A method for detecting weld defects

Abstract

This invention relates to the field of weak magnetic field detection technology and proposes a method for detecting weld defects. The method includes: arranging six weak magnetic field sensors in an L-shape, stacked in pairs. The second and fifth weak magnetic field sensors, forming the first stack, are located directly above the weld. The first and fourth weak magnetic field sensors, forming the second stack, are located directly behind the first stack. The third and sixth weak magnetic field sensors, forming the third stack, are arranged side-by-side and aligned on one side of the first stack. Specifically, the difference between two magnetic field induction intensity signal values ​​within the same stack is calculated to obtain magnetic field induction intensity data after removing the background field. Based on the characteristics of the weld and the principle of weak magnetic field detection, at least two conditions for judging defects are set. Using the principle of redundancy detection, the magnetic field induction intensity data after removing the background field is substituted into the conditions for judging defects to identify defect features. Based on probabilistic statistical principles, the correct probability of defect judgment is calculated, which can improve the accuracy of weld defect judgment results.

Description

Technical Field

[0001] This invention relates to the field of weak magnetic field detection technology, and in particular to a method for detecting weld defects. Background Technology

[0002] Non-destructive testing is an indispensable technical means in modern industry, and its importance is self-evident.

[0003] As an emerging passive non-destructive testing method, weak magnetic field detection technology has the following advantages: (1) No additional excitation source required: It works in the geomagnetic field environment and does not require a special magnetic excitation device; (2) Non-contact detection: It can detect coated parts without special treatment of the surface of the workpiece being inspected; (3) No special shape requirements: It has no special requirements for the shape of the workpiece being inspected and is highly adaptable; (4) Fast detection speed and simple operation: It uses a high-precision fluxgate sensor for scanning and data processing is rapid.

[0004] Its working principle is based on the stress and atomic structure changes caused by material defects, which lead to the spontaneous generation of abnormal magnetic fields. This abnormal signal is detected by a high-precision magnetic sensor, thus characterizing the material defects. Assuming the permeability of the bulk material is μ, and the permeability of the discontinuous region within the workpiece is μ', if the relative permeability of the discontinuous region is greater than that of the bulk material (μ' > μ), then the magnetic flux density curve will show an upward convexity when the magnetic sensor passes through this region. Conversely, if the relative permeability of the discontinuous region is less than that of the bulk material (μ' < μ), then the magnetic flux density curve will show a downward concaveness when the magnetic sensor passes through this region. Weak magnetic field detection technology utilizes a high-precision magnetic sensor to detect this abrupt change in magnetic flux density, thereby characterizing material defects.

[0005] Currently, weld defect detection based on the principle of weak magnetic field mainly relies on sensors to directly scan and detect the weld surface. It is difficult to suppress external interference, resulting in large quantitative errors in defect detection. Furthermore, the limited data also leads to a single defect judgment condition, which in turn results in errors in the final defect judgment result. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method for detecting weld defects, which can improve the accuracy of weld defect judgment results.

[0007] This invention provides a method for detecting weld defects, comprising: stacking six weak magnetic sensors in pairs in an L-shape, wherein the second and fifth weak magnetic sensors in the first stack are located directly above the weld, the first and fourth weak magnetic sensors in the second stack are located directly behind the first stack, and the third and sixth weak magnetic sensors in the third stack are arranged side by side and aligned on one side of the first stack, thereby acquiring multiple magnetic induction intensity signal values ​​of the weld and the substrate material; preprocessing the magnetic induction intensity signal values, i.e., calculating the difference between two magnetic induction intensity signal values ​​in the same group to obtain magnetic induction intensity data after removing the background field; setting at least two conditions for judging defects based on the characteristics of the weld and the principle of weak magnetic detection; using the principle of redundancy detection, substituting the magnetic induction intensity data after removing the background field into the conditions for judging defects to identify defect features; and calculating the correct probability of defect judgment based on probability statistics.

[0008] Optionally, the conditions for determining a defect include: condition 1, condition 2, and condition 3; among which,

[0009] Condition 1 includes: determining whether there is an abnormality in the difference between the magnetic induction intensity signal value of the second weak magnetic sensor after removing the background field and the magnetic induction intensity signal value of the first weak magnetic sensor after removing the background field;

[0010] Condition 2 includes: determining whether there is an abnormality in the difference between the magnetic induction intensity signal value of the second weak magnetic sensor after removing the background field and the magnetic induction intensity signal value of the third weak magnetic sensor after removing the background field;

[0011] Condition 3 includes: determining whether there is an anomaly in the magnetic induction intensity signal value of the second weak magnetic sensor after removing the background field.

[0012] Optionally, the preprocessing includes: connecting the magnetic induction intensity signal values ​​detected by the same weak magnetic sensor to form a line, obtaining six corresponding real-time curves, and performing differential calculation on the real-time curves corresponding to two weak magnetic sensors stacked together to obtain three curves with background field removed. The curve with background field removed corresponding to the weak magnetic sensors stacked together in the first group is denoted as Rebf1, the curve with background field removed corresponding to the weak magnetic sensors stacked together in the second group is denoted as Rebf2, and the curve with background field removed corresponding to the weak magnetic sensors stacked together in the third group is denoted as Rebf3.

