Submarine cable aluminum sheath weld defect detection method, system and industrial personal computer

By obtaining the surface point cloud data of the aluminum sheathed surface of the submarine cable and using deep learning models to identify weld defects, the problem of inefficient traditional manual detection is solved, the accuracy and efficiency of weld detection is improved, and the quality and safety of submarine cable is ensured. .

CN118941531APending Publication Date: 2024-11-12中国电气装备集团科学技术研究院有限公司
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202410999547.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

Traditional weld quality inspection relies on artificial sight, is inefficient and is susceptible to human factors, resulting in missed inspection or misjudgment, and it is impossible to effectively detect defects in submarine cable aluminum sheath welds.

Method used

By obtaining the surface point cloud data of the aluminum sheathed seam in the sea cable, weld profile features are extracted, and defect detection is performed using a pre-trained defect recognition model, combining deep learning and traditional point cloud algorithms to identify weld defects.

Benefits of technology

It improves the efficiency and accuracy of weld inspection, ensures the quality of submarine cable aluminum sheath, ensures the sealing and long-term stability of submarine cables, reduces maintenance costs, and ensures the safety and reliability of marine communications and energy transmission.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118941531A_ABST
    Figure CN118941531A_ABST
Patent Text Reader

Abstract

According to the technical scheme of the embodiment of the invention, the first point cloud data representing the surface contour of the submarine cable aluminum sheath is acquired, and the weld joint contour features of the submarine cable aluminum sheath are determined for the weld joint point cloud data representing the weld joint contour on the submarine cable aluminum sheath and extracted from the first point cloud data; and performing defect detection on the welding seam point cloud data according to the welding seam contour features, and performing defect detection on second point cloud data except the welding seam point cloud data in the first point cloud data through a defect identification model. According to the scheme of the embodiment of the invention, a traditional point cloud algorithm is combined with a deep learning algorithm, the problem that the current weld defect detection is manually executed is overcome, the detection efficiency is improved, the detection accuracy is improved, the aluminum sheath quality in the submarine cable quality can be improved, the sealability and the long-term use stability of the submarine cable are ensured, and the service life of the submarine cable is prolonged. Expenditure caused by damage of maintenance cost is reduced, and safety and reliability of ocean communication and energy transmission are guaranteed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of weld detection, and in particular to a method, system and industrial control computer for detecting weld defects of a submarine cable aluminum sheath. Background Art

[0002] As an important communication equipment laid on the seabed to connect oil platforms with inland areas, and cross-sea facilities with inland areas, submarine cables are crucial to the development of energy mining and other aspects. There is an aluminum sheath in the submarine cable, and the quality of the aluminum sheath is directly related to the sealing, corrosion resistance and long-term stability of the submarine cable. Since the aluminum sheath needs to be wrapped with welding technology, the welding quality of the aluminum sheath weld seriously affects the protective performance of the sheath.

[0003] Traditional weld quality inspection methods rely on manual visual inspection, which is inefficient and susceptible to human factors, resulting in missed inspections or misjudgments.

[0004] In summary, a method for detecting weld defects of aluminum sheath of submarine cables with higher detection quality and efficiency is needed. Summary of the invention

[0005] The present invention provides a method, a system and an industrial control computer for detecting defects in a submarine cable aluminum sheath weld, which can achieve higher detection quality and efficiency in detecting defects in a submarine cable aluminum sheath weld.

[0006] According to one aspect of the present invention, a method for detecting weld defects of a submarine cable aluminum sheath is provided, comprising:

[0007] Acquire first point cloud data representing the surface profile of the submarine cable aluminum sheath;

[0008] Extracting weld point cloud data representing a weld contour on the aluminum sheath of the submarine cable from the first point cloud data, determining weld contour features according to the weld point cloud data, and performing defect detection on the weld point cloud data according to the weld contour features;

[0009] The second point cloud data obtained by removing the weld point cloud data from the first point cloud data is input into a pre-trained defect recognition model for defect detection.

[0010] Optionally, before extracting the weld point cloud data representing the weld contour on the aluminum sheath of the submarine cable from the first point cloud data, at least one of the following operations is also included:

[0011] Smoothing the point cloud data and removing noise points in the first point cloud data by Gaussian filtering;

[0012] Downsampling the first point cloud data through a voxel grid of a preset size;

[0013] A target plane in the first point cloud data is identified by a plane fitting algorithm, and the first point cloud data corresponding to the target plane is removed.

[0014] Optionally, extracting weld point cloud data representing a weld contour on the aluminum sheath of the submarine cable from the first point cloud data includes:

[0015] Determine the curvature of each data point in the first point cloud data, and take the data point whose curvature meets a preset condition as a candidate edge point;

[0016] The candidate edge points are fitted into a weld contour to obtain the weld point cloud data.

