A method and system for non-destructive testing of weld defects in a metal structure

Abstract

This invention relates to the field of weld defect detection technology, specifically to a non-destructive testing method and system for weld defects in metal structural components. The method includes: acquiring A-scan waveforms at different angles along the weld centerline; determining feature points from the A-scan waveforms; determining the degree of a single peak based on the signal difference, sequence interval, and amplitude of the feature point compared to its adjacent maxima; identifying other feature points at different angles that are closest in sampling time to any given feature point, as control points; determining the defect stability of the feature point based on the difference in the degree of a single peak and the difference in sampling sequence interval between the feature point and its control points; determining the defect manifestation probability of each feature point based on the defect stability and degree of a single peak of all feature points in the A-scan waveform at the same angle; and selecting feature points for non-fusion defects based on the defect manifestation probability, thereby achieving defect detection. This invention can effectively distinguish between non-fusion defects and geometric boundary reflection signals, improving the accuracy of the detection results.

Description

Technical Field

[0001] This invention relates to the field of weld defect detection technology, specifically to a non-destructive testing method and system for weld defects in metal structural components. Background Technology

[0002] The quality of welds in metal structural components directly affects the strength, safety, and service life of the engineering structure. Ultrasonic non-destructive testing (NDT) technology can detect potential defects in a timely manner without compromising structural integrity. Incomplete fusion in welds is a common and highly dangerous type of welding defect in metal structural components. This occurs when the weld metal fails to fully fuse with the base metal or the previous weld pass, forming an unbonded interface. This defect typically extends along the weld sidewall or between layers, exhibiting planar characteristics that are difficult to detect visually. However, it severely weakens the mechanical properties of the weld, becoming a source of crack propagation and easily leading to fracture failure under stress, especially in load-bearing components or fatigue load conditions. Therefore, accurate detection and identification of incomplete fusion defects are crucial to ensuring structural safety.

[0003] In existing technologies, incomplete fusion defects often occur at the interface between the weld metal and the base metal or the previous weld pass, close to the geometric boundaries of the weld, such as the bevel sidewall or interlayer interface. Their reflection characteristics are very similar to those of geometric boundaries in ultrasonic testing, potentially generating high-amplitude, planar echo signals, leading to misjudgments or missed detections. Therefore, in nondestructive testing, incomplete fusion defects are easily confused with the reflection signals of geometric boundaries, resulting in poor accuracy of the test results. Summary of the Invention

[0004] To address the technical problem in nondestructive testing (NDT) where incomplete fusion defects are easily confused with geometric boundary reflection signals, leading to poor accuracy of test results, this invention provides a nondestructive testing method and system for weld defects in metal structural components. The specific technical solution adopted is as follows:

[0005] This invention proposes a non-destructive testing method for weld defects in metal structural components, the method comprising:

[0006] At the measuring point on the center line of the weld, an ultrasonic detector is used to scan along different directions to obtain A-scan waveforms at different angles.

[0007] The maximum points are determined from the A-scan waveform to form a maximum point sequence. From the maximum point sequence, the maximum points in the sequence are further determined to obtain the feature points. Based on the signal difference between the feature points and adjacent maximum points in the sequence and the sequence interval, the sharpness at the location of the feature point is determined. Combining the sharpness and the amplitude of the feature point, the degree of single peak of the feature point is determined.

[0008] Identify other feature points that are closest to any feature point in the sampling time sequence at different angles, and use them as control points; determine the defect stability of the feature point based on the difference in the degree of single peak and the difference in the sampling sequence interval between the feature point and its control point; determine the defect performance probability of each feature point based on the defect stability and degree of single peak of all feature points in the A-scan waveform at the same angle.

[0009] By combining the probability of defect manifestations, feature points of unfused defects are selected to achieve defect detection.