[0013] Optionally, based on the redundancy detection principle, the three curves after removing the background field are differentially processed to generate two judgment curves. Specifically, the difference between Rebf1 and Rebf2 is used to obtain the judgment curve Decide1, and the difference between Rebf2 and Rebf3 is used to obtain the judgment curve Decide2. The gradient curve of the second weak magnetic sensor 2 is denoted as Decide3. The point values ​​in the judgment curves Decide1, Decide2, and Decide3 are the array of judgment defect data.

[0014] Optionally, it is necessary to combine two or more conditions for judging defects and calculate the probability of it being a defect according to the formula before it can be determined whether a defect exists at a certain location. When a certain condition for judging defects shows that there is no defect at this location or the calculated result is that the probability of a defect at this location is small, then it is determined that there is no defect at this location. The method for judging defects is as follows: a) If there are two conditions, then both conditions must be true to judge that there is a defect at this location; b) If there are three conditions, under the premise that condition 1 judges that there is a defect, if one of condition 2 or condition 3 judges that there is a defect at this location, then it can be judged that there is a defect at this location.

[0015] Optionally, the probability of a defect can be calculated using the following formula:

[0016] Probability=A1*0.6+A2*0.2+A3*0.2

[0017] In this context, A1 corresponds to condition 1, A2 corresponds to condition 2, and A3 corresponds to condition 3. The values ​​of A1, A2, and A3 are either 0 or 1. 0 represents no anomaly, and 1 represents an anomaly. When Probability ≥ 0.8, it can be determined that there is a defect here.

[0018] Optionally, the method for judging anomalies between magnetic induction intensity signal values ​​is the 3σ method, where the methods for judging signal anomalies in conditions 1 and 2 are the same. The 3σ method includes:

[0019] First, extract the magnetic field gradient values ​​of the two judgment curves. The magnetic field gradient value is the first derivative of the value at each corresponding point on each of the two judgment curves. Store the obtained magnetic field gradient values ​​in the array Decide-f[i,j], as shown in the formula below:

[0020] Decide-f[i,j]=Decide[i,j+1]-Decide[i,j]

[0021] Where Decide-f[i,j] is the weld magnetic induction intensity gradient data array, i is the number of channels, j is the data sequence number, where the number of channels corresponding to the first weak magnetic sensor is channel 0, the data sequence number corresponding to the first weak magnetic sensor is 1, and so on.

[0022] After obtaining the magnetic field gradient curve, the threshold range of each judgment curve is further obtained. in, σ is the average value of the obtained magnetic field gradient, and σ is the standard deviation of the obtained magnetic field gradient.

[0023] Optional, The calculation methods for σ are as follows:

[0024]

[0025] Where n is the total number of data points on the judgment curve, and ΔB(i) is the magnetic field gradient value of the i-th data point on the judgment curve. When K is 3, the magnetic field gradient value is calculated to obtain the threshold line of the interval with a confidence probability of 99.73%. The magnetic field gradient value corresponding to the judgment curve is compared with the threshold line. The sampling point location corresponding to the magnetic field gradient value that exceeds the threshold range is determined to be an abnormal area.

[0026] The technical solution provided by the embodiments of the present invention has the following advantages compared with the prior art:

[0027] The weld defect detection method provided in this invention arranges six weak magnetic sensors in two L-shaped stacks to perform precise weld detection. Specifically, the second and fifth weak magnetic sensors in the first group are located directly above the weld to detect the magnetic induction intensity signal value directly above the weld. The first and fourth weak magnetic sensors in the second group are located directly behind the first group and also detect the magnetic induction intensity signal value directly above the weld. The third and sixth weak magnetic sensors in the third group are arranged side-by-side and aligned on one side of the first group to detect the magnetic induction intensity signal value of the substrate material. To effectively reduce detection errors, the difference between two magnetic induction intensity signal values ​​from the same group is calculated. Then, based on the characteristics of the weld and the principle of weak magnetic detection, at least two conditions for judging defects are set. Combining the principle of redundant detection, the magnetic induction intensity data after removing the background field is substituted into the conditions for judging defects to identify defects. Based on the principle of probability statistics, the probability of correct defect judgment is calculated. An array-type weak magnetic sensor is used and its layout is optimized. Data differential technology is used to suppress external interference. Multiple conditions for judging defects can be set through multiple sets of data, thereby improving the reliability of the detection results and the accuracy of weld defect judgment results. Attached Figure Description

[0028] Figure 1 This is a structural schematic diagram showing the relative positional relationship of multiple weak magnetic sensors provided in an embodiment of the present invention;

[0029] Figure 2 A flowchart provided for an embodiment of the present invention;

[0030] Figure 3 The original curve simulation diagram provided for the embodiments of the present invention;

[0031] Figure 4 This is a simulation diagram of the background removal field curve provided in an embodiment of the present invention;

[0032] Figure 5 The simulation diagram of the judgment curve provided in the embodiment of the present invention;

[0033] Figure 6 This is a schematic diagram of a weak magnetic field detection structure provided in an embodiment of the present invention;

[0034] Figure 7 This is a conceptual diagram of a weak magnetic field detection fixture provided in an embodiment of the present invention;

[0035] Figure 8 This is a cross-sectional view of the mounting slot for the weak magnetic field sensor provided in an embodiment of the present invention;

[0036] Figure 9 This is a schematic diagram of the probe detection movement path provided in an embodiment of the present invention;

[0037] Figure 10 The original curve for scanning weld seams provided in this embodiment of the invention;

[0038] Figure 11 Background removal field curve provided for embodiments of the present invention;

[0039] Figure 12 The defect judgment curve provided in the embodiments of the present invention;

[0040] Figure 13 Judgment condition 1 provided for embodiments of the present invention;

[0041] Figure 14 Judgment condition 2 provided for embodiments of the present invention;

[0042] Figure 15 Judgment condition 3 provided for embodiments of the present invention;

[0043] Figure 16 This is a defect judgment diagram provided for an embodiment of the present invention.