[0017] Optionally, the performing defect detection on the weld point cloud data according to the weld contour feature includes:

[0018] In the case where the weld contour feature matches the contour defect feature corresponding to a preset standard defect type, the standard defect type is used as the defect type of the weld point cloud data.

[0019] Optionally, the weld contour feature includes a height difference, an area width, and an area slope, wherein the height difference, the area width, and the area slope respectively characterize the height difference between the highest point and the lowest point of the weld contour corresponding to the weld point cloud data, the maximum distance of the edge, and the degree of inclination; accordingly, when the weld contour feature matches the contour defect feature corresponding to a preset standard defect type, the standard defect type is used as the defect type of the weld point cloud data, including:

[0020] When the height difference, area width and area slope of the weld contour feature are all within the height difference range, area width range and area slope range corresponding to a preset standard defect type, it is determined that the weld contour feature matches the contour defect feature corresponding to the preset standard defect type, and the standard defect type is used as the defect type of the current weld point cloud data.

[0021] Optionally, the defect recognition model includes: an input layer, an upsampling layer, a convolution layer, and an activation layer; accordingly, the second point cloud data after removing the weld point cloud data from the first point cloud data is input into the pre-trained defect recognition model for defect detection, including:

[0022] Extracting a first feature image of each of the second point cloud data through an encoder at the input layer;

[0023] Upsampling the first feature image at the upsampling layer, processing the first feature image into a second feature image having the same size as the weld point cloud data, and performing feature fusion on the first feature image and the second feature image to obtain a third feature image;

[0024] Performing a preset number of convolutions at the convolution layer based on current weights and current biases to map the third feature image to a number of channels that matches the predicted segmentation mask;

[0025] The binary classification result of each pixel in the mapped third feature image is determined in the activation layer by a preset activation function, and the defect type of the second point cloud data is determined according to the binary classification result.

[0026] Optionally, the method further includes:

[0027] The defect type obtained by the defect recognition model is verified, and the defect recognition model is updated in a preset manner according to the verification result; wherein the preset manner includes adjusting the parameters of a preset activation function of the defect recognition model and / or increasing the amount of training data of the defect recognition model through at least one data enhancement method; data enhancement includes at least one of random rotation, flipping and scaling processing.

[0028] According to another aspect of the present invention, there is provided an industrial computer, comprising:

[0029] A point cloud data acquisition unit, which acquires first point cloud data representing the surface profile of the aluminum sheath of the submarine cable;

[0030] a weld point cloud data extraction unit, configured to extract weld point cloud data representing a weld contour on the aluminum sheath of the submarine cable from the first point cloud data, determine weld contour features according to the weld point cloud data, and perform defect detection on the weld point cloud data according to the weld contour features;

[0031] The point cloud data input unit is used to input second point cloud data representing the weld point cloud data removed from the first point cloud data into a pre-trained defect recognition model for defect detection.

[0032] According to another aspect of the present invention, a submarine cable aluminum sheath weld defect detection system is provided, the detection system comprising: a box provided with a hole for the submarine cable to enter and exit, a structured light sensor, a motion device, and the industrial control computer described in the above embodiment; the inside of the hole is set as a detection area; the structured light sensor and the motion device are both arranged in the box, the structured light sensor is connected to the motion device, and the structured light sensor and the motion device are both connected to the industrial control computer for communication;

[0033] The structured light sensor is used to determine the distance difference between the depth of field range and the aluminum sheath of the submarine cable in the detection area; when the submarine cable enters the detection area from the hole, point cloud data of the surface of the aluminum sheath of the submarine cable in the detection area is collected and transmitted to the industrial computer;

[0034] The industrial computer is further used to control the motion device to drive the structured light sensor to move according to the distance difference.

[0035] Optionally, it also includes: a shading device;

[0036] The shading device is used to move to a set position when the submarine cable enters from the hole to block the ambient light that affects the structured light sensor.

[0037] The technical solution of the embodiment of the present invention obtains the first point cloud data representing the surface contour of the aluminum sheath of the submarine cable, determines the weld contour features for the weld point cloud data representing the weld contour on the aluminum sheath of the submarine cable that can be extracted from the first point cloud data, performs defect detection on the weld point cloud data according to the weld contour features, and then performs defect detection through a defect recognition model for the second point cloud data other than the weld point cloud data in the first point cloud data. The solution of the embodiment of the present invention combines the traditional point cloud algorithm with the deep learning algorithm, overcomes the problem of manual execution of weld defect detection at present, improves the detection efficiency and the accuracy of detection, can improve the quality of the aluminum sheath in the submarine cable, ensures the sealing and long-term stability of the submarine cable, reduces the maintenance cost due to damage, and ensures the safety and reliability of marine communications and energy transmission.

