Weld joint quality detection system based on combination of visual identification and current and voltage
By combining visual identification and current voltage detection technology to calculate the internal and surface quality scores of the weld, the problem of difficulty in detecting internal defects of welds and evaluating surface defects in the prior art is solved, and the accuracy of weld quality detection is improved.
Patent Information
- Application Number
- CN202410968339.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-08
AI Technical Summary
The prior art is difficult to detect internal defects of welds, resulting in inaccurate weld quality detection results and the severity of the weld surface defects and the size of defect areas cannot be comprehensively evaluated.
Using a weld quality detection system based on visual recognition and current voltage, the data processing module collects the electrical parameters and image recognition algorithm of the weld to extract surface defect characteristics, calculates the internal and surface quality scores of the weld, and analyzes defects in combination with welding parameters. reason.
Accurate detection of the internal and surface quality of the weld seam is achieved, the accuracy of the quality inspection of the weld seam is improved, and the internal and surface quality conditions of the weld seam are accurately reflected.
Smart Images

Figure CN118914196A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of welding quality detection, and in particular is a welding seam quality detection system based on the combination of visual recognition and current and voltage. Background Art
[0002] As welding technology is increasingly used in the manufacturing industry, the quality of its welds is directly related to the performance, safety and service life of the product. Therefore, efficient and accurate defect detection of welds has become an important part of ensuring weld quality.
[0003] The prior art collects the original weld image of the weld, pre-processes the original weld image and extracts features to obtain defect features, and analyzes the defect feature image to obtain the quality of the weld. However, although the prior art can detect the type and location of surface defects of the weld, it is difficult to detect internal defects of the weld by visual recognition technology, resulting in inaccurate results of weld quality inspection.
[0004] The present invention proposes a weld quality detection system based on the combination of visual recognition and current and voltage to solve the above technical problems. Summary of the invention
[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a weld quality detection system based on a combination of visual recognition and current and voltage, which is used to solve the problem that the prior art does not comprehensively evaluate the surface quality of the weld according to the severity of the surface defect type and the size of the defect area, resulting in inaccurate results of the weld quality detection; in addition, most of the prior art only uses visual recognition to detect weld quality, and visual recognition is difficult to detect internal defects of the weld, resulting in inaccurate results of the weld quality detection.
[0006] To achieve the above-mentioned object, the first aspect of the present invention provides a weld quality detection system based on visual recognition combined with current and voltage, comprising: a data processing module, and a data acquisition module and a quality detection module connected thereto;
[0007] The data acquisition module is used to acquire the original weld image and welding parameters of the weld to be tested, and to acquire the electrical parameters of several detection areas in the weld to be tested; wherein the welding parameters include the welding current, welding voltage, welding speed and welding angle during the welding process, and the electrical parameters include the detection current and detection voltage;
[0008] The data processing module is used to calculate the internal quality score of the weld to be tested based on the electrical parameters, and to extract the surface defect features in the original weld image based on the image recognition algorithm, and to calculate the surface quality score of the weld to be tested based on the surface defect features; wherein the surface defect features include the surface defect type and the corresponding pixel ratio, and the surface defect type includes burn-through, incomplete welding, undercut, surface pores, surface cracks and weld nodules;
[0009] The quality inspection module: evaluates the quality of the weld to be tested based on the surface quality score and the internal quality score of the weld to be tested, and obtains the weld inspection result; wherein the weld inspection result includes the surface quality inspection result and the internal quality inspection result; and,
[0010] The causes of weld defects are analyzed according to the weld detection results and the welding parameters; wherein the causes of weld defects include abnormal welding current, abnormal welding voltage, abnormal welding speed and abnormal welding angle.
[0011] Preferably, the division of several detection areas includes:
[0012] The weld to be tested is divided into several detection areas with equal distances along the weld direction, and the detection areas are marked as i; wherein i=1, 2, ..., n, and n is the total number of detection areas.
