A method for imaging and quantifying cracks on a metal surface excited by an arbitrarily oriented beam of eddy currents

Abstract

The application discloses a metal surface crack imaging and quantification method excited by an induced current beam in an arbitrary direction, and belongs to the field of nondestructive testing. The method applies a uniform in-plane induced current beam in an arbitrary direction on a metal surface to be detected, arranges a magnetic sensor array perpendicularly to the induced current and scans along the parallel direction of the induced current, and collects a normal component disturbance signal of an induced magnetic field caused by a crack. A pure crack magnetic field feature is extracted by subtracting a no-crack reference data from a to-be-detected data, a high-resolution feature matrix is constructed by a double cubic spline interpolation, and a surface imaging map is generated. A first-order difference, threshold screening and non-maximum suppression are used to extract crack end point polarity feature points, feature point pairing is completed by combining physical constraints such as polarity and orthogonal azimuth angle, and the crack length and strike angle are accurately calculated. The application does not need to preposition the crack strike, can realize multi-crack imaging and quantitative identification through one-time scanning, has high detection efficiