Pipeline welding seam quality detection method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a pipeline welding seam quality detection method. The method comprises the following steps: aiming at each of at least two to-be-detected pipelines, acquiring a welding seam image of a welding seam on the to-be-detected pipeline, the at least two to-be-detected pipelines corresponding to the same ultrasonic phased array scanning device; detecting the weld joint image to obtain a first detection result representing the external welding condition of the weld joint, and acquiring an ultrasonic phased array image of the weld joint based on the ultrasonic phased array scanning device under the condition of determining that the weld joint has external defects based on the first detection result; and detecting the ultrasonic phased array image to obtain a second detection result representing the internal welding condition of the welding seam, and detecting the quality of the welding seam based on the second detection result. According to the technical scheme provided by the embodiment of the invention, the accurate detection of the welding seam quality of the pipeline can be realized under the condition of saving the detection cost.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of image processing, and in particular to a pipeline weld quality detection method, device, electronic equipment and storage medium. Background Art

[0002] In practical applications, plastic pipe products are widely connected by hot-melt butt welding, which heats the joint surfaces of the plastic materials, causing them to melt and then be squeezed and butted to form a weld.

[0003] The quality of the weld is directly related to the safety and reliability of the entire plastic pipe pressure system. However, the existing weld quality detection methods are difficult to balance the detection accuracy and detection cost, which needs to be solved urgently. Summary of the invention

[0004] The embodiments of the present invention provide a pipeline weld quality detection method, device, electronic equipment and storage medium, so as to realize accurate detection of pipeline weld quality while saving detection cost.

[0005] According to one aspect of the present invention, a pipeline weld quality detection method is provided, which may include:

[0006] For each of at least two pipelines to be tested, a weld image of the weld on the pipeline to be tested is obtained, wherein the at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device;

[0007] Detecting the weld image to obtain a first detection result characterizing an external welding condition of the weld, and acquiring an ultrasonic phased array image of the weld based on an ultrasonic phased array scanning device when it is determined that the weld has an external defect based on the first detection result;

[0008] The ultrasonic phased array image is inspected to obtain a second inspection result characterizing the internal welding condition of the weld, and the quality of the weld is inspected based on the second inspection result.

[0009] According to another aspect of the present invention, a pipeline weld quality detection device is provided, which may include:

[0010] A weld image acquisition module, used for acquiring a weld image of a weld on each of at least two pipelines to be tested, wherein at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device;

[0011] An ultrasonic phased array image acquisition module is used to detect the weld image to obtain a first detection result characterizing the external welding condition of the weld, and when it is determined based on the first detection result that the weld has an external defect, obtain an ultrasonic phased array image of the weld based on an ultrasonic phased array scanning device;

[0012] The quality detection module is used to detect the ultrasonic phased array image to obtain a second detection result characterizing the internal welding condition of the weld, and perform quality detection on the weld based on the second detection result.

[0013] According to another aspect of the present invention, there is provided an electronic device, which may include:

[0014] at least one processor; and

[0015] a memory communicatively connected to at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by at least one processor. The computer program is executed by at least one processor so that the pipeline weld quality detection method provided by any embodiment of the present invention is implemented when the at least one processor executes the computer program.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, on which computer instructions are stored, and the computer instructions are used to enable a processor to implement the pipeline weld quality detection method provided by any embodiment of the present invention when executed.

[0018] According to the technical solution in the embodiment of the present invention, for each of the at least two pipelines to be tested, a weld image of the weld on the pipeline to be tested is obtained, wherein the at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device. Compared with respectively setting an ultrasonic phased array scanning device for each pipeline to be tested, the same ultrasonic phased array scanning device is set for multiple pipelines to be tested, thereby reducing the detection cost; the weld image is detected to obtain a first detection result characterizing the external welding condition of the weld, and when it is determined based on the first detection result that the weld has an external defect, the ultrasonic phased array image of the weld is obtained based on the ultrasonic phased array scanning device, and when the external defect exists, the ultrasonic phased array image is obtained again, thereby improving the detection efficiency when there are fewer ultrasonic phased array scanning devices; the ultrasonic phased array image is detected to obtain a second detection result characterizing the internal welding condition of the weld, and based on the second detection result, the weld is quality inspected, and the accuracy of the quality inspection can be improved by re-inspecting using the ultrasonic phased array image. The above technical solution detects external defects of the weld through weld images, and then uses an ultrasonic phased array scanning device to further perform internal inspection on welds with external defects, which can achieve accurate inspection of pipeline weld quality while saving inspection costs.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flow chart of a pipeline weld quality detection method provided according to an embodiment of the present invention;

[0022] Figure 2 is a flow chart of another pipeline weld quality detection method provided according to an embodiment of the present invention;

[0023] Figure 3 is a flow chart of another pipeline weld quality detection method provided according to an embodiment of the present invention;

[0024] Figure 4 is a schematic diagram of weld quality detection according to a specific example of another pipeline weld quality detection method provided in an embodiment of the present invention;

[0025] Figure 5 is a model architecture diagram of a specific example of another pipeline weld quality detection method provided in an embodiment of the present invention;

[0026] Figure 6 is a structural block diagram of a pipeline weld quality detection device provided according to an embodiment of the present invention;

[0027] Figure 7 It is a structural schematic diagram of an electronic device for implementing the pipeline weld quality detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. The situations of "target", "original", etc. are similar and will not be repeated here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] Before introducing the embodiment of the present invention, the relevant contents described in the background technology are first exemplarily described to better understand the pipeline weld quality detection method described in the embodiment of the present invention.

