follows:

[0028] Based on the differences in various degrees of defects between the historical fault samples of each fault and the clustering cluster center, determine the clustering clusters of each fault type; record the corrected membership degree of the fault sample to be recognized belonging to the first fault clustering cluster as , and its formula form is: , where represents the initial membership degree of the fault sample to be recognized belonging to the first fault clustering cluster, represents the correlation between the first fault and the first degree of defect, represents the square of the difference in the first degree of defect between the fault sample to be recognized and the cluster center of the first fault clustering cluster, represents the correlation between the first fault and the second degree of defect, represents the square of the difference in the second degree of defect between the fault sample to be recognized and the cluster center of the first fault clustering cluster, represents the correlation between the first fault and the third degree of defect, represents the square of the difference in the third degree of defect between the fault sample to be recognized and the cluster center of the first fault clustering cluster.

[0029] Among them, the specific process of determining the clustering clusters of each fault type is as follows:

[0030] Obtain the cluster center of each clustering cluster, calculate the difference between the various degrees of defect of the clustering cluster center and the average value of the various degrees of defect of all historical fault samples of various faults, and select the clustering cluster with the smallest difference as the clustering cluster of various faults.

[0031] Among them, the steps of determining the fault type of the fault sample to be recognized include:

[0032] Take the fault type of the clustering cluster with the largest corrected membership degree as the fault type of the fault sample to be recognized.

[0033] This application has at least the following beneficial effects:

[0034] This application first collects X-ray data of BGA soldering fault samples, extracts edge information, and determines the contour. Based on the distribution characteristics between the contours, various degrees of defects are analyzed. The beneficial effect is that by analyzing the characteristics of the causes of known BGA soldering occasional faults, the accuracy of subsequent fault identification based on these degrees of defects is improved. Based on the distribution differences of various degrees of defects in the historical fault sample data of each fault after threshold segmentation, the correlation between each fault and various degrees of defects is determined. The beneficial effect is that analyzing the distribution of different fault types at each degree of defect can more accurately identify the occurrence of specific faults, thereby improving the accuracy of fault identification. Cluster all degrees of defects of all historical fault samples and the fault samples to be identified for each fault type, and combine the correlation between each fault and various degrees of defects to correct the membership degree of the fault samples to be identified belonging to the clustering clusters of each fault type. By correcting the membership degree of the samples to be identified, the ambiguity of boundary samples can be reduced, making the clustering results more accurate and improving the accuracy of the clustering algorithm. This helps to accurately classify those fault types that are difficult to distinguish, especially for fault samples with high similarity, and improves the efficiency of fault identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flowchart of the method for identifying occasional BGA soldering faults based on a clustering algorithm provided by this application;

[0036] Figure 2 is a schematic diagram for obtaining various degrees of defects provided by this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] In the description of the embodiments of this application, words such as "exemplary", "or", "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary", "or", "for example" is intended to present related concepts in a specific manner.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used in the description of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0039] In addition, it should be noted that the terms "first" and "second" in this application and its accompanying drawings are used to distinguish similar objects, rather than to describe a specific order or sequence. In the methods disclosed in the embodiments of this application or shown in the flowcharts, including one or more steps for implementing the methods, without departing from the scope of protection of this application, the execution order of multiple steps can be interchanged with each other, and some steps can also be deleted.

[0040] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs.

[0041] This application first proposes a method for identifying occasional faults in BGA soldering defects based on a clustering algorithm, which is applied to the field of data processing technology. Referring to the attached Figure 1 , the method includes the following steps:

[0042] S1: Collect X-ray data of BGA soldering fault samples.

[0043] In actual production, this application considers extracting some samples from each batch of produced circuit boards and using X-ray equipment for detection to evaluate the overall soldering quality. Specifically, this application mainly considers fault identification of fault samples, so X-ray data of each fault sample is obtained, and median filtering is used to denoise the X-ray data of each fault sample. Among them, median filtering is a well-known existing technology, and this application will not elaborate on it.

