Electron beam welding quality detection method for metal composite material

Abstract

The invention relates to an electron beam welding quality detection method for a metal composite material. The method comprises the steps that a first metal material and a second metal material are machined so that the composite interface of the first metal material and the composite interface of the second metal material can have the same size; the first metal material and the second metal material are assembled in a composite interface-to-composite interface mode to obtain a metal composite material assembly; after solid-state vacuum treatment is conducted on the metal composite material assembly, electron beam welding is conducted on the metal composite material assembly to obtain a welded metal composite material blank; and welding quality detection is conducted on the welded metal composite blank. Therefore, whether the welding quality meets the preset standard or not can be intelligently judged, further, cracks, air holes, incomplete penetration and other defects in the welding seam can be found out by conducting welding detection on the welding metal composite blank, and structural failure caused by the welding defects is avoided.

Description

Technical Field

[0001] The present disclosure relates to the technical field of electron beam welding, and in particular to a method for detecting the quality of electron beam welding of metal composite materials. Background Art

[0002] Metal composites are widely sought for high-performance structural components due to their unique combination of properties, such as high strength, high stiffness, good wear resistance and corrosion resistance. These materials are usually composed of two or more metal materials with different physical and chemical properties in order to obtain comprehensive properties that are difficult to achieve with a single material.

[0003] Traditional welding techniques, such as laser welding and arc welding, present many challenges when processing metal composites. For example, the difference in thermophysical properties between different metals may lead to uneven thermal stress during welding, which may cause cracks, deformation and other problems. In addition, some materials in the composite material may be very sensitive to heat treatment, resulting in performance degradation.

[0004] Electron beam welding is a precision welding technology that achieves local melting and connection of materials by bombarding the surface of materials with high-energy electron beams. Although electron beam welding has many advantages, it also has some problems and defects. During the electron beam welding process, defects such as weld porosity, cracks, weld subsidence, and undercut may occur. These defects may affect the quality and performance of the welded joint.

[0005] Therefore, an optimized electron beam welding quality inspection method for metal composites is expected. Summary of the invention

[0006] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0007] In a first aspect, the present disclosure provides a method for detecting the quality of electron beam welding of a metal composite material, the method comprising: Mechanically processing the first metal material and the second metal material so that a composite interface of the first metal material and the second metal material has a consistent size; Assembling the first metal material and the second metal material in a composite interface-to-composite interface manner to obtain a metal composite material assembly; After subjecting the metal composite material blank to solid-state vacuum treatment, the metal composite material blank is subjected to electron beam welding to obtain a welded metal composite material blank; The welding quality of the welded metal composite material blank is inspected.

[0008] Optionally, the welding quality inspection of the welded metal composite material blank includes: acquiring a welding area image captured by a camera; performing grayscale processing on the welding area image to obtain a welding area grayscale image; performing welding state feature extraction and foreground mapping visualization on the welding area grayscale image to obtain a foreground-significant welding area state feature map; and determining the inspection result based on the foreground-significant welding area state feature map.

[0009] Optionally, welding state feature extraction and foreground mapping visualization are performed on the welding area grayscale image to obtain a foreground-salient welding area state feature map, including: passing the welding area grayscale image through a welding area state feature extractor based on a DenseNet model to obtain a welding area state feature map; feature dispersion along the channel dimension of the welding area state feature map to obtain a set of welding area pixel channel feature vectors; passing the set of welding area pixel channel feature vectors through a welding area pixel-level semantic association encoder based on a Transformer layer to obtain a set of context welding area pixel channel feature vectors; feature aggregation of the set of context welding area pixel channel feature vectors along a feature dispersion pattern to obtain an optimized welding area state feature map; passing the optimized welding area state feature map through a foreground mapping network based on an attention mechanism to obtain the foreground-salient welding area state feature map.

[0010] Optionally, the optimized welding area state feature map is passed through a foreground mapping network based on an attention mechanism to obtain the foreground-salient welding area state feature map, including: calculating the welding area spatial attention feature matrix of the optimized welding area state feature map; masking the welding area spatial attention feature matrix based on a predetermined threshold to obtain a masked welding area spatial attention feature matrix; and fusing the masked welding area spatial attention feature matrix and the optimized welding area state feature map to obtain the foreground-salient welding area state feature map.