[0038] 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

[0039] 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.

[0040] Figure 1 This is a flow chart of a method for detecting weld defects of an aluminum sheath of a submarine cable provided in Embodiment 1 of the present invention;

[0041] Figure 2 It is a structural schematic diagram of an industrial computer provided by Embodiment 2 of the present invention;

[0042] Figure 3Schematic diagram of a submarine cable aluminum sheath weld defect detection system provided by Embodiment 3 of the present invention;

[0043] Figure 4 Schematic diagram of a submarine cable aluminum sheath weld defect detection system applicable to Embodiment 3 of the present invention;

[0044] Figure 5 It is a structural schematic diagram of an electronic device for implementing the method for detecting weld defects of aluminum sheath of a submarine cable according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] 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.

[0046] 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 interchanged 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. In addition, the terms "including" and "having" and any variations thereof 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.

[0047] Embodiment 1

[0048] Figure 1 This is a flow chart of a method for detecting defects in the weld of a submarine cable aluminum sheath provided by the first embodiment of the present invention. This embodiment is applicable to the case of detecting defects in the weld of a rotten aluminum sheath. The method can be executed by an industrial computer, which can be implemented in the form of hardware and / or software. The industrial computer device can be configured in a submarine cable aluminum sheath production line. Figure 1 As shown, the method includes:

[0049] S110, obtaining first point cloud data representing the surface profile of the aluminum sheath of the submarine cable.

[0050] As an important communication equipment laid on the seabed and connecting oil platforms with inland areas, cross-sea facilities with inland areas, the safety and reliability of submarine cables are crucial to the development of energy extraction and other aspects. There is an aluminum sheath in the submarine cable, and the quality of this sheath is directly related to the sealing, corrosion resistance and long-term stability of the submarine cable. Since the aluminum sheath needs to be wrapped with the internal cable through welding technology, the welding quality seriously affects the protective performance of the sheath.

[0051] Point cloud data refers to a set of vectors in a three-dimensional coordinate system. Therefore, the first point cloud data can reflect the surface profile of the aluminum sheath of the submarine cable through a set of data points. The acquisition of the first point cloud data can be achieved through sensing devices such as structured light sensors and millimeter wave radars, which are not specifically limited in the embodiments of the present invention.

[0052] In an embodiment of the present invention, the method may further include at least one of the following operations:

[0053] Smoothing the point cloud data and removing noise points in the first point cloud data by Gaussian filtering;

[0054] Downsampling the first point cloud data through a voxel grid of a preset size;

[0055] A target plane in the first point cloud data is identified by a plane fitting algorithm, and the first point cloud data corresponding to the target plane is removed.

[0056] The above operation is to preprocess the first point cloud data to improve image quality and reduce noise.

[0057] Gaussian filtering is a linear smoothing filter that is suitable for eliminating Gaussian noise and is widely used in the noise reduction process of image processing. Gaussian filtering is the process of weighted averaging the entire image. The value of each pixel is obtained by weighted averaging the value of itself and other pixels in the neighborhood. The specific operation of Gaussian filtering is: use a template (or convolution, mask) to scan each pixel in the image, and use the weighted average grayscale value of the pixels in the neighborhood determined by the template to replace the value of the central pixel of the template.

[0058] Set the voxel grid size and use voxel grid downsampling to reduce the point cloud data density while preserving key features.

[0059] The plane fitting algorithm can use the RANSAC algorithm to identify and remove planes in the point cloud in order to highlight the shape and structure of the weld. The RANSAC algorithm is an iterative algorithm used to estimate a mathematical model from a data set that contains a lot of noise and outliers. Its main principle is to randomly select some data points from the data set and use only these data points to estimate the model parameters. Then, the obtained model is used to measure other data points, and the points less than the threshold are regarded as inliers. Repeat the above process many times, and select the point set with the most inliers to calculate the accurate model parameters. This algorithm can not only effectively eliminate a large number of mismatched points, but also obtain more robust estimation results.

[0060] S120. Extract weld point cloud data representing a weld contour on the aluminum sheath of the submarine cable from the first point cloud data, determine weld contour features based on the weld point cloud data, and perform defect detection on the weld point cloud data based on the weld contour features.

[0061] Among them, the weld point cloud data in the first point cloud data can be extracted through edge detection, and the weld contour features can be further determined based on the data point distribution in the weld point cloud data, and defect detection can be performed based on the weld contour features.