[0044] S2: Extract the edges in the X-ray data, screen the closed edges, analyze the similarity between each closed edge and other closed edges to obtain the target contours and irregular contours; determine the solder ball contours and bubble contours according to the positional relationship between the target contours; obtain the first defect degree based on the number of bubble contours contained in each solder ball contour and the quantity distribution of the target contours; perform clustering on all closed edges, perform anomaly detection on the lengths of the closed edges in the clustering results to determine the second defect degree; analyze the distance distribution between the pixel points on all irregular contours and the pixel points on their corresponding maximum inscribed circles, and combine the quantity characteristics of the irregular contours to obtain the third defect degree.

[0045] The common causes of soldering defects can be mainly divided into three categories: one is the open solder caused by the presence of bubbles inside the solder ball, denoted as the first fault; the second is the open solder caused by the enlargement of BGA solder balls, denoted as the second fault; the third is the open solder caused by insufficient solder volume due to vias, denoted as the third fault. Therefore, it is necessary to detect the BGA soldering quality and perform feature analysis and clustering according to these three fault causes, so as to realize the identification and diagnosis of faults.

[0046] Specifically, the Canny edge detection algorithm is used to obtain all the edges in the X-ray data of each fault sample. In the embodiments of the present application, the Canny edge detection algorithm is adopted. Then, each edge is judged to determine whether it is a closed edge. In this embodiment, the method for judging a closed edge is as follows: for each edge, if each edge pixel point has adjacent edge pixel points, then the edge is a continuous edge; if all the edge pixel points on the continuous edge have more than two edge pixel points, the edge is a closed edge. The fact that the edge is a closed edge indicates that the edge is a solder ball on the BGA solder joint, and subsequent analysis is performed based on the closed edge.

[0047] Since the contours of the solder balls on the BGA solder joints and the contours of the air bubbles inside the solder balls are usually in a closed shape and their appearances are close to circular, their morphologies have high similarity. Therefore, the shape context algorithm is adopted in the present application to analyze these closed edges: obtain the similarity between all pairwise combinations of the closed edges in each fault sample. The stronger the similarity between a closed edge and the contours of other closed edges, the greater the possibility that the closed edge is the target contour, that is, the greater the possibility that it is the solder ball contour or the air bubble contour; otherwise, it may be other irregular contours. For each closed edge, calculate the average similarity between it and all other edges, set a similarity threshold, and regard the closed edges with the average similarity greater than or equal to the similarity threshold as the target contours, and record the number of target contours as P; regard the remaining closed edges as irregular contours.

[0048] Extract the center of each target contour, that is, the average coordinate of all pixel points in each target contour; use the fitting circle algorithm to fit each target contour and extract the edge radius of each target contour; compare the distance between the centers of the pairwise combinations of the target contours with the radii of the corresponding target contours to obtain the positional relationship between the pairwise combinations of the target contours: when the distance between the centers of two target contours is less than the maximum value of the edge radii of the two target contours, the two target contours are in an inclusion positional relationship. If the edge radius of target contour A is greater than the edge radius of target contour B, then target contour A includes target contour B; it should be understood that if a target contour includes a target contour, the target contour is the solder ball contour and the included target contour is the air bubble contour; since the same solder ball contour may include multiple air bubble contours, extract the number of solder ball contours that include air bubble contours and record it as L; further, record the number of air bubble contours included in the i-th solder ball contour that includes air bubble contours as .

[0049] Based on the number of air bubble contours included in each solder ball contour and the distribution of the number of target contours, obtain the first degree of defect Gr, and its formula form is: ; where represents the length of the j-th bubble contour in the i-th solder ball contour containing the bubble contour, and norm() represents the normalization function.

[0050] It should be understood that when the ratio of the solder balls containing bubbles in the X-ray data to the number of target contours is larger, that is, the larger it is, and the more bubbles in the solder ball, and the larger the corresponding bubbles, it indicates that the degree of solder ball bubble defect is larger, that is, the first defect degree is larger.