[0011] Optionally, calculating the welding area spatial attention feature matrix of the optimized welding area state feature map includes: calculating the global mean pooling of the optimized welding area state feature map along the channel dimension to obtain a welding state pooling feature matrix; passing the welding state pooling feature matrix through an activation unit based on a sigmoid function to obtain the welding area spatial attention feature matrix.

[0012] Optionally, the welding area spatial attention feature matrix is ​​masked based on a predetermined threshold to obtain a masked welding area spatial attention feature matrix, including: setting eigenvalues ​​in the welding area spatial attention feature matrix that are less than or equal to the predetermined threshold to zero to obtain the masked welding area spatial attention feature matrix.

[0013] Optionally, the masked welding area spatial attention feature matrix and the optimized welding area state feature map are fused to obtain the foreground-salient welding area state feature map, including: calculating the position point multiplication between each feature matrix of the optimized welding area state feature map and the masked welding area spatial attention feature matrix to obtain the foreground-salient welding area state feature map.

[0014] Optionally, determining the detection result based on the foreground-salient welding area state characteristic map includes: passing the foreground-salient welding area state characteristic map through a classifier-based welding quality detector to obtain the detection result, wherein the detection result is used to indicate whether the welding quality meets a predetermined standard.

[0015] Optionally, the method further includes a training step: training the welding area state feature extractor based on the DenseNet model, the welding area pixel-level semantic association encoder based on the Transformer layer, the foreground mapping network based on the attention mechanism, and the welding quality detector based on the classifier.

[0016] Optionally, the training step includes: obtaining a training welding area image captured by a camera, and a true value of whether the welding quality meets a predetermined standard; grayscale processing the training welding area image to obtain a training welding area grayscale image; passing the training welding area grayscale image through the welding area state feature extractor based on the DenseNet model to obtain a training welding area state feature map; performing feature dispersion along the channel dimension on the training welding area state feature map to obtain a set of training welding area pixel channel feature vectors; passing the set of training welding area pixel channel feature vectors through the welding area pixel-level semantic association encoder based on the Transformer layer to obtain a set of training context welding area pixel channel feature vectors; and performing feature aggregation on the set of training context welding area pixel channel feature vectors along the feature dispersion mode. To obtain a training optimized welding area state feature map; pass the training optimized welding area state feature map through the foreground mapping network based on the attention mechanism to obtain a training foreground salient welding area state feature map; pass the training foreground salient welding area state feature map through the classifier-based welding quality detector to obtain a detection result; calculate the cross entropy loss function value between the detection result and the true value to obtain a classification loss function value; based on the classification loss function value, train the welding area state feature extractor based on the DenseNet model, the welding area pixel-level semantic association encoder based on the Transformer layer, the foreground mapping network based on the attention mechanism and the welding quality detector based on the classifier, and when updating the model parameters through gradient back propagation based on the classification loss function, further introduce a noise distribution loss function.

[0017] The above technical solution is adopted, by mechanically processing the first metal material and the second metal material so that the composite interface of the first metal material and the second metal material has the same size; the first metal material and the second metal material are assembled in the manner of composite interface to composite interface to obtain a metal composite material assembly; after the metal composite material assembly is subjected to solid vacuum treatment, the metal composite material assembly is subjected to electron beam welding to obtain a welded metal composite material blank; the welded metal composite material blank is subjected to welding quality inspection. In this way, it is possible to intelligently judge whether the welding quality meets the predetermined standard. Furthermore, welding inspection of the welded metal composite material blank can find defects such as cracks, pores, and incomplete penetration in the weld, thereby avoiding structural failure caused by these welding defects.

[0018] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings: Figure 1 The present invention is a flow chart of a method for detecting the quality of electron beam welding of a metal composite material according to an exemplary embodiment.

[0020] Figure 2 is based on Figure 1 The illustrated embodiment shows a flow chart of step S104 of a method for detecting quality of electron beam welding of a metal composite material.

[0021] Figure 3 is a schematic diagram of the penetration depth after welding according to an exemplary embodiment.

[0022] Figure 4 is a block diagram of an electron beam welding system for metal composite materials according to an exemplary embodiment.

[0023] Figure 5 It is a block diagram of an electronic device according to an exemplary embodiment.