[0062] In an embodiment of the present invention, extracting weld point cloud data representing the weld contour on the aluminum sheath of the submarine cable from the first point cloud data includes:

[0063] Determine the curvature of each data point in the first point cloud data, and take the data point whose curvature meets a preset condition as a candidate edge point;

[0064] The candidate edge points are fitted into a weld contour to obtain the weld point cloud data.

[0065] For each data point P in the first point cloud data i , whose normal vector Curvature Among them, λ1, λ2 are the principal curvatures, and N i The characteristic value of The points are taken as candidate edge points, and the candidate edge points are connected by the least square method to fit the straight line to form the weld contour; Gaussian filtering or B-spline fitting is used to smooth the edge line to make it smoother.

[0066] In an embodiment of the present invention, the step of performing defect detection on the weld point cloud data according to the weld contour feature includes:

[0067] In the case where the weld contour feature matches the contour defect feature corresponding to a preset standard defect type, the standard defect type is used as the defect type of the weld point cloud data.

[0068] Since weld defects mainly include cracks, pores, undercuts, lack of fusion and lack of penetration, the characteristics of different defects are very different. For example, hole defects usually appear as irregular high holes in the weld section, undercut defects form a shape similar to tooth marks at the edge of the weld, and crack defects appear as elongated holes in the weld area. In addition, weld deviation defects may cause the weld texture to suddenly narrow at a certain section. For welds with obvious features, large height or width differences, and discontinuous point clouds, their features can be identified through point cloud processing technology, and the weld contour features can be compared with the contour defect features of different defects to determine which standard defect type the weld belongs to.

[0069] In an embodiment of the present invention, the weld contour feature includes a height difference, an area width, and an area slope, wherein the height difference, the area width, and the area slope respectively characterize the height difference between the highest point and the lowest point of the weld contour corresponding to the weld point cloud data, the maximum distance of the edge, and the degree of inclination; accordingly, when the weld contour feature matches the contour defect feature corresponding to the preset standard defect type, the standard defect type is used as the defect type of the weld point cloud data, including:

[0070] When the height difference, area width and area slope of the weld contour feature are all within the height difference range, area width range and area slope range corresponding to a preset standard defect type, it is determined that the weld contour feature matches the contour defect feature corresponding to the preset standard defect type, and the standard defect type is used as the defect type of the current weld point cloud data.

[0071] The weld contour features of the extracted weld point cloud data are calculated to calculate the weld contour height difference h, defect area width w, and defect area slope k, where the point cloud in the contour is calculated point by point in order, and the defect area width w represents the difference in the horizontal direction of the defect boundary point; the weld surface defect depth h represents the maximum height difference in the area; the slope k represents the slope between the defect boundary point and the point with the maximum height difference. The defect type is determined according to the contour features. For welds with obvious defect area features, the defect classification is performed based on the obtained contour feature parameters and the threshold value set for the defect type. For example, for concave defects, since the width w and the height difference h are both large, the threshold values ​​w>5mm and h>2mm are set; for porosity defects, the slope, i.e., the height difference, is large, so the threshold values ​​w<2mm, h>2mm, and k>10 are set; therefore, if the weld contour feature w is greater than 5mm and h is greater than 2mm, the weld has a concave defect, and the same is true for porosity defects. If the weld profile feature does not belong to the profile defect feature corresponding to any standard defect type, the weld profile is within the normal range and there is no defect.

[0072] S130, inputting the second point cloud data obtained by removing the weld point cloud data from the first point cloud data into a pre-trained defect recognition model for defect detection.

[0073] In an embodiment of the present invention, the defect recognition model includes: an input layer, an upsampling layer, a convolution layer, and an activation layer; accordingly, the second point cloud data after removing the weld point cloud data from the first point cloud data is input into the pre-trained defect recognition model for defect detection, including:

[0074] Extracting a first feature image of each of the second point cloud data through an encoder at the input layer;

[0075] Upsampling the first feature image at the upsampling layer, processing the first feature image into a second feature image having the same size as the weld point cloud data, and performing feature fusion on the first feature image and the second feature image to obtain a third feature image;

[0076] Performing a preset number of convolutions at the convolution layer based on current weights and current biases to map the third feature image to a number of channels that matches the predicted segmentation mask;

[0077] The binary classification result of each pixel in the mapped third feature image is determined in the activation layer by a preset activation function, and the defect type of the second point cloud data is determined according to the binary classification result.

[0078] Among them, the defect recognition model can be implemented based on the U-Net model. For example, the U-Net model includes: five convolution layer blocks and one pooling layer, each convolution layer block includes two point cloud convolution layers; upsampling layer: bilinear interpolation method is used for upsampling to restore the feature map size of the encoder to the size of the original input image. Feature fusion is performed to stack two feature layers into one, and when the feature layer sizes are inconsistent, the smaller layer is filled; convolution layer: two convolutions are performed to map the feature map to the number of channels that match the predicted segmentation mask; activation layer: Sigmoid function is used to generate binary classification results for each pixel.

