A welding porosity prediction method based on temperature field RGB gradient fusion

Abstract

This invention discloses a method for predicting weld porosity based on RGB gradient fusion of temperature field. This method innovatively combines welding time-series temperature field data with deep learning image recognition technology, achieving its goal through the following technical path: First, establishing a mapping relationship between welding process parameters and porosity, and using finite element simulation to obtain temperature thermal cycle data; second, compressing and enhancing the temperature data at three feature points, transforming the time-series temperature curve into an R, G, B three-channel color matrix; then, generating a color gradient image through multi-source data fusion; finally, constructing a deep neural network model to achieve image-based intelligent hierarchical prediction of weld porosity. Compared with traditional physical experimental methods, this method significantly reduces experimental costs through simulation technology, and the use of chromatographic gradient coding technology can effectively characterize the spatiotemporal evolution characteristics of the temperature field. This technology provides a new digital solution for welding quality prediction and has significant value in the field of welding engineering applications.

Description

Technical Field

[0001] This invention belongs to the field of welding technology, and in particular relates to a method for predicting welding porosity based on RGB gradient fusion of temperature field. Background Technology

[0002] Welding technology is widely used in the automotive, aerospace, shipbuilding, and energy industries. Its high efficiency and reliability in joining various materials and structures make it an indispensable process in modern industrial manufacturing. However, traditional welding techniques suffer from high heat input, a large heat-affected zone, and are prone to defects such as weld slag, deformation, porosity, and hot cracks. Porosity, in particular, is the most easily generated and difficult to control, limiting further improvements in welding quality.

[0003] Especially in the welding of aluminum alloys, porosity remains a critical problem that urgently needs to be addressed in industrial applications. These pores are mainly caused by the trapping or incomplete evacuation of gases within the molten pool, typically including hydrogen, air, or oxide particles. The presence of pores significantly reduces the mechanical properties of the weld joint, particularly fatigue strength and tensile properties. To solve this problem, many experts and scholars have conducted in-depth research on the formation mechanism of pores and proposed some innovative methods for pore suppression, such as incorporating laser oscillation to improve the molten pool dynamics.

[0004] Despite significant advancements in laser welding technology for aluminum alloys, predicting porosity remains a complex and challenging problem. Researchers have developed multiphysics numerical simulation models using high-performance computing, incorporating processes such as heat conduction, fluid dynamics, and metal evaporation to simulate gas behavior within the molten pool. However, most of these models only simulate the porosity formation mechanism and cannot accurately determine porosity under different process parameters. Furthermore, fluid simulation models are complex and time-consuming. Therefore, finding a simple technique to predict porosity solely through the temperature field is essential, as it could save process engineers considerable time and costs and has direct engineering practical value. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for predicting weld porosity based on RGB gradient fusion of temperature field. The method aims to convert the multi-point thermal cycling temperature field process into a computer-recognizable image using RGB chromatographic gradient fusion, so as to build a deep learning model based on chromatographic gradient and thereby achieve the goal of predicting weld porosity.

[0006] The technical solution of this invention is: a method for predicting welding porosity based on RGB gradient fusion of temperature field, comprising the following steps:

[0007] Step A: Determine the actual physical welding experiment scheme to obtain the original modeling basis data, take different welding process parameters as input, and use porosity as the output target for excellent classification.

[0008] Step B: Perform finite element simulation verification based on the physical test data from Step A, and extract temperature field data;

[0009] Step C: Process the temperature field data extracted in step B. The specific steps are as follows:

[0010] Step C1: Extract temperature thermal cycling probe data using a finite element simulation model, and compress the data points of each temperature thermal cycling data set by random sampling according to a normal distribution. The probability density S(x) of the sampled temperature data points follows a normal distribution with a mean of μ and a standard deviation of σ, i.e., S(x) ~ N(μ,σ), where x is the temperature data point, and the mean μ and standard deviation σ are determined based on the overall temperature range distribution. Its probability density is:

[0011]

[0012] Step C2: Perform data augmentation on each group of compression temperature data. The data augmentation process can add noise and shift the original data.

[0013] Step C3: Align the three different data points in each group and convert them into red, green and blue color matrix images respectively using the R, G and B channels. The temperature data is determined according to the appropriate temperature range based on the material welding process. Align the data to the RGB color description range [0, 255] and then rearrange the data into a 15×15 matrix image.

[0014] Step C4: Fuse each group of RGB channel images into a color matrix image. The color matrix image generated by fusing the three RGB channels covers the information of the three-point thermal cycle and the temperature gradient information between the points.

[0015] Step D: Use each set of image data finally obtained in step C as the input parameters of the deep learning training set, and use the corresponding porosity grade as the output parameters of the deep learning model. Based on the finally trained deep learning model, predict the porosity of the welded workpiece.