[0024] Figure 6 It is an application scenario diagram of a method for detecting quality of electron beam welding of a metal composite material according to an exemplary embodiment. DETAILED DESCRIPTION

[0025] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0026] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0027] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0028] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0029] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0030] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0031] In order to solve the above problems, the present disclosure provides a method for detecting the quality of electron beam welding of metal composite materials, which includes mechanically processing a first metal material and a second metal material so that the composite interface of the first metal material and the second metal material has a consistent size, the first metal material and the second metal material are both solid metals, and remain solid at the vacuum treatment temperature of electron beam welding; assembling the first metal material and the second metal material in a composite interface-to-composite interface manner to obtain a metal composite material assembly; after the metal composite material assembly is subjected to solid vacuum treatment, the metal composite material assembly is subjected to electron beam welding to obtain a welded metal composite material blank; and the welded metal composite material blank is subjected to welding quality detection. In this way, it is possible to intelligently determine whether the welding quality meets the predetermined standard. Furthermore, welding detection of the welded metal composite material blank can detect defects such as cracks, pores, and incomplete penetration in the weld, thereby avoiding structural failure caused by these welding defects.

[0032] The specific embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0033] Figure 1 is a flow chart of a method for detecting the quality of electron beam welding of a metal composite material according to an exemplary embodiment. Figure 1 As shown, the method includes: Step 101, machining a first metal material and a second metal material so that a composite interface of the first metal material and the second metal material has a consistent size; Step 102, assembling the first metal material and the second metal material in a composite interface-to-composite interface manner to obtain a metal composite material assembly; Step 103, after performing solid-state vacuum treatment on the metal composite material assembly, electron beam welding is performed on the metal composite material assembly to obtain a welded metal composite material body; Step 104: Perform welding quality inspection on the welded metal composite material blank.

[0034] It should be understood that in the actual application process of the present application, by performing welding inspection on the welded metal composite material blank, defects such as cracks, pores, and incomplete penetration in the weld can be found to avoid structural failure caused by these welding defects. However, some existing inspection technologies may not be able to detect all types of defects. For example, minor defects may be difficult to detect accurately. In addition, some inspection technologies require highly skilled operators to perform, and this welding quality inspection method has high requirements on the operator's experience and skill level.

[0035] In response to the above technical problems, the technical concept of the present application is to introduce deep learning algorithms and computer vision technology in the process of welding quality inspection, extract image features from the welding area image to capture the state characteristics of the welding area, and highlight the significant features in the welding area through the attention mechanism, so as to intelligently judge whether the welding quality meets the predetermined standards.

[0036] Figure 2 is based on Figure 1 The embodiment shown is a flow chart of step S104 of a method for detecting the quality of electron beam welding of a metal composite material. Figure 2 The step 104 of performing welding quality inspection on the welded metal composite material blank comprises: Step 1041, obtaining a welding area image captured by a camera; Step 1042, performing grayscale processing on the welding area image to obtain a welding area grayscale image; Step 1043, performing welding state feature extraction and foreground mapping visualization on the welding area grayscale image to obtain a foreground-visualized welding area state feature map; Step 1044: Determine the detection result based on the foreground-salient welding area state feature map.

[0037] Specifically, in step S104, the welding quality inspection of the welded metal composite material blank includes: first, obtaining a welding area image captured by a camera. Here, the welding area image contains rich information of the welding area, such as the shape and size change of the weld, etc. This rich information is of great significance for identifying welding anomalies such as cracks, pores, undercuts and other defects.

[0038] Then, the welding area image is gray-scaled to obtain a welding area gray-scale image. Gray-scale processing can convert a color image or a multi-band image into a gray-scale image. In a gray-scale image, the value of each pixel is represented as a single value of brightness, usually ranging from 0 (black) to 255 (white). A gray-scale image is a single-channel image, which contains only brightness information but no color information compared to a color image. In this way, the color information in the welding area image is removed by gray-scale processing, and only the key structural information in the welding area image is retained, which helps to simplify the subsequent processing process.

[0039] In one embodiment of the present disclosure, welding state feature extraction and foreground mapping visualization are performed on the welding area grayscale image to obtain a foreground-salient welding area state feature map, including: passing the welding area grayscale image through a welding area state feature extractor based on a DenseNet (Densely Connected Convolutional Networks) model to obtain a welding area state feature map; performing feature dispersion along the channel dimension on the welding area state feature map to obtain a set of welding area pixel channel feature vectors; passing the set of welding area pixel channel feature vectors through a welding area pixel-level semantic association encoder based on a Transformer layer to obtain a set of context welding area pixel channel feature vectors; performing feature aggregation on the set of context welding area pixel channel feature vectors along a feature dispersion pattern to obtain an optimized welding area state feature map; and passing the optimized welding area state feature map through a foreground mapping network based on an attention mechanism to obtain the foreground-salient welding area state feature map.