[0013] Preferably, the calculating the internal quality score of the weld to be tested based on the electrical parameters comprises:
[0014] A1: Extract the detection current I of the detection area i i and the detection voltage V i ;
[0015] A2: By formula R i =V i / I i Calculate the sense resistor R i ;
[0016] A3: Set the resistance range threshold [DY1, DY2], and DY1 <DY2;
[0017] A4: Determine the detection resistance R i Is it less than the minimum value DY1 of the resistance range threshold? If yes, mark the detection area i as a normal area; if no, jump to A5;
[0018] A5: Determine the detection resistance R i Whether it is within the resistance range threshold; if yes, the detection area i is marked as a slightly abnormal area; if no, the detection area i is marked as a seriously abnormal area;
[0019] A6: Calculate the proportion of slightly abnormal areas ZB1 by the formula ZB1=q / n; where q is the number of slightly abnormal areas;
[0020] A7: Calculate the proportion of severe abnormal areas ZB2 by the formula ZB2=y / n; where y is the number of severe abnormal areas;
[0021] A8: The internal quality score NP of the weld is calculated by the formula NP = α × ZB1 + β × ZB2; where α and β are proportional coefficients greater than 0.
[0022] It should be noted that, the smaller the internal quality score NP of the weld, the better the internal quality of the weld; conversely, the larger the internal quality score NP of the weld, the worse the internal quality of the weld.
[0023] The present invention collects the detection current and detection voltage of the detection area in the weld to be tested, calculates the detection resistance according to the detection current and the detection voltage, marks the detection area as a normal area, a slightly abnormal area and a seriously abnormal area based on the detection resistance and the resistance range threshold, calculates the proportion of the slightly abnormal area and the seriously abnormal area in all the detection areas, and calculates the internal quality score of the weld according to the corresponding proportion, which can accurately reflect the internal quality of the weld and is beneficial to improving the accuracy of detecting the quality of the weld.
[0024] Preferably, the extracting of surface defect features in the original weld image based on an image recognition algorithm includes:
[0025] X1: preprocess the original weld image to obtain a standard weld image;
[0026] X2: Get the gray value of all pixels in the standard weld image;
[0027] X3: Compare the gray value of each pixel with the preset gray threshold to obtain the defect area image;
[0028] X4: Identify the corresponding surface defect type from the defect area image based on the image recognition algorithm;
[0029] X5: Calculate the pixel ratio of the defect area image in the standard weld image.
[0030] Preferably, the preprocessing of the original weld image includes:
[0031] Y1: extract the original weld image;
[0032] Y2: The original weld image is subjected to denoising processing by an image denoising algorithm to obtain weld image 1;
[0033] Y3: Improve the contrast of weld image 1 to obtain weld image 2;
[0034] Y4: Perform grayscale processing on weld image 2 to obtain a standard weld image.
[0035] The present invention obtains weld image one by performing noise reduction processing on the original weld image, which can reduce the influence of image noise on the subsequent extraction of weld surface defect features; improving the contrast of weld image one enhances the texture details of the weld surface defect features, which is beneficial to improving the accuracy of subsequent extraction of weld surface defect features; performing grayscale processing on weld image two can reduce the amount of calculation, which is beneficial to improving the efficiency of subsequent extraction of weld surface defect features.
[0036] Preferably, comparing the grayscale value of each pixel with a preset grayscale threshold comprises:
[0037] Z1: Determine whether the grayscale value of each pixel is greater than a preset grayscale threshold; if yes, mark the corresponding pixel as a defective pixel; if no, mark the corresponding pixel as a normal pixel;
[0038] Z2: Mark the area surrounded by all normal pixels as the normal area image;
[0039] Z3: Mark the area surrounded by all defective pixels as the defective area image.
[0040] Preferably, the calculating the surface quality score of the weld to be tested based on the surface defect characteristics comprises:
[0041] B1: Extract the surface defect types and corresponding pixel proportions in the surface defect features of the weld;
[0042] B2: Mark the surface defect type of the weld as k; where k = 1, 2, ..., m, and m is the number of surface defect types;
[0043] B3: By formula Calculate the surface quality score BP of the weld; where BFk is the score coefficient corresponding to the kth surface defect type, and XBk is the sum of the pixel proportions corresponding to the kth surface defect type.
[0044] It should be noted that the smaller the surface quality score BP of the weld, the better the surface quality of the weld; conversely, the larger the surface quality score BP of the weld, the worse the surface quality of the weld.
[0045] The present invention sets corresponding scoring coefficients for surface defect types of welds, multiplies the sum of the pixel proportions of each surface defect type by the corresponding scoring coefficient to obtain a score for each surface defect type, and sums the scores of all surface defect types to obtain a surface quality score for the weld. This can accurately reflect the surface quality of the weld, thereby facilitating improving the accuracy of subsequent quality inspections of the weld.