[0016] In step A, the original modeling data includes welding trajectory, welding power, welding speed, welding oscillation frequency, welding oscillation amplitude, and porosity statistically calculated based on the weld cross-section.

[0017] In step B, the specific steps for performing finite element simulation verification based on the physical test data from step A and extracting temperature field data are as follows:

[0018] Step B1: Create a 3D model of the workpiece and import it into the finite element analysis software;

[0019] Step B2: In the finite element simulation software, analyze and set the material parameters, initial conditions, and boundary conditions for the welding model. The material properties include the material density, coefficient of thermal expansion, Young's modulus, Poisson's ratio, yield strength, thermal conductivity, and specific heat. The specific mesh generation type is selected according to the structure of the welded workpiece, choosing an appropriate tetrahedral or hexahedral mesh. The initial and boundary conditions include the initial ambient temperature, thermal convection heat transfer coefficient, and constraint force conditions.

[0020] Step B3: Determine the heat source model and verify the finite element simulation heat source model. The heat source includes electron beam and laser, and the heat source model includes Gaussian heat source model, double ellipsoidal heat source model and hemispherical heat source model. The heat source parameters include welding voltage, welding current, laser power and heat source moving speed.

[0021] Step B4: Extract temperature thermal cycling probe data from three specified points for each finite element simulation model. These three points have temperature gradient characteristics in the temperature field.

[0022] In step D, the deep learning model is a CNN-based convolutional neural network model. By inputting the RGB color matrix image in step C4 under any welding process parameters into the trained deep learning model, the porosity under the corresponding welding process parameters can be graded and predicted.

[0023] The beneficial effects of this invention are as follows: This invention proposes a method for predicting welding porosity based on the RGB gradient fusion of the temperature field. It is the first time that the temperature gradient curve has been converted into an RGB color spectrum and applied to the welding field, and has been applied in actual engineering cases. This solves the problem of high detection costs and difficulty in predicting welding porosity defects, and helps enterprises select low-cost, high-efficiency, and high-quality welding process solutions. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of a welding porosity prediction method based on temperature field RGB gradient fusion in this invention.

[0025] Figure 2 This is a schematic diagram of the welding method in this invention.

[0026] Figure 3 This is a schematic diagram of the welding trajectory in this invention.

[0027] Figure 4 This is a schematic diagram of the combined heat source in this invention.

[0028] Figure 5 This is a schematic diagram of the heat source verification in this invention.

[0029] Figure 6 This is a schematic diagram of the sampling points in this invention.

[0030] Figure 7 This is the original data curve in this invention.

[0031] Figure 8 This is the compressed data curve in this invention.

[0032] Figure 9 This is the RGB temperature gradient fusion image in this invention.

[0033] Figure 10 This is a diagram of the deep learning framework in this invention. Detailed Implementation

[0034] To make the objectives and technical solutions of this invention clearer, detailed explanations are provided below with reference to specific examples. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this invention.

[0035] like Figure 1 The diagram shown is a schematic flowchart illustrating the application of a welding porosity prediction method based on RGB gradient fusion of temperature field in laser oscillating welding. The application of this welding porosity prediction method based on RGB gradient fusion of temperature field in laser oscillating welding includes the following steps:

[0036] Step A: Determine the actual physical welding experiment scheme to obtain the original modeling basis data, take different welding process parameters as input, and use porosity as the output target for excellent classification.

[0037] Step B: Perform finite element simulation verification based on the physical test data from Step A, and extract temperature field data;

[0038] Step C: Process the temperature field data extracted in step B. The specific steps are as follows:

[0039] Step C1: Extract temperature thermal cycling probe data using a finite element simulation model, and compress the data points of each temperature thermal cycling data set by random sampling according to a normal distribution. The probability density S(x) of the sampled temperature data points follows a normal distribution with a mean of μ and a standard deviation of σ, i.e., S(x) ~ N(μ,σ), where x is the temperature data point, and the mean μ and standard deviation σ are determined based on the overall temperature range distribution. Its probability density is:

[0040]

[0041] Step C2: Perform data augmentation on each group of compression temperature data. The data augmentation process can include adding noise, shifting, etc., to the original data.

[0042] Step C3: Align the three different data points in each group and convert them into red, green and blue color matrix images using the R, G and B channels respectively. The temperature data is based on the appropriate temperature range for the material welding process, such as [22℃-1000℃] for aluminum alloy. Align the data to the RGB color description range [0, 255] and then rearrange the data into a 15×15 matrix image.

[0043] Step C4: Fuse each group of RGB channel images into a color matrix image. The color matrix image generated by fusing the three RGB channels covers the information of the three-point thermal cycle and the temperature gradient information between the points.

