Secondary water supply equipment welding spot quality detection method

Through image and vibration signal analysis, combined with convolutional neural network detection model, the problem of subjectivity and low efficiency of traditional solder joint detection methods is solved, the precise quality evaluation of solder joints of secondary water supply equipment is realized, and the safety of equipment is improved and reliability.

CN119090836AInactive Publication Date: 2024-12-06GUIZHOU ZHUWEI SMART WATER CO LTD

Patent Information

Application Number
CN202411180048.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The traditional welding joint quality inspection method relies on the experience of the operator, has subjectivity and inconsistency, and is low in efficiency, so it is impossible to effectively evaluate the welding joint quality of secondary water supply equipment.

Method used

Welding joint images and vibration signals are collected, defect detection models are constructed through image crack recognition and convolutional neural network, and the soldering joint quality is judged based on surface and internal detection results.

Benefits of technology

Accurate inspection of welding joint quality is achieved, economic losses caused by welding problems are avoided, and the safety and reliability of the equipment are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090836A_ABST
    Figure CN119090836A_ABST
Patent Text Reader

Abstract

The invention discloses a secondary water supply equipment welding spot quality detection method, which belongs to the technical field of welding spot detection, and comprises the following steps: collecting a welding spot image and a vibration signal of secondary water supply equipment; performing crack identification on the welding spot image to obtain a surface detection result of the welding spot of the secondary water supply equipment; constructing a defect detection model based on a convolutional neural network, and identifying the vibration signal through the defect detection model to obtain an internal detection result of the welding spot of the secondary water supply equipment; and combining the surface detection result and the internal detection result of the welding spot of the secondary water supply equipment to obtain the quality of the welding spot of the secondary water supply equipment. Through the combination of the image and the vibration signal, whether the welding spot of the secondary water supply equipment has cracks or not can be effectively detected, and economic loss caused by unqualified welding quality is prevented.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of solder joint detection, and in particular relates to a method for detecting solder joint quality of secondary water supply equipment. Background Art

[0002] Secondary water supply equipment plays a vital role in the urban water supply system. It not only guarantees the daily water needs of residents, but also ensures the safety and sanitation of water quality. However, the operating stability and safety of the equipment are directly related to the water safety of users. Therefore, every link in its manufacturing and maintenance process is particularly important.

[0003] Welding is a key connection technology in the manufacturing and maintenance of secondary water supply equipment. The quality of its welds directly affects the overall performance and service life of the equipment. A high-quality weld can not only ensure the sealing and durability of the equipment, but also avoid problems such as leakage and corrosion caused by poor welding, thereby ensuring the normal operation of the water supply system.

[0004] Traditional solder joint quality inspection methods such as visual inspection and pressure testing, although they can evaluate the quality of solder joints to a certain extent, often rely on the operator's experience and have a certain degree of subjectivity. This means that different operators may have different judgments on the quality of the same solder joint, resulting in inconsistent inspection results. In addition, these traditional methods usually take a long time to complete and are relatively inefficient. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a method for detecting the quality of welding points of secondary water supply equipment to solve the problems existing in the above prior art.

[0006] To achieve the above object, the present invention provides a method for detecting the quality of solder joints of secondary water supply equipment, comprising the following steps:

[0007] Collect welding point images and vibration signals of secondary water supply equipment;

[0008] Performing crack recognition on the weld spot image to obtain surface detection results of the weld spots of the secondary water supply equipment;

[0009] A defect detection model is constructed based on a convolutional neural network, and the vibration signal is identified by the defect detection model to obtain internal detection results of the welding points of the secondary water supply equipment;

[0010] The quality of the secondary water supply equipment solder joints is obtained by combining the surface inspection results and the internal inspection results of the secondary water supply equipment solder joints.

[0011] Preferably, the method for performing crack identification on the weld spot image comprises:

[0012] Grayscale the solder joint image to obtain a grayscale image;

[0013] Calculating the gradients of the grayscale image in the horizontal direction and the vertical direction by first-order partial derivatives to obtain the gradient information of each row and each column in the grayscale image;

[0014] A segmentation threshold is determined according to the gradient information, and the grayscale image is discretely segmented based on the segmentation threshold to obtain a crack segmentation map of the secondary water supply equipment.