[0040] Next, the grayscale image of the welding area is passed through a welding area state feature extractor based on the DenseNet model to obtain a welding area state feature map. DenseNet is a deep convolutional neural network that improves the information flow and parameter efficiency of the network by introducing direct connections between each layer, and is particularly suitable for image feature extraction. Specifically, the connection method of DenseNet allows the network to fuse features from all previous layers during the forward propagation process, which helps to capture richer and more semantically powerful feature representations. Since the grayscale image of the welding area may contain a variety of complex structural information, including different defect types and background noise, the powerful feature extraction capability of DenseNet can handle these complexities.

[0041] Considering that the welding area state feature map is limited by the convolution kernel during the feature extraction process, it may only extract the state features of the local neighborhood pixel space of the welding area grayscale image, while ignoring the contextual connection between local area blocks. Therefore, in the technical solution of the present application, the welding area state feature map is feature-dispersed along the channel dimension to obtain a set of welding area pixel channel feature vectors; and the set of welding area pixel channel feature vectors is passed through a welding area pixel-level semantic association encoder based on the Transformer layer to obtain a set of context welding area pixel channel feature vectors. In this way, the spatial association relationship and contextual semantic information of the welding area state feature map are extracted and integrated to understand the complex welding defect feature pattern.

[0042] Since there may be a large amount of background information in the welding area image, this background information may be retained when the feature extraction of the welding area image is performed, and it may be deepened and amplified as the network level deepens. In other words, there may be a large amount of noise interference feature information in the set of the context welding area pixel channel feature vectors, and this noise interference feature information is not important for the learning and recognition of defect feature patterns, and may even interfere with the target task. Therefore, in the technical solution of the present application, the set of the context welding area pixel channel feature vectors is further feature aggregated along the feature dispersion pattern to obtain an optimized welding area state feature map; and the optimized welding area state feature map is passed through a foreground mapping network based on an attention mechanism to obtain a foreground salient welding area state feature map.

[0043] In one embodiment of the present disclosure, the optimized welding area state feature map is passed through a foreground mapping network based on an attention mechanism to obtain the foreground-salient welding area state feature map, including: calculating a welding area spatial attention feature matrix of the optimized welding area state feature map; masking the welding area spatial attention feature matrix based on a predetermined threshold to obtain a masked welding area spatial attention feature matrix; and fusing the masked welding area spatial attention feature matrix and the optimized welding area state feature map to obtain the foreground-salient welding area state feature map.

[0044] Specifically, in one embodiment of the present disclosure, the welding area spatial attention feature matrix of the optimized welding area state feature map is calculated, including: calculating the global mean pooling of the optimized welding area state feature map along the channel dimension to obtain the welding state pooling feature matrix; passing the welding state pooling feature matrix through an activation unit based on a sigmoid function to obtain the welding area spatial attention feature matrix.

[0045] More specifically, in one embodiment of the present disclosure, the welding area spatial attention feature matrix is ​​masked based on a predetermined threshold to obtain a masked welding area spatial attention feature matrix, including: setting the eigenvalues ​​in the welding area spatial attention feature matrix that are less than or equal to the predetermined threshold to zero to obtain the masked welding area spatial attention feature matrix.

[0046] More specifically, in one embodiment of the present disclosure, the masked welding area spatial attention feature matrix and the optimized welding area state feature map are fused to obtain the foreground salient welding area state feature map, including: calculating the position point multiplication between each feature matrix of the optimized welding area state feature map and the masked welding area spatial attention feature matrix to obtain the foreground salient welding area state feature map.

[0047] Here, the pixel channel feature vectors of each context welding area with discrete feature distribution are aggregated into the optimized welding area state feature map through feature aggregation, and the foreground mapping network based on the attention mechanism that introduces the spatial attention mechanism and the masking mechanism is used to highlight the importance of the foreground information in the optimized welding area state feature map. Specifically, the foreground mapping network based on the attention mechanism uses the high-level image semantic feature information in the optimized welding area state feature map to extract the corresponding foreground target feature information of interest, such as important defect areas, and marks and purifies the foreground target feature information of interest and filters the background information through mask processing, so that the network module can focus on the foreground target of the welding area image.

