An AI Vision Intelligent Diagnosis Method and System for Ultrasonic Detection Results of Weld Nuggets

Abstract

The present invention provides a method and system for AI vision intelligent diagnosis of ultrasonic inspection results of solder joint nuggets. By combining AI vision intelligent diagnosis with ultrasonic inspection of solder joints, the AI is used to identify and extract the effective nugget area from the ultrasonic inspection result image, automatically calibrate the detection circle of the nugget, calculate the pixel proportion of the effective nugget area based on the pixel extraction information in the ultrasonic inspection result image, calculate the effective nugget diameter according to a preset algorithm, and finally determine the quality of the solder joint according to the preset intelligent diagnosis rules, avoiding the risk of subjective misjudgment that occurs during the manual judgment of the qualification of the inspection result image, improving the diagnostic reliability and inspection efficiency, and saving labor costs.

Description

Technical Field

[0001] The present invention relates to the field of intelligent automotive welding, and specifically, to a method and system for AI vision intelligent diagnosis of ultrasonic inspection results of solder joint nuggets. Background Art

[0002] In the current industrial inspection field, the appearance inspection technology of solder joints / weld seams / adhesive application based on artificial intelligence has formed a mature application system. The technical implementation path mainly includes the following links: First, digital modeling of qualified standard workpieces is carried out through three-dimensional scanning technology to construct a high-precision three-dimensional model including geometric morphology, dimensional accuracy and other features; Subsequently, the standard model is converted into multi-view two-dimensional image data and input into the AI inspection system for deep learning training, and finally, the appearance quality judgment standard of solder joints / weld seams / adhesive application based on image feature matching is established.

[0003] It should be particularly pointed out that the traditional vision inspection method has an essential technical limitation: there is no strict correlation between the surface morphology of the solder joint and its internal quality parameters (such as the effective diameter of the nugget, the density of internal pores, etc.). To break through this technical bottleneck, ultrasonic detection technology is generally used in the industry for supplementary detection. This technology can accurately quantify and characterize the internal structure features of the solder joint through the acoustic wave reflection signal, significantly improving the comprehensiveness and accuracy of quality assessment.

[0004] However, there is still significant room for optimization in the current ultrasonic detection system: The C-scan images generated by the detection need to rely on manual experience for calculating the nugget area and evaluating the pore density. This process has two defects: on the one hand, the subjective judgment of people is likely to lead to deviations in the implementation of the standard, and on the other hand, the efficiency of manual image analysis is difficult to meet the high-speed detection requirements of modern production lines. This technical breakpoint seriously restricts the overall reliability and detection versatility of the quality detection system. Summary of the Invention

[0005] Aiming at the defects in the prior art, the purpose of the present invention is to provide a method and system for AI vision intelligent diagnosis of ultrasonic inspection results of solder joint nuggets, which combines AI vision intelligence with the ultrasonic inspection results of solder joints, and performs intelligent and accurate calculation of the area ratio according to the graphics and makes a standard judgment, avoiding the risk of subjective misjudgment or inaccuracy in the manual judgment of the ultrasonic inspection results of solder joint nuggets, reducing the escape probability of the body welding quality risk problem, and improving the quality inspection and diagnosis efficiency.

[0006] To achieve the above technical effects, the present invention adopts the following technical solutions:

[0007] According to the first aspect of the present invention, there is provided a method for AI vision intelligent diagnosis of ultrasonic inspection results of solder joint nuggets, including the following steps:

[0008] S1. Image generation and extraction of the effective nugget area: Generate an ultrasonic inspection result image of the nugget through ultrasonic scanning of the solder joint, and automatically identify and label the effective nugget area in the image; the effective nugget area refers to the core area of the solder joint that meets the quality requirements.

[0009] S2. Automatic calibration of the nugget detection circle: The system extracts the contour of the effective nugget area and fits a detection circle of the effective nugget at the geometric center of the effective nugget area using a preset algorithm. The diameter d of the detection circle represents the nugget detection diameter, where the preset algorithm is a circular detection algorithm optimized for the fuzzy contour and noise interference of the effective nugget.