[0079] Specifically, after extracting the weld point cloud data from the first point cloud data, there may still be welds that cannot be identified in the first point cloud data. In this case, it is necessary to further perform defect detection through the defect recognition model. Exemplarily, the method of performing weld defect detection through the U-Net model may specifically include: Input layer: The encoder gradually reduces the image size and feature dimension through convolution and pooling operations to extract high-level features of the image. It includes five convolution layer blocks and one pooling layer, and each convolution layer block includes two point cloud convolution layers. Upsampling layer: Upsampling is performed using a bilinear interpolation method to restore the size of the encoder's feature map to the size of the original input image. Feature fusion is performed to stack the two feature layers into one, and when the feature layer sizes are inconsistent, the smaller layer is filled. Convolution layer: Perform two convolutions to map the feature map to the number of channels that matches the predicted segmentation mask. Activation layer: Use the Sigmoid function to generate a binary classification result for each pixel. The Sigmoid function maps the predicted value of each pixel to the range of [0,1], which can be interpreted as the probability that the pixel belongs to the target.

[0080] In an embodiment of the present invention, the method may further include the following steps:

[0081] The defect type obtained by the defect recognition model is verified, and the defect recognition model is updated in a preset manner according to the verification result; wherein the preset manner includes adjusting the parameters of a preset activation function of the defect recognition model and / or increasing the amount of training data of the defect recognition model through at least one data enhancement method; data enhancement includes at least one of random rotation, flipping and scaling processing.

[0082] After obtaining the defect type through the defect recognition model, the corresponding weld can be reviewed to determine whether the defect recognition model is accurate. If the recognition is wrong, the verification result indicates that the accuracy of the defect recognition model is insufficient and the defect recognition model needs to be updated in a preset way. Specifically, various data enhancement operations are performed on the training data, such as random rotation, flipping, scaling, etc., to increase the diversity of training samples and improve the generalization ability and robustness of the model. The training data can be data collected by simulating weld defects or some public data sets of weld defects. During the training process, stochastic gradient descent (SGD) is used to optimize the model parameters. Adjust the weights and biases in the defect recognition model.

[0083] The solution of the embodiment of the present invention provides a defect detection algorithm suitable for the aluminum sheath weld of the submarine cable. It uses the traditional point cloud algorithm combined with the deep learning method to measure the defect area characteristics at multiple positions according to the defect characteristics of the weld, and judge the type and degree of the defect based on the measurement results; in view of the small size of the weld defect characteristics and other problems, the U-Net network is improved to retain more weld defect detail information, improve the accuracy and quality of segmentation and detection accuracy. This invention can improve the quality of the aluminum sheath in the submarine cable, ensure the sealing and long-term stability of the submarine cable, reduce the maintenance cost caused by damage, and ensure the safety and reliability of marine communications and energy transmission.

[0084] Embodiment 2

[0085] Figure 2 is a structural diagram of an industrial computer provided by Embodiment 2 of the present invention, such as Figure 2 As shown, the industrial computer includes:

[0086] A point cloud data acquisition unit 210 acquires first point cloud data representing a surface profile of the aluminum sheath of the submarine cable;

[0087] A weld point cloud data extraction unit 220 is used to extract weld point cloud data representing the weld contour on the aluminum sheath of the submarine cable from the first point cloud data, determine weld contour features according to the weld point cloud data, and perform defect detection on the weld point cloud data according to the weld contour features;

[0088] The point cloud data input unit 230 is used to input the second point cloud data representing the weld point cloud data removed from the first point cloud data into a pre-trained defect recognition model for defect detection.

[0089] Optionally, the device may further include: a preprocessing unit 240, the preprocessing unit 240 being specifically configured to perform at least one of the following operations:

[0090] Smoothing the point cloud data and removing noise points in the first point cloud data by Gaussian filtering;

[0091] Downsampling the first point cloud data through a voxel grid of a preset size;

[0092] A target plane in the first point cloud data is identified by a plane fitting algorithm, and the first point cloud data corresponding to the target plane is removed.

[0093] Optionally, the weld point cloud data extraction unit 220, when extracting weld point cloud data representing a weld contour on the aluminum sheath of the submarine cable from the first point cloud data, specifically performs:

[0094] Determine the curvature of each data point in the first point cloud data, and take the data point whose curvature meets a preset condition as a candidate edge point;

[0095] The candidate edge points are fitted into a weld contour to obtain the weld point cloud data.

