Post-welding detection method based on automatic laser welding and related equipment

Abstract

The invention relates to the technical field of welding, in particular to a post-welding detection method and related equipment based on automatic laser welding, and the method comprises the following steps: acquiring a positioning image and an initial image of a welding spot area based on a preset shooting rule; acquiring a standard position based on the positioning image; preprocessing the initial image, and obtaining a target image; obtaining target geometric information of the edge contour of the chip based on the target image; matching the target geometric information with preset standard geometric information, and obtaining a position offset; based on the position offset and the standard position, acquiring a pixel position of a bonding pad in the target image; and obtaining a welding spot detection result based on the target image, the pixel position and a preset deep learning model. According to the invention, the efficiency and accuracy of welding spot detection can be improved.

Description

Technical Field

[0001] The present application relates to the field of welding technology, and in particular to a post-weld detection method based on automatic laser welding and related equipment. Background Art

[0002] At present, in the packaging and assembly process of electronic products, solder is needed to connect different electronic components. For example, in the flip chip process of connecting the chip to the printed circuit board, the surface mounting (SMT) process of connecting the BGA packaged product to the circuit board, etc., solder balls are needed to connect the chip and the printed circuit board or to connect the BGA (ball grid array) package and the circuit board.

[0003] The quality of welding directly affects not only the appearance, but also the quality of electronic products. Therefore, after welding, the quality of the solder joints needs to be inspected and evaluated. The traditional method of inspecting and evaluating the quality of solder joints is mainly through manual visual inspection. However, manual visual inspection is not only time-consuming and labor-intensive, but also inefficient. In addition, the inspection results mainly rely on the technical experience of the inspectors, which is highly subjective. If the inspectors have poor technical experience, it is easy to lead to low accuracy of the inspection results. Summary of the invention

[0004] In order to help improve the efficiency and accuracy of weld spot detection, the present application provides a post-weld detection method based on automatic laser welding and related equipment.

[0005] In the first aspect, the present application provides a post-weld detection method based on automatic laser welding, which adopts the following technical solution: A post-weld detection method based on automatic laser welding, comprising: Based on the preset shooting rules, the positioning image and the initial image of the welding spot area are obtained; Based on the positioning image, obtaining a standard position; Preprocessing the initial image and acquiring a target image; Based on the target image, acquiring target geometric information of the chip edge contour; Matching the target geometric information with preset standard geometric information and obtaining a position offset; Based on the position offset and the standard position, obtaining the pixel position of the pad in the target image; Based on the target image, the pixel position and a preset deep learning model, a solder joint detection result is obtained.

[0006] By adopting the above technical solution, the standard position is first obtained according to the positioning image, and then the target geometric information of the chip edge contour is obtained according to the target image, and it is matched with the preset standard geometric information to obtain the position offset, and then the pixel position of the pad in the target image is calculated according to the position offset and the standard position, and finally the target image and the pixel position are input into the preset deep learning model, and the solder joints in the target image are analyzed and detected by the preset deep learning model, so as to obtain the solder joint detection results; Before inspecting the solder joints, they are first accurately positioned to provide an accurate chip pad inspection area for image inspection, which helps to more accurately identify the solder joint locations, thereby improving the accuracy of the solder joint inspection results. By inspecting the target image through a preset deep learning model, the quality of the solder joints can be detected without manual inspection. This not only helps to improve inspection efficiency, but also helps to reduce dependence on the experience and technology of inspectors, reducing the subjectivity of manual inspection, thereby helping to further improve the accuracy of inspection results.

[0007] Optionally, obtaining the solder joint detection result based on the target image, the pixel position and a preset deep learning model includes: Inputting the target image and the pixel position into the preset deep learning model, and obtaining the solder joint features; Comparing the solder joint feature with a preset standard feature to generate a comparison result; Based on the comparison result and the preset scoring standard, obtaining a single score corresponding to the solder joint feature; Based on the single scoring and the preset scoring criteria, a solder joint detection result is obtained.

[0008] Optionally, obtaining a single score corresponding to the solder joint feature based on the comparison result and a preset scoring standard includes: Based on the comparison result, obtaining the matching degree corresponding to the different solder joint features; Determine whether the matching degree is lower than a preset matching threshold; If the matching degree is lower than the preset matching threshold, determining whether the solder joint feature is a designated feature; If the solder joint feature is the specified feature, the single item score corresponding to the solder joint feature is set to 0 points; If the solder joint feature is not the designated feature, a single score corresponding to the solder joint feature is obtained based on the matching degree and the preset scoring standard.