[0031] At present, the inspection methods for pipeline welds are mainly divided into two categories: destructive inspection and non-destructive inspection. Among them, destructive inspection methods usually evaluate the physical properties of welds through mechanical testing, but these methods require destroying the pipeline and cannot be applied in real time on the production line.

[0032] Nondestructive testing technology does not damage the material structure and is suitable for online testing. Common methods include ultrasonic phased array testing (PAUT), ultrasonic time-of-flight diffraction (TOFD), visual inspection, X-ray inspection, microwave inspection, and infrared thermal imaging inspection.

[0033] For non-destructive testing technology, the most widely used is visual inspection, and typical items of visual inspection include cracks, poor shapes, notches and scratches, welding surface mismatch, welding surface angle mismatch, weld flash type, weld flange width, pores or bubbles and poor joints, etc. However, visual inspection usually relies on the experience of operators to judge the quality of welds, lacks intelligent and automated means, and may lead to inconsistent inspection results and low efficiency in actual production.

[0034] The second is ultrasonic phased array detection, which is a technology that uses an array of multiple transmitting and receiving elements to transmit and receive ultrasonic waves. By controlling the transmission timing and amplitude of each element, the beam can be formed and controlled. Phased array technology can adjust the direction, focus and scanning range of the beam to achieve all-round detection of welds. However, the price of ultrasonic phased array detection systems is relatively high, and the economic pressure of widespread application is relatively high. It is generally used in the field of high-end pipelines and it is difficult to cover all production lines.

[0035] Therefore, how to achieve accurate detection of weld quality while saving costs needs to be solved urgently.

[0036] Figure 1 1 is a flow chart of a pipeline weld quality detection method provided in an embodiment of the present invention. This embodiment is applicable to the case of accurately detecting the quality of pipeline welds. The method can be executed by a pipeline weld quality detection device provided in an embodiment of the present invention, which can be implemented by software and / or hardware, and the device can be integrated in an electronic device, which can be various user terminals or servers.

[0037] See also Figure 1 The method of the embodiment of the present invention specifically includes the following steps:

[0038] S110. For each of the at least two pipelines to be tested, a weld image of the weld on the pipeline to be tested is obtained, wherein the at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device.

[0039] The pipeline to be tested can be understood as a pipeline on a pipeline welding production line that is to be subjected to weld inspection. Optionally, there may be multiple pipeline welding production lines, and each pipeline welding production line may have multiple pipelines to be tested.

[0040] The weld image can be understood as an image obtained by photographing the weld of the pipeline. Optionally, an industrial camera can be installed on each pipeline welding production line. When the pipeline to be tested passes through the industrial camera, the industrial camera automatically photographs the weld of the pipeline to be tested to obtain a weld image.

[0041] The ultrasonic phased array scanning device can be understood as a device for performing ultrasonic phased array scanning on the weld position to obtain an ultrasonic phased array image. Optionally, the ultrasonic phased array scanning device can be composed of a host, a probe, and a scanner, etc., which has a high cost. Using one ultrasonic phased array scanning device to correspond to at least two pipelines to be tested can effectively reduce the detection cost.

[0042] S120, detecting the weld image to obtain a first detection result characterizing the external welding condition of the weld, and when it is determined based on the first detection result that the weld has external defects, acquiring an ultrasonic phased array image of the weld based on an ultrasonic phased array scanning device.

[0043] The first detection result can be understood as the result reflecting the external welding condition of the weld after the weld image of the weld is detected, such as irregular weld width, weld cracks, grooves and notches on the pipe surface, weld depressions, overheated welding, mismatched joint surface angles, foreign matter inclusions, pores and normal welds, etc. Optionally, the weld image can be detected using image detection technologies such as deep learning models, machine learning models and template matching.

[0044] Ultrasonic phased array images can be understood as images obtained by scanning welds with external defects using an ultrasonic phased array scanning device. By controlling the emission timing and amplitude of each element of the ultrasonic phased array scanning device, beam formation and control can be achieved, and by adjusting the direction, focus and scanning range of the beam, all-round detection of the weld can be achieved. Optionally, an ultrasonic phased array scanning device can be used to form a fan-shaped scanning range of a certain angle at a certain incident point of the weld to obtain an ultrasonic phased array S-scan image with a fan-shaped display.

[0045] The weld image can be first inspected using an image inspection method with a lower inspection cost to obtain a first inspection result characterizing the external welding condition of the weld. When it is determined based on the first inspection result that the weld has external defects, the weld can be scanned using an ultrasonic phased array scanning device with a higher inspection cost to obtain an ultrasonic phased array image of the weld, thereby reducing the inspection cost.

[0046] S130, detecting the ultrasonic phased array image to obtain a second detection result characterizing the internal welding condition of the weld, and performing quality detection on the weld based on the second detection result.

[0047] The second test result can be understood as the result reflecting the internal condition of the weld obtained by testing the ultrasonic phased array image of the weld with external defects, such as whether there are holes and inclusions inside the weld. The weld can be further tested by using the ultrasonic phased array image to obtain the second test result, and the weld quality can be tested based on the second test result. Through two tests, the accuracy of quality testing can be improved.

[0048] According to the technical solution in the embodiment of the present invention, for each of the at least two pipelines to be tested, a weld image of the weld on the pipeline to be tested is obtained, wherein the at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device. Compared with respectively setting an ultrasonic phased array scanning device for each pipeline to be tested, the same ultrasonic phased array scanning device is set for multiple pipelines to be tested, thereby reducing the detection cost; the weld image is detected to obtain a first detection result characterizing the external welding condition of the weld, and when it is determined based on the first detection result that the weld has an external defect, the ultrasonic phased array image of the weld is obtained based on the ultrasonic phased array scanning device, and when the external defect exists, the ultrasonic phased array image is obtained again, thereby improving the detection efficiency when there are fewer ultrasonic phased array scanning devices; the ultrasonic phased array image is detected to obtain a second detection result characterizing the internal welding condition of the weld, and based on the second detection result, the weld is quality inspected, and the accuracy of the quality inspection can be improved by re-inspecting using the ultrasonic phased array image. The above technical solution detects external defects of the weld through weld images, and then uses an ultrasonic phased array scanning device to further perform internal inspection on welds with external defects, which can achieve accurate inspection of pipeline weld quality while saving inspection costs.