[0096] Optionally, the weld point cloud data extraction unit 220, when performing defect detection on the weld point cloud data according to the weld contour feature, specifically performs:

[0097] In the case where the weld contour feature matches the contour defect feature corresponding to a preset standard defect type, the standard defect type is used as the defect type of the weld point cloud data.

[0098] Optionally, the weld contour feature includes a height difference, an area width, and an area slope, wherein the height difference, the area width, and the area slope respectively characterize a height difference between a highest point and a lowest point of a weld contour corresponding to the weld point cloud data, a maximum distance of an edge, and a degree of inclination; accordingly, the weld point cloud data extraction unit 220, when performing matching between the weld contour feature and a contour defect feature corresponding to a preset standard defect type, takes the standard defect type as the defect type of the weld point cloud data, specifically performs:

[0099] When the height difference, area width and area slope of the weld contour feature are all within the height difference range, area width range and area slope range corresponding to a preset standard defect type, it is determined that the weld contour feature matches the contour defect feature corresponding to the preset standard defect type, and the standard defect type is used as the defect type of the current weld point cloud data.

[0100] Optionally, the defect recognition model includes: an input layer, an upsampling layer, a convolution layer and an activation layer; accordingly, the point cloud data input unit 230, when inputting the second point cloud data after removing the weld point cloud data from the first point cloud data into the pre-trained defect recognition model for defect detection, specifically performs:

[0101] Extracting a first feature image of each of the second point cloud data through an encoder at the input layer;

[0102] Upsampling the first feature image at the upsampling layer, processing the first feature image into a second feature image having the same size as the weld point cloud data, and performing feature fusion on the first feature image and the second feature image to obtain a third feature image;

[0103] Performing a preset number of convolutions at the convolution layer based on current weights and current biases to map the third feature image to a number of channels that matches the predicted segmentation mask;

[0104] The binary classification result of each pixel in the mapped third feature image is determined in the activation layer by a preset activation function, and the defect type of the second point cloud data is determined according to the binary classification result.

[0105] Optionally, the device may also include: a model updating module 250; the model updating module 250 is used to execute: verifying the defect type obtained by the defect recognition model, and updating the defect recognition model in a preset manner according to the verification result; wherein the preset manner includes adjusting the parameters of the preset activation function of the defect recognition model and / or increasing the amount of training data of the defect recognition model through at least one data enhancement method; data enhancement includes at least one of random rotation, flipping and scaling processing.

[0106] The industrial control computer provided in the embodiment of the present invention can execute the submarine cable aluminum sheath weld defect detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0107] Embodiment 3

[0108] Figure 3 Schematic diagram of a submarine cable aluminum sheath weld defect detection system provided by Embodiment 3 of the present invention. Figure 3 As shown, the system includes:

[0109] A box 310 is provided with a hole 300 for the entry and exit of a submarine cable, a structured light sensor 320, a motion device 330, and an industrial computer 340 provided in the above embodiment; the inside of the hole 300 is set as a detection area; the structured light sensor 320 and the motion device 330 are both arranged in the box 310, the structured light sensor 320 is connected to the motion device 330, and the structured light sensor 320 and the motion device 330 are both connected to the industrial computer 340 for communication;

[0110] The structured light sensor 320 is used to determine the distance difference between the depth of field range and the aluminum sheath of the submarine cable in the detection area; when the submarine cable enters the detection area from the hole, point cloud data of the surface of the aluminum sheath of the submarine cable in the detection area is collected and transmitted to the industrial computer 340;

[0111] The industrial computer 340 is further configured to control the motion device 330 to drive the structured light sensor 320 to move according to the distance difference.

[0112] Preferably, in order to ensure the firmness of the box, the box can be a vertical cabinet, which is made of structural steel and sheet metal welding and bolted. The box is a supporting structure of the whole equipment, and there is a hole in the box, so that the submarine cable can pass through it. The box can be set in the installation or production line of the aluminum sheath of the submarine cable, so that the moving device in the assembly line drives the submarine cable from the hole into the detection area of ​​the box. Since the depth of field range of the structured light sensor is fixed, and only when the aluminum sheath of the submarine cable is within the depth of field range can effective point cloud data be obtained, a motion module is set. When the aluminum sheath of the submarine cable enters the detection area, due to the different diameters of the aluminum sheaths of different submarine cables, the distance from the depth of field range is also different. At this time, the structured light sensor determines whether the distance to the aluminum sheath of the submarine cable is within the depth of field range. If it is not within the range, the distance h required to be moved in the vertical direction is determined and fed back to the industrial control computer. The industrial control computer controls the motion device to drive the structured light sensor to move a distance h in the vertical direction, and the structured light sensor starts to collect the first point cloud data. In addition, the distance difference can also be determined by an external distance sensor, which is fixed on one side of the structured light sensor to measure the distance to the aluminum sheath of the submarine cable.