[0010] Further, determining the sharpness at the location of a feature point based on the signal difference between the feature point and its adjacent maxima and the sequence interval includes:

[0011] The two nearest maxima points on either side of the maximum point sequence are taken as adjacent points;

[0012] Based on the difference in signal amplitude between the feature point and each adjacent point and the sequence interval, the sharpness analysis index of the feature point and its corresponding adjacent points is determined.

[0013] The mean of the sharpness index of the feature point and all its neighboring points is used as the sharpness at the location of the feature point.

[0014] Further, the step of determining the sharpness analysis index of the feature point and its corresponding neighboring points based on the signal amplitude difference and sequence interval between the feature point and each neighboring point includes:

[0015] Calculate the signal amplitude difference and sequence interval number between the feature point and each adjacent point. Normalize the ratio of the signal amplitude difference to the sequence interval number and use it as the sharpness analysis index between the feature point and its corresponding adjacent points.

[0016] Furthermore, determining the degree of monophasic peak of a feature point by combining the sharpness and the amplitude of the feature point includes:

[0017] The ratio of the amplitude of the feature point to the maximum amplitude in the A-scan waveform is used as the amplitude coefficient of the feature point.

[0018] Calculate the product of the sharpness and the amplitude coefficient, and normalize it to obtain the single peak degree of the feature point.

[0019] Furthermore, determining the defect stability of a feature point based on the difference in the degree of single peak and the difference in sampling sequence interval between the feature point and its control point includes:

[0020] The absolute value of the difference between the single peak degree of the feature point and any control point is calculated and normalized to serve as the peak influence index between the feature point and the corresponding control point.

[0021] The sampling sequence interval between the feature point and any control point is determined, and the negative number of the sampling sequence interval is normalized and used as a time series influence index.

[0022] By combining the peak influence index and time-series influence index of the feature point with each control point, the defect stability of the feature point is determined.

[0023] Furthermore, the determination of the defect stability of the feature point by combining the peak influence index and the time-series influence index of each control point includes:

[0024] Calculate the product of the peak influence index and the time-series influence index of the feature point and the same control point, and normalize it to obtain the defect analysis coefficient of the feature point and the corresponding control point.

[0025] The mean of the defect analysis coefficients of the feature point and all control points is used as the defect stability of the feature point.

[0026] Furthermore, determining the defect manifestation probability of each feature point based on the defect stability and single peak degree of all feature points in the A-scan waveform at the same angle includes:

[0027] The product of the defect stability and the single peak degree of the feature point is normalized and used as the defect manifestation probability of the feature point.

[0028] Furthermore, the feature points for screening non-fusion defects by combining defect performance probabilities include:

[0029] Feature points whose defect manifestation probability is greater than a preset probability threshold are used as feature points of unfused defects.

[0030] Furthermore, the ultrasonic detector is located directly above the measuring point and includes an adjustable angle probe. The angle is the deflection angle of the ultrasonic detector towards the direction of travel of the weld, and the angle range is from 45° to 70°. An A-scan waveform is acquired every 1°.

[0031] On the other hand, it also includes a non-destructive testing system for weld defects in metal structural components, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the method as described in any of the foregoing.

[0032] The present invention has the following beneficial effects:

[0033] This invention provides an embodiment of A-scan analysis, which obtains A-scan waveforms at different angles. Then, based on the wave characteristics of the A-scan waveforms, specific analysis is performed to identify feature points. The sharpness of the feature point's location is determined based on the signal difference and sequence interval between the feature point and its adjacent maxima. The single-peak intensity of the feature point is determined by combining the sharpness and the amplitude of the feature point. Subsequently, multi-angle feature analysis is performed on the feature points at different angles to identify other feature points with the closest sampling time sequence to any feature point at different angles, serving as control points. The defect stability of the feature point is determined based on the difference in single-peak intensity and sampling time sequence between the feature point and its control points. The defect performance probability of each feature point is determined based on the defect stability and single-peak intensity of all feature points in the A-scan waveform at the same angle. Finally, feature points with unfused defects are selected based on the defect performance probability to achieve defect detection. Compared to traditional ultrasonic testing, which cannot accurately distinguish between unfused defects and geometric boundaries, this application analyzes the fluctuations of peaks at the same angle and the changes at different angles to accurately identify unfused defects and geometric boundaries, effectively distinguishing unfused defects and avoiding confusion between the reflection signals of unfused defects and geometric boundaries, thereby improving the accuracy of the test results. Attached Figure Description