[0048] Then, the foreground-salient welding area state feature map is passed through a welding quality detector based on a classifier to obtain a detection result, and the detection result is used to indicate whether the welding quality meets a predetermined standard.

[0049] In one embodiment of the present disclosure, a detection result is determined based on the foreground-salient welding area state feature map, including: passing the foreground-salient welding area state feature map through a classifier-based welding quality detector to obtain the detection result, wherein the detection result is used to indicate whether the welding quality meets a predetermined standard.

[0050] Furthermore, in one embodiment of the present disclosure, the electron beam welding quality detection method for metal composite materials further includes a training step: training the welding area state feature extractor based on the DenseNet model, the welding area pixel-level semantic association encoder based on the Transformer layer, the foreground mapping network based on the attention mechanism, and the welding quality detector based on the classifier. The training step includes: obtaining a training welding area image captured by a camera, and a true value of whether the welding quality meets a predetermined standard; grayscale processing the training welding area image to obtain a training welding area grayscale image; passing the training welding area grayscale image through the welding area state feature extractor based on the DenseNet model to obtain a training welding area state feature map; performing feature dispersion along the channel dimension on the training welding area state feature map to obtain a set of training welding area pixel channel feature vectors; passing the set of training welding area pixel channel feature vectors through the welding area pixel-level semantic association encoder based on the Transformer layer to obtain a set of training context welding area pixel channel feature vectors; performing feature aggregation on the set of training context welding area pixel channel feature vectors along the feature dispersion mode to obtain to train and optimize the welding area state feature map; pass the training optimized welding area state feature map through the foreground mapping network based on the attention mechanism to obtain a training foreground salient welding area state feature map; pass the training foreground salient welding area state feature map through the welding quality detector based on the classifier to obtain a detection result; calculate the cross entropy loss function value between the detection result and the true value to obtain a classification loss function value; based on the classification loss function value, train the welding area state feature extractor based on the DenseNet model, the welding area pixel-level semantic association encoder based on the Transformer layer, the foreground mapping network based on the attention mechanism and the welding quality detector based on the classifier, and when updating the model parameters through gradient back propagation based on the classification loss function, further introduce a noise distribution loss function.

[0051] In a preferred example, when updating the model parameters by gradient back propagation based on the classification loss function, a noise distribution loss function is further introduced, and the calculation of the noise distribution loss function includes the following steps: Obtain the class probability value of the training prospect-significant welding area state feature map through the classifier and calculate a value minus the class probability value to obtain an inverse probability value; Calculate the feature mean and feature variance of the training foreground-significant welding area state feature map; Calculate the product term difference between each feature value in the training prospect salient welding area state feature map and the first product term of the class probability value and the second product term of the inverse probability value and the feature mean; The squares of the product term differences corresponding to each feature value in the training prospect significant welding area state feature map are summed and divided by the scale of the training prospect significant welding area state feature map to obtain the first difference term of the noise distribution loss function, wherein the scale of the training prospect significant welding area state feature map is the product of the width and height of each feature matrix of the training prospect significant welding area state feature map and the number of channels of the training prospect significant welding area state feature map; Calculating the square sum of the class probability value and the inverse probability value and multiplying the sum by the feature variance to obtain a second difference term of the noise distribution loss function; The difference between the first difference term and the second difference term is calculated to obtain the noise distribution loss function.

[0052] Here, in order to solve the problem that after the dense image semantic feature representation of the training welding area grayscale image is associated through channel context, the spatial distribution focusing of the local image semantic features in the foreground mapping network based on the attention mechanism causes the image semantic distribution out-noise of the training foreground significant welding area state feature map, which further leads to the overall class distribution noise of the training foreground significant welding area state feature map, thereby affecting the accuracy of the classification result, the applicant of the present application further introduces the above-mentioned noise distribution loss function.

[0053] Specifically, the noise distribution loss function sets the expected characteristics of the approximate variance of the out-of-distribution noise of the feature set of the training foreground significant welding area state feature map relative to the Bernoulli sampling instances of the feature set based on the class probability Bernoulli distribution characteristics expressed by the class probability value and inverse probability value of the classifier of the training foreground significant welding area state feature map.

[0054] Therefore, the noise distribution loss function is used to measure the distribution feature noise through statistical noise distribution based on the class probability distribution characteristics expressed by the class probability value and the inverse probability value. The possible sampling noise characteristics can be degraded based on the probability crossover random selection characteristics of the variance expectation for each feature value of the training foreground significant welding area state feature map during the classification process, thereby realizing the class distribution noise reduction of self-supervised learning in the model parameter updating process, so as to improve the accuracy of the detection results of the training foreground significant welding area state feature map obtained by the classifier-based welding quality detector.