[0049] An optional technical solution performs quality inspection on the weld based on a second test result, including: when it is determined based on the second test result that the weld has internal defects, based on the first test result and the second test result, determining the defect type of the weld to achieve quality inspection of the weld; and / or, when it is determined based on the second test result that the weld does not have internal defects, determining that the quality of the weld is qualified.

[0050] Among them, after further testing the weld by ultrasonic phased array imaging, when the second test result obtained indicates that the weld has internal defects, the external defect type of the weld can be determined based on the first test result and the internal defect type of the weld can be determined based on the second test result. The defect type of the weld and the cause of the defect can be further determined by the external defect type and the external defect type, thereby achieving weld quality inspection. When the second test result obtained indicates that the weld does not have internal defects, although there are external defects in the surface weld, it does not affect normal use, and the quality of the weld can be determined to be qualified.

[0051] The above technical solution can realize accurate detection of weld quality by jointly determining whether the weld quality is qualified through two detection results.

[0052] Figure 21 is a flow chart of another pipeline weld quality detection method provided in an embodiment of the present invention. This embodiment is optimized based on the above-mentioned technical solutions. In this embodiment, optionally, the weld image is detected to obtain a first detection result characterizing the external welding condition of the weld, including: obtaining a pre-trained weld image detection model, and using the weld image detection model to detect the weld image to obtain a first detection result characterizing the external welding condition of the weld.

[0053] The explanations of the terms that are the same as or corresponding to the above embodiments are not repeated here.

[0054] See also Figure 2 The method of this embodiment may specifically include the following steps:

[0055] S210. For each of at least two pipelines to be tested, obtain a weld image of the weld on the pipeline to be tested, wherein the at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device.

[0056] S220, obtaining a pre-trained weld image detection model, and using the weld image detection model to detect the weld image to obtain a first detection result characterizing an external welding condition of the weld, and when it is determined based on the first detection result that the weld has external defects, obtaining an ultrasonic phased array image of the weld based on an ultrasonic phased array scanning device.

[0057] The weld image detection model can be understood as a pre-trained deep learning model for detecting weld images. Optionally, a pre-built weld image training dataset can be used to train the pre-built initial weld image detection model to obtain a weld image detection model. Optionally, the weld image training dataset can be obtained by:

[0058] 1. Based on the image data in actual production, an image dataset containing various weld states is constructed. The sources of the image dataset can include photos taken by mobile phones (which contain a large number of non-weld photos) and weld photos taken by industrial cameras (most of which are weld images), which have different backgrounds and sizes.

[0059] 2. Perform data augmentation on the photos in the image dataset, such as compressing the image to 640×640 pixels and using step boxes and random boxes to select 640×640 image areas.

[0060] 3. Manually select and preliminarily annotate images containing weld features from the image dataset after data enhancement. Then, further manually screen and remove invalid weld photos (such as images without valid features due to screenshots) to obtain datasets with clean features under different backgrounds. Further data enhancement processing is performed on the dataset with clean features, such as rotation, flipping, adding noise, adjusting contrast, changing lighting, blurring and color jittering. Finally, the image annotation tool is used to manually annotate the defect features, generate the corresponding file and convert the file into a weld image training dataset suitable for initial weld image detection model training.

[0061] By using a pre-trained weld image detection model to detect weld images and detecting weld images through automated means, errors caused by the experience and judgment of detection personnel can be avoided, the influence of human factors can be reduced, the accuracy of weld image detection can be improved, and the first detection result that characterizes the external welding condition of the weld can be obtained.

[0062] S230, detecting the ultrasonic phased array image to obtain a second detection result characterizing the internal welding condition of the weld, and performing quality detection on the weld based on the second detection result.

[0063] The technical solution of the embodiment of the present invention detects weld images through a weld image detection model, which can improve the accuracy and efficiency of weld image detection compared to manual detection.

[0064] An optional technical solution, the weld image detection model includes a convolution module, the convolution module includes a convolution layer, a batch normalization layer and an activation function layer, the convolution layer includes an attention mechanism unit and an adaptive feature extraction unit; using the weld image detection model, the weld image is detected to obtain a first detection result characterizing the external welding condition of the weld, including: using the attention mechanism unit to determine the feature weight corresponding to the weld image, and using the adaptive feature extraction unit to extract features of the weld image based on the feature weight to obtain a first feature image; using the batch normalization layer to normalize the first feature image to obtain a second feature image; using the activation function layer to perform nonlinear feature enhancement on the second feature image to obtain a first output feature image, so as to obtain a first detection result characterizing the external welding condition of the weld based on the first output feature image.