[0009] Optionally, the solder joint inspection result includes excellent, good and unqualified; and obtaining the solder joint inspection result based on the single scoring and the preset scoring standard includes: Determine whether there is a single item score of 0; If any of the single item scores is 0, the solder joint inspection result is set as unqualified; If there is no single item score of 0, obtaining a first quantity based on the single item score and a first score threshold; Determining whether the first quantity is greater than or equal to a first quantity threshold; If the first number is greater than or equal to a first number threshold, the solder joint detection result is set to be excellent.

[0010] Optionally, after determining whether the first quantity is greater than or equal to a first quantity threshold, the method further includes: If the first quantity is less than the first quantity threshold, obtaining a second quantity based on the single item score and a second score threshold; Determining whether the second quantity is greater than a second quantity threshold; If the second number is greater than the second number threshold, setting the solder joint detection result as unqualified; If the second number is less than or equal to the second number threshold, obtaining a total score of the solder joint based on the single item score; Get the feature quantity corresponding to the solder joint feature; Based on the total score of the solder joint and the number of features, obtaining an average score for each item; Determine whether the single item average score is greater than or equal to the second scoring threshold; If the single item average score is greater than or equal to the second scoring threshold, the solder joint detection result is set as good; If the single item average score is less than the second scoring threshold, the solder joint inspection result is set as unqualified.

[0011] Optionally, also include: Obtain the number of solder joints and the number of unqualified solder joints corresponding to the test results; Based on the preset label corresponding to the welding point, obtaining the target device number corresponding to the detection device; Based on the number of solder joints and the number of unqualified solder joints, obtaining a failure rate; Determining whether the failure rate exceeds a third quantity threshold; If the unqualified rate exceeds the third quantity threshold, re-inspecting the solder joint corresponding to the target device number by other detection devices, and generating a re-inspection unqualified rate; Based on the re-inspection unqualified rate and the unqualified rate, obtaining the unqualified rate difference; Based on the failure rate difference, a flag is generated.

[0012] Optionally, the specific step of generating a mark based on the non-conforming rate difference includes: Determining whether the failure rate is less than a preset failure rate threshold; If the unqualified rate is less than the preset unqualified rate threshold, marking the welding device corresponding to the welding point as an abnormal welding device; If the unqualified rate is greater than or equal to the preset unqualified rate threshold, the detection device corresponding to the welding point is marked as an abnormal detection device.

[0013] In the second aspect, the present application also discloses a post-weld detection system based on automatic laser welding, which adopts the following technical solution: A post-weld inspection system based on automatic laser welding, comprising: A first acquisition module, used to acquire a positioning image and an initial image of a welding spot area based on a preset shooting rule; A second acquisition module, used for acquiring a standard position based on the positioning image; A third acquisition module, used for preprocessing the initial image and acquiring a target image; A fourth acquisition module, used for acquiring target geometric information of a chip edge contour based on the target image; A fifth acquisition module, used to match the target geometric information with preset standard geometric information and obtain a position offset; A sixth acquisition module, used for acquiring a pixel position of the pad in the target image based on the position offset and the standard position; The seventh acquisition module is used to obtain the solder joint detection result based on the target image, the pixel position and the preset deep learning model.

[0014] By adopting the above technical solution, the standard position is first obtained according to the positioning image, and then the target geometric information of the chip edge contour is obtained according to the target image, and it is matched with the preset standard geometric information to obtain the position offset, and then the pixel position of the pad in the target image is calculated according to the position offset and the standard position, and finally the target image and the pixel position are input into the preset deep learning model, and the solder joints in the target image are analyzed and detected by the preset deep learning model, so as to obtain the solder joint detection results; Before inspecting the solder joints, they are first accurately positioned to provide an accurate chip pad inspection area for image inspection, which helps to more accurately identify the solder joint locations, thereby improving the accuracy of the solder joint inspection results. By inspecting the target image through a preset deep learning model, the quality of the solder joints can be detected without manual inspection. This not only helps to improve inspection efficiency, but also helps to reduce dependence on the experience and technology of inspectors, reducing the subjectivity of manual inspection, thereby helping to further improve the accuracy of inspection results.