[0055] In summary, the above scheme is adopted. By introducing deep learning algorithm and computer vision technology in the process of welding quality inspection, image features of the welding area image are extracted to capture the state characteristics of the welding area, and the significant features in the welding area are highlighted through the attention mechanism, so as to intelligently judge whether the welding quality meets the predetermined standards.

[0056] Furthermore, with the increasingly urgent demand for large-scale equipment manufacturing and diversified functions, ultra-thick steel plates and heterogeneous metal composite materials are mostly produced by explosive welding, large ingot casting, riveting and other processes. There have always been low-magnification defects such as unstable composite interface quality, low yield, component segregation and center looseness.

[0057] The "solid-state vacuum treatment + electron beam welding" process, combined with the subsequent rolling process, can produce high-quality metal composite materials and achieve the following goals: ① Composite materials of the same material to meet the current thickness and size of continuous casting billet rolling thick steel plate process restrictions (compression ratio ≥ 3). ② According to the physical properties of different metals such as thermal conductivity, linear expansion coefficient, melting point, etc., determine reasonable solid-state vacuum treatment and electron beam welding process parameters to produce heterogeneous materials such as carbon steel + stainless steel, carbon steel + nickel-based alloy, steel + aluminum, copper + steel, titanium + steel and other metal composite materials. ③ Establish an evaluation mechanism and test method for the composite interface quality and performance of composite materials.

[0058] Specifically, in one embodiment of the present disclosure, the process of non-oxidation surface machining of the quasi-composite metal material → surface treatment with organic solvents → solid-state vacuum treatment of the composite material → vacuum chamber electron beam welding → non-destructive testing of the weld → subsequent rolling and pressure processing to produce the composite material (plate, strip, tube, rod) → interface performance testing is adopted.

[0059] Among them, in the step of machining the non-oxidized surface of metal materials, the "low-speed milling machine + local inert gas protection of the machined surface" process is adopted to carry out non-oxidized processing of the metal composite interface, so that the thickness of the composite interface oxide layer, surface flatness and surface roughness meet the composite requirements.

[0060] In the solid-state vacuum treatment step of composite materials, in order to avoid the presence of impurities such as air, water vapor, and dust in the composite interface layer during composite welding, which affects the quality of the composite interface layer, the composite materials need to be solid-state vacuum treated. The treatment vacuum degree, vacuum holding time, vacuum degree at the beginning of welding, and welding seam cooling speed under vacuum conditions are the core process parameters of this process.

[0061] In the electron beam welding process research steps, according to the metal properties of different composite metal materials such as thermal conductivity, linear expansion coefficient, melting point, melting rate, thickness storage, etc., reasonable surface focusing method, welding beam and welding speed are determined to achieve the best match between the melting depth, melting width, weld surface quality and internal quality of the composite materials, thus ensuring the quality of subsequent processing.

[0062] In the evaluation mechanism and test steps of composite interface quality and performance, a scientific composite interface quality evaluation and test method is established. According to relevant standards, a series of methods such as ultrasonic flaw detection (UT), interface tensile test, interface torsion delamination test, and interface shear test are planned to be used to evaluate the composite effect.

[0063] Furthermore, the "low-speed milling machine + local inert gas protection of the machined surface" process is used to carry out oxidation-free processing of the solid metal composite interface, and the composite material is subjected to solid-state vacuum treatment before electron beam welding. By studying the focusing method, welding beam, welding speed, etc. of the composite interface surface, the depth of penetration, width of penetration, weld surface quality and internal quality of the composite material after welding are optimally matched, thereby ensuring the subsequent rolling process and achieving high-quality composite of metal materials. Based on this type of research and practice, an evaluation mechanism and testing method for the quality and performance of the composite interface are established.

[0064] In another embodiment of the present disclosure, a vacuum welding method for asymmetric composite of ferritic stainless steel and carbon steel is provided, wherein a composite interface is machined without beveling, and electron beam welding is used, and the weld has a high aspect ratio and strength.