[0113] Optionally, the system may further include: a shading device 350; the shading device 350 is used to move to a set position when the submarine cable enters from the hole to block ambient light that affects the structured light sensor.

[0114] Preferably, the shading device is an I / O control mechanism of a cylinder-driven shading plate, which is arranged above the hole of the box body and realizes the shading action before the structured light sensing image is collected by reciprocating extension and contraction, thereby avoiding the interference of arc light on the structured light sensor.

[0115] After all positions of the submarine cable aluminum sheath welds are inspected, the submarine cable stops moving and a test result report including all positions is generated, such as cracks appearing at the aluminum sheath at the xth meter of the submarine cable and pores appearing at the aluminum sheath at the yth meter.

[0116] In addition, the system may also include an alarm device and a display screen. Once a defect is detected, the alarm device is triggered to sound an alarm and display relevant information on the display screen, such as a weld leak on the aluminum sheath of the zth meter of the submarine cable.

[0117] For example, Figure 4 Schematic diagram of a submarine cable aluminum sheath weld defect detection system applicable to the third embodiment of the present invention. Figure 4The visual inspection module shown includes a structured light sensor and an autofocus module. The aluminum sheath of the submarine cable is placed in the cable placement position in the inspection system, and the shading motion mechanism automatically moves to the preset position to reduce the influence of arc light. The distance sensor in the visual inspection module starts to measure the distance of the cable protective sheath, compares the distance with the depth of field of the structured light sensor, and calculates the height difference h. The motion control mechanism drives the structured light sensor to move in the vertical direction, adjusts the position of the structured light sensor according to the height difference, and realizes autofocus. The structured light sensor starts to collect data, and the submarine cable moves a distance after the collection is completed. At the same time, the algorithm starts to analyze and process the collected data, and detects the weld quality of the cable protective sheath in real time. In particular, the defect detection process is specifically as follows: preprocessing: filtering and downsampling the weld point cloud image; feature parameter calculation: extracting the weld point cloud after edge detection of the preprocessed point cloud, and calculating the weld shape parameters such as the height, width, and continuity of the weld; defect recognition: according to the calculated weld parameters, using traditional point cloud algorithms and U-Net models for classification and recognition. When the defect detection is completed, the first point cloud data and the identified defect information are saved in the local industrial computer to facilitate the subsequent weld defect problem investigation. Once a defect is detected, the alarm light will flash and the buzzer will be displayed on the display. After the detection of all positions of the submarine cable aluminum sheath weld is completed, the submarine cable stops moving and the system generates a test result report including all positions.

[0118] Embodiment 4

[0119] Figure 5 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, mainframe computers, 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 required herein.

[0120] like Figure 5As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, 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, the ROM 12, and the 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.

[0121] A number of 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.

[0122] The processor 11 may be a variety of 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, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as a method for detecting defects in a submarine cable aluminum sheath weld.

[0123] In some embodiments, the method for detecting defects in the weld of the aluminum sheath of a submarine cable 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 method for detecting defects in the weld of the aluminum sheath of a submarine cable described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the method for detecting defects in the weld of the aluminum sheath of a submarine cable by any other appropriate means (e.g., by means of firmware).

[0124] Various implementations of the systems and techniques described above herein can 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 implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can 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.

[0125] Computer programs for implementing the methods 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 special-purpose 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 and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0126] 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 combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. 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.

[0127] 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 the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the 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).

[0128] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by 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), a blockchain network, and the Internet.

[0129] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding 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 the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0130] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0131] 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 method for detecting defects in weld seams of aluminum sheaths of submarine cables. It is characterized in that include: Acquire first point cloud data representing the surface profile of the submarine cable aluminum sheath; Extracting weld point cloud data representing a weld contour on the aluminum sheath of the submarine cable from the first point cloud data, determining weld contour features according to the weld point cloud data, and performing defect detection on the weld point cloud data according to the weld contour features; The second point cloud data obtained by removing the weld point cloud data from the first point cloud data is input into a pre-trained defect recognition model for defect detection.

2. The method according to claim 1, It is characterized in that Before extracting the weld point cloud data representing the weld contour on the aluminum sheath of the submarine cable from the first point cloud data, at least one of the following operations is also included: Smoothing the point cloud data and removing noise points in the first point cloud data by Gaussian filtering; Downsampling the first point cloud data through a voxel grid of a preset size; A target plane in the first point cloud data is identified by a plane fitting algorithm, and the first point cloud data corresponding to the target plane is removed.