[0044] Explanation of reference numerals in the attached figures:

[0045] 1. First magnetic field weakening sensor; 2. Second magnetic field weakening sensor; 3. Third magnetic field weakening sensor; 4. Fourth magnetic field weakening sensor; 5. Fifth magnetic field weakening sensor; 6. Sixth magnetic field weakening sensor; 7. Workpiece; 8. Weld seam; 9. Magnetic field weakening probe; 10. Sensor mounting slot; 11. Adapter device; 12. Cable bundle; 13. Aviation connector. Detailed Implementation

[0046] A specific embodiment of the present invention is described in detail below with reference to the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific embodiment.

[0047] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the technical solution of this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0048] Non-destructive testing (NDT) is an indispensable technology in modern industry, and its importance is self-evident. The following is a detailed introduction to current mainstream NDT methods and their limitations, as well as weak magnetic field testing technology:

[0049] Radiographic Testing (RT): Widely used for complex structural components, but unsuitable for porous materials and unable to detect subsurface defects. Furthermore, RT has a long turnaround time, high cost, and requires radiation protection for personnel. Ultrasonic Testing (UT): Widely used for near-surface defect detection, but requires specific material roughness and shape, and the material must be suitable for ultrasonic wave propagation. Magnetic Particle Testing (MT): Suitable for layered, near-surface, and surface defects, but has limited ability to detect subsurface defects in non-ferromagnetic materials. Penetrant Testing (PT): Only suitable for surface defect detection, unsuitable for porous materials, and improper operation can easily cause environmental pollution. Eddy Current Testing (ECT): A detection method based on the eddy current principle, requiring an excitation coil to generate magnetic field lines, but suffers from induced magnetic field interference and a trade-off between the effective detection area and efficiency.

[0050] As an emerging passive non-destructive testing method, weak magnetic field testing technology mainly uses sensors to directly scan and detect weld surfaces. It is difficult to suppress external interference, resulting in large quantitative errors in defect detection. Furthermore, the limited data also leads to a single defect judgment condition, which in turn results in errors in the final defect judgment.

[0051] Therefore, embodiments of the present invention provide a method for detecting weld defects, which can improve the accuracy of weld defect judgment results.

[0052] At least one embodiment of the present invention provides a method for detecting weld defects, comprising: stacking six weak magnetic sensors in pairs in an L-shape, wherein the second and fifth weak magnetic sensors, which form a first stack, are located directly above the weld; the first and fourth weak magnetic sensors, which form a second stack, are located directly behind the first stack; and the third and sixth weak magnetic sensors, which form a third stack, are arranged side-by-side and aligned on one side of the first stack, thereby acquiring multiple magnetic induction intensity signal values ​​of the weld and the substrate material; preprocessing the magnetic induction intensity signal values, i.e., calculating the difference between two magnetic induction intensity signal values ​​in the same group to obtain magnetic induction intensity data after removing the background field; setting at least two conditions for judging defects based on the characteristics of the weld and the principle of weak magnetic detection; using the principle of redundancy detection, substituting the magnetic induction intensity data after removing the background field into the conditions for judging defects to identify defect features; and calculating the correct probability of defect judgment based on probability statistics.

[0053] In the weld defect detection method provided in the above embodiments of the present invention, the reliability of the detection results is improved by arranging multiple sets of weak magnetic sensors in L-shaped positions, combining the principle of redundancy detection, and fusing multiple judgment methods.

[0054] The present invention will be described below through several specific embodiments. To keep the following description of the embodiments clear and concise, detailed descriptions of known functions and components may be omitted. When any component of an embodiment of the present invention appears in more than one drawing, the component may be represented by the same reference numerals in each drawing.

[0055] refer to Figure 1 , Figure 7 and Figure 8 , Figure 1 This is a structural schematic diagram showing the relative positional relationship of multiple weak magnetic sensors provided in an embodiment of the present invention. Figure 7 This is a conceptual diagram of a weak magnetic field detection fixture provided in an embodiment of the present invention. Figure 8 This is a cross-sectional view of the mounting slot for the weak magnetic sensor provided in an embodiment of the present invention, as shown below. Figure 1 , Figure 7 and Figure 8As shown, this embodiment of the invention provides a weld defect detection method, comprising: stacking six weak magnetic sensors in pairs in an L-shape, wherein the second and fifth weak magnetic sensors, forming the first stack, are located directly above the weld; the first and fourth weak magnetic sensors, forming the second stack, are located directly behind the first stack; and the third and sixth weak magnetic sensors, forming the third stack, are arranged side-by-side and aligned on one side of the first stack, thereby acquiring multiple magnetic induction intensity signal values ​​of the weld and the substrate material; preprocessing the magnetic induction intensity signal values, i.e., calculating the difference between two magnetic induction intensity signal values ​​in the same group to obtain magnetic induction intensity data after removing the background field; setting at least two conditions for judging defects based on the characteristics of the weld and the principle of weak magnetic detection; using the principle of redundancy detection, substituting the magnetic induction intensity data after removing the background field into the conditions for judging defects to identify defect features; and calculating the correct probability of defect judgment based on probability statistics.