[0015] In a third aspect, the present application provides a computer device, which adopts the following technical solution: An intelligent terminal comprises a memory and a processor, wherein the memory is used to store a computer program that can be run on the processor, and when the processor loads the computer program, the method of the first aspect is executed.

[0016] By adopting the above technical solution, a computer program is generated based on the method of the first aspect and stored in a memory to be loaded and executed by a processor, thereby manufacturing an intelligent terminal based on the memory and the processor for user convenience.

[0017] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium stores a computer program, and when the computer program is loaded by a processor, the method of the first aspect is executed.

[0018] By adopting the above technical solution, a computer program is generated based on the method of the first aspect and stored in a computer-readable storage medium so as to be loaded and executed by a processor. The computer-readable storage medium facilitates the readability and storage of the computer program.

[0019] In summary, this application includes the following beneficial technical effects: Before inspecting the solder joints, they are first accurately positioned to provide an accurate chip pad inspection area for image inspection, which helps to more accurately identify the solder joint locations, thereby improving the accuracy of the solder joint inspection results. By inspecting the target image through a preset deep learning model, the quality of the solder joints can be detected without manual inspection. This not only helps to improve inspection efficiency, but also helps to reduce dependence on the experience and technology of inspectors, reducing the subjectivity of manual inspection, thereby helping to further improve the accuracy of inspection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a main flow chart of a post-weld detection method based on automatic laser welding according to an embodiment of the present application; Figure 2is a flowchart of steps S201 to S204; Figure 3 is a flowchart of steps S301 to S305; Figure 4 is a flowchart of steps S401 to S405; Figure 5 is a flowchart of steps S501 to S509; Figure 6 is a flowchart of steps S601 to S607; Figure 7 is a flowchart of steps S701 to S703; Figure 8 This is a module diagram of a post-weld inspection system based on automatic laser welding in an embodiment of the present application.

[0021] Description of reference numerals: 1. First acquisition module; 2. Second acquisition module; 3. Third acquisition module; 4. Fourth acquisition module; 5. Fifth acquisition module; 6. Sixth acquisition module; 7. Seventh acquisition module. DETAILED DESCRIPTION

[0022] In a first aspect, the present application discloses a post-weld inspection method based on automatic laser welding.

[0023] Reference Figure 1 , a post-weld detection method based on automatic laser welding, comprising steps S101 to S107: Step S101: based on preset shooting rules, a positioning image and an initial image of the welding spot area are acquired.

[0024] Specifically, the preset shooting rules are the rules pre-set for shooting the positioning image and the initial image. The positioning image is used to locate the solder joints in the initial image. The initial image is the initial shot image containing the solder joints. In this embodiment, the precise placement of each chip to be inspected has a certain degree of randomness. Before inspecting the solder pad, the solder pad needs to be accurately positioned. After controlling the ring light source to be adjusted to uniform illumination and appropriate brightness, an image is taken with the post-weld inspection camera as the positioning image.

[0025] After taking the positioning image, four initial images at different angles are taken. The ring light source in the computer vision module is turned on in the order of front, back, left, and right, and the camera shoots the solder joints from a vertical direction. Under the illumination of light sources at different angles, the solder will have different pixel grayscale value distributions on the pad, showing different image features.

[0026] Step S102: Based on the positioning image, obtain the standard position.

[0027] Specifically, in this embodiment, the standard position is a reference position determined according to the positioning image.

[0028] Step S103: pre-process the initial image and obtain the target image.

[0029] Specifically, in this embodiment, through contrast enhancement, histogram equalization, filtering and other algorithms, the contrast of the welding points is improved, the noise is reduced, and it is easier to distinguish; the target image is the initial image after preprocessing.

[0030] Step S104: Based on the target image, obtain target geometric information of the chip edge contour.

[0031] Specifically, in this embodiment, the geometric information includes edge shape, size, position, etc.

[0032] Step S105: Match the target geometric information with the preset standard geometric information and obtain the position offset.

[0033] Specifically, the preset standard geometric information is the pre-set standard geometric information. By comparing the preset standard geometric information with the target geometric information, the position offset is calculated. In this embodiment, the position offset is the position offset between the edge contour of the geometric chip in the target image and the standard edge contour corresponding to the pre-set standard geometric information, including the offset distance and the offset direction.