[0034] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a flowchart of a non-destructive testing method for weld defects in metal structural components, provided as an embodiment of the present invention. Detailed Implementation

[0036] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a non-destructive testing method and system for weld defects in metal structural components proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0038] The following describes in detail, with reference to the accompanying drawings, a specific scheme for a non-destructive testing method for weld defects in metal structural components provided by the present invention.

[0039] Please see Figure 1 The diagram illustrates a flowchart of a non-destructive testing method for weld defects in metal structural components according to an embodiment of the present invention. The method includes:

[0040] S101: At the measuring point on the center line of the weld, use an ultrasonic detector to scan along different directions to obtain A-scan waveforms at different angles.

[0041] The quality of welds in metal structural components directly affects the strength, safety, and service life of the engineering structure. Ultrasonic non-destructive testing (NDT) technology can detect potential defects in a timely manner without compromising structural integrity. Incomplete fusion in welds is a common and highly dangerous type of welding defect in metal structural components. This occurs when the weld metal fails to fully fuse with the base metal or the previous weld pass, forming an unbonded interface. This defect typically extends along the weld sidewall or between layers, exhibiting planar characteristics that are difficult to detect visually. However, it severely weakens the mechanical properties of the weld, becoming a source of crack propagation and easily leading to fracture failure under stress, especially in load-bearing components or fatigue load conditions. Therefore, accurate detection and identification of incomplete fusion defects are crucial to ensuring structural safety.

[0042] In existing technologies, incomplete fusion defects often occur at the interface between the weld metal and the base metal or the previous weld pass, close to the geometric boundaries of the weld, such as the bevel sidewall or interlayer interface. Their reflection characteristics are very similar to those of geometric boundaries in ultrasonic testing, potentially generating high-amplitude, planar echo signals, leading to misjudgments or missed detections. Therefore, in nondestructive testing, incomplete fusion defects are easily confused with the reflection signals of geometric boundaries, resulting in poor accuracy of the test results.

[0043] To address the aforementioned issues, this invention analyzes the differences between unfused defects and geometric boundaries to effectively distinguish unfused defects, thereby improving the accuracy of detection results.

[0044] In this embodiment of the invention, the weld is a long, straight weld. The weld is constructed by constructing a minimum bounding rectangle, and the center line in the welding direction is determined parallel to the long side. The main scanning path is laid out along the center line of the weld, and a measuring point is set every 10 mm. In this embodiment of the invention, the ultrasonic detector is located directly above the measuring point and includes an adjustable angle probe. The angle is the deflection angle of the ultrasonic detector in the direction of the weld, and the angle range is 45° to 70°. An A-scan waveform is acquired every 1°.

[0045] That is, when the ultrasonic probe is vertically downward, the angle is 0°. Then, it is rotated in the direction of welding and A-scan waveform is acquired every 1° within the range of 45° to 70°.

[0046] S102: Determine the maximum points from the A-scan waveform to form a maximum point sequence. Further determine the maximum points in the sequence from the maximum point sequence to obtain the feature points. Based on the signal difference between the feature points and adjacent maximum points in the sequence and the sequence interval, determine the sharpness at the location of the feature points. Combine the sharpness and the amplitude of the feature points to determine the degree of single peak of the feature points.

[0047] Lack of fusion defects typically appear as planar distributions along weld sidewalls or interlayer interfaces. They are geometrically regular and have relatively smooth surfaces, similar to the geometric boundaries of bevel sidewalls and weld toes, and can all form obvious interface reflections. In ultrasonic testing, when a sound beam is incident on these interfaces at a specific angle, a similar specular reflection phenomenon occurs, causing both to appear as high-amplitude, sharp echo signals in A-scan.