3. The method according to claim 1, It is characterized in that The step of extracting weld point cloud data representing the weld contour on the aluminum sheath of the submarine cable from the first point cloud data comprises: Determine the curvature of each data point in the first point cloud data, and take the data point whose curvature meets a preset condition as a candidate edge point; The candidate edge points are fitted into a weld contour to obtain the weld point cloud data.

4. The method according to claim 1, It is characterized in that The performing defect detection on the weld point cloud data according to the weld contour feature comprises: In the case where the weld contour feature matches the contour defect feature corresponding to a preset standard defect type, the standard defect type is used as the defect type of the weld point cloud data.

5. The method according to claim 4, It is characterized in that The weld contour feature includes a height difference, an area width and an area slope, wherein the height difference, the area width and the area slope respectively characterize the height difference between the highest point and the lowest point of the weld contour corresponding to the weld point cloud data, the maximum distance of the edge and the degree of inclination; accordingly, when the weld contour feature matches the contour defect feature corresponding to the preset standard defect type, the standard defect type is used as the defect type of the weld point cloud data, including: When the height difference, area width and area slope of the weld contour feature are all within the height difference range, area width range and area slope range corresponding to a preset standard defect type, it is determined that the weld contour feature matches the contour defect feature corresponding to the preset standard defect type, and the standard defect type is used as the defect type of the current weld point cloud data.

6. The method according to claim 1, It is characterized in that The defect recognition model includes: an input layer, an upsampling layer, a convolution layer and an activation layer; accordingly, the second point cloud data after removing the weld point cloud data from the first point cloud data is input into the pre-trained defect recognition model for defect detection, including: Extracting a first feature image of each of the second point cloud data through an encoder at the input layer; Upsampling the first feature image at the upsampling layer, processing the first feature image into a second feature image having the same size as the weld point cloud data, and performing feature fusion on the first feature image and the second feature image to obtain a third feature image; Performing a preset number of convolutions at the convolution layer based on current weights and current biases to map the third feature image to a number of channels that matches the predicted segmentation mask; The binary classification result of each pixel in the mapped third feature image is determined in the activation layer by a preset activation function, and the defect type of the second point cloud data is determined according to the binary classification result.

7. The method according to claim 6, It is characterized in that Also includes: Verifying the defect type obtained by the defect recognition model, and updating the defect recognition model in a preset manner according to the verification result; wherein the preset manner includes adjusting parameters of a preset activation function of the defect recognition model and / or increasing the amount of training data of the defect recognition model by at least one data enhancement method; The data augmentation includes at least one of random rotation, flipping and scaling.

8. An industrial computer, It is characterized in that include: A point cloud data acquisition unit, which acquires first point cloud data representing the surface profile of the aluminum sheath of the submarine cable; a weld point cloud data extraction unit, configured to extract weld point cloud data representing a weld contour on the aluminum sheath of the submarine cable from the first point cloud data, determine weld contour features according to the weld point cloud data, and perform defect detection on the weld point cloud data according to the weld contour features; The point cloud data input unit is used to input second point cloud data representing the weld point cloud data removed from the first point cloud data into a pre-trained defect recognition model for defect detection.

9. A submarine cable aluminum sheath weld defect detection system, It is characterized in that The detection system comprises: a box provided with a hole for the entry and exit of a submarine cable, a structured light sensor, a motion device, and the industrial computer according to claim 8; the interior of the hole is set as a detection area; the structured light sensor and the motion device are both arranged in the box, the structured light sensor is connected to the motion device, and the structured light sensor and the motion device are both connected to the industrial computer for communication; The structured light sensor is used to determine the distance difference between the depth of field range and the aluminum sheath of the submarine cable in the detection area; when the submarine cable enters the detection area from the hole, point cloud data of the surface of the aluminum sheath of the submarine cable in the detection area is collected and transmitted to the industrial computer; The industrial computer is further used to control the motion device to drive the structured light sensor to move according to the distance difference.

10. The system according to claim 9, It is characterized in that Also includes: Shading device; The shading device is used to move to a set position when the submarine cable enters from the hole to block the ambient light that affects the structured light sensor.

Citation Information

Patent Citations

  • Weld defect real-time detection method and system based on 3D point cloud

    CN114354639A

  • Sealing nail welding seam quality detection method in complex environment

    CN116109587A

  • Method for automatically identifying defects of corrugated aluminum sheath of submarine cable

    CN116452513A

  • Weld joint surface quality detection method and computer readable storage medium

    CN116664508A

  • Weld seam internal defect intelligent detection device and method, and medium

    WO2022053001A1