[0015] Preferably, the grayscale processing method includes: performing grayscale processing by a grayscale function, and the expression of the grayscale function is:

[0016] f(x,y)=I(x,y);

[0017] Where I represents the intensity function of the grayscale image, and f(x, y) represents the welding point image of the secondary water supply equipment.

[0018] Preferably, the method for determining the segmentation threshold according to the gradient information comprises: calculating the gradient amplitude through the gradient information, obtaining a local maximum value based on the gradient amplitude, and using the local maximum value as the segmentation threshold;

[0019] The calculation expression of the gradient amplitude is:

[0020]

[0021] In the formula, represents the gradient magnitude, G x Represents the horizontal gradient, G y Represents the vertical gradient.

[0022] Preferably, the expression of the gradient amplitude is:

[0023] θ = atan2(Gy,Gx);

[0024] Where atan2 represents the inverse tangent function of two parameters.

[0025] Preferably, the method for constructing a defect detection model based on a convolutional neural network includes:

[0026] Select the pre-trained AlexNet network to build the initial defect detection model;

[0027] Collect the vibration signal of the welding point and convert it into a spectrum diagram;

[0028] An initial defect detection model is trained based on the frequency spectrum to obtain a defect detection model.

[0029] Preferably, the loss function for training the initial defect detection model is:

[0030] Loss=-(ylog(p)+(1-y)log(1-p));

[0031] Where y is the true label and p is the probability that the model predicts that the sample contains cracks.

[0032] Preferably, before training the initial defect detection model based on the spectrum graph, the spectrum graph is subjected to wavelet transform processing, and the wavelet transform processing method includes:

[0033] Select the wavelet basis function and set the decomposition scale according to the vibration signal acquisition method;

[0034] Discretization sampling is performed through discrete wavelet transform to obtain sampling data;

[0035] Perform inverse wavelet transform on the sampled data to obtain the original signal.

[0036] Compared with the prior art, the present invention has the following advantages and technical effects:

[0037] The present invention discloses a method for detecting the quality of solder joints of secondary water supply equipment, which belongs to the technical field of solder joint detection, and comprises the following steps: collecting solder joint images and vibration signals of secondary water supply equipment; performing crack recognition on the solder joint images to obtain surface detection results of the solder joints of secondary water supply equipment; constructing a defect detection model based on a convolutional neural network, identifying the vibration signals through the defect detection model, and obtaining internal detection results of the solder joints of secondary water supply equipment; combining the surface detection results and internal detection results of the solder joints of secondary water supply equipment to obtain the quality of the solder joints of secondary water supply equipment. The present invention combines the two detection results to comprehensively judge the quality of the solder joints. This method effectively combines image and vibration signal analysis, can accurately detect whether there are cracks in the solder joints, avoids economic losses caused by welding problems, and improves the safety and reliability of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0039] Figure 1 This is a flow chart of a method for detecting quality of solder joints of secondary water supply equipment according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0041] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0042] Embodiment 1

[0043] like Figure 1 As shown, this embodiment provides a method for detecting the quality of welding points of secondary water supply equipment, comprising the following steps:

[0044] Collect welding point images and vibration signals of secondary water supply equipment;

[0045] Perform crack recognition on the weld spot image to obtain the surface inspection results of the weld spot of the secondary water supply equipment;

[0046] A defect detection model is built based on a convolutional neural network. The vibration signal is identified through the defect detection model to obtain the internal detection results of the welding points of the secondary water supply equipment.

[0047] The quality of the solder joints of the secondary water supply equipment is obtained by combining the surface inspection results and internal inspection results of the solder joints of the secondary water supply equipment.