[0048] In A-scan waveforms, unfused defects and geometric boundaries often exhibit high-amplitude, single-peak reflected waves with sharp waveforms. Therefore, the suspected defect locations are first identified based on the common characteristics of unfused defects and geometric boundaries.

[0049] For any angle of any measurement point, the A-scan waveform is fluctuating. In order to observe the feature points that may be defects, all the maximum points of the A-scan waveform are first obtained. Since the purpose of this application is to distinguish between unfused defects and geometric boundaries, the two may coexist and both exhibit the same single peak feature. Therefore, in the sequence of maximum points formed by obtaining all the maximum values, the maximum points in the sequence are obtained and recorded as feature points.

[0050] Feature points may express unfused defect characteristics, geometric boundary characteristics, or no characteristics at all. Therefore, the degree of monoclinic activity of each feature point is first calculated to distinguish between feature points that express characteristics and those that do not.

[0051] Furthermore, in some embodiments of the present invention, determining the sharpness at the location of a feature point based on the signal difference between the feature point and its adjacent maxima and the sequence interval includes: designating the two maxima closest to the feature point on both sides of the maxima sequence as adjacent points; determining the sharpness analysis index between the feature point and each adjacent point based on the signal amplitude difference and the sequence interval; and taking the average of the sharpness analysis indices between the feature point and all adjacent points as the sharpness at the location of the feature point.

[0052] The sharpness is manifested in the waveform as a sudden appearance with a high amplitude. In order to analyze this feature, in this embodiment of the invention, adjacent points are first determined, and then the feature point and each adjacent point are analyzed.

[0053] Furthermore, in some embodiments of the present invention, the sharpness analysis index of the feature point and its corresponding adjacent points is determined based on the signal amplitude difference and sequence interval between the feature point and each adjacent point, including: calculating the signal amplitude difference and sequence interval between the feature point and each adjacent point, and normalizing the ratio of the signal amplitude difference to the sequence interval as the sharpness analysis index between the feature point and its corresponding adjacent points.

[0054] The sequence interval number represents the sequence interval between the feature point and its neighboring points in the sequence. For example, if the feature point is the 5th point in the sequence and the neighboring point is the 7th point in the sequence, the sequence interval number is 2.

[0055] In this embodiment of the invention, since the feature point is the maximum among the maximum points, the signal amplitude of the feature point is greater than that of its neighboring points. The larger the difference in signal amplitude between the feature point and its neighboring points, the greater the amplitude drop between the feature point and its neighboring points, meaning that the signal amplitude difference at the feature point is more pronounced. The sequence interval number is used as a weight for analysis; the smaller the sequence interval number, the closer the feature point is to its corresponding neighboring point, meaning that the signal amplitude difference is more obvious.

[0056] Based on the above characteristics, this embodiment of the invention calculates the ratio of the signal amplitude difference to the number of sequence intervals, and normalizes it as a sharpness analysis index between feature points and their corresponding neighboring points.

[0057] Then, the mean of the sharpness analysis index of the feature point and all its adjacent points is taken as the sharpness at the feature point location. The larger the sharpness value, the faster the peak value decreases, which is consistent with the rapid descent trend of unfused defects and geometric boundaries. This feature point is more sharply reflected in the A-scan waveform.

[0058] Furthermore, in some embodiments of the present invention, the single peak degree of a feature point is determined by combining the sharpness and the amplitude of the feature point, including: using the ratio of the amplitude of the feature point to the maximum amplitude in the A-scan waveform as the amplitude coefficient of the feature point; calculating the product of the sharpness and the amplitude coefficient, and normalizing the product to obtain the single peak degree of the feature point.

[0059] Among them, the sharpness is analyzed using amplitude analysis, which mainly involves the difference in amplitude between adjacent values. According to prior experience, unfused defects and geometric boundaries themselves also have high amplitude characteristics, so they need to be analyzed together.