[0034] Step S106: based on the position offset and the standard position, the pixel position of the pad in the target image is obtained.

[0035] Specifically, the pixel position of the pad in the target image can be calculated through the standard position and the position offset. In this embodiment, the pixel position of the pad can be used as the pixel position of the solder point.

[0036] Step S107: Obtain solder joint detection results based on the target image, pixel position and a preset deep learning model.

[0037] Specifically, first, the convolutional neural network layer is used to extract features from the manually pre-labeled pad image data; then, these feature vectors are classified and learned through the fully connected neural network layer; during the classification learning process, the back propagation algorithm is used to update the node weights of the neural network. Once the model completes automatic learning, the preset deep learning model will be verified on the image test set, and the tested model will be applied to the post-soldering inspection of chip pads. The solder joint inspection results include excellent, good and unqualified.

[0038] The post-weld inspection method based on automatic laser welding provided in this embodiment first obtains the standard position according to the positioning image, then obtains the target geometric information of the edge contour of the chip according to the target image, and matches it with the preset standard geometric information to obtain the position offset, and then calculates the pixel position of the pad in the target image according to the position offset and the standard position, and finally inputs the target image and the pixel position into a preset deep learning model, and analyzes and detects the solder joints in the target image through the preset deep learning model, thereby obtaining the solder joint inspection result; Before inspecting the solder joints, they are first accurately positioned to provide an accurate chip pad inspection area for image inspection, which helps to more accurately identify the solder joint locations, thereby improving the accuracy of the solder joint inspection results. By inspecting the target image through a preset deep learning model, the quality of the solder joints can be detected without manual inspection. This not only helps to improve inspection efficiency, but also helps to reduce dependence on the experience and technology of inspectors, reducing the subjectivity of manual inspection, thereby helping to further improve the accuracy of inspection results.

[0039] Reference Figure 2 In one implementation of this embodiment, step S107 obtains the solder joint detection result based on the target image, pixel position and the preset deep learning model, including steps S201 to S204: Step S201: input the target image and pixel position into a preset deep learning model, and obtain the solder joint features.

[0040] Specifically, in this embodiment, a convolutional neural network in deep learning is used to perform classification learning on pre-labeled image data, and a trained neural network model is used to evaluate the solder joint features on the chip. The solder joint features include solder leaks, tin coverage, solder ball offset, bridging, and overall solder joint morphology.

[0041] Step S202: Compare the solder joint features with the preset standard features to generate a comparison result.

[0042] Specifically, the preset standard feature is a pre-set standard feature, and the solder joint represented by the standard feature is a perfect solder joint. In this embodiment, the comparison result is the result generated after comparing the solder joint feature with the preset standard feature.

[0043] Step S203: Based on the comparison result and the preset scoring standard, the single score corresponding to the solder joint feature is obtained.

[0044] Specifically, in this embodiment, the preset scoring standard is a pre-set standard for generating corresponding scores according to the comparison results. Different comparison results correspond to different single scores, and the single scores are scores corresponding to different solder joint features.

[0045] Step S204: Based on the individual scoring and the preset scoring criteria, the solder joint inspection result is obtained.

[0046] Specifically, in this implementation, the corresponding solder joint detection results are generated according to the single item scores. The post-weld inspection method based on automatic laser welding provided in this embodiment compares the weld feature with the preset standard feature to generate a comparison result, and then obtains the single score corresponding to the weld feature according to different comparison results and preset scoring standards. Finally, the corresponding weld inspection result is generated according to the single score, and the inspection result is digitized and objectified, which reduces the subjectivity of manual inspection and helps to improve the accuracy of the weld inspection result.

[0047] Reference Figure 3 In one implementation of this embodiment, step S203 obtains a single score corresponding to the solder joint feature based on the comparison result and the preset scoring standard, including steps S301 to S305: Step S301: based on the comparison result, obtaining the matching degree corresponding to different solder joint features.

[0048] Specifically, in this embodiment, the matching degree refers to the matching degree between different weld characteristics and preset standard characteristics. The matching degree varies according to different weld characteristics, for example, including the matching degree of weld perfection, the matching degree of weld coverage, and the matching degree of weld position.

[0049] Step S302: Determine whether there is a match degree lower than a preset match threshold.