[0044] Step D: Use each set of image data finally obtained in step C as the input parameters of the deep learning training set, and use the corresponding porosity grade as the output parameters of the deep learning model. Based on the finally trained deep learning model, predict the porosity of the welded workpiece.

[0045] In step A, the original modeling base data includes welding trajectory, welding power, welding speed, welding oscillation frequency, welding oscillation amplitude, and porosity statistically calculated based on the weld cross-section, wherein the welding trajectory includes, but is not limited to, the shapes of “〇”, “8”, and “∞”.

[0046] This invention takes laser oscillating welding of flat plates as an example. The three-dimensional dimensions of the workpiece are as follows: Figure 2 As shown, the overall dimensions are 40mm in length, 40mm in width, and 8mm in height; the welding trajectory is an "O" shape, as shown. Figure 3 As shown, the porosity classification is based on GB / T 22085.2—2008 "Guideline for Defect Quality Classification of Electron Beam and Laser Welded Joints Part 2: Aluminum and Aluminum Alloys" with B / C / D classification, and the porosity >10% index is added to the classification as unqualified grade E.

[0047] Table 1 Basic Data Table

[0048] Serial Number trajectory shape laser power Welding speed Swing amplitude Oscillation frequency Porosity Defect Classification 1 ○ 4 1.8 1 100 37.3 E 2 ○ 4 1.8 1 200 28.7 E 3 ○ 4 1.8 1 300 22.3 E 4 ○ 4 1.8 2 100 27.4 E 5 ○ 4 1.8 2 200 10.4 D 6 ○ 4 1.8 2 250 5.2 C 7 ○ 4 1.8 2 300 0 B 8 ○ 4 1.8 3 300 0 B

[0049] In step B, the specific steps for performing finite element simulation verification based on the physical test data from step A and extracting temperature field data are as follows:

[0050] B1. Create a 3D model of the workpiece and import it into the finite element analysis software;

[0051] B2. In the finite element simulation software, analyze and set the material parameters, initial conditions, and boundary conditions of the welding model. The material properties include the material density, coefficient of thermal expansion, Young's modulus, Poisson's ratio, yield strength, thermal conductivity, and specific heat. The specific mesh generation type is selected according to the structure of the welded workpiece, choosing an appropriate tetrahedral or hexahedral mesh. The initial and boundary conditions include the initial ambient temperature, thermal convection heat transfer coefficient, and constraint force conditions.

[0052] In step B2, the specific material properties of the welded workpiece are set as follows: the material is set to 5A06, the density is shown in Table 2, the thermal conductivity is shown in Table 3, and the specific heat is shown in Table 4.

[0053] Table 2 Density Table

[0054] Temperature (C) 20 100 200 300 400 600 800 1200 1500 <![CDATA[Density (kg·m -3 )]]> 2660 2660 2550 2500 2350 2350 2300 2300 2300

[0055] Table 3 Thermal conductivity coefficients

[0056] Temperature (C) 20 100 200 300 400 600 800 1200 1500 <![CDATA[Thermal conductivity (W·m -1 ·°C -1 )]]> 118 121 126 130 138 118 116 122 128

[0057] Table 4 Specific Heat Table

[0058] Temperature (C) 20 100 200 300 400 600 800 1200 1500 <![CDATA[Specific heat (W·m -1 ·C -1 )]]> 924 921 1005 1047 1089 1390 1200 1010 1005

[0059] The mesh generation type was determined to be hexahedral elements, with a mesh size between 0.2mm and 1.4mm, and localized mesh refinement around the weld. Initial environmental conditions were set at an initial temperature of 22℃ and a thermal convection heat transfer coefficient of 20W / m². 2 ·℃.

[0060] B3. Determine the heat source model and verify the finite element simulation heat source model. The heat source includes electron beam and laser, and the heat source model includes Gaussian heat source model, double ellipsoidal heat source model and hemispherical heat source model, etc. The heat source parameters include, but are not limited to, welding voltage, welding current, laser power, heat source moving speed, etc.

[0061] In step B3, the heat source is a oscillating laser, and the selected heat source model is a combination of a Gaussian surface heat source and a volume heat source, such as... Figure 4 As shown; and the heat source model is verified, as follows. Figure 5 As shown;

[0062]

[0063] Q η =Q m +Q t

[0064] B4. Extract temperature thermal cycling probe data at three specified points for each finite element simulation model. The three points have temperature gradient characteristics in the temperature field.