[0060] First, the ratio of the amplitude of the feature point to the maximum amplitude in the A-scan waveform is used as the amplitude coefficient of the feature point; that is, the amplitude coefficient is obtained by normalizing the maximum amplitude in the A-scan waveform. The larger the amplitude coefficient, the more it conforms to the high amplitude feature.

[0061] In this embodiment of the invention, the single peak degree of the feature point is obtained by directly calculating the product of the sharpness and the amplitude coefficient and normalizing the result. The larger the value of the single peak degree, the more it reflects the sharp waveform characteristics of the single-peak reflected wave of the feature point.

[0062] S103: Determine other feature points that are closest to any feature point in the sampling time sequence at different angles, and use them as control points; determine the defect stability of the feature point based on the difference in the degree of single peak and the difference in the sampling sequence interval between the feature point and its control point; determine the defect performance probability of each feature point based on the defect stability and degree of single peak of all feature points in the A-scan waveform at the same angle.

[0063] The above steps determine the degree of single peak of feature points based on the common characteristics of unfused defects and geometric boundaries, and judge feature points that may represent unfused defects and geometric boundaries in the A-scan waveform. Since unfused defects are defect features, while geometric boundaries are normal features of metal structural parts, it is necessary to distinguish between the two when performing defect detection on metal structural parts, so as to achieve the accuracy of defect identification.

[0064] Unfused defects typically exhibit strong directionality, forming sharp, high-amplitude specular reflections only at specific incident angles. These reflections show a sudden increase or instantaneous peak as the angle changes, while rapidly attenuating or even disappearing at other angles, demonstrating clear angle selectivity. In contrast, geometric boundaries, due to their continuous and regular transition surfaces, have more dispersed reflected energy. Their waveform amplitude shows a smooth upward or downward trend as the probe angle changes, without exhibiting sharp peaks.

[0065] Based on the above characteristics, it is first necessary to analyze the influence of changes at different angles. For any A-scan waveform at any angle of the measuring point, any feature point is recorded as the target feature point. In the A-scan waveform at other angles of the measuring point, the feature point whose horizontal coordinate is closest to the target feature point is obtained (the logic of this approach is that the unfused defect exhibits stability on the horizontal axis, specifically, it appears repeatedly at a specific depth, that is, in the process of continuous angle changes, it exhibits defect features at irregular intervals, but the horizontal coordinate position of the defect feature is stable), and is recorded as the reference point of the target feature point.

[0066] When the target feature point represents the location of an unfused defect, the characteristics of this defect exhibit a sharp increase or instantaneous peak with changing angles, followed by rapid attenuation or even disappearance at other angles. Therefore, among all the reference feature points for the target feature point, the disappearance of the defect feature at some angles means that the reference feature point obtained at those angles is not actually the feature point representing the defect. In other words, the stability of the horizontal coordinate position is disrupted, causing fluctuations in the horizontal coordinate of the reference feature point. Therefore, a control point is selected for specific analysis.

[0067] Furthermore, in some embodiments of the present invention, the defect stability of a feature point is determined based on the difference in the degree of single peak between the feature point and its control point and the difference in the sampling sequence interval. This includes: calculating the absolute value of the difference in the degree of single peak between the feature point and any control point, and normalizing it as a peak influence index between the feature point and the corresponding control point; determining the sampling sequence interval between the feature point and any control point, and normalizing the negative of the sampling sequence interval as a time-series influence index; and combining the peak influence index and the time-series influence index between the feature point and each control point to determine the defect stability of the feature point.

[0068] In this embodiment of the invention, the peak influence index represents the difference in the degree of single peak between the control point and the feature point. The larger the value, the greater the peak change between the control point and the feature point, that is, the more sudden the peak change at the feature point location.