[0065] Among them, when the weld image detection model is used to detect the weld image, the convolution module in the weld image detection model can be used to perform convolution operation first, gradually reduce the resolution of the weld image, increase the number of channels, and extract features at different levels layer by layer. Optionally, the convolution layer in the convolution module can be a KWConv structure, and the attention mechanism unit in the KWConv structure is used to determine the feature weight corresponding to the weld image. Optionally, the attention module constructed by the adaptive average pooling and the double-layer fully connected network in the attention mechanism unit can be used to adaptively learn the feature weights and highlight important features. Then, the adaptive feature extraction unit in the KWConv structure is used to extract features from the weld image based on the feature weights to obtain a first feature image. After obtaining the first feature image, the batch normalization layer in the convolution module can be used to combine one-dimensional and two-dimensional batch normalization to normalize the first feature image to obtain a second feature image, thereby effectively improving the expression ability of the first feature image and the generalization performance of the model. Finally, the activation function layer in the convolution module is used to perform nonlinear feature enhancement on the second feature image to achieve efficient feature extraction and conversion, thereby obtaining a first output feature image. Based on the first output feature image, subsequent model processing is performed to obtain a first detection result characterizing the external welding condition of the weld.

[0066] The above technical solution, through the convolution module in the weld image detection model, can achieve accurate feature extraction of weld images, efficient attention allocation, enhance feature expression capabilities and improve the generalization performance of the weld image detection model.

[0067] Another optional technical solution, the weld image detection model includes a first module and a channel attention module, the channel attention module includes a dual pooling layer, a channel compression layer, a nonlinear activation function layer, a channel recovery layer and an activation function normalization layer, the dual pooling layer includes an average pooling layer and a maximum pooling layer; using the weld image detection model, the weld image is detected to obtain a first detection result characterizing the external welding condition of the weld, including: using the first module to obtain a first input feature image based on the weld image; using the average pooling layer to obtain global statistical features in the first input feature image, and using the maximum pooling layer to obtain key features in the first input feature image; using the channel compression layer to perform channel compression on the global statistical features and key features; using the nonlinear activation function layer to activate the global statistical features and key features after channel compression to obtain activated features; using the channel recovery layer to perform channel recovery on the activated features; using the activation function normalization layer to activate the activated features after channel recovery again to obtain a second output feature image, so as to obtain a first detection result characterizing the external welding condition of the weld based on the second output feature image.

[0068] Among them, the first input feature image can be understood as the feature image obtained by processing the module before the channel attention module in the weld image detection model; the first module can be understood as the module that receives the first input feature image in the weld image detection model.

[0069] The channel attention module can be understood as a module that strengthens the weight of the feature channel and enhances the expression of useful information. Optionally, the channel attention module can be a channel attention mechanism-hierarchical scale-based feature pyramid network (ChannelAttention-Hierarchical Scale-based Feature Pyramid Network, ChannelAttention-HSFPN) module. In the channel attention module, the average pooling layer is first used to obtain the global statistical features in the first input feature image, and the maximum pooling layer is used to obtain the key features in the first input feature image; secondly, the channel compression layer is used to perform channel compression on the global statistical features and key features, for example, compressing 256 channels into 64 channels; then, the nonlinear activation function layer is used to activate the global statistical features and key features after channel compression through a double nonlinear activation strategy to obtain activated features; then, the channel recovery layer is used to perform channel recovery on the activated features, for example, restoring 64 channels into 256 channels; finally, the activation function normalization layer, such as the Sigmoid function, is used to activate the activated features after channel recovery again to obtain the second output feature image, and subsequent model processing is performed based on the second output feature image to obtain the first detection result characterizing the external welding condition of the weld.

[0070] The above technical solution can comprehensively inform and express the first input feature image through the channel attention module, enhance the feature selection ability of the weld image detection model, and better realize the feature information interaction between channels.

[0071] Another optional technical solution is that the weld image detection model includes a second module and a dynamic sampling module. The weld image detection model is used to detect the weld image to obtain a first detection result characterizing the external welding condition of the weld, including: using the second module to obtain a second input feature image based on the weld image; using the dynamic sampling module to perform channel feature mapping processing on the second input feature image to obtain a displacement feature, and determining a sampling position for sampling the second input feature image based on the displacement feature, sampling the second input feature image based on the sampling position, and obtaining a first detection result characterizing the external welding condition of the weld based on a third output feature image obtained by sampling.

[0072] The second input feature image can be understood as a feature image obtained by a module before the dynamic sampling module in the weld image detection model. The second module can be understood as a module in the dynamic sampling module that receives the second input feature image.

[0073] The dynamic sampling module can be understood as a module for dynamic sampling operations. The sampling accuracy can be improved by dynamically downsampling the second input feature image. It is particularly suitable for small target detection. Optionally, the dynamic sampling module can be a DySample (Dynamic Sampling) module. The DySample module flexibly adjusts the sampling position of the convolution kernel by generating a sampling displacement (offset), and can realize adaptive sampling of the second input feature image, so that the convolution can flexibly capture different spatial features. In this way, when detecting objects with variable shapes or positions, such as polyethylene weld defects, the irregular feature distribution will make it difficult for ordinary convolution to effectively learn its features. By dynamically adjusting the sampling position, the adaptability of the weld image detection model to deformation and complex backgrounds can be enhanced, especially for small targets and defects with unclear boundaries. The dynamic sampling module in the weld image detection model can be used to perform channel feature mapping processing on the second input feature image to obtain displacement features, for example, mapping the input 64-channel features into 32-channel displacement features; then, the sampling position for sampling the second input feature image is determined based on the displacement features, and the second input feature image is sampled based on the sampling position, so that the convolution kernel can automatically fine-tune the sampling position on the second input feature image, thereby capturing local features with uncertainty and obtaining a third output feature image; subsequent model processing is performed based on the third output feature image obtained by sampling to obtain a first detection result characterizing the external welding condition of the weld.

[0074] The above technical solution can enhance the flexibility of the weld image detection model and its ability to capture small targets through a dynamic sampling module.