[0050] Specifically, the preset matching threshold is a pre-set judgment standard for judging whether different solder joint features are excellent, good or unqualified.

[0051] Step S303: If the matching degree is lower than the preset matching threshold, it is determined whether the solder joint feature is a designated feature.

[0052] Specifically, in this embodiment, the designated feature is a pre-set feature that has a very large or very important impact on welding, such as missing welding.

[0053] Step S304: If the solder joint feature is a specified feature, the single score corresponding to the solder joint feature is set to 0 points.

[0054] Specifically, in this embodiment, when the weld point feature is a specified feature, and the degree of match between the weld point feature and the preset standard feature is lower than the preset matching threshold, it indicates that the weld point has a greater risk or has a greater impact on the use effect after welding. Because the weld point may cause the corresponding equipment to fail to work properly, the single score corresponding to the weld point feature is directly set to 0 points.

[0055] Step S305: If the solder joint feature is not a designated feature, a single score corresponding to the solder joint feature is obtained based on the matching degree and a preset scoring standard.

[0056] Specifically, in this embodiment, if the weld feature is not a specified feature, it indicates that there is a risk in the weld or it affects the use effect after welding, but the weld will not cause the corresponding equipment to be unable to be used normally. Therefore, the single score corresponding to the weld feature is further obtained based on the matching degree and the preset scoring criteria.

[0057] The post-weld inspection method based on automatic laser welding provided in this embodiment selects the corresponding method of obtaining a single score according to different situations, which helps to reduce the situation where the corresponding equipment cannot be used normally due to ignoring or underestimating the risks of certain specified features, and further improves the accuracy of the weld spot inspection results.

[0058] Reference Figure 4 In one implementation of this embodiment, step S204 obtains the solder joint inspection result based on the single item scoring and the preset scoring standard, including steps S401 to S405: Step S401: Determine whether there is a single item with a score of 0.

[0059] Step S402: If there is a single item score of 0, the solder joint inspection result is set as unqualified.

[0060] Specifically, in this embodiment, if there is a single score of 0, it indicates that there is a very large difference between the solder joint feature corresponding to the single score and the preset standard feature, which also indicates that the solder joint feature is very poor, so the solder joint detection result is set to unqualified.

[0061] Step S403: If there is no single item score of 0, a first quantity is obtained based on the single item score and a first score threshold.

[0062] Specifically, in this embodiment, the first scoring threshold is a pre-set criterion for determining whether the solder joint feature is an excellent feature, and the first data is the number of items whose single-item scores exceed the first scoring threshold.

[0063] Step S404: Determine whether the first quantity is greater than or equal to a first quantity threshold.

[0064] Specifically, the first quantity threshold is a pre-set criterion for determining whether the solder joint result is a priority. In this embodiment, the first quantity threshold may be set to the total number of items corresponding to the solder joint features.

[0065] Step S405: If the first number is greater than or equal to the first number threshold, the solder joint detection result is set to be excellent.

[0066] Specifically, in this embodiment, if the first number is greater than or equal to the first number threshold, it indicates that all weld features are very close to the corresponding preset standard features, that is, the weld is very close to or is the standard weld, so the weld detection result is set to excellent.

[0067] The post-weld inspection method based on automatic laser welding provided in this embodiment first determines whether there is a single item with a score of 0. If so, it indicates that there is a very large difference between at least one weld feature of the weld and the preset standard feature, and the problem is relatively serious. Therefore, the weld inspection result is determined to be unqualified. If not, the corresponding weld inspection result is further obtained based on the relationship between the first quantity and the first quantity threshold. The corresponding method is selected according to different situations to generate the corresponding weld inspection result, which helps to improve the accuracy of the weld inspection result.

[0068] Reference Figure 5 In one implementation of this embodiment, after determining whether the first quantity is greater than or equal to the first quantity threshold in step S404, steps S501 to S509 are also included: Step S501: If the first quantity is less than the first quantity threshold, a second quantity is obtained based on the single item score and the second score threshold.

[0069] Specifically, in this embodiment, the second scoring threshold is a pre-set criterion for determining whether the solder joint feature is an unqualified feature, and the second data is the number of items whose single item scores are lower than the second scoring threshold.

[0070] Step S502: Determine whether the second quantity is greater than a second quantity threshold.