[0069] Among them, the time sequence influence index represents the sampling sequence interval between the control point and the feature point. Specifically, the smaller the sampling sequence interval, the more abrupt anomaly features the control point and the feature point exhibit at the same position under different angles. On the other hand, the larger the sampling sequence interval, the more it conforms to the transition characteristics of the continuous and regular geometric boundary structure. Therefore, in this embodiment of the invention, the negative number of the sampling sequence interval is normalized as the time sequence influence index. The larger the value of the time sequence influence index, the more it conforms to the abrupt anomaly characteristics of the non-fusion defect.

[0070] Therefore, in this embodiment of the invention, the defect stability of a feature point is determined by combining the peak influence index and the time-series influence index of the feature point with each control point. This includes: calculating the product of the peak influence index and the time-series influence index of the feature point and the same control point, normalizing it to obtain the defect analysis coefficient of the feature point and the corresponding control point; and taking the mean of the defect analysis coefficients of the feature point and all control points as the defect stability of the feature point.

[0071] The degree of defect stability indicates the stability of the feature point at all angles. The larger the value, the more obvious the sudden anomaly of the feature point, which is more in line with the sudden anomaly characteristics of the non-fusion defect.

[0072] Furthermore, in some embodiments of the present invention, the defect performance probability of each feature point is determined based on the defect stability and single peak degree of all feature points in the A-scan waveform at the same angle, including: normalizing the product of the defect stability and single peak degree of the feature point as the defect performance probability of the feature point.

[0073] In this embodiment of the invention, the larger the value of the defect stability degree, the more obvious the sudden anomaly of the feature point under different angles of analysis, that is, the more it conforms to the sudden anomaly characteristics of the non-fusion defect. The larger the value of the single peak degree, the more it reflects the sharp waveform characteristics of the single-peak reflected wave of the feature point. Therefore, the product of the defect stability degree and the single peak degree is directly calculated and normalized to obtain the defect performance probability of the feature point, that is, the probability value of the corresponding feature point being a non-fusion defect.

[0074] S104: Combine the probability of defect manifestation to screen feature points of unfused defects to achieve defect detection.

[0075] In this embodiment of the invention, feature points with a defect manifestation probability greater than a preset probability threshold are used as feature points of unfused defects.

[0076] The preset probability threshold is a threshold value for the probability of defect manifestation. In this embodiment of the invention, the preset probability threshold can be, for example, 0.7. That is, feature points with a defect manifestation probability greater than 0.7 are used as feature points of unfused defects.

[0077] This invention provides an embodiment of A-scan analysis, which obtains A-scan waveforms at different angles. Then, based on the wave characteristics of the A-scan waveforms, specific analysis is performed to identify feature points. The sharpness of the feature point's location is determined based on the signal difference and sequence interval between the feature point and its adjacent maxima. The single-peak intensity of the feature point is determined by combining the sharpness and the amplitude of the feature point. Subsequently, multi-angle feature analysis is performed on the feature points at different angles to identify other feature points with the closest sampling time sequence to any feature point at different angles, serving as control points. The defect stability of the feature point is determined based on the difference in single-peak intensity and sampling time sequence between the feature point and its control points. The defect performance probability of each feature point is determined based on the defect stability and single-peak intensity of all feature points in the A-scan waveform at the same angle. Finally, feature points with unfused defects are selected based on the defect performance probability to achieve defect detection. Compared to traditional ultrasonic testing, which cannot accurately distinguish between unfused defects and geometric boundaries, this application analyzes the fluctuations of peaks at the same angle and the changes at different angles to accurately identify unfused defects and geometric boundaries, effectively distinguishing unfused defects and avoiding confusion between the reflection signals of unfused defects and geometric boundaries, thereby improving the accuracy of the test results.

[0078] On the other hand, it also includes a non-destructive testing system for weld defects in metal structural components, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the non-destructive testing method for weld defects in metal structural components as described in any of the preceding claims.