[0046] Preferably, the quality of the weld to be tested is evaluated based on the surface quality score and the internal quality score of the weld to be tested, including:
[0047] T1: Extract the surface quality score and internal quality score of the weld;
[0048] T2: Determine whether the surface quality score of the weld is less than the set surface quality score threshold; if yes, mark the surface quality test result as qualified; if not, mark the surface quality test result as unqualified.
[0049] T3: Determine whether the internal quality score of the weld is less than the set internal quality score threshold; if yes, mark the internal quality test result as qualified; if not, mark the internal quality test result as unqualified.
[0050] Preferably, analyzing the cause of the weld defect according to the weld inspection result and the welding parameters includes:
[0051] The weld inspection results and welding parameters are extracted and integrated into standard input data, which are then input into the weld defect analysis model to obtain the causes of weld defects. The weld defect analysis model is trained based on an artificial intelligence model, and the causes of weld defects include abnormal welding current, abnormal welding voltage, abnormal welding speed and abnormal welding angle.
[0052] Preferably, the weld defect analysis model is constructed based on an artificial intelligence model, including:
[0053] The weld inspection results and welding parameters are obtained through the database; the weld inspection results and welding parameters are integrated into several groups of training data and several groups of inspection data; the artificial intelligence model is trained using the training data; the trained artificial intelligence model is inspected using the inspection data, and the artificial intelligence model is adjusted according to the inspection results; finally, the input is the weld inspection results and welding parameters, and the output is the cause of the weld defect; wherein the artificial intelligence model includes a BP neural network model or a RBF neural network model.
[0054] It should be noted that during the construction of the artificial intelligence model, both its input data and output data need to be adaptively processed, such as normalized, to ensure the parameter requirements of the artificial intelligence model. In the present invention, a weld defect label can be output, and the corresponding weld defect cause can be matched according to the weld defect label.
[0055] Compared with the prior art, the present invention has the following beneficial effects:
[0056] 1. The present invention collects the detection current and detection voltage of the detection area in the weld to be tested, calculates the detection resistance according to the detection current and the detection voltage, marks the detection area as a normal area, a slightly abnormal area and a seriously abnormal area based on the detection resistance and the resistance range threshold, calculates the proportion of the slightly abnormal area and the seriously abnormal area in all the detection areas, and calculates the internal quality score of the weld according to the corresponding proportion, which can accurately reflect the internal quality of the weld and is conducive to improving the accuracy of detecting the quality of the weld.
[0057] 2. The present invention sets corresponding scoring coefficients for surface defect types of welds, multiplies the sum of the pixel proportions of each surface defect type by the corresponding scoring coefficient to obtain a score for each surface defect type, and sums the scores of all surface defect types to obtain a surface quality score for the weld, which can accurately reflect the surface quality of the weld, thereby helping to improve the accuracy of weld quality detection. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 paying creative work.
[0059] Figure 1 It is a schematic diagram of the principle of the weld quality detection system of the present invention;
[0060] Figure 2 It is a flow chart of the weld quality detection system of the present invention;
[0061] Figure 3 A flow chart for calculating the internal quality score of a weld in the present invention;
[0062] Figure 4 The figure is a flow chart for detecting weld quality in the present invention. DETAILED DESCRIPTION
[0063] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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 are within the scope of protection of the present invention.
[0064] See also Figure 1-Figure 4, the first aspect of the present invention provides a welding seam quality detection system based on the combination of visual recognition and current and voltage, including: a data processing module, and a data acquisition module and a quality detection module connected thereto;
[0065] The data acquisition module is used to acquire the original weld image and welding parameters of the weld to be tested, and to acquire the electrical parameters of several detection areas in the weld to be tested; wherein the welding parameters include the welding current, welding voltage, welding speed and welding angle during the welding process, and the electrical parameters include the detection current and detection voltage;
[0066] The data processing module is used to calculate the internal quality score of the weld to be tested based on the electrical parameters, and to extract the surface defect features in the original weld image based on the image recognition algorithm, and to calculate the surface quality score of the weld to be tested based on the surface defect features; wherein the surface defect features include the surface defect type and the corresponding pixel ratio, and the surface defect type includes burn-through, incomplete welding, undercut, surface pores, surface cracks and weld nodules;
[0067] The quality inspection module: evaluates the quality of the weld to be tested based on the surface quality score and the internal quality score of the weld to be tested, and obtains the weld inspection result; wherein the weld inspection result includes the surface quality inspection result and the internal quality inspection result; and,
[0068] The causes of weld defects are analyzed according to the weld detection results and the welding parameters; wherein the causes of weld defects include abnormal welding current, abnormal welding voltage, abnormal welding speed and abnormal welding angle.