[0051] S202: Analyze the defect feature of open soldering caused by the enlargement of BGA solder balls to obtain the second defect degree.

[0052] Specifically, perform K-means clustering on all closed edges. In this embodiment, the absolute value of the difference in the lengths of the closed edges is used as the clustering metric, and the number of clusters is set to 2. All the closed edges in the cluster with the largest number of elements are recorded as normal contours, and the average value of the lengths of all normal contours is used as the standard solder ball contour length; all the remaining contours are recorded as abnormal contours.

[0053] Since the fault feature is that the solder balls at the fault point are larger than other solder balls, the lengths of all closed edges and the length of the standard solder ball contour are analyzed using the Local Outlier Factor (LOF) algorithm to obtain the LOF outlier values of each abnormal contour. The average value of the LOF outlier values of all abnormal contours is used as the second defect degree, denoted as PLOF; it should be understood that the larger the value of the second defect degree, the greater the possibility that the fault of the fault sample is open soldering.

[0054] S203: Analyze the defect feature of insufficient solder volume caused by vias to obtain the third defect degree.

[0055] The third type of defect usually occurs when there are vias in the solder joints. During the reflow soldering process, some solder balls may flow into the vias due to capillary action, resulting in insufficient solder volume. Sometimes, similar problems may also occur in the area of the via close to the pad. This situation usually shows as a decrease in the volume of the solder ball in the X-ray data. When too much solder is sucked away by the via, it may lead to open soldering at the solder joint.

[0056] The closed contours are divided into target contours and irregular contours. The irregular contours may be caused by capillary action. Therefore, it is necessary to calculate the degree of irregularity of these edge contours. These irregular contours are usually caused by the deformation of the originally nearly circular solder balls during the soldering process. In this embodiment, first, the largest inscribed circle of the irregular contour is extracted. Then, it is analyzed by calculating the distance between the irregular contour and the nearest edge on its largest inscribed circle. The closer the distance, the smaller the degree of irregularity. On the contrary, the larger the distance, the greater the degree of irregularity. The specific calculation method is as follows: Obtain the minimum distance between each pixel point on each irregular contour and the edge pixel points on the largest inscribed circle of the same irregular contour; Calculate the ratio of all irregular contours to all target contours, denoted as the irregular proportion; Take the normalized result of the positive fusion of the sum of the minimum distances of all irregular contours and the irregular proportion as the third defect degree. In this embodiment, the positive fusion of multiple variables adopts a multiplication calculation method.

[0057] It should be understood that when the irregular proportion is larger and the distance distribution between the pixel points on the corresponding irregular contour and the largest inscribed circle is also larger, it indicates that the degree of irregularity of the solder balls in the image is greater, and thus the third defect degree is greater.

[0058] Among them, the schematic diagrams for obtaining various defect degrees are as Figure 2 shown.

[0059] S3: Based on the distribution differences of various defect degrees of the historical fault sample data of each fault after threshold segmentation, determine the correlation between each fault and various defect degrees; Form a triple with all the defect degrees of each fault sample, cluster the triples of all historical fault samples and the to-be-identified fault sample of each fault type, and combine the correlation between each fault and various defect degrees to correct the membership degree of the to-be-identified fault sample belonging to the clustering cluster of each fault type, and determine the fault type of the to-be-identified fault sample.

[0060] According to the three defect degrees corresponding to the three known fault causes of the fault samples, form a triple with the three defect degrees of each historical fault sample, and cluster based on these three features. In this embodiment, the fuzzy clustering (Fuzzy C-Mean, FCM) algorithm is used for clustering the historical fault samples, and the number of clustering clusters is set to 3. Since there is an association between different defect degrees and fault types when a fault occurs, the importance of each defect degree needs to be considered during clustering to correct the obtained membership degree.