[0065] In step C, the temperature field data extracted in step B is processed. The specific steps are as follows:

[0066] C1. The coordinates of the selected temperature sampling points are as follows: Figure 6 As shown, the data points of each temperature thermal cycling result are compressed. Based on the temperature thermal cycling probe data extracted from the finite element simulation model, the data point set is compressed by normal distribution random sampling. The probability density S(x) of the sampled temperature data points follows a normal distribution random variable with mean μ = 0.4 and standard deviation σ = 1000, i.e., S(x) ~ N(μ,σ), where x is a temperature data point, and its probability density is:

[0067]

[0068] The compressed data uses a normal distribution to sample the original data at the center location, as illustrated in the diagram below. Figure 7 As shown, the compressed data is as follows Figure 8 As shown.

[0069] C2. Perform data augmentation on each group of compression temperature data. The data augmentation process involves adding noise and shifting the original data to increase the sample size.

[0070] C3. Align the three different data points in each group, and scale the compressed data of sampling points 1-3 from the temperature range [22-1000] to the RGB color gamut range [0-255]. Then rearrange the data into a 15×15 matrix image and convert it into red, green, and blue color matrix images respectively using the R, G, and B channels. Sampling point 1 corresponds to the R channel, sampling point 2 corresponds to the G channel, and sampling point 3 corresponds to the B channel.

[0071] C4. Fuse each group of RGB channel images into a single color matrix image. The color matrix image generated by fusing the RGB three channels encompasses information about the thermal cycle at the three points and the temperature gradient information between the points, such as... Figure 9 As shown;

[0072] In step D, deep learning training is performed using the improved Googlenet framework, as illustrated below. Figure 10 As shown, for the trained deep learning model, by inputting the RGB color matrix image in step C4 under any welding process parameters, the porosity under the corresponding welding process parameters can be predicted in a graded manner.

[0073] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for predicting weld porosity based on RGB gradient fusion of temperature field, characterized in that, Includes the following steps: Step A: Determine the actual physical welding experiment scheme to obtain the original modeling basis data, take different welding process parameters as input, and use porosity as the output target for excellent classification. Step B: Perform finite element simulation verification based on the physical test data from Step A, and extract temperature field data; Step C: Process the temperature field data extracted in step B. The specific steps are as follows: Step C1: Extract temperature thermal cycling probe data using a finite element simulation model, and compress the data points of each temperature thermal cycling data set by random sampling according to a normal distribution. The probability density S(x) of the sampled temperature data points follows a normal distribution with a mean of μ and a standard deviation of σ, i.e., S(x) ~ N(μ,σ), where x is the temperature data point, and the mean μ and standard deviation σ are determined based on the overall temperature range distribution. Its probability density is: Step C2: Perform data augmentation on each group of compression temperature data. The data augmentation process can add noise and shift the original data. Step C3: Align the three different data points in each group and convert them into red, green and blue color matrix images respectively using the R, G and B channels. The temperature data is determined according to the appropriate temperature range based on the material welding process. Align the data to the RGB color description range [0, 255] and then rearrange the data into a 15×15 matrix image. Step C4: Fuse each group of RGB channel images into a color matrix image. The color matrix image generated by fusing the three RGB channels covers the information of the three-point thermal cycle and the temperature gradient information between the points. Step D: Use each set of image data finally obtained in step C as the input parameters of the deep learning training set, and use the corresponding porosity grade as the output parameters of the deep learning model. Based on the finally trained deep learning model, predict the porosity of the welded workpiece.

2. The method as described in claim 1, characterized in that, In step A, the original modeling data includes welding trajectory, welding power, welding speed, welding oscillation frequency, welding oscillation amplitude, and porosity statistically calculated based on the weld cross-section.

3. The method as described in claim 1, characterized in that, In step B, the specific steps for performing finite element simulation verification based on the physical test data from step A and extracting temperature field data are as follows: Step B1: Create a 3D model of the workpiece and import it into the finite element analysis software; Step B2: In the finite element simulation software, analyze and set the material parameters, initial conditions, and boundary conditions for the welding model. The material properties include the material density, coefficient of thermal expansion, Young's modulus, Poisson's ratio, yield strength, thermal conductivity, and specific heat. The specific mesh generation type is selected according to the structure of the welded workpiece, choosing an appropriate tetrahedral or hexahedral mesh. The initial and boundary conditions include the initial ambient temperature, thermal convection heat transfer coefficient, and constraint force conditions. Step B3: Determine the heat source model and verify the finite element simulation heat source model. The heat source includes electron beam and laser, and the heat source model includes Gaussian heat source model, double ellipsoidal heat source model and hemispherical heat source model. The heat source parameters include welding voltage, welding current, laser power and heat source moving speed. Step B4: Extract temperature thermal cycling probe data from three specified points for each finite element simulation model. These three points have temperature gradient characteristics in the temperature field.

4. The method as described in claim 1, characterized in that, In step D, the deep learning model is a CNN-based convolutional neural network model. By inputting the RGB color matrix image in step C4 under any welding process parameters into the trained deep learning model, the porosity under the corresponding welding process parameters can be graded and predicted.

Images