[0056] It is important to understand that two weak magnetic sensors in the same group must be adjacent to each other. This can be understood as follows: a set of weak magnetic sensors stacked together detects the magnetic induction intensity signal in that vertical direction. This magnetic induction intensity signal is generated by the test specimen. Therefore, the signal of the weak magnetic sensor located at the bottom of the stack will be stronger than the signal of the weak magnetic sensor located at the top of the stack. As the distance between the weak magnetic sensor and the test specimen below increases, the influence of the test specimen on the sensor detection signal will become smaller and smaller.

[0057] The weld defect detection method provided in this invention arranges six weak magnetic sensors in two L-shaped stacks to perform precise weld detection. Specifically, the second and fifth weak magnetic sensors in the first group are located directly above the weld to detect the magnetic induction intensity signal value directly above the weld. The first and fourth weak magnetic sensors in the second group are located directly behind the first group and also detect the magnetic induction intensity signal value directly above the weld. The third and sixth weak magnetic sensors in the third group are arranged side-by-side and aligned on one side of the first group to detect the magnetic induction intensity signal value of the substrate material. To effectively reduce detection errors, the difference between two magnetic induction intensity signal values ​​from the same group is calculated. Then, based on the characteristics of the weld and the principle of weak magnetic detection, at least two conditions for judging defects are set. Combining the principle of redundant detection, the magnetic induction intensity data after removing the background field is substituted into the conditions for judging defects to identify defects. Based on the principle of probability statistics, the probability of correct defect judgment is calculated. An array-type weak magnetic sensor is used and its layout is optimized. Data differential technology is used to suppress external interference. Multiple conditions for judging defects can be set through multiple sets of data, thereby improving the reliability of the detection results and the accuracy of weld defect judgment results.

[0058] Specifically, the gap between two stacked weak magnetic sensors is less than or equal to 1 millimeter.

[0059] Specifically, the spacing between two adjacent sets of weak magnetic sensors is less than or equal to 4 millimeters. Considering that signal lines need to be led out from the weak magnetic sensors, the spacing between sets is larger than the stacking range.

[0060] Optionally, the conditions for judging defects include: Condition 1, Condition 2, and Condition 3. Condition 1 includes: judging whether there is an anomaly in the difference between the magnetic induction intensity signal value of the second weak magnetic sensor 2 after removing the background field and the magnetic induction intensity signal value of the first weak magnetic sensor 1 after removing the background field; Condition 2 includes: judging whether there is an anomaly in the difference between the magnetic induction intensity signal value of the second weak magnetic sensor 2 after removing the background field and the magnetic induction intensity signal value of the third weak magnetic sensor 3 after removing the background field; Condition 3 includes: judging whether there is an anomaly in the magnetic induction intensity signal value of the second weak magnetic sensor 2 after removing the background field. Here, the anomaly judgments for Conditions 1, 2, and 3 are all based on the 3σ principle. The magnetic field gradient is calculated from the condition curve, and then 3σ is calculated to see if the gradient curve exceeds 3σ. Condition 1 has the highest importance and is therefore a rejection condition for judging defects; Conditions 2 and 3 are optional. There are other conditions, but their importance decreases progressively.

[0061] Optionally, the preprocessing includes: connecting the magnetic induction intensity signal values ​​detected by the same weak magnetic sensor to form a line, obtaining six corresponding real-time curves, and performing differential calculation on the real-time curves corresponding to two weak magnetic sensors stacked together to obtain three curves with background field removed. The curve with background field removed corresponding to the weak magnetic sensors stacked together in the first group is denoted as Rebf1, the curve with background field removed corresponding to the weak magnetic sensors stacked together in the second group is denoted as Rebf2, and the curve with background field removed corresponding to the weak magnetic sensors stacked together in the third group is denoted as Rebf3.

[0062] Optionally, based on the redundancy detection principle, the three curves after removing the background field are differentially processed to generate two judgment curves. Specifically, the difference between Rebf1 and Rebf2 yields the judgment curve Decide1, and the difference between Rebf2 and Rebf3 yields the judgment curve Decide2. The gradient curve of the second weak magnetic sensor 2 is denoted as Decide3. The gradient curve is obtained by differentiating the curve after removing the background field. Connecting the gradient values ​​to form a line yields the gradient curve. The point values ​​in the judgment curves Decide1, Decide2, and Decide3 constitute the array of judgment defect data.

[0063] Current weak magnetic field testing techniques typically rely on the 3σ principle to determine defects based on test data. However, in actual testing, this relatively simple method often leads to misjudgments.

[0064] Therefore, this invention employs array-type sensors and optimizes their layout, utilizes data differential technology to suppress external interference, and uses intelligent detection algorithms to optimize defect judgment conditions, thereby effectively reducing detection errors and improving the reliability of the detection system, providing strong support for the widespread application of weak magnetic field detection technology in weld defect detection.