[0061] First, obtain the parameter values of all defect degrees of historical fault samples of known fault types, analyze the parameter characteristics reflected in different faults, and analyze the importance of various defect degree parameters in different faults. Since the analysis method for each fault type is the same, in the embodiment of the present application, the fault of having air bubbles in solder balls is recorded as the first fault, and taking the first fault and the first defect degree as an example, the correlation between the first defect degree and the first fault is recorded as the first correlation. The specific acquisition method is as follows: perform threshold segmentation on the first defect degree data of all historical fault samples with the fault being the first fault using the Otsu threshold algorithm to obtain the first segmentation threshold; record the mean value of all first defect degrees in the category with the largest number of elements after segmentation as the first mean value; record the mean value of all first defect degrees in the other category of data as the second mean value; calculate the absolute value of the difference between the first mean value and the second mean value, and record it as the first absolute difference value; calculate the absolute value of the difference between the number of elements in the two categories of data after segmentation, and record it as the second absolute difference value; the first correlation has a negative correlation with both the first absolute difference value and the second absolute difference value.

[0062] In this embodiment, the first absolute difference value is recorded as T, the second absolute difference value is recorded as S, and the specific formula form of the first correlation is: ; where exp() represents the exponential function with the natural constant as the base. It should be understood that the greater the difference between the two categories of data after analyzing the threshold segmentation, the smaller the correlation, and the smaller the difference, the greater the correlation, indicating that the first fault can be accurately judged only based on this parameter.

[0063] Adopt the same calculation steps as the first correlation to calculate the correlation between any fault and any defect degree.

[0064] Next, perform FCM clustering on all triples of the fault sample to be identified and the historical fault samples, obtain the cluster center of each cluster, calculate the difference between the various defect degrees of the cluster center of the cluster and the average value of the various defect degrees of all historical fault samples of the first fault, and select the cluster with the smallest difference as the first fault cluster. Then, the correction formula for the membership degree of the data to be analyzed belonging to the first fault cluster is specifically: , where represents the corrected membership degree of the fault sample to be identified belonging to the first fault cluster, represents the initial membership degree of the fault sample to be identified belonging to the first fault cluster, represents the correlation between the first fault and the first defect degree, represents the square of the difference in the first defect degree between the fault sample to be identified and the cluster center of the first fault cluster, represents the correlation between the first fault and the second defect degree, It represents the square of the difference in the second defect degree between the fault sample to be recognized and the cluster center of the first fault cluster. It represents the correlation between the first fault and the third defect degree. It represents the square of the difference in the third defect degree between the fault sample to be recognized and the cluster center of the first fault cluster.

[0065] It should be understood that when judging whether a fault sample belongs to the first fault, the larger the initial membership degree in its initial clustering result, and the smaller the square of the difference in various defect degrees between the corresponding fault sample to be recognized and the cluster center of the first fault cluster, the more likely the fault type that the fault sample to be recognized may have is the first fault.

[0066] Correspondingly, obtain the corrected membership degrees of the fault sample to be recognized belonging to the second fault cluster and the third fault cluster. 、 Normalize all the corrected membership degrees of each fault sample to be recognized. The specific calculation method is to use the sum of all the corrected membership degrees as the denominator and the corresponding membership degree as the numerator. Thus, the correction of the membership degrees of the data to be detected belonging to various fault type clusters is completed.

[0067] Take the fault type corresponding to the cluster with the largest membership degree after normalization of the fault sample to be recognized as the fault type of the fault sample to be recognized. Thus, the identification of the occasional fault of BGA soldering defect is completed.