[0010] S3. Calculation of the pixel ratio of the effective nugget area: The AI intelligent diagnosis system counts the number of pixels in the effective nugget area within the detection circle and the total number of pixels in the detection circle, and calculates the effective nugget pixel ratio K; K = the number of pixels in the effective nugget area / the total number of pixels within the detection circle.

[0011] S4. Calculation of the effective nugget diameter: The AI intelligent diagnosis system calculates the effective nugget diameter D according to the following formula, D² = d 2 ×K, where d is the diameter of the nugget detection circle and K is the effective nugget pixel ratio in step S3.

[0012] S5. Solder joint quality determination: The AI intelligent diagnosis system compares the effective nugget diameter D with the preset standard diameter and minimum diameter of the solder joint to determine the quality of the solder joint. The determination rules are:

[0013] (1) Solder joints with an effective nugget diameter D greater than the standard diameter are qualified solder joints.

[0014] (2) Solder joints with an effective nugget diameter D within the range of the standard diameter and the minimum diameter are solder joints with a smaller nugget.

[0015] (3) Solder joints with an effective nugget diameter D less than the minimum diameter are unqualified solder joints.

[0016] Among them, the standard diameter and the minimum diameter are preset calibration ranges of the solder joint diameter determined according to the thickness of different spot welding materials.

[0017] Optionally, in step S1, it also includes using Gaussian filtering and median filtering methods to reduce noise interference in the ultrasonic inspection result image.

[0018] Optionally, in step S2, an edge detection algorithm is used to extract the contour of the effective nugget area. The edge detection algorithm includes: Canny algorithm or Sobel algorithm.

[0019] Optionally, the preset algorithms for fitting the detection circle of the effective nugget in the geometric center of the effective nugget region in step S2 include: the least squares circle fitting algorithm or the Hough circle transformation algorithm.

[0020] Optionally, the accuracy of the pixel ratio K of the effective nugget region within the detection circle region is accurate to four decimal places.

[0021] Optionally, the standard diameter is the standard value of the weld nugget diameter under the preset material thickness in the Ford quality standard, and the minimum diameter is the lower threshold of the allowable weld nugget diameter under the preset material thickness in the Ford quality standard.

[0022] According to the second aspect of the present invention, there is provided an AI vision intelligent diagnosis system for ultrasonic detection results of weld nuggets, which is used to implement the method for AI vision intelligent diagnosis of ultrasonic detection results of weld nuggets described above.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] The present invention gives full play to the advantages of AI vision intelligence, combines AI vision intelligence with ultrasonic detection of weld nuggets, and according to the preset algorithm and judgment rules, uses AI intelligent diagnosis to determine the quality of the weld nuggets in the ultrasonic detection result image, which well avoids the subjective misjudgment risk that may occur during the manual judgment of the qualification of the detection result image, improves the diagnosis reliability and inspection efficiency, and saves labor costs. Description of the Drawings

[0025] By reading the following detailed description of the non-limiting embodiments with reference to the accompanying drawings, other features, purposes, and advantages of the present invention will become more obvious:

[0026] Figure 1 It is a flowchart of the steps of the method for AI vision intelligent diagnosis of ultrasonic detection results of weld nuggets described in the first embodiment;

[0027] Figure 2 It is an ultrasonic detection result image of the weld nugget described in the first embodiment. Detailed Embodiments

[0028] To make the purposes, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Usually, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0029] Accordingly, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but merely represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts fall within the scope of protection of the present application. Further, the descriptions involving "first", "second", etc. in the application are only for descriptive purposes and cannot be construed as indicating or implying their relative importance or implicitly specifying the quantity of the indicated technical features.

[0030] First Embodiment

[0031] As Figure 1 shown, this embodiment provides a method for AI vision intelligent diagnosis of ultrasonic detection results of solder joint fusion cores, which combines AI vision intelligent diagnosis and ultrasonic detection of solder joints. Through AI recognition, information is collected from the ultrasonic detection result image and calculated according to a preset algorithm. Finally, the quality of the solder joint is determined according to the preset intelligent diagnosis rules, avoiding the risk of subjective misjudgment that may occur during the manual judgment of the qualification of the detection result image. The specific steps include:

[0032] Step S1. Image recognition and extraction of the effective fusion core area.