[0065] Therefore, this invention provides an improved method. In this embodiment, it is necessary to combine two or more conditions for judging defects and calculate the probability of a defect according to the formula before determining whether a defect exists at a certain location. When a certain condition for judging defects indicates that there is no defect at this location or the calculated result shows that the probability of a defect at this location is small, then it is determined that there is no defect at this location. The method for judging defects is as follows: a) If there are two conditions, both conditions must be met to determine that there is a defect at this location; b) If there are three conditions, under the premise that condition 1 judges that there is a defect, if one of condition 2 or condition 3 judges that there is a defect at this location, then it can be judged that there is a defect at this location. The reliability of the detection result is improved by fusing multiple judgment methods.

[0066] Specifically, the formula for calculating the probability of a defect is:

[0067] Probability=A1*0.6+A2*0.2+A3*0.2

[0068] In this context, A1 corresponds to condition 1, A2 corresponds to condition 2, and A3 corresponds to condition 3. The values ​​of A1, A2, and A3 are either 0 or 1. 0 represents no anomaly, and 1 represents an anomaly. When Probability ≥ 0.8, it can be determined that there is a defect here.

[0069] Specifically, the method for judging anomalies in magnetic flux density signal values ​​is the 3σ method, where conditions 1 and 2 are judged using the same method. The 3σ method includes:

[0070] First, extract the magnetic field gradient values ​​of the two judgment curves. The magnetic field gradient value is the first derivative of the value at each corresponding point on each of the two judgment curves. Store the obtained magnetic field gradient values ​​in the array Decide-f[i,j], as shown in the formula below:

[0071] Decide-f[i,j]=Decide[i,j+1]-Decide[i,j]

[0072] Where Decide-f[i,j] is the weld magnetic induction intensity gradient data array, i is the number of channels, j is the data sequence number, where the number of channels corresponding to the first weak magnetic sensor 1 is channel 0, the data sequence number corresponding to the first weak magnetic sensor 1 is 1, and so on.

[0073] After obtaining the magnetic field gradient curve, the threshold range of each judgment curve is further obtained. in, σ is the average value of the obtained magnetic field gradient, and σ is the standard deviation of the obtained magnetic field gradient.

[0074] Optional, The calculation methods for σ are as follows:

[0075]

[0076] Where n is the total number of data points on the judgment curve, and ΔB(i) is the magnetic field gradient value of the i-th data point on the judgment curve. When K is 3, the magnetic field gradient value is calculated to obtain the threshold line of the interval with a confidence probability of 99.73%. The magnetic field gradient value corresponding to the judgment curve is compared with the threshold line. The sampling point location corresponding to the magnetic field gradient value that exceeds the threshold range is determined to be an abnormal area.

[0077] Utilizing weak magnetic field detection technology, multiple weak magnetic field sensors arranged in an L-shape are used to perform precise inspection of the weld. Combining data acquisition, preprocessing, defect feature recognition, and two-dimensional imaging technologies, automatic defect identification and visualization of the weld are achieved. The signals obtained by the six sensors during detection are voltage signals. The detection areas of each sensor are as follows: the first weak magnetic field sensor 1 and the second weak magnetic field sensor 2 detect directly above the weld, with the second weak magnetic field sensor 2 in front and the first weak magnetic field sensor 1 behind; the third weak magnetic field sensor 3 detects the base material; the fourth weak magnetic field sensor 4 detects the weld directly above the first weak magnetic field sensor 1; the fifth weak magnetic field sensor 5 detects the weld directly above the second weak magnetic field sensor 2; and the sixth weak magnetic field sensor 6 detects the base material directly above the third weak magnetic field sensor 3.

[0078] like Figure 1 The diagram shows the relative arrangement of six weak magnetic sensors. This invention determines whether a defect exists based on the corresponding calculations of the detection data between the sensors.

[0079] refer to Figure 2 , Figure 2 The flowchart provided for the embodiments of the present invention describes the detection operation process as follows: Figure 2As shown, the inspection fixture is used to scan the weld workpiece 7. The magnetic induction intensity signal is collected by a weak magnetic sensor, and the collected signal is stored. The stored signal is then read and preprocessed, that is, the background field of the collected magnetic induction intensity curve is removed (to eliminate environmental noise and magnetic field interference generated by non-defect areas of the workpiece itself). Finally, based on the characteristics of the weld and the principle of weak magnetic detection, two or more conditions for judging defects are set (such as sudden changes in magnetic induction intensity, abnormal distribution, etc.). Using the principle of redundancy detection, defect feature identification is performed for each judgment condition, and the probability of correct defect judgment is calculated according to the principle of probability statistics to improve the accuracy and reliability of the detection.

[0080] The specific steps are as follows:

[0081] 1. Preprocessing of magnetic induction intensity curve

[0082] Figure 1 This is a simulation of the original curves from six-channel probes (CH1, CH2, CH3, CH4, CH5, CH6) scanning for defects in precast welds. The weak magnetic field acquisition circuit board decodes the magnetic flux density data collected by the weak magnetic field sensors, and then performs equivalent calculations based on the sensor calibration values ​​to obtain the magnetic flux density collected by each sensor during the detection process. The host computer software, programmed in C#, uses a corresponding drawing interface to plot the magnetic flux density values ​​corresponding to each acquisition point. Connecting the points with lines yields the result... Figure 3 The real-time curves of the six probe channels are shown.