[0075] Figure 3 It is a flow chart of another pipeline weld quality detection method provided in an embodiment of the present invention. This embodiment is optimized based on the above-mentioned technical solutions. In this embodiment, optionally, the ultrasonic phased array image is detected to obtain a second detection result characterizing the internal welding condition of the weld, including: obtaining a pre-trained ultrasonic phased array image detection model, and using the ultrasonic phased array image detection model to detect the ultrasonic phased array image, to obtain a second detection result characterizing the internal welding condition of the weld. Among them, the explanations of the terms that are the same or corresponding to the above-mentioned embodiments are not repeated here.

[0076] See also Figure 3 The method of this embodiment may specifically include the following steps:

[0077] S310. For each of at least two pipelines to be tested, obtain a weld image of the weld on the pipeline to be tested, wherein the at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device.

[0078] S320, detecting the weld image to obtain a first detection result characterizing the external welding condition of the weld, and when it is determined based on the first detection result that the weld has external defects, acquiring an ultrasonic phased array image of the weld based on an ultrasonic phased array scanning device.

[0079] S330, obtaining a pre-trained ultrasonic phased array image detection model, and using the ultrasonic phased array image detection model to detect the ultrasonic phased array image, to obtain a second detection result characterizing the internal welding condition of the weld, and based on the second detection result, perform quality inspection on the weld.

[0080] The ultrasonic phased array image detection model can be understood as a pre-trained deep learning model for detecting ultrasonic phased array images. Optionally, a pre-constructed initial ultrasonic phased array image detection model can be trained using a pre-constructed ultrasonic phased array image training data set to obtain an ultrasonic phased array image detection model. Optionally, the ultrasonic phased array image training data set can be constructed in the following ways:

[0081] 1. Obtain a screenshot of the ultrasonic phased array S-scan image of the ultrasonic phased array detection software interface. Perform data enhancement on the ultrasonic phased array S-scan image, such as compressing the image to 640×640 pixels and using a random box to select a 640×640 image area. Then manually select and annotate the images containing weld features from the data-enhanced image.

[0082] 2. Use image annotation tools to manually annotate the two defects of holes and inclusions, and generate corresponding ultrasonic phased array image files. Convert the ultrasonic phased array image files into a file format suitable for the initial ultrasonic phased array image detection model training to obtain the ultrasonic phased array image training data set.

[0083] Using the pre-trained ultrasonic phased array image detection model to detect ultrasonic phased array images can effectively avoid the instability and errors judged by the inspection personnel based on experience, reduce the influence of human factors, improve the accuracy of ultrasonic phased array image detection, and obtain a second detection result that characterizes the internal welding condition of the weld.

[0084] The technical solution of the embodiment of the present invention detects the ultrasonic phased array image through the ultrasonic phased array image detection model, which can improve the accuracy and efficiency of detection compared with manual detection.

[0085] In order to better understand the above-mentioned technical solutions, the following is an exemplary description in combination with a specific example. In this specific example, the pipeline to be tested is a polyethylene pipeline. The schematic diagram of the weld quality detection in this specific example is as follows: Figure 4 As shown, the specific steps are as follows:

[0086] 1. Training of weld image detection model

[0087] 1. Construct a weld image training dataset

[0088] The weld image training dataset includes normal welds and weld samples with defects, with a total of 9 labels: 0-8 represents irregular weld width, weld cracks, notches and score marks, weld flash notches, overheat weld, angular mismatch of joint face, foreign matter inclusions, pores, and normal weld bead. These labels correspond to common visually detectable defects in the polyethylene welding process. The labels and quantities of the weld image training dataset are shown in the following table:

[0089] Table 1: Weld image training data table

[0090] No Labels Qty 0 Irregular weld bead width 502 1 Cracks 1204 2 Notches and score marks 792 3 Welding flash notches 989 4 Overheat weld 526 5 Angular mismatch of joint face 301 6 Foreign matter inclusions 1878 7 Pores 592 8 Normal weld bead 405

[0091] 2. Train the initial weld image detection model to obtain the weld image detection model

[0092] 1) Environment preparation: Set up a deep learning environment, including installing the necessary libraries and frameworks for model training.

[0093] 2) Preparation of weld image training data set: Perform data preparation according to the construction of weld image training data set. Ensure that the weld image training data set is organized in the format required by the initial weld image detection model, including image files and corresponding annotation files. The weld image training data set is divided into training set, validation set and test set in a ratio of 7:2:1 or other appropriate ratios.

[0094] 3) Configuration file preparation: Create a configuration file for the initial weld image detection model. This file defines the basic parameters of the initial weld image detection model, including: label type and model size, etc.

[0095] 4) Initial weld image detection model training:

[0096] The initial weld image detection model is improved on the basis of the YOLOV8n model. The model architecture diagram of the initial weld image detection model is as follows: Figure 5 The specific model processing flow is as follows:

[0097] Step 1: Training weld image input

[0098] The input training weld images enter the initial weld image detection model in the format of (3,640,640), representing a three-channel RGB image with a size of 640x640.

[0099] Step 2: The backbone network extracts multi-scale features of the image.

[0100] The KWConv layer performs convolution operations, which is responsible for gradually reducing the resolution of the image, increasing the number of channels, and extracting features at different levels layer by layer. The C2f_KW module performs feature extraction and uses cross-channel connection operations to maintain the connection between spatial information and different feature levels. The SPPF spatial pyramid pooling layer is used in the 9th layer to further aggregate features, so that the model can obtain multi-scale information.

[0101] Step 3: Head further processes the multi-level features extracted by Backbone and outputs the classification and bounding box information of the target

[0102] The ChannelAttention-HSFPN module is responsible for strengthening the weight of feature channels and enhancing the expression of useful information. The DySample module is used for dynamic sampling operations, which improves accuracy through dynamic downsampling and is suitable for small target detection. The Conv2d convolution layer is used to further process the feature map, reduce its dimension, and output a 256-channel feature map. The Multiply and Add operations are used to fuse features, enhance important information, and improve detection accuracy.