[0033] By scanning the solder joint with ultrasonic waves, an ultrasonic detection result image of the solder joint fusion core as Figure 2 shown is generated. The AI intelligent diagnosis system automatically recognizes the "effective fusion core" area (i.e., the core part of the solder joint that meets the quality requirements) in the ultrasonic detection result image according to color features (such as preset red, green, and blue ranges), and marks it with a specific color. Methods such as Gaussian filtering and median filtering are used to reduce noise interference in the ultrasonic detection result image.

[0034] Step S2. Automatic calibration of the fusion core detection circle.

[0035] The AI intelligent diagnosis system applies a preset algorithm and uses edge detection algorithms such as Canny and Sobel to extract the solder joint contour, filling holes or smoothing edges through dilation and erosion. And the extracted solder joint contour is fitted into a detection circle of the effective fusion core using the least squares circle fitting algorithm or the Hough circle transformation algorithm, and a detection circle is automatically generated at the geometric center position of the fusion core in the ultrasonic detection result image, that is, Figure 2 the dashed circle in, and finally the radius and center coordinates of the circle are adaptively adjusted according to the actual size and position of the solder joint to ensure the geometric matching of the detection circle with the actual contour of the effective fusion core. The diameter of the detection circle represents the detection diameter d of the fusion core.

[0036] Among them, both the "least squares circle fitting method" and the "Hough circle transformation" are circular detection algorithms optimized for problems such as fuzzy fusion core contours and noise interference.

[0037] Step S3. Calculate the pixel area ratio of the nugget area in the detection circle.

[0038] Within the range of the generated detection circle by the AI intelligent diagnosis system, count the number of pixels in the "effective nugget" area marked with a specific color and the total number of pixels within the detection circle, and calculate the pixel ratio K of the effective nugget area within the detection circle area, with the precision accurate to four decimal places (0.01%). During this process, noise points in the image need to be filtered to ensure accurate data. K = the number of pixels in the effective nugget area / the total number of pixels within the detection circle.

[0039] Step S4. Calculate the effective nugget diameter.

[0040] The AI intelligent diagnosis system calculates the effective nugget diameter D of the measured solder joint according to the following formula:

[0041] According to the formula: D² = 4 × S 圆 × K / π to obtain, D 2 = d 2 × K, thus obtaining the effective nugget diameter D.

[0042] Among them, S 圆 is the total area of the detection circle (calculated by the detection diameter d of the detection circle), and K is the ratio of the effective nugget to the total number of pixels in the detection circle area.

[0043] Step S5. Judge the quality of the solder joint according to the effective nugget diameter D and the intelligent diagnosis standard:

[0044] The AI intelligent diagnosis system judges the quality of the solder joint through the following diagnosis standard:

[0045] (1) The solder joint with the effective nugget diameter D greater than the standard diameter is a solder joint with qualified quality;

[0046] (2) The solder joint with the effective nugget diameter D within the range of the standard diameter and the minimum diameter is a solder joint with a smaller nugget;

[0047] (3) The solder joint with the effective nugget diameter D less than the minimum diameter is a solder joint with unqualified quality.

[0048] Among them, both the standard diameter and the minimum diameter are preset calibration values. As shown in Table 1 below, the "standard diameter" is the qualified standard value of the weld nugget diameter under a specific material thickness according to Ford quality standards, and the "minimum diameter" is the lower threshold value of the allowable weld nugget diameter under a specific material thickness according to Ford quality standards. That is, according to Ford quality standards, at a specific material thickness, the same weld has comparison calibration values of "standard diameter" and "minimum diameter". In this embodiment, the AI intelligent diagnosis system automatically compares the effective weld nugget diameter of the weld at a specific material thickness with its corresponding "standard diameter" and "minimum diameter" to determine the quality of the weld, avoiding the risk of subjective misjudgment or inaccuracy in the manual judgment of the ultrasonic inspection results of the weld nugget, reducing the escape probability of body welding quality risk problems, and improving the quality inspection and diagnosis efficiency by at least 30%.