[0069] In this embodiment, the division of several detection areas includes:
[0070] The weld to be tested is divided into several detection areas with equal distances along the weld direction, and the detection areas are marked as i; wherein i=1, 2, ..., n, and n is the total number of detection areas.
[0071] Exemplarily, the weld to be tested is divided into five detection areas at equal distances along the weld direction, and the detection areas are marked as i; wherein i=1, 2, ..., n, n=5.
[0072] In this embodiment, the internal quality score of the weld to be tested is calculated based on the electrical parameters, including:
[0073] A1: Extract the detection current I of the detection area i i and the detection voltage V i ;
[0074] A2: By formula R i =V i / I i Calculate the sense resistor Ri ;
[0075] A3: Set the resistance range threshold [DY1, DY2], and DY1 <DY2;
[0076] A4: Determine the detection resistance R i Is it less than the minimum value DY1 of the resistance range threshold? If yes, mark the detection area i as a normal area; if no, jump to A5;
[0077] A5: Determine the detection resistance R i Whether it is within the resistance range threshold [DY1, DY2]; if yes, the detection area i is marked as a slightly abnormal area; if no, the detection area i is marked as a severely abnormal area;
[0078] A6: Calculate the proportion of slightly abnormal areas ZB1 by the formula ZB1=q / n; where q is the number of slightly abnormal areas;
[0079] A7: Calculate the proportion of severe abnormal areas ZB2 by the formula ZB2=y / n; where y is the number of severe abnormal areas;
[0080] A8: The internal quality score NP of the weld is calculated by the formula NP = α × ZB1 + β × ZB2; where α and β are proportional coefficients greater than 0.
[0081] It should be noted that, the smaller the internal quality score NP of the weld, the better the internal quality of the weld; conversely, the larger the internal quality score NP of the weld, the worse the internal quality of the weld.
[0082] For example, α=50, β=80 are set;
[0083] Set the detection current of detection area 1 to I 1 =4.12A, detection voltage V 1 =22.36V, the detection resistance R is calculated by the formula 1 ≈5.43Ω;
[0084] Set the detection current of detection area 2 to I 2 =4.21A, detection voltage V 2 =22.54V, the detection resistance R is calculated by the formula 2 ≈5.35Ω;
[0085] Set the detection current of detection area 3 to I 3 =4.21A, detection voltage V 3 =34.72V, the detection resistance R is calculated by the formula 3 ≈8.25Ω;
[0086] Set the detection current of detection area 4 to I 4 =4.24A, detection voltage V 4 =53.67V, the detection resistance R is calculated by the formula 4 ≈12.66Ω;
[0087] Set the detection current of detection area 5 to I 5 =4.19A, detection voltage V 5 =33.82V, the detection resistance R is calculated by the formula 5 ≈8.07Ω;
[0088] Set DY1 = 6, DY2 = 9, then the resistance range threshold [DY1, DY2] = [6, 9]; due to the detection resistor R 1 and the detection resistor R 2 The minimum value DY1=6 is smaller than the resistance range threshold, so detection area 1 and detection area 2 are marked as normal areas; 3 and the detection resistor R 5 is within the resistance range threshold [6, 9], so detection area 3 and detection area 5 are marked as slightly abnormal areas; 4 The maximum value DY2=9, which is greater than the resistance range threshold, marks the detection area 4 as a serious abnormal area;
[0089] The proportion of mildly abnormal areas is calculated by the formula ZB1 = 0.4;
[0090] The proportion of severely abnormal areas is calculated by the formula ZB2 = 0.2;
[0091] The internal quality score NP=36 was calculated by the formula.
[0092] The present invention collects the detection current and detection voltage of the detection area in the weld to be tested, calculates the detection resistance according to the detection current and the detection voltage, marks the detection area as a normal area, a slightly abnormal area and a seriously abnormal area based on the detection resistance and the resistance range threshold, calculates the proportion of the slightly abnormal area and the seriously abnormal area in all the detection areas, and calculates the internal quality score of the weld according to the corresponding proportion, which can accurately reflect the internal quality of the weld and is beneficial to improving the accuracy of detecting the quality of the weld.