[0103] Step 4: P3, P4, and P5 detection heads for feature capture

[0104] C2f-KW processes P4 / 16, P3 / 8, and P5 / 32 feature images respectively to capture objects of different scales. The Detect layer merges feature images from different levels for final object detection.

[0105] Step 5: Output the test results

[0106] The final Detect layer outputs the detection results, including the category and location information of the target.

[0107] Step 6: Adjust the parameters of the initial weld image detection model based on the annotation and detection results to obtain the weld image detection model

[0108] Continuously monitor the training curve, including the change of loss function and the improvement of accuracy, to ensure the smooth progress of the training process of the initial weld image detection model. At the same time, save the model checkpoints regularly so that you can resume training or select the best model when needed. This training process is iterative and requires multiple attempts and adjustments to the parameters of the initial weld image detection model to achieve the best results.

[0109] Step 7: Evaluation of weld image detection model

[0110] After the weld image detection model training is completed, the test set is used to evaluate the final performance of the weld image detection model. The evaluation indicators include: mAP (mean Average Precision), recall rate (Recall) and precision (Precision) F1 score. For the set 9 categories, these indicators are calculated separately and the average value is calculated. The verification results are shown in the following table:

[0111] Table 2: Weld image detection model verification results

[0112]

[0113] It can be seen that compared with other models in the YOLO series, the weld image detection model shows excellent detection performance and robustness. Under various working conditions, the weld image detection model maintains high accuracy and stability, with an average mAP of 73.1% and an FPS of 333, which meets the needs of industrial detection and has significantly improved detection speed and accuracy.

[0114] 2. Training of Ultrasonic Phased Array Image Detection Model

[0115] 1. Constructing ultrasonic phased array image training dataset

[0116] The ultrasonic phased array image training data set includes weld samples with defects, and two labels are set: 0-1 represents holes (Holes) and inclusions (Conclusions). These labels correspond to common defects in the polyethylene pipe welding process. The ultrasonic phased array image training data set includes: 815 files for holes, 1348 annotations converted; 889 files for inclusions, 911 annotations converted. The ultrasonic phased array image training data set is divided into: 1192 training sets, 340 validation sets, and 172 test sets.

[0117] 2. Based on the ultrasonic phased array image training data set, the ultrasonic phased array image detection model is trained

[0118] Obtain the initial ultrasonic phased array image detection model built based on YOLO11, train the initial ultrasonic phased array image detection model based on the ultrasonic phased array image training data set, and obtain the ultrasonic phased array image detection model. The test results of the ultrasonic phased array image detection model built based on YOLO11 in the test set are shown in the following table:

[0119] Table 3: Ultrasonic phased array image detection model verification results

[0120] Model Parameter quantity mAP FPS YOLO V10 2695196 0.992 238 YOLOV8 2684758 0.989 263 YOLO11 2582542 0.994 232 yolov5 2182054 0.982 192 YOLO11n-hspan-dysample-Kernelwarehouse 2020118 0.99 213 YOLO11-starnet-C3k2-Star-LSCD 1737365 0.993 200

[0121] It can be seen that all the benchmark models and improved models have good accuracy performance. Considering the accuracy, speed and parameter size comprehensively, it is planned to introduce the YOLO11n standard version into the ultrasonic phased array image detection model for use. The detection accuracy of this model on the test set can reach 99.4%, and the FPS exceeds 200, which can effectively perform online detection.

[0122] 3. System Integration

[0123] The trained and verified weld image detection model and ultrasonic phased array image detection model are encapsulated, wherein the weld image detection model (i.e., external weld visual detection) is integrated with the factory's manufacturing execution system (MES). The system automatically captures the weld images generated during the production process, and inputs these weld images into the weld image detection model for detection, and feeds back the results to the MES system, which has an alarm mechanism and can perform real-time monitoring.

[0124] The ultrasonic phased array image detection model (i.e., internal weld detection) is combined with the ultrasonic phased array host system. After opening the ultrasonic phased array software, the detection begins. The ultrasonic phased array image detection model is started. The program will automatically capture dynamic ultrasonic phased array images and provide real-time feedback on the detection results when defects are detected, assisting the inspector in making judgments and improving the accuracy and speed of detection.

[0125] In this specific example, after the weld is inspected for external defects using the weld image detection model, the ultrasonic phased array image detection model is used to perform further internal inspection on the ultrasonic phased array image of the weld with external defects, thereby achieving accurate inspection of the quality of the pipeline weld while saving inspection costs.

[0126] Figure 6This is a structural block diagram of a pipeline weld quality detection device provided in an embodiment of the present invention, and the device is used to execute the pipeline weld quality detection method provided in any of the above embodiments. The device and the pipeline weld quality detection method in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the pipeline weld quality detection device, please refer to the embodiments of the above pipeline weld quality detection method. Figure 6 The device may specifically include: a weld image acquisition module 410, an ultrasonic phased array image acquisition module 420 and a quality detection module 430.

[0127] The weld image acquisition module 410 is used to acquire a weld image of a weld on each of at least two pipelines to be tested, wherein the at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device;

[0128] The ultrasonic phased array image acquisition module 420 is used to detect the weld image to obtain a first detection result characterizing the external welding condition of the weld, and when it is determined based on the first detection result that the weld has an external defect, obtain an ultrasonic phased array image of the weld based on the ultrasonic phased array scanning device;

[0129] The quality detection module 430 is used to detect the ultrasonic phased array image to obtain a second detection result characterizing the internal welding condition of the weld, and perform quality detection on the weld based on the second detection result.