[0068] The embodiment of the present application provides a method for identifying occasional faults of BGA soldering defects based on a clustering algorithm. The method includes: First, collect X-ray data of BGA soldering fault samples, extract edge information, and determine the contour; based on the distribution characteristics between the contours, analyze various defect degrees, and its beneficial effect is to improve the accuracy of subsequent fault identification according to these defect degrees by analyzing the characteristics of the known causes of occasional BGA soldering defects; based on the distribution differences of various defect degrees of the historical fault sample data of each fault after threshold segmentation, determine the correlation between each fault and various defect degrees, and its beneficial effect is to analyze the distribution of different fault types at each defect degree, which can more accurately identify the occurrence of specific faults, thereby improving the accuracy of fault identification; cluster all the defect degrees of all historical fault samples of each fault type and the fault samples to be recognized, and combine the correlation between each fault and various defect degrees to correct the membership degrees of the fault samples to be recognized belonging to the cluster of each fault type. By correcting the membership degrees of the samples to be recognized, the ambiguity of boundary samples can be reduced, making the clustering result more accurate and improving the accuracy of the clustering algorithm. This helps to accurately classify those fault types that are difficult to distinguish, especially for fault samples with high similarity, and improves the efficiency of fault identification.

[0069] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code that contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions noted in the blocks may occur in a different order than that noted in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. Each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0070] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included within the protection scope of the present application.

Claims

1. A method for identifying occasional failures of BGA soldering defects based on a clustering algorithm, characterized in that The method includes the following steps: S1: Collect X-ray data of BGA welding fault samples; S2: Extract the edges in the X-ray data, screen the closed edges, analyze the similarity between each closed edge and other closed edges, obtain the target contours and irregular contours; Determine the solder ball contours and bubble contours according to the positional relationship between the target contours; Based on the number of bubble contours contained in each solder ball contour and the number distribution of the target contours, obtain the first defect degree; Cluster all the closed edges, perform anomaly detection on the lengths of the closed edges in the clustering result, and determine the second defect degree; Analyze the distance distribution between the pixel points on all the irregular contours and the pixel points on their corresponding maximum inscribed circles, and combine the quantity characteristics of the irregular contours to obtain the third defect degree; S3: Based on the distribution differences of various defect degrees of the historical fault sample data of each fault after threshold segmentation, determine the correlation between each fault and various defect degrees; Combine all the defect degrees of each fault sample into a triple, cluster the triples of all the historical fault samples and the to-be-identified fault sample of each fault type, and combine the correlation between each fault and various defect degrees to correct the membership degrees of the to-be-identified fault sample belonging to the clustering clusters of each fault type, and determine the fault type of the to-be-identified fault sample; The process of correcting the membership degrees of the to-be-identified fault sample belonging to the clustering clusters of each fault type is as follows: Determine the clustering clusters of each fault type based on the differences in the degrees of various defects between the historical fault samples of each fault and the clustering cluster centers; denote the corrected membership degree of the fault sample to be identified belonging to the first fault clustering cluster as Its formula form is: , where represents the initial membership degree of the fault sample to be identified belonging to the first fault clustering cluster, represents the correlation between the first fault and the first degree of defect, represents the square of the difference in the first degree of defect between the fault sample to be identified and the cluster center of the first fault clustering cluster, represents the correlation between the first fault and the second degree of defect, represents the square of the difference in the second degree of defect between the fault sample to be identified and the cluster center of the first fault clustering cluster, represents the correlation between the first fault and the third degree of defect, represents the square of the difference in the third degree of defect between the fault sample to be identified and the cluster center of the first fault clustering cluster, represents the normalization function.

2. The method for identifying occasional failures of BGA soldering defects based on the clustering algorithm according to claim 1, wherein, The specific process of analyzing the similarity between each closed edge and other closed edges to obtain the target contours and irregular contours is as follows: Use the shape context algorithm to obtain the similarity between all pairwise combinations of closed edges in each fault sample; For each closed edge, calculate the mean similarity between it and all other edges, set a similarity threshold, and regard the closed edges with similarity means greater than or equal to the similarity threshold as target contours, and record the remaining closed edges as irregular contours.