[0065] Vacuum electron beam welding is to machine the continuous casting billet and the stainless steel plate to make the composite interface size consistent, then assemble the stainless steel and carbon steel as shown in the figure above, put them into the vacuum chamber and start to evacuate. After the chamber reaches a high vacuum, the welding process and parameters are set according to the chemical and physical properties of the materials. After welding, the penetration depth can reach Figure 3 As shown, it is about 50mm. Vacuum electron beam welding does not require beveling and uses mechanical processing, which has high production efficiency. Compared with symmetrical compounding, vacuum electron beam welding does not require isolation agents and seals. Compared with explosive compounding, it solves the problems of low efficiency, high pollution, and small product size of explosive compounding. Therefore, vacuum electron beam welding has the advantages of high production efficiency, low cost, high yield rate, and high welding quality and performance.

[0066] Specifically, vacuum electron beam welding uses a high-speed electron beam to bombard the metal, converting kinetic energy into heat energy, melting the metal to be welded, and forming a weld after rapid cooling and solidification. The billets are assembled as required using a centering assembly device, and the process parameters are set according to the material properties and composition to weld under high vacuum. After welding, rolling is performed, and the plate shape is flat and the performance meets the standards after rolling.

[0067] In this way, the working time of beveling can be reduced. Compared with symmetrical composite welding, no isolation agent is required and welding is performed in a high vacuum environment, which reduces working time, improves efficiency, reduces costs, and reduces material consumption.

[0068] Figure 4 FIG. 1 is a block diagram of an electron beam welding system for metal composite materials according to an exemplary embodiment. Figure 4 As shown, the electron beam welding system 200 includes: A machining module 201, for machining the first metal material and the second metal material so that the composite interface of the first metal material and the second metal material has a consistent size; An assembly module 202 is used to assemble the first metal material and the second metal material in a composite interface-to-composite interface manner to obtain a metal composite material assembly; An electron beam welding module 203 is used for performing electron beam welding on the metal composite material assembly after performing solid-state vacuum treatment on the metal composite material assembly to obtain a welded metal composite material body; The welding quality detection module 204 is used to perform welding quality detection on the welded metal composite material blank.

[0069] Reference below Figure 5 , which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present disclosure. The terminal device in the embodiment of the present disclosure may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0070] like Figure 5As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0071] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0072] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0073] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0074] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0075] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0076] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0077] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0078] The modules involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a module does not limit the module itself in some cases. For example, a test parameter acquisition module may also be described as a "module for acquiring device test parameters corresponding to a target device".

[0079] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0080] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0081] Figure 6 FIG. 1 is an application scenario diagram of a method for detecting the quality of electron beam welding of a metal composite material according to an exemplary embodiment. Figure 6 As shown, in this application scenario, first, an image of the welding area captured by a camera is obtained (for example, Figure 6 Then, the obtained welding area image is input to a server (for example, Figure 6 In S) as shown in FIG. 1 , the server is capable of processing the welding area image based on an electron beam welding algorithm for metal composite materials to determine a detection result.

[0082] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0083] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

Claims

1. A method for detecting the quality of electron beam welding of metal composite materials, characterized in that: include: Mechanically processing the first metal material and the second metal material so that the composite interface of the first metal material and the second metal material has a consistent size, the first metal material and the second metal material are both solid metals and remain solid at a vacuum processing temperature of electron beam welding; Assembling the first metal material and the second metal material in a composite interface-to-composite interface manner to obtain a metal composite material assembly; After subjecting the metal composite material blank to solid-state vacuum treatment, the metal composite material blank is subjected to electron beam welding to obtain a welded metal composite material blank; The welding quality of the welded metal composite material blank is inspected.

2. The method for detecting the quality of electron beam welding of metal composite materials according to claim 1, characterized in that: The welding quality inspection of the welded metal composite material blank comprises: Acquire the welding area image captured by the camera; Performing grayscale processing on the welding area image to obtain a welding area grayscale image; Extracting welding state features and performing foreground mapping visualization on the welding area grayscale image to obtain a foreground-visualized welding area state feature map; The detection result is determined based on the foreground-salient welding area state characteristic map.