[0083] Figure 4 The three curves (Rebf1, Rebf2, Rebf3) with the background field removed are obtained by differential operation of the weak magnetic field sensor data in six channels. The original curve data of the six channels are stored in the array Data[i,j], where Data[0,j] represents the data of channel CH1, Data[1,j] represents the data of channel CH2, Data[2,j] represents the data of channel CH3, Data[3,j] represents the data of channel CH4, Data[4,j] represents the data of channel CH5, and Data[5,j] represents the data of channel CH6. The selected channels are then processed as follows: (The data value of each channel is the magnetic induction intensity value after decoding the voltage signal. The magnetic induction intensity value of each channel is stored in the array Data[i,j] for calculation, where i is the channel number and j is the data sequence number;

[0084] The magnetic field strength value of channel 1 minus that of channel 4 corresponds to sensor 1 and sensor 4.

[0085] The magnetic field strength value of channel 2 minus that of channel 5 corresponds to sensor 2 and sensor 5.

[0086] The magnetic field strength value of channel 3 minus that of channel 6 corresponds to sensor 3 and sensor 6.

[0087] Rebf[i,j]=Data[i,j]-Data[i+3,j]

[0088] (4-1)

[0089] Where Rebf[i,j] is an array storing the background field data (corresponding channels (1 and 4, 2 and 5, 3 and 6) have the same trend of data change; subtracting the data can effectively extract the change and suppress noise, which can be considered as removing the background field data), i is the number of channels, and j is the data sequence number. Plotting points into lines in the drawing interface yields the following result: Figure 4 The three curves Rebf1, Rebf2, and Rebf3 shown are after the background field has been removed. The reason for arranging the six sensors in this way is that the second magnetic field weakening sensor 2 is chosen as the core, and the conditions for defect judgment are 2-1, 2-2, and 2-3, respectively. If the third magnetic field weakening sensor 3 is aligned with the first magnetic field weakening sensor 1, there will be a positional difference between the second and third magnetic field weakening sensors 2 and 3, resulting in a significant error in their signal difference. Of course, this arrangement can also be based on the first magnetic field weakening sensor 1, in which case the judgment conditions should be 1-2, 1-1, and 1-3.

[0090] Figure 5 Based on the principle of redundancy detection, three curves with the background field removed are used to improve the accuracy of the judgment by using data redundancy and different channel data. Two judgment curves are generated through differential operations, along with the gradient curves Decide1, Decide2, and Decide3 of the sensor signal located directly above the weld. For conditions 1 and 2, the background-removed data Rebf[i,j] is processed as follows, where i is 0 and 1:

[0091] Decide[i,j]=Rebf[i,j]-Rebf[i+1,j]

[0092] (4-2)

[0093] For condition 3, the data Rebf[2,j] after removing the background field is processed as follows:

[0094] Decide[2,j]=Rebf[2,j+1]-Rebf[2,j]

[0095] (4-3)

[0096] Here, Decide[i,j] is an array that stores the data for judging defects, where i is the number of channels and j is the data sequence number.

[0097] A determination of whether a defect exists requires considering two or more judgment conditions and calculating the probability of a defect using a formula. If one of the judgment conditions indicates that there is no defect at this location, or if the calculated probability of a defect at this location is low, then it is determined that there is no defect at this location.

[0098] There are generally three conditions for determining whether a weld defect exists: (1) whether there is an abnormality in the difference between the detection signals of the two sensors located directly above the weld; (2) whether there is an abnormality in the comparison signal between the weld and the base material; and (3) whether there is an abnormality in the detection signal gradient of the first sensor located directly above the weld.

[0099] The probability calculation method for judging defects is as follows: (1) If there are two conditions, both conditions must be true before it can be judged as a defect; (2) If there are three conditions, if condition 1 is judged as a defect, and one of condition 2 or condition 3 is judged as a defect, then it can be judged as a defect.

[0100] The formula for calculating the defect probability is:

[0101] Probability=A1*0.6+A2*0.2+A3*0.2

[0102] (4-4)

[0103] The values ​​of A1, A2, and A3 are either 0 or 1, representing the presence or absence of abnormal signals. When Probability ≥ 0.8, a defect can be identified. Through comprehensive calculations based on multiple conditions, the reliability of the detection system is greatly improved.

[0104] 2. Conditional Judgment

[0105] (1) Detection signal

[0106] The method for judging magnetic signal anomalies is the 3σ method. Conditions 1 and 2 are the same.

[0107] First, extract the magnetic field gradient value of the "judgment curve" and store the obtained magnetic field gradient value in the array Decide-f[i,j]. The formula is shown below:

[0108] Decide-f[i,j]=Decide[i,j+1]-Decide[i,j]

[0109] (4-5)

[0110] Where Decide-f[i,j] is the array of magnetic induction intensity gradient data of the weld, i is the number of channels, and j is the data sequence number.

[0111] After obtaining the magnetic field gradient curve, the threshold range of each judgment curve is further obtained using formulas 4-6 and 4-7. in, σ represents the average value of the obtained magnetic field gradient, and σ is the standard deviation of the obtained magnetic field gradient. Specifically... The calculation methods for σ are as follows:

[0112]

[0113] Where n is the total number of data points on the judgment curve, and ΔB(i) is the magnetic field gradient value of the i-th data point on the judgment curve. When K is 3, the magnetic field gradient value is used to calculate the threshold line (i.e., the upper and lower limits of the threshold range) with a confidence probability of 99.73%. The magnetic field gradient value corresponding to the judgment curve is compared with the threshold line, and the sampling point location corresponding to the magnetic field gradient value that exceeds the threshold range is determined to be an abnormal area.