[0048] Furthermore, the method for performing crack recognition on a solder joint image includes:

[0049] When processing the solder joint image, the first step is to grayscale it. The purpose of this step is to convert the original color image into a grayscale image for subsequent analysis and processing. After obtaining the grayscale image, the next task is to calculate the horizontal and vertical gradients of the grayscale image through the first-order partial derivative. This process is to obtain the gradient information of each row and column in the grayscale image, which is crucial for subsequent image segmentation.

[0050] After obtaining the gradient information, the next step is to determine the segmentation threshold based on this information. The selection of the segmentation threshold is a key step because it directly affects the effect of image segmentation. Once the segmentation threshold is determined, the grayscale image can be discretely segmented based on this threshold. The discrete segmentation process is to divide the pixels in the image into different categories or regions based on the relationship between their grayscale values ​​and the segmentation threshold, and obtain the crack segmentation map of the secondary water supply equipment.

[0051] Furthermore, the grayscale processing method includes: performing grayscale processing by a grayscale function, and the expression of the grayscale function is:

[0052] f(x,y)=I(x,y);

[0053] Where I represents the intensity function of the grayscale image, and f(x, y) represents the welding point image of the secondary water supply equipment.

[0054] Furthermore, the method for determining the segmentation threshold according to the gradient information includes: calculating the gradient amplitude according to the gradient information, obtaining the local maximum value based on the gradient amplitude, and using the local maximum value as the segmentation threshold;

[0055] The calculation expression of the gradient amplitude is:

[0056]

[0057] In the formula, represents the gradient magnitude, G x Represents the horizontal gradient, G y Represents the vertical gradient.

[0058] Furthermore, the expression of the gradient amplitude is:

[0059] θ=atan2(G y ,G x );

[0060] Where atan2 represents the inverse tangent function of two parameters.

[0061] Furthermore, the method of constructing a defect detection model based on a convolutional neural network includes:

[0062] In the process of building a defect detection model, it is first necessary to select a pre-trained convolutional neural network as a basis. In this embodiment, the AlexNet network is selected to build the initial defect detection model. AlexNet is a deep convolutional neural network that has a wide range of applications and excellent performance in the field of image recognition.

[0063] Next, the vibration signals of the solder joints need to be collected. These signals contain various information about the solder joints during operation, including their working status and whether there are defects. In order to make better use of this information, the vibration signals need to be converted into a spectrogram. A spectrogram is a method of displaying the frequency components of a signal in a graphical form. It allows us to see the various frequency components in the signal more intuitively, thereby better understanding the characteristics of the signal.

[0064] With the spectrograms, the initial defect detection model can be trained based on these spectrograms. In this process, a large number of spectrograms and the corresponding solder joint status (whether there is a defect) are used as training data, and the parameters of the model are continuously adjusted through algorithms such as back propagation, so that the model can better identify defective solder joints.

[0065] After a period of training, a defect detection model with good performance can be obtained.

[0066] Furthermore, the loss function for training the initial defect detection model is:

[0067] Loss=-(ylog(p)+(1-y)log(1-p));

[0068] Where y is the true label and p is the probability that the model predicts that the sample contains cracks.

[0069] Furthermore, before training the initial defect detection model based on the spectrum graph, the spectrum graph is subjected to wavelet transform processing, and the wavelet transform processing method includes:

[0070] First, a wavelet basis function needs to be selected. In this embodiment, Haar wavelet is selected as the wavelet basis function. Haar wavelet is a simple and effective wavelet basis function, which has good time-frequency local characteristics and can effectively extract the characteristic information of the signal.

[0071] Next, the decomposition scale needs to be set according to the vibration generation method of the vibration signal. This step is to ensure the accuracy and effectiveness of the wavelet transform. By setting the appropriate decomposition scale, the vibration signal can be better analyzed and processed.

[0072] Then, discrete sampling is performed through discrete wavelet transform to obtain sampled data. Discrete wavelet transform is a method of converting continuous time signals into discrete time signals, which can effectively reduce the complexity of data processing and improve processing speed.