[0093] In this embodiment, the surface defect features in the original weld image are extracted based on the image recognition algorithm, including:
[0094] X1: preprocess the original weld image to obtain a standard weld image;
[0095] X2: Get the gray value of all pixels in the standard weld image;
[0096] X3: Compare the gray value of each pixel with the preset gray threshold to obtain the defect area image;
[0097] X4: Identify the corresponding surface defect type from the defect area image based on the image recognition algorithm;
[0098] X5: Calculate the pixel ratio of the defect area image in the standard weld image.
[0099] In this embodiment, the original weld image is preprocessed, including:
[0100] Y1: extract the original weld image;
[0101] Y2: The original weld image is subjected to denoising processing by an image denoising algorithm to obtain weld image 1;
[0102] Y3: Improve the contrast of weld image 1 to obtain weld image 2;
[0103] Y4: Perform grayscale processing on weld image 2 to obtain a standard weld image.
[0104] The present invention obtains weld image one by performing noise reduction processing on the original weld image, which can reduce the influence of image noise on the subsequent extraction of weld surface defect features; improving the contrast of weld image one enhances the texture details of the weld surface defect features, which is beneficial to improving the accuracy of subsequent extraction of weld surface defect features; performing grayscale processing on weld image two can reduce the amount of calculation, which is beneficial to improving the efficiency of subsequent extraction of weld surface defect features.
[0105] In this embodiment, the grayscale value of each pixel is compared with a preset grayscale threshold, including:
[0106] Z1: Determine whether the grayscale value of each pixel is greater than a preset grayscale threshold; if yes, mark the corresponding pixel as a defective pixel; if no, mark the corresponding pixel as a normal pixel;
[0107] Z2: Mark the area surrounded by all normal pixels as the normal area image;
[0108] Z3: Mark the area surrounded by all defective pixels as the defective area image.
[0109] In this embodiment, the surface quality score of the weld to be tested is calculated based on the surface defect characteristics, including:
[0110] B1: Extract the surface defect types and corresponding pixel proportions in the surface defect features of the weld;
[0111] B2: Mark the surface defect type of the weld as k; where k = 1, 2, ..., m, and m is the number of surface defect types;
[0112] B3: By formula Calculate the surface quality score BP of the weld; where BFk is the score coefficient corresponding to the kth surface defect type, and XBk is the sum of the pixel proportions corresponding to the kth surface defect type.
[0113] It should be noted that the smaller the surface quality score BP of the weld, the better the surface quality of the weld; conversely, the larger the surface quality score BP of the weld, the worse the surface quality of the weld.
[0114] Exemplarily, the surface defect types of burn-through, incomplete welding, undercut, surface pores, surface cracks and weld nodules are marked as 1, 2, 3, 4 and 5, respectively, and the corresponding pixel proportions are set to XB1=1.26%, XB2=2.61%, XB3=0.26%, XB4=3.74%, XB5=2.39%, XB6=5.21%, respectively, and the corresponding scoring coefficients are set to BF1=1000, BF2=800, BF3=700, BF4=500, BF5=900, BF6=600, respectively. The weld surface quality score BP=106.77 is calculated by the formula.
[0115] The present invention sets corresponding scoring coefficients for surface defect types of welds, multiplies the sum of the pixel proportions of each surface defect type by the corresponding scoring coefficient to obtain a score for a single surface defect type, and sums the scores of all surface defect types to obtain a surface quality score for the weld. This can accurately reflect the surface quality of the weld, thereby facilitating improving the accuracy of subsequent quality inspections of the weld.
[0116] In this embodiment, the quality of the weld to be tested is evaluated based on the surface quality score and the internal quality score of the weld to be tested, including:
[0117] T1: Extract the surface quality score and internal quality score of the weld;
[0118] T2: Determine whether the surface quality score of the weld is less than the set surface quality score threshold; if yes, mark the surface quality test result as qualified; if not, mark the surface quality test result as unqualified.
[0119] T3: Determine whether the internal quality score of the weld is less than the set internal quality score threshold; if yes, mark the internal quality test result as qualified; if not, mark the internal quality test result as unqualified.