[0114] This invention arranges weak magnetic sensors in an L-shape and generates a detection curve containing various detection information through numerical calculations between data from multiple sensor sets. Anomaly identification of the detection curve yields multiple defect judgment information, and a comprehensive evaluation method based on these multiple judgment information improves the accuracy of defect judgment.

[0115] The corresponding structural units of the weld inspection method based on weak magnetic field technology designed in this invention are as follows: Figure 6 As shown, it mainly consists of a host computer, a weak magnetic signal acquisition system, a probe, and tooling.

[0116] The weak magnetic field detection instrument uses a fluxgate sensor as the magnetic field measurement probe. The resolution of this sensor can reach 1nT, which enables high-precision detection of the magnetic induction intensity signal on the surface of the test piece.

[0117] probe fixtures such as Figure 7 As shown, it includes: a sensor mounting slot 10, an adapter 11, a cable management box 12, and an aviation connector 4. The interior of the sensor mounting slot 10 is as follows... Figure 8 As shown, six weak magnetic sensors are arranged in the mounting slot. The first weak magnetic sensor 1, the second weak magnetic sensor 2, and the third weak magnetic sensor 3 form an "L" arrangement, and the fourth weak magnetic sensor 4, the fifth weak magnetic sensor 5, and the sixth weak magnetic sensor 6 form another "L" arrangement. The adapter 11 connects the sensor mounting slot 10 and the cable bundle 12. The cable bundle 12 provides structural support for the entire probe fixture, ensuring its stability. Sensor replacement can be easily achieved by disassembling and assembling the cable bundle 12. The aviation connector 13 is used to fix the aviation connector cable and prevent it from shaking. To avoid interference with sensor signal acquisition due to excessive magnetic permeability of the fixture material, a non-magnetic non-metallic material was selected for the fixture in the design.

[0118] The probe detection path diagram is as follows: Figure 9 As shown, the arrow indicates the direction of probe movement. During the inspection, the weld workpiece 7 is placed flat on the test table, and the fixture scans the defects at a uniform speed from right to left, starting from the leftmost end of the workpiece. During the inspection, the first weak magnetic sensor 1, the second weak magnetic sensor 2, the fourth weak magnetic sensor 4, and the fifth weak magnetic sensor 5 are positioned directly above the weld, and the third weak magnetic sensor 3 and the sixth weak magnetic sensor 6 are positioned on the substrate. Specific Implementation

[0120] 1. Explanation of experimental principle:

[0121] Taking the inspection of a weld as an example, the surface of the workpiece 7 to be tested is regarded as a flat surface, and 6 weak magnetic sensors are selected according to... Figure 8 The sensors are placed in the mounting slot 10 in a relative arrangement and scanned at a constant speed along the weld seam 8. The distance between the probe and the weld seam 8 should be as small as possible, and the relative position of the probe and the weld seam should be kept constant.

[0122] After scanning the weld seam using a probe fixture, the system stores continuous detection data from six sensor channels, generating an original curve with the number of acquisition points as the X-axis and magnetic induction intensity as the Y-axis. Background field removal is performed on the original curve using Formula 4-1; three defect judgment conditions are obtained using Formulas 4-2 and 4-3, combined with the defect probability calculation formula 4-4.

[0123] The three methods for identifying defects are the same: Since the change in the weak magnetic signal at the defect location is very subtle, it is difficult to determine the defect solely by comparing the signals. Therefore, gradient values ​​are extracted from the differential curve to obtain a gradient value curve reflecting the abrupt change in the intensity of the weak magnetic signal, and an appropriate threshold line is set. The spatial magnetic field gradient value of each sampling point is compared with the threshold line; the sampling point region corresponding to the spatial magnetic field gradient value exceeding the threshold range is determined to be a defect. The final determination of a defect is contingent upon the first judgment condition identifying a defect, and at least one of the second or third judgment conditions identifying a defect; otherwise, it is considered a false positive. After determining the defect location, the detection result is represented in two dimensions using a spline interpolation algorithm.

[0124] 2. Example illustration:

[0125] The workpiece 7 being inspected is a flat weld, made of austenitic stainless steel. Three defects were created in the weld. The surface of the workpiece is considered flat, and the process is as follows: Figure 9 A schematic diagram of the probe's movement path is shown, scanning weld seam 8 at a uniform speed. Based on the experimental principle, the detected magnetic signal is preprocessed to obtain judgment conditions for defect identification and result display in a two-dimensional image.

[0126] The original curve of the weld is as follows:

[0127] Figure 10 , 11 Figures 1 and 12 represent the original curve, the background field removal curve, and the defect judgment curve obtained from scanning the weld template with a relative probe. Based on the experimental principle, the background field is removed from the original curve. The background field removal condition curve, based on the redundancy detection principle, yields two defect judgment curves, completing the preprocessing operation.

[0128] according to Figure 15 Ultimately, three defective areas were identified. Judgment condition 1 was displayed in four areas at sampling points 55-65, 13245, 200-210, and 26375. The first three areas met judgment condition 2 or condition 3, confirming the existence of defects. However, in the fourth area, neither condition 2 nor condition 3 indicated a defect, so this area was judged to be defect-free.

[0129] The above inventions are merely a few specific embodiments of the present invention. However, the embodiments of the present invention are not limited thereto, and any variations that can be conceived by those skilled in the art should fall within the protection scope of the present invention.