[0130] An optional ultrasonic phased array image acquisition module 420 includes: a first detection result obtaining submodule, which is used to obtain a pre-trained weld image detection model, and use the weld image detection model to detect the weld image to obtain a first detection result characterizing the external welding condition of the weld.

[0131] On this basis, an optional weld image detection model includes a convolution module, the convolution module includes a convolution layer, a batch normalization layer and an activation function layer, the convolution layer includes an attention mechanism unit and an adaptive feature extraction unit; the first detection result obtaining submodule includes:

[0132] A first feature image obtaining unit is used to determine a feature weight corresponding to the weld image by using the attention mechanism unit, and to extract features of the weld image based on the feature weight by using the adaptive feature extraction unit to obtain a first feature image;

[0133] A second feature image obtaining unit is used to perform a normalization process on the first feature image by using batch normalization to obtain a second feature image;

[0134] The first output feature image obtaining unit is used to use the activation function layer to perform nonlinear feature enhancement on the second feature image to obtain the first output feature image, so as to obtain a first detection result characterizing the external welding condition of the weld based on the first output feature image.

[0135] Alternatively, the weld image detection model includes a first module and a channel attention module, the channel attention module includes a dual pooling layer, a channel compression layer, a nonlinear activation function layer, a channel recovery layer and an activation function normalization layer, and the dual pooling layer includes an average pooling layer and a maximum pooling layer; the first detection result obtaining submodule includes:

[0136] A first input feature image obtaining unit, used for obtaining a first input feature image based on the weld image by using the first module;

[0137] A key feature acquisition unit, used to acquire global statistical features in the first input feature image using an average pooling layer, and to acquire key features in the first input feature image using a maximum pooling layer;

[0138] A channel compression unit, used for performing channel compression on global statistical features and key features using a channel compression layer;

[0139] The activation feature obtaining unit is used to activate the global statistical features and key features after channel compression using a nonlinear activation function layer to obtain activation features;

[0140] A channel restoration unit, used for performing channel restoration on the activated features using a channel restoration layer;

[0141] The second output feature image obtaining unit is used to use the activation function normalization layer to reactivate the activated features after channel restoration to obtain the second output feature image, so as to obtain a first detection result characterizing the external welding condition of the weld based on the second output feature image.

[0142] In another alternative, the weld image detection model includes a second module and a dynamic sampling module, and the first detection result obtaining submodule includes:

[0143] A second input feature image obtaining unit, used for obtaining a second input feature image based on the weld image by using the second module;

[0144] The third output feature image obtaining unit is used to use the dynamic sampling module to perform channel feature mapping processing on the second input feature image to obtain a displacement feature, and determine a sampling position for sampling the second input feature image based on the displacement feature, and sample the second input feature image based on the sampling position to obtain a first detection result characterizing the external welding condition of the weld based on the third output feature image obtained by sampling.

[0145] Alternatively, the quality inspection module 430 includes: a second inspection result obtaining submodule, which is used to obtain a pre-trained ultrasonic phased array image detection model, and use the ultrasonic phased array image detection model to detect the ultrasonic phased array image to obtain a second inspection result characterizing the internal welding condition of the weld.

[0146] Alternatively, the quality inspection module 430 includes: a defect type determination submodule for determining the defect type of the weld based on the first inspection result and the second inspection result when it is determined that the weld has an internal defect based on the second inspection result, so as to implement quality inspection of the weld; and / or

[0147] The quality determination submodule is used to determine that the quality of the weld is qualified when it is determined based on the second detection result that the weld has no internal defects.

[0148] In the pipeline weld quality detection device in the embodiment of the present invention, a weld image of a weld on the pipeline to be tested is obtained for each pipeline to be tested in at least two pipelines to be tested through a weld image acquisition module, wherein at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device. Compared with setting an ultrasonic phased array scanning device for each pipeline to be tested, the same ultrasonic phased array scanning device is set for multiple pipelines to be tested, thereby reducing the detection cost; the weld image is detected through the ultrasonic phased array image acquisition module to obtain a first detection result characterizing the external welding condition of the weld, and when it is determined based on the first detection result that the weld has an external defect, an ultrasonic phased array image of the weld is obtained based on the ultrasonic phased array scanning device, and when the external defect exists, an ultrasonic phased array image is obtained again, thereby improving the detection efficiency when there are fewer ultrasonic phased array scanning devices; the ultrasonic phased array image is detected through the quality detection module to obtain a second detection result characterizing the internal welding condition of the weld, and based on the second detection result, the weld is quality detected, and the accuracy of the quality detection can be improved by re-inspecting using the ultrasonic phased array image. The above-mentioned pipeline weld quality detection device detects external defects of the weld through the weld image, and then uses the ultrasonic phased array scanning device to further perform internal detection on the weld with external defects, which can achieve accurate detection of pipeline weld quality while saving detection costs.