[0120] Exemplarily, the surface quality score threshold is set to 50 and the internal quality score threshold is set to 20; since the surface quality score BP=106.77 of the weld is greater than the surface quality score threshold 50, the surface quality inspection result is marked as unqualified; since the internal quality score NP=36 of the weld is greater than the internal quality score threshold 20, the internal quality inspection result is marked as unqualified.
[0121] In this embodiment, analyzing the cause of the weld defect according to the weld inspection result and the welding parameters includes:
[0122] The weld inspection results and welding parameters are extracted and integrated into standard input data, which are then input into the weld defect analysis model to obtain the causes of weld defects. The weld defect analysis model is trained based on an artificial intelligence model, and the causes of weld defects include abnormal welding current, abnormal welding voltage, abnormal welding speed and abnormal welding angle.
[0123] In this embodiment, the weld defect analysis model is constructed based on an artificial intelligence model, including:
[0124] The weld inspection results and welding parameters are obtained through the database; the weld inspection results and welding parameters are integrated into several groups of original data, 80% of the original data are used as training data, and 20% are used as test data; the artificial intelligence model is trained using the training data; the trained artificial intelligence model is tested using the test data, and the artificial intelligence model is adjusted according to the test results; finally, the input is the weld inspection results and welding parameters, and the output is the cause of the weld defect; wherein the artificial intelligence model includes a BP neural network model or a RBF neural network model.
[0125] It should be noted that during the construction of the artificial intelligence model, both its input data and output data need to be adaptively processed, such as normalized, to ensure the parameter requirements of the artificial intelligence model. In the present invention, a weld defect label can be output, and the corresponding weld defect cause can be matched according to the weld defect label.
[0126] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.
[0127] Working principle of the present invention:
[0128] The present invention collects the original weld image and welding parameters of the weld to be tested, and collects the electrical parameters of several detection areas in the weld to be tested; calculates the internal quality score of the weld to be tested based on the electrical parameters, and extracts the surface defect features in the original weld image based on an image recognition algorithm, and calculates the surface quality score of the weld to be tested based on the surface defect features; evaluates the quality of the weld to be tested based on the surface quality score and the internal quality score of the weld to be tested to obtain a weld detection result; and analyzes the cause of the weld defect according to the weld detection result and the welding parameters.
[0129] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. A weld quality detection system based on visual recognition combined with current and voltage, comprising: The data processing module, and the data acquisition module and quality detection module connected thereto are characterized in that: The data acquisition module is used to acquire the original weld image and welding parameters of the weld to be tested, and to acquire the electrical parameters of several detection areas in the weld to be tested; wherein the welding parameters include the welding current, welding voltage, welding speed and welding angle during the welding process, and the electrical parameters include the detection current and detection voltage; The data processing module is used to calculate the internal quality score of the weld to be tested based on the electrical parameters, and to extract the surface defect features in the original weld image based on the image recognition algorithm, and to calculate the surface quality score of the weld to be tested based on the surface defect features; wherein the surface defect features include the surface defect type and the corresponding pixel ratio, and the surface defect type includes burn-through, incomplete welding, undercut, surface pores, surface cracks and weld nodules; The quality inspection module: evaluates the quality of the weld to be tested based on the surface quality score and the internal quality score of the weld to be tested, and obtains the weld inspection result; wherein the weld inspection result includes the surface quality inspection result and the internal quality inspection result; and, The causes of weld defects are analyzed according to the weld detection results and the welding parameters; wherein the causes of weld defects include abnormal welding current, abnormal welding voltage, abnormal welding speed and abnormal welding angle.
2. A weld quality detection system based on visual recognition combined with current and voltage according to claim 1, characterized in that: The division of several detection areas includes: The weld to be tested is divided into several detection areas with equal distances along the weld direction, and the detection areas are marked as i; wherein i=1, 2, ..., n, and n is the total number of detection areas.
3. A weld quality detection system based on visual recognition combined with current and voltage according to claim 2, characterized in that: The step of calculating the internal quality score of the weld to be tested based on the electrical parameters comprises: A1: Extract the detection current I of the detection area i i and the detection voltage V i ; A2: By formula R i =V i / I i Calculate the sense resistor R i ; A3: Set the resistance range threshold [DY1, DY2], and DY1 <DY2; A4: Determine the detection resistance R i Is it less than the minimum value DY1 of the resistance range threshold? If yes, mark the detection area i as a normal area; if no, jump to A5; A5: Determine the detection resistance R i Whether it is within the resistance range threshold; if yes, the detection area i is marked as a slightly abnormal area; if no, the detection area i is marked as a seriously abnormal area; A6: Calculate the proportion of slightly abnormal areas ZB1 by the formula ZB1=q / n; where q is the number of slightly abnormal areas; A7: Calculate the proportion of severe abnormal areas ZB2 by the formula ZB2=y / n; where y is the number of severe abnormal areas; A8: The internal quality score NP of the weld is calculated by the formula NP=α×ZB1+β×ZB2; where α and β are proportional coefficients greater than 0.