[0049]

[0050] Table 1 Qualified standard diameter and minimum diameter of weld nuggets of specific metal sheet thickness under Ford quality standards

[0051] Example 2

[0052] This embodiment provides a system for AI vision intelligent diagnosis of ultrasonic inspection results of weld nuggets, which is used to implement the method for AI vision intelligent diagnosis of ultrasonic inspection results of weld nuggets described in the first embodiment.

[0053] The specific embodiments of the present invention have been described above. Through the above description, relevant staff can make various changes and modifications without departing from the technical idea of the present invention.

Claims

1. An AI visual intelligent diagnosis method for ultrasonic detection results of weld nuggets, characterized in that: The following steps are involved: S1. Image generation and extraction of effective nugget area: Generate an image of the nugget ultrasonic detection result by ultrasonically scanning the weld point, and automatically identify and mark the effective nugget area in the image; the effective nugget area refers to the core area of ​​the weld point that meets the quality requirements; S2. Automatic calibration of the nugget detection circle: The system extracts the contour of the effective nugget area, and applies a preset algorithm to fit a detection circle of the effective nugget at the geometric center of the effective nugget area, wherein the diameter d of the detection circle represents the nugget detection diameter, wherein the preset algorithm is a circular detection algorithm optimized for blurring of the effective nugget contour and noise interference; S3. Calculation of pixel ratio of effective nugget area: The AI ​​intelligent diagnosis system counts the number of pixels in the effective nugget area within the detection circle and the total number of pixels in the detection circle, and calculates the effective nugget pixel ratio K; K = number of pixels in the effective nugget area / total number of pixels in the detection circle; S4. Calculation of effective nugget diameter: The AI ​​intelligent diagnosis system calculates the effective nugget diameter D according to the following formula: D²=d 2 ×K, where d is the diameter of the weld nugget detection circle, and K is the effective weld nugget pixel ratio in step S3; S5. Welding spot quality determination: The AI ​​intelligent diagnosis system compares the effective nugget diameter D with the preset welding spot standard diameter and minimum diameter to determine the welding spot quality. The determination rules are as follows: (1) The weld with effective nugget diameter D larger than the standard diameter is a qualified weld; (2) The weld with effective nugget diameter D in the range of standard diameter and minimum diameter is a weld with small nugget; (3) The welds whose effective nugget diameter D is smaller than the minimum diameter are considered to be unqualified welds; The standard diameter and the minimum diameter are preset calibration ranges of the spot diameter determined according to different spot welding material thicknesses.

2. The AI ​​visual intelligent diagnosis method for ultrasonic detection results of weld nuggets according to claim 1 is characterized in that: Step S1 also includes using Gaussian filtering and median filtering methods to reduce noise interference on the ultrasonic detection result image.

3. The AI ​​visual intelligent diagnosis method for ultrasonic detection results of weld nuggets according to claim 1 is characterized in that: In step S2, an edge detection algorithm is used to extract the contour of the effective nugget area. The edge detection algorithm includes: Canny algorithm or Sobel algorithm.

4. The AI ​​visual intelligent diagnosis method for ultrasonic detection results of weld nuggets according to claim 1 is characterized in that: The preset algorithm for fitting a detection circle of the effective nugget at the geometric center of the effective nugget region in step S2 includes: a least squares circle fitting algorithm or a Hough circle transform algorithm.

5. The AI ​​visual intelligent diagnosis method for ultrasonic detection results of weld nuggets according to claim 1 is characterized in that: The pixel ratio K of the effective nugget area within the detection circle area is accurate to four decimal places.

6. The AI ​​visual intelligent diagnosis method for ultrasonic detection results of weld nuggets according to claim 1 is characterized in that: The standard diameter is the standard value of the weld nugget diameter at a preset material thickness under Ford quality standards, and the minimum diameter is the lower limit threshold of the weld nugget diameter allowed at a preset material thickness under Ford quality standards.

7. An AI visual intelligent diagnosis system for ultrasonic detection results of weld nuggets, characterized in that: A method for implementing the AI ​​visual intelligent diagnosis of weld nugget ultrasonic detection results as described in any one of claims 1 to 6.

Images