Claims

1. A method for detecting weld defects, characterized in that, include: Six weak magnetic sensors are stacked in pairs in an L-shape. The second weak magnetic sensor (2) and the fifth weak magnetic sensor (5) of the first stack are located directly above the weld. The first weak magnetic sensor (1) and the fourth weak magnetic sensor (4) of the second stack are located directly behind the first group. The third weak magnetic sensor (3) and the sixth weak magnetic sensor (6) of the third stack are arranged side by side and aligned on one side of the first group, thereby obtaining multiple magnetic induction intensity signal values ​​of the weld and the base material. The magnetic field strength signal values ​​are preprocessed, that is, the difference between two magnetic field strength signal values ​​in the same group is calculated to obtain the magnetic field strength data after removing the background field. Based on the characteristics of the weld and the principle of weak magnetic field detection, at least two conditions for judging defects are set. Using the redundancy detection principle, the magnetic induction intensity data after removing the background field is substituted into the conditions for judging defects to identify defects, and the probability of correct defect judgment is calculated according to the principle of probability statistics. The conditions for determining defects include: condition 1, condition 2, and condition 3; wherein, Condition 1 includes: determining whether there is an abnormality in the difference between the magnetic induction intensity signal value of the second weak magnetic sensor (2) after removing the background field and the magnetic induction intensity signal value of the first weak magnetic sensor (1) after removing the background field; Condition 2 includes: determining whether there is an abnormality in the difference between the magnetic induction intensity signal value of the second weak magnetic sensor (2) after removing the background field and the magnetic induction intensity signal value of the third weak magnetic sensor (3) after removing the background field; Condition 3 includes: determining whether there is an abnormality in the magnetic induction intensity signal value of the second weak magnetic sensor (2) after removing the background field.

2. The weld defect detection method as described in claim 1, characterized in that, The preprocessing includes: connecting the magnetic induction intensity signal values ​​detected by the same weak magnetic sensor to form a line, obtaining six corresponding real-time curves, and performing differential calculation on the real-time curves corresponding to two weak magnetic sensors stacked together to obtain three curves for removing the background field. The curve for removing the background field corresponding to the weak magnetic sensors stacked together in the first group is denoted as Rebf1, the curve for removing the background field corresponding to the weak magnetic sensors stacked together in the second group is denoted as Rebf2, and the curve for removing the background field corresponding to the weak magnetic sensors stacked together in the third group is denoted as Rebf3.

3. The weld defect detection method as described in claim 2, characterized in that, Based on the redundancy detection principle, the three curves after removing the background field are differentially processed to generate two judgment curves. The difference between Rebf1 and Rebf2 is used to obtain the judgment curve Decide1, and the difference between Rebf2 and Rebf3 is used to obtain the judgment curve Decide2. The gradient curve of the second weak magnetic sensor (2) is denoted as Decide3. The point values ​​in the judgment curves Decide1, Decide2 and Decide3 are the array of judgment defect data.

4. The weld defect detection method as described in claim 3, characterized in that, It is necessary to combine two or more conditions for judging defects and calculate the probability of it being a defect according to the formula before it can be determined whether a defect exists. When a certain condition for judging defects shows that there is no defect here or the calculated result is that the probability of a defect here is small, then it is judged that there is no defect here. The method for judging defects is as follows: a) If there are two conditions, both conditions must be true to judge that there is a defect here; b) If there are three conditions, under the premise that condition 1 judges that there is a defect, if one of condition 2 or condition 3 judges that there is a defect here, then it can be judged that there is a defect here.

5. The weld defect detection method as described in claim 4, characterized in that, The probability calculation formula for the defect is as follows: Probability=A1*0.6+A2*0.2+A3*0.2 In this context, A1 corresponds to condition 1, A2 corresponds to condition 2, and A3 corresponds to condition 3. The values ​​of A1, A2, and A3 are either 0 or 1. 0 represents no anomaly, and 1 represents an anomaly. When Probability ≥ 0.8, it can be determined that there is a defect here.

6. The weld defect detection method as described in claim 5, characterized in that, The method for judging anomalies between magnetic induction intensity signal values ​​is the 3σ method, wherein the methods for judging signal anomalies in conditions 1 and 2 are the same, and the 3σ method includes: First, extract the magnetic field gradient values ​​from the two judgment curves. The magnetic field gradient value is the first derivative of the value at each corresponding point on each of the two judgment curves. Store the obtained magnetic field gradient values ​​into an array. In Chinese, the formula is as follows: in, This is an array of magnetic induction intensity gradient data for weld seams. For the number of channels, The data sequence number is 0, where the number of channels corresponding to the first weak magnetic sensor (1) is channel 0, the data sequence number corresponding to the first weak magnetic sensor (1) is 1, and so on. After obtaining the magnetic field gradient value curve, the threshold range of each judgment curve is further obtained. ),in, σ is the average value of the obtained magnetic field gradient, and σ is the standard deviation of the obtained magnetic field gradient.

7. The weld defect detection method as described in claim 6, characterized in that, The The calculation methods for σ are as follows: in, To determine the total number of data points on the curve, To determine the magnetic field gradient value of the i-th data point on the curve, when K is 3, the magnetic field gradient value is calculated to obtain a threshold line with a confidence probability of 99.73%. The magnetic field gradient value corresponding to the judgment curve is compared with the threshold line. The sampling point location corresponding to the magnetic field gradient value that exceeds the threshold range is determined to be an abnormal area.

Images