4. A weld quality detection system based on visual recognition combined with current and voltage according to claim 1, characterized in that: The method of extracting surface defect features from the original weld image based on an image recognition algorithm includes: X1: preprocess the original weld image to obtain a standard weld image; X2: Get the gray value of all pixels in the standard weld image; X3: Compare the gray value of each pixel with the preset gray threshold to obtain the defect area image; X4: Identify the corresponding surface defect type from the defect area image based on the image recognition algorithm; X5: Calculate the pixel ratio of the defect area image in the standard weld image.
5. A weld quality detection system based on visual recognition combined with current and voltage according to claim 4, characterized in that: The preprocessing of the original weld image comprises: Y1: extract the original weld image; Y2: The original weld image is subjected to denoising processing by an image denoising algorithm to obtain weld image 1; Y3: Improve the contrast of weld image 1 to obtain weld image 2; Y4: Perform grayscale processing on weld image 2 to obtain a standard weld image.
6. A weld quality detection system based on visual recognition combined with current and voltage according to claim 4, characterized in that: The step of comparing the grayscale value of each pixel with a preset grayscale threshold value includes: Z1: Determine whether the grayscale value of each pixel is greater than a preset grayscale threshold; if yes, mark the corresponding pixel as a defective pixel; if no, mark the corresponding pixel as a normal pixel; Z2: Mark the area surrounded by all normal pixels as the normal area image; Z3: Mark the area surrounded by all defective pixels as the defective area image.
7. The weld quality detection system based on visual recognition combined with current and voltage according to claim 1 is characterized in that: The calculating the surface quality score of the weld to be tested based on the surface defect characteristics comprises: B1: Extract the surface defect types and corresponding pixel proportions in the surface defect features of the weld; B2: Mark the surface defect type of the weld as k; where k = 1, 2, ..., m, and m is the number of surface defect types; B3: By formula Calculate the surface quality score BP of the weld; where BFk is the score coefficient corresponding to the kth surface defect type, and XBk is the sum of the pixel proportions corresponding to the kth surface defect type.
8. The weld quality detection system based on visual recognition combined with current and voltage according to claim 1 is characterized in that: The quality of the weld to be tested is evaluated based on the surface quality score and the internal quality score of the weld to be tested, including: T1: Extract the surface quality score and internal quality score of the weld; T2: Determine whether the surface quality score of the weld is less than the set surface quality score threshold; if yes, mark the surface quality test result as qualified; if no, mark the surface quality test result as unqualified; T3: Determine whether the internal quality score of the weld is less than the set internal quality score threshold; if yes, mark the internal quality test result as qualified; if not, mark the internal quality test result as unqualified.
9. The weld quality detection system based on visual recognition combined with current and voltage according to claim 1 is characterized in that: Analyzing the cause of the weld defect according to the weld inspection result and the welding parameters includes: The weld inspection results and welding parameters are extracted and integrated into standard input data, and the standard input data is input into the weld defect analysis model to obtain the cause of the weld defect; wherein, the weld defect analysis model is obtained based on artificial intelligence model training.
10. A weld quality detection system based on visual recognition combined with current and voltage according to claim 9, characterized in that: The weld defect analysis model is constructed based on an artificial intelligence model and includes: The weld inspection results and welding parameters are obtained through the database; the weld inspection results and welding parameters are integrated into several groups of training data and several groups of inspection data; the artificial intelligence model is trained using the training data; the trained artificial intelligence model is inspected using the inspection data, and the artificial intelligence model is adjusted according to the inspection results; finally, the input is the weld inspection results and welding parameters, and the output is the cause of the weld defect; wherein the artificial intelligence model includes a BP neural network model or a RBF neural network model.
Citation Information
Cited By
Welding quality control system and method for rear axle of electro-tricycle
CN119515192A
Weld defect detection system and method based on target recognition
CN119672017A