[0073] Finally, the sampled data is inversely transformed to obtain the original signal. Inverse wavelet transform is the inverse process of wavelet transform, which can restore the data processed by wavelet transform to its original state, thus obtaining the required signal.

[0074] The present invention considers the contrast of extreme local features and is less affected by the pre-processing process. Therefore, it is not necessary to pre-process the image during segmentation, thus avoiding image deformation caused by pre-processing and saving data resources to the greatest extent.

[0075] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for detecting the quality of solder joints of secondary water supply equipment, characterized in that: The following steps are involved: Collect welding point images and vibration signals of secondary water supply equipment; Performing crack recognition on the weld spot image to obtain surface detection results of the weld spots of the secondary water supply equipment; A defect detection model is constructed based on a convolutional neural network, and the vibration signal is identified by the defect detection model to obtain internal detection results of the welding points of the secondary water supply equipment; The quality of the secondary water supply equipment solder joints is obtained by combining the surface inspection results and the internal inspection results of the secondary water supply equipment solder joints.

2. The secondary water supply equipment solder joint quality detection method according to claim 1, characterized in that: The method for performing crack recognition on the solder joint image includes: Grayscale the solder joint image to obtain a grayscale image; Calculating the gradients of the grayscale image in the horizontal direction and the vertical direction by first-order partial derivatives to obtain the gradient information of each row and each column in the grayscale image; A segmentation threshold is determined according to the gradient information, and the grayscale image is discretely segmented based on the segmentation threshold to obtain a crack segmentation map of the secondary water supply equipment.

3. The secondary water supply equipment solder joint quality detection method according to claim 2, characterized in that: The grayscale processing method includes: performing grayscale processing by a grayscale function, and the expression of the grayscale function is: f(x,y)=I(x,y); Where I represents the intensity function of the grayscale image, and f(x, y) represents the welding point image of the secondary water supply equipment.

4. The secondary water supply equipment solder joint quality detection method according to claim 2, characterized in that: The method for determining the segmentation threshold according to the gradient information includes: calculating the gradient amplitude through the gradient information, obtaining a local maximum value based on the gradient amplitude, and using the local maximum value as the segmentation threshold; The calculation expression of the gradient amplitude is: In the formula, represents the gradient magnitude, G x Represents the horizontal gradient, G y Represents the vertical gradient.

5. The secondary water supply equipment solder joint quality detection method according to claim 4, characterized in that: The expression of the gradient amplitude is: θ=atan2(G y ,G x ); Where atan2 represents the inverse tangent function of two parameters.

6. The secondary water supply equipment solder joint quality detection method according to claim 1, characterized in that: Methods for building defect detection models based on convolutional neural networks include: Select the pre-trained AlexNet network to build the initial defect detection model; Collect the vibration signal of the welding point and convert it into a spectrum diagram; An initial defect detection model is trained based on the frequency spectrum to obtain a defect detection model.

7. The secondary water supply equipment solder joint quality detection method according to claim 6, characterized in that: The loss function for training the initial defect detection model is: Loss=-(ylog(p)+(1-y)log(1-p)); Where y is the true label and p is the probability that the model predicts that the sample contains cracks.

8. The secondary water supply equipment solder joint quality detection method according to claim 6, characterized in that: Before training the initial defect detection model based on the spectrum graph, the spectrum graph is subjected to wavelet transform processing, and the wavelet transform processing method includes: Select the wavelet basis function and set the decomposition scale according to the vibration signal acquisition method; Discretization sampling is performed through discrete wavelet transform to obtain sampling data; Perform inverse wavelet transform on the sampled data to obtain the original signal.

Citation Information

Patent Citations

  • Welding spot defect detection system based on machine vision

    CN109030502A

  • Concrete structure internal defect detection method based on deep learning

    CN110045015A

  • Convolutional neural network training method and welding spot defect detection method and device

    CN110930347A

  • SMT welding spot defect detection method based on iForest model verification

    CN111145175A

  • Welding seam defect detection method and system, storage medium, computer equipment and terminal

    CN111932489A

Cited By

  • Welding method and system based on deep learning

    CN121147645A