[0071] Specifically, the second quantity threshold is a pre-set criterion for determining whether the solder joint result is unqualified. In this embodiment, the second quantity threshold may be 60% of the total number of items corresponding to the solder joint features or other values.

[0072] Step S503: If the second number is greater than the second number threshold, the solder joint detection result is set as unqualified.

[0073] Specifically, if the second number is greater than the second number threshold, it indicates that a larger number of solder joint features among all solder joint features have large differences with corresponding preset standard features, and therefore the solder joint detection result is set as unqualified.

[0074] Step S504: if the second number is less than or equal to the second number threshold, obtaining a total score of the solder joint based on the individual scoring items.

[0075] Specifically, in this embodiment, the total score of the solder joint is the sum of all individual scores.

[0076] Step S505: Obtain the feature quantity corresponding to the solder joint feature.

[0077] Specifically, in this embodiment, the feature quantity is the total number of items corresponding to the solder joint features.

[0078] Step S506: Based on the total score of the solder joints and the number of features, obtain the average score of each item.

[0079] Specifically, in this embodiment, the single item average score is the value obtained by dividing the total score of the solder joint by the number of features.

[0080] Step S507: Determine whether the single item average score is greater than or equal to the second scoring threshold.

[0081] Step S508: If the single item average score is greater than or equal to the second scoring threshold, the solder joint inspection result is set as good.

[0082] Step S509: If the single item average score is less than the second scoring threshold, the solder joint inspection result is set as unqualified.

[0083] The post-weld inspection method based on automatic laser welding provided in this embodiment selects corresponding methods to generate corresponding weld spot inspection results according to different situations, which helps to improve the accuracy of the weld spot inspection results.

[0084] Reference Figure 6 In one implementation of this embodiment, steps S601 to S607 are also included: Step S601: Obtain the number of solder joints and the number of unqualified solder joints corresponding to the inspection results.

[0085] Specifically, in this embodiment, the number of welds is the total number of all welds, and the number of unqualified welds is the number of welds corresponding to all welds for which weld inspection results are unqualified.

[0086] Step S602: Based on the preset label corresponding to the welding point, the target device number corresponding to the detection device is obtained.

[0087] Specifically, in this embodiment, the preset label is a label pre-marked on the weld, and the label includes a number and a corresponding detection device and welding device, etc.; when performing weld detection, the preset label corresponding to the weld is identified, and all the contents in the preset label can be known; the detection device is a device for detecting the weld, and different weld devices have corresponding device numbers. The target device number is the device number of the detection device that detects the batch of welds.

[0088] Step S603: Based on the number of solder joints and the number of unqualified ones, obtain the unqualified rate.

[0089] Specifically, in this embodiment, the unqualified rate is the final value obtained by dividing the number of unqualified welds by the number of welds and then multiplying the value by 100%.

[0090] Step S604: Determine whether the failure rate exceeds a third quantity threshold.

[0091] Specifically, the third quantity threshold is a criterion for determining whether the failure rate is too high. In this embodiment, the third quantity threshold can be set to 30% or other values.

[0092] Step S605: If the unqualified rate exceeds the third quantity threshold, re-inspect the solder joint corresponding to the target device number through other inspection devices, and generate a re-inspection unqualified rate.

[0093] Specifically, if the unqualified rate exceeds the third quantity threshold, it indicates that the unqualified rate is too high. In order to improve the accuracy of the weld detection results, the part of the welds are retested using other detection devices except the detection device corresponding to the target equipment number, and the retest unqualified rate is obtained. The retest unqualified rate is the unqualified rate obtained by the retest. In this embodiment, when the number of welds is large, retesting can also be carried out by random sampling. It is worth noting that other detection devices corresponding to higher qualified rates are preferentially selected for retesting.

[0094] Step S606: Based on the re-inspection unqualified rate and the unqualified rate, obtain the unqualified rate difference.

[0095] Specifically, in this embodiment, the difference in unqualified rate is the absolute value of the difference obtained by subtracting the retested unqualified rate from the unqualified rate.

[0096] Step S607: Generate a mark based on the non-conforming rate difference.

[0097] The post-weld inspection method based on automatic laser welding provided in this embodiment, when the unqualified rate is too large, in order to further improve the accuracy of the weld spot inspection results, the batch of weld spots are retested, and it is determined whether the situation is caused by an abnormality in the inspection device based on the unqualified rate difference between the unqualified rate and the retested unqualified rate.