[0079] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A non-destructive testing method for weld defects in metal structural components, characterized in that, The method includes: At the measuring point on the center line of the weld, an ultrasonic detector is used to scan along different directions to obtain A-scan waveforms at different angles. The maximum points are determined from the A-scan waveform to form a maximum point sequence. From the maximum point sequence, the maximum points in the sequence are further determined to obtain the feature points. Based on the signal difference between the feature points and adjacent maximum points in the sequence and the sequence interval, the sharpness at the location of the feature point is determined. Combining the sharpness and the amplitude of the feature point, the degree of single peak of the feature point is determined. Identify other feature points that are closest to any feature point in the sampling time sequence at different angles, and use them as control points; determine the defect stability of the feature point based on the difference in the degree of single peak and the difference in the sampling sequence interval between the feature point and its control point; determine the defect performance probability of each feature point based on the defect stability and degree of single peak of all feature points in the A-scan waveform at the same angle. By combining the probability of defect manifestations to screen feature points of non-fusion defects, defect detection can be achieved; The method of determining the single peak degree of a feature point by combining the sharpness and the amplitude of the feature point includes: taking the ratio of the amplitude of the feature point to the maximum amplitude in the A-scan waveform as the amplitude coefficient of the feature point; calculating the product of the sharpness and the amplitude coefficient, and normalizing the product to obtain the single peak degree of the feature point. The step of determining the defect stability of a feature point based on the difference in the degree of single peak between the feature point and its control point and the difference in the sampling sequence interval includes: calculating the absolute value of the difference in the degree of single peak between the feature point and any control point, and normalizing it as a peak influence index between the feature point and the corresponding control point; determining the sampling sequence interval between the feature point and any control point, and normalizing the negative number of the sampling sequence interval as a time-series influence index; and combining the peak influence index and the time-series influence index between the feature point and each control point to determine the defect stability of the feature point. The method of determining the defect stability of a feature point by combining the peak influence index and time-series influence index of the feature point with each control point includes: calculating the product of the peak influence index and time-series influence index of the feature point and the same control point, normalizing it to obtain the defect analysis coefficient of the feature point and the corresponding control point; and taking the mean of the defect analysis coefficients of the feature point and all control points as the defect stability of the feature point.

2. The non-destructive testing method for weld defects in metal structural components as described in claim 1, characterized in that, The step of determining the sharpness at the location of a feature point based on the signal difference between the feature point and its adjacent maxima and the sequence interval includes: The two nearest maxima points on either side of the maximum point sequence are taken as adjacent points; Based on the difference in signal amplitude between the feature point and each adjacent point and the sequence interval, the sharpness analysis index of the feature point and its corresponding adjacent points is determined. The mean of the sharpness index of the feature point and all its neighboring points is used as the sharpness at the location of the feature point.

3. The non-destructive testing method for weld defects in metal structural components as described in claim 2, characterized in that, The step of determining the sharpness analysis index of a feature point and its corresponding neighboring points based on the signal amplitude difference and sequence interval between the feature point and each neighboring point includes: Calculate the signal amplitude difference and sequence interval number between the feature point and each adjacent point. Normalize the ratio of the signal amplitude difference to the sequence interval number and use it as the sharpness analysis index between the feature point and its corresponding adjacent points.

4. The non-destructive testing method for weld defects in metal structural components as described in claim 1, characterized in that, The step of determining the defect manifestation probability of each feature point based on the defect stability and single peak degree of all feature points in the A-scan waveform at the same angle includes: The product of the defect stability and the single peak degree of the feature point is normalized and used as the defect manifestation probability of the feature point.

5. The non-destructive testing method for weld defects in metal structural components as described in claim 1, characterized in that, The feature points for screening non-fusion defects by combining defect performance probabilities include: Feature points whose defect manifestation probability is greater than a preset probability threshold are used as feature points of unfused defects.

6. The non-destructive testing method for weld defects in metal structural components as described in claim 1, characterized in that, The ultrasonic detector is located directly above the measuring point and includes an adjustable angle probe. The angle is the deflection angle of the ultrasonic detector towards the direction of the weld, and the angle range is from 45° to 70°. An A-scan waveform is acquired every 1°.

7. A non-destructive testing system for weld defects in metal structural components, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

Images