[0149] The pipeline weld quality detection device provided in the embodiment of the present invention can execute the pipeline weld quality detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0150] It is worth noting that in the embodiment of the above-mentioned pipeline weld quality inspection device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0151] Figure 7 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein. Figure 7As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11 in communication, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, ROM 12 and RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14. Multiple components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks. The processor 11 may be various general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs the various methods and processes described above, such as the pipeline weld quality detection method. In some embodiments, the pipeline weld quality detection method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the pipeline weld quality detection method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the pipeline weld quality detection method in any other appropriate manner (e.g., by means of firmware). Various embodiments of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof.These various embodiments may include: being implemented in one or more computer programs, which may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general programmable processor, which may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device. The computer program for implementing the method of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a dedicated computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server. In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or device. Computer-readable storage media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. Alternatively, the computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media may include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the above. In order to provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) through which a user can provide input to the electronic device. Other types of devices may also be used to provide interaction with a user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).The systems and techniques described herein may be implemented in a computing system including a backend component (e.g., as a data server), or a computing system including a middleware component (e.g., an application server), or a computing system including a frontend component (e.g., a user computer with a graphical user interface or a web browser, through which a user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by digital data communication (e.g., a communication network) in any form or medium. Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet. A computing system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client and server relationship is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in a cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services. It should be understood that the various forms of the process shown above may be used to reorder, add, or delete steps. For example, the steps described in the present invention may be executed in parallel, sequentially, or in different orders, as long as the expected results of the technical solution of the present invention can be achieved, and this document does not impose any limitations thereon.

[0152] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A pipeline weld quality detection method, characterized in that: include: For each of at least two pipelines to be tested, a weld image of the weld on the pipeline to be tested is acquired, wherein at least two of the pipelines to be tested correspond to the same ultrasonic phased array scanning device; Detecting the weld image to obtain a first detection result characterizing an external welding condition of the weld, and acquiring an ultrasonic phased array image of the weld based on the ultrasonic phased array scanning device when it is determined that the weld has an external defect based on the first detection result; The ultrasonic phased array image is detected to obtain a second detection result characterizing an internal welding condition of the weld, and based on the second detection result, a quality detection is performed on the weld.

2. The method according to claim 1, characterized in that The detecting the weld image to obtain a first detection result characterizing an external welding condition of the weld includes: A pre-trained weld image detection model is obtained, and the weld image is detected using the weld image detection model to obtain a first detection result characterizing an external welding condition of the weld.

3. The method according to claim 2, characterized in that The weld image detection model includes a convolution module, which includes a convolution layer, a batch normalization layer and an activation function layer, and the convolution layer includes an attention mechanism unit and an adaptive feature extraction unit; The method of using the weld image detection model to detect the weld image to obtain a first detection result representing an external welding condition of the weld includes: Using the attention mechanism unit, determining the feature weight corresponding to the weld image, and using the adaptive feature extraction unit, extracting features from the weld image based on the feature weight to obtain a first feature image; Using the batch normalization layer, the first feature image is normalized to obtain a second feature image; The activation function layer is used to perform nonlinear feature enhancement on the second feature image to obtain a first output feature image, so as to obtain a first detection result characterizing the external welding condition of the weld based on the first output feature image.

4. The method according to claim 2, characterized in that: The weld image detection model includes a first module and a channel attention module, wherein the channel attention module includes a dual pooling layer, a channel compression layer, a nonlinear activation function layer, a channel recovery layer, and an activation function normalization layer, and the dual pooling layer includes an average pooling layer and a maximum pooling layer; The method of using the weld image detection model to detect the weld image to obtain a first detection result representing an external welding condition of the weld includes: Using the first module, obtaining a first input feature image based on the weld image; Using the average pooling layer, obtaining global statistical features in the first input feature image, and using the maximum pooling layer, obtaining key features in the first input feature image; Using the channel compression layer, channel compression is performed on the global statistical features and the key features; Activating the global statistical features and the key features after channel compression using the nonlinear activation function layer to obtain activation features; Using the channel recovery layer, performing channel recovery on the activation feature; The activation function normalization layer is used to reactivate the activated features after channel restoration to obtain a second output feature image, so as to obtain a first detection result characterizing the external welding condition of the weld based on the second output feature image.

5. The method according to claim 2, characterized in that: The weld image detection model includes a second module and a dynamic sampling module. The weld image detection model is used to detect the weld image to obtain a first detection result characterizing the external welding condition of the weld, including: Using the second module, obtaining a second input feature image based on the weld image; The dynamic sampling module is used to perform channel feature mapping processing on the second input feature image to obtain a displacement feature, and a sampling position for sampling the second input feature image is determined based on the displacement feature. The second input feature image is sampled based on the sampling position, so as to obtain a first detection result characterizing the external welding condition of the weld based on a third output feature image obtained by sampling.

6. The method according to claim 1, characterized in that The detecting of the ultrasonic phased array image to obtain a second detection result characterizing the internal welding condition of the weld includes: A pre-trained ultrasonic phased array image detection model is obtained, and the ultrasonic phased array image is detected using the ultrasonic phased array image detection model to obtain a second detection result characterizing an internal welding condition of the weld.

7. The method according to claim 1, characterized in that The step of performing quality inspection on the weld based on the second inspection result includes: In the case where it is determined based on the second test result that the weld has internal defects, the defect type of the weld is determined based on the first test result and the second test result to implement quality inspection of the weld; and / or, When it is determined based on the second detection result that the weld has no internal defects, the quality of the weld is determined to be qualified.

8. A pipeline weld quality detection device, characterized in that: include: A weld image acquisition module, used for acquiring a weld image of a weld on each of at least two pipelines to be tested, wherein at least two pipelines to be tested correspond to the same ultrasonic phased array scanning device; an ultrasonic phased array image acquisition module, configured to detect the weld image to obtain a first detection result characterizing an external welding condition of the weld, and, when it is determined based on the first detection result that the weld has an external defect, obtain an ultrasonic phased array image of the weld based on the ultrasonic phased array scanning device; The quality detection module is used to detect the ultrasonic phased array image to obtain a second detection result characterizing the internal welding condition of the weld, and perform quality detection on the weld based on the second detection result.

9. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor executes the pipeline weld quality detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the pipeline weld quality detection method as described in any one of claims 1-7 when executed.

Images