[0098] Reference Figure 7 In one implementation of this embodiment, step S607 generates a marking step based on the non-conforming rate difference, including steps S701 to S703: Step S701: Determine whether the failure rate is less than a preset failure rate threshold.

[0099] Specifically, in this embodiment, the preset failure rate threshold is a pre-set judgment standard for judging whether the failure rate is accurate.

[0100] Step S702: If the unqualified rate is less than a preset unqualified rate threshold, the welding device corresponding to the welding point is marked as an abnormal welding device.

[0101] Specifically, in this embodiment, if the unqualified rate is less than the preset unqualified rate threshold, it indicates that the difference between the re-tested unqualified rate and the unqualified rate is small, and therefore indicates that the unqualified rate is accurate. Therefore, there must be an abnormality in the welding device used to weld the weld, resulting in an excessively high unqualified rate, and therefore the welding device is marked as an abnormal device.

[0102] Step S703: If the unqualified rate is greater than or equal to the preset unqualified rate threshold, the detection device corresponding to the welding point is marked as an abnormal detection device.

[0103] Specifically, in this embodiment, if the unqualified rate is greater than or equal to the preset unqualified rate threshold, it indicates that there is a large difference between the re-tested unqualified rate and the unqualified rate, and therefore indicates that the unqualified rate is inaccurate. Therefore, there must be an abnormality in the detection device that detects the weld, that is, the detection device corresponding to the target device number, which leads to an excessively high unqualified rate. Therefore, the detection device is marked as an abnormal device.

[0104] The post-weld inspection method based on automatic laser welding provided in this embodiment can determine whether the failure rate is too high due to an abnormality in the detection device or the welding device, according to different situations, and mark the abnormality, which helps the staff to repair the abnormal device in time, thereby helping to improve the accuracy of the weld spot inspection results.

[0105] In a second aspect, the present application also discloses a post-weld inspection system based on automatic laser welding.

[0106] Reference Figure 8 , a post-weld inspection system based on automatic laser welding, comprising: A first acquisition module, used to acquire a positioning image and an initial image of a welding spot area based on a preset shooting rule; A second acquisition module is used to acquire a standard position based on the positioning image; The third acquisition module is used to pre-process the initial image and acquire the target image; A fourth acquisition module is used to acquire target geometric information of the chip edge contour based on the target image; A fifth acquisition module, used to match the target geometric information with the preset standard geometric information and obtain a position offset; A sixth acquisition module, used for acquiring a pixel position of the pad in the target image based on the position offset and the standard position; The seventh acquisition module is used to obtain the solder joint detection results based on the target image, pixel position and preset deep learning model.

[0107] In a third aspect, an embodiment of the present application discloses an intelligent terminal, including a memory and a processor, wherein the memory is used to store a computer program that can be run on the processor, and when the processor loads the computer program, a post-weld detection method based on automatic laser welding of the above embodiment is executed.

[0108] In a fourth aspect, an embodiment of the present application discloses a computer-readable storage medium, and a computer program is stored in the computer-readable storage medium, wherein when the computer program is loaded by a processor, a post-weld inspection method based on automatic laser welding of the above embodiment is executed.

[0109] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A post-weld inspection method based on automatic laser welding, characterized in that: include: Based on the preset shooting rules, the positioning image and the initial image of the welding spot area are obtained; Based on the positioning image, obtaining a standard position; Preprocessing the initial image and acquiring a target image; Based on the target image, acquiring target geometric information of the chip edge contour; Matching the target geometric information with preset standard geometric information and obtaining a position offset; Based on the position offset and the standard position, obtaining the pixel position of the pad in the target image; Based on the target image, the pixel position and a preset deep learning model, a solder joint detection result is obtained.

2. A post-weld detection method based on automatic laser welding according to claim 1, characterized in that: The obtaining of solder joint detection results based on the target image, the pixel position and the preset deep learning model includes: Inputting the target image and the pixel position into the preset deep learning model, and obtaining the solder joint features; Comparing the solder joint feature with a preset standard feature to generate a comparison result; Based on the comparison result and the preset scoring standard, obtaining a single score corresponding to the solder joint feature; Based on the single scoring and the preset scoring criteria, a solder joint detection result is obtained.