3. The method for identifying occasional faults of BGA soldering defects based on the clustering algorithm according to claim 1, characterized in that, The specific process of determining the solder ball contours and bubble contours according to the positional relationship between the target contours is as follows: Extract the center of each target contour, that is, the coordinate mean of all pixel points in each target contour; Use the circle fitting algorithm to fit each target contour and extract the edge radius of each target contour; When the distance between the centers of two target contours is less than the maximum value of the edge radii of the two target contours, the two target contours are in an inclusion positional relationship; Regard the target contour with the largest edge radius among the two target contours with an inclusion relationship as the solder ball contour, and the target contour with the smallest edge radius as the bubble contour.

4. The method for identifying the occasional failure of BGA soldering defects based on the clustering algorithm according to claim 1, wherein, The specific formula for obtaining the first degree of defect based on the number of bubble contours included in each solder ball contour and the number distribution of target contours is as follows: ; where represents the first degree of defect; represents the number of target contours; L represents the number of solder ball contours containing bubble contours; represents the number of bubble contours contained in the i-th solder ball contour containing bubble contours; represents the length of the j-th bubble contour in the i-th solder ball contour containing bubble contours, represents the normalization function.

5. The method for identifying occasional faults of BGA soldering defects based on the clustering algorithm according to claim 1, wherein, The specific process of clustering all the closed edges, performing anomaly detection on the lengths of the closed edges in the clustering result, and determining the second defect degree is as follows: Regard all the closed edges in the clustering cluster with the largest number of elements as normal contours, and take the average length of all the normal contours as the standard solder ball contour length; Record all the remaining contours as abnormal contours; Perform anomaly detection on the lengths of all the closed edges and the length of the standard solder ball contour, and take the mean value of the anomaly values of all the abnormal contours as the second defect degree.

6. The method for identifying occasional faults of BGA soldering defects based on the clustering algorithm according to claim 1, wherein Analyze the distance distribution between the pixel points on all irregular contours and the pixel points on their corresponding maximum inscribed circles, and combine the quantitative characteristics of the irregular contours to obtain the third degree of defect. Specifically: Obtain the minimum distance between each pixel point on each irregular contour and the edge pixel points on the maximum inscribed circle of the same irregular contour; calculate the ratio of all irregular contours to all target contours, denoted as the irregular ratio; use the normalized result of the positive fusion of the sum of the minimum distances of all irregular contours and the irregular ratio as the third degree of defect.

7. The method for identifying occasional faults of BGA soldering defects based on the clustering algorithm according to claim 1, characterized in that, The process of determining the correlation between each type of fault and various degrees of defect based on the distribution differences of various degrees of defect in the historical fault sample data of each type of fault after threshold segmentation is specifically as follows: Perform threshold segmentation on any degree-of-defect data of all historical fault samples of each type of fault using the Otsu threshold algorithm to obtain the first segmentation threshold; Denote the mean value of all the above-mentioned any degree of defect in the category with the largest number of elements after segmentation as the first mean value; Denote the mean value of all the above-mentioned any degree of defect in the other category of data as the second mean value; calculate the absolute value of the difference between the first mean value and the second mean value, denoted as the first absolute difference value; Calculate the absolute value of the difference between the number of elements in the two categories of data after segmentation, denoted as the second absolute difference value; The correlation is negatively correlated with both the first absolute difference value and the second absolute difference value.

8. The method for identifying the occasional failure of BGA soldering defects based on the clustering algorithm according to claim 1, characterized in that, The specific process of determining the clustering clusters of each type of fault is as follows: Obtain the cluster center of each clustering cluster, calculate the difference between the various degrees of defect of the clustering cluster center and the average value of the various degrees of defect of all historical fault samples of various types of faults, and select the clustering cluster with the smallest difference as the clustering cluster of various types of faults.

9. The method for identifying occasional faults of BGA soldering defects based on the clustering algorithm according to claim 1, characterized in that, The steps of determining the type of fault of the fault sample to be identified include: Take the type of fault of the clustering cluster with the largest corrected membership degree as the type of fault of the fault sample to be identified.

Images