3. The method for detecting the quality of electron beam welding of metal composite materials according to claim 2, characterized in that: Extracting welding state features and visualizing foreground mapping on the welding area grayscale image to obtain a foreground-visualized welding area state feature map includes: The welding area grayscale image is passed through a welding area state feature extractor based on a DenseNet model to obtain a welding area state feature map; Performing feature dispersion on the welding area state feature map along the channel dimension to obtain a set of welding area pixel channel feature vectors; Passing the set of welding area pixel channel feature vectors through a welding area pixel-level semantic association encoder based on a Transformer layer to obtain a set of context welding area pixel channel feature vectors; Performing feature aggregation on the set of pixel channel feature vectors of the context welding area along a feature dispersion mode to obtain an optimized welding area state feature map; The optimized welding area state feature map is passed through a foreground mapping network based on an attention mechanism to obtain the foreground-salient welding area state feature map.

4. The method for detecting the quality of electron beam welding of metal composite materials according to claim 3, characterized in that: The optimized welding area state feature map is passed through a foreground mapping network based on an attention mechanism to obtain the foreground salient welding area state feature map, including: Calculating a welding area spatial attention feature matrix of the optimized welding area state feature map; Performing masking processing on the welding region spatial attention feature matrix based on a predetermined threshold to obtain a masked welding region spatial attention feature matrix; The masked welding region spatial attention feature matrix and the optimized welding region state feature map are fused to obtain the foreground salient welding region state feature map.

5. The method for detecting the quality of electron beam welding of metal composite materials according to claim 4, characterized in that: Calculating the welding area spatial attention feature matrix of the optimized welding area state feature map includes: Calculating the global mean pooling of the optimized welding area state feature map along the channel dimension to obtain a welding state pooling feature matrix; The welding state pooling feature matrix is ​​passed through an activation unit based on a sigmoid function to obtain the welding area spatial attention feature matrix.

6. The method for detecting the quality of electron beam welding of metal composite materials according to claim 5, characterized in that: The method of performing masking processing on the welding region spatial attention feature matrix based on a predetermined threshold to obtain a masked welding region spatial attention feature matrix comprises: The eigenvalues ​​in the welding region spatial attention feature matrix that are less than or equal to the predetermined threshold are set to zero to obtain the masked welding region spatial attention feature matrix.

7. The method for detecting the quality of electron beam welding of metal composite materials according to claim 6, characterized in that: The masked welding region spatial attention feature matrix and the optimized welding region state feature map are integrated to obtain the foreground salient welding region state feature map, including: The foreground-salient welding region state feature map is obtained by calculating the position point multiplication between each feature matrix of the optimized welding region state feature map and the masked welding region space attention feature matrix.

8. The method for detecting the quality of electron beam welding of metal composite materials according to claim 7, characterized in that: Determining the detection result based on the foreground-salient welding area state feature map includes: The foreground-salient welding area state feature map is passed through a welding quality detector based on a classifier to obtain the detection result, and the detection result is used to indicate whether the welding quality meets the predetermined standard.

9. The method for detecting the quality of electron beam welding of metal composite materials according to claim 8, characterized in that: It also includes a training step: training the welding area state feature extractor based on the DenseNet model, the welding area pixel-level semantic association encoder based on the Transformer layer, the foreground mapping network based on the attention mechanism, and the welding quality detector based on the classifier.

10. The method for detecting the quality of electron beam welding of metal composite materials according to claim 9, characterized in that: The training step comprises: Obtaining the training welding area image captured by the camera, and the real value of whether the welding quality meets the predetermined standard; Performing grayscale processing on the training welding area image to obtain a training welding area grayscale image; Passing the training welding area grayscale image through the welding area state feature extractor based on the DenseNet model to obtain a training welding area state feature map; Performing feature dispersion along the channel dimension on the training welding area state feature map to obtain a set of training welding area pixel channel feature vectors; Passing the set of training welding area pixel channel feature vectors through the welding area pixel level semantic association encoder based on the Transformer layer to obtain a set of training context welding area pixel channel feature vectors; Performing feature aggregation on the set of pixel channel feature vectors of the training context welding area along a feature dispersion mode to obtain a training optimized welding area state feature map; Passing the training optimized welding area state feature map through the foreground mapping network based on the attention mechanism to obtain a training foreground salient welding area state feature map; Passing the training prospect salient welding area state feature map through the classifier-based welding quality detector to obtain a detection result; Calculating a cross entropy loss function value between the detection result and the true value to obtain a classification loss function value; The welding area state feature extractor based on the DenseNet model, the welding area pixel-level semantic association encoder based on the Transformer layer, the foreground mapping network based on the attention mechanism and the welding quality detector based on the classifier are trained based on the classification loss function value, and the noise distribution loss function is further introduced when the model parameters are updated through gradient back propagation based on the classification loss function.

Images