3. A post-weld detection method based on automatic laser welding according to claim 2, characterized in that: The obtaining of a single score corresponding to the solder joint feature based on the comparison result and the preset scoring standard includes: Based on the comparison result, obtaining the matching degree corresponding to the different solder joint features; Determine whether the matching degree is lower than a preset matching threshold; If the matching degree is lower than the preset matching threshold, determining whether the solder joint feature is a designated feature; If the solder joint feature is the specified feature, the single item score corresponding to the solder joint feature is set to 0 points; If the solder joint feature is not the designated feature, a single score corresponding to the solder joint feature is obtained based on the matching degree and the preset scoring standard.

4. The post-weld detection method based on automatic laser welding according to claim 2 is characterized in that: The solder joint inspection results include excellent, good and unqualified; the obtaining of solder joint inspection results based on the single scoring and the preset scoring criteria includes: Determine whether there is a single item score of 0; If any of the single item scores is 0, the solder joint inspection result is set as unqualified; If there is no single item score of 0, obtaining a first quantity based on the single item score and a first score threshold; Determining whether the first quantity is greater than or equal to a first quantity threshold; If the first number is greater than or equal to a first number threshold, the solder joint detection result is set to be excellent.

5. The post-weld detection method based on automatic laser welding according to claim 4 is characterized in that: After determining whether the first quantity is greater than or equal to a first quantity threshold, the method further includes: If the first quantity is less than the first quantity threshold, obtaining a second quantity based on the single item score and a second score threshold; Determining whether the second quantity is greater than a second quantity threshold; If the second number is greater than the second number threshold, setting the solder joint detection result as unqualified; If the second number is less than or equal to the second number threshold, obtaining a total score of the solder joint based on the single item score; Get the feature quantity corresponding to the solder joint feature; Based on the total score of the solder joint and the number of features, obtaining an average score for each item; Determine whether the single item average score is greater than or equal to the second scoring threshold; If the single item average score is greater than or equal to the second scoring threshold, the solder joint detection result is set as good; If the single item average score is less than the second scoring threshold, the solder joint inspection result is set as unqualified.

6. The post-weld detection method based on automatic laser welding according to claim 1 is characterized in that: Also includes: Obtain the number of solder joints and the number of unqualified solder joints corresponding to the test results; Based on the preset label corresponding to the welding point, obtaining the target device number corresponding to the detection device; Based on the number of solder joints and the number of unqualified solder joints, obtaining a failure rate; Determining whether the failure rate exceeds a third quantity threshold; If the unqualified rate exceeds the third quantity threshold, re-inspecting the solder joint corresponding to the target device number by other detection devices, and generating a re-inspection unqualified rate; Based on the re-inspection unqualified rate and the unqualified rate, obtaining the unqualified rate difference; Based on the failure rate difference, a flag is generated.

7. A post-weld inspection method based on automatic laser welding according to claim 6, characterized in that: The specific steps of generating a mark based on the non-conforming rate difference include: Determining whether the failure rate is less than a preset failure rate threshold; If the unqualified rate is less than the preset unqualified rate threshold, marking the welding device corresponding to the welding point as an abnormal welding device; If the unqualified rate is greater than or equal to the preset unqualified rate threshold, the detection device corresponding to the welding point is marked as an abnormal detection device.

8. A post-weld inspection system based on automatic laser welding, characterized in that: include: A first acquisition module, used to acquire a positioning image and an initial image of a welding spot area based on a preset shooting rule; A second acquisition module, used for acquiring a standard position based on the positioning image; A third acquisition module, used for preprocessing the initial image and acquiring a target image; A fourth acquisition module, used for acquiring target geometric information of a chip edge contour based on the target image; A fifth acquisition module, used to match the target geometric information with preset standard geometric information and obtain a position offset; A sixth acquisition module, used for acquiring a pixel position of the pad in the target image based on the position offset and the standard position; The seventh acquisition module is used to obtain the solder joint detection result based on the target image, the pixel position and the preset deep learning model.

9. An intelligent terminal, comprising a memory and a processor, characterized in that: The memory is used to store a computer program that can be run on the processor, and when the processor loads the computer program, it executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored therein, characterized in that: When the computer program is loaded by a processor, the method according to any one of claims 1 to 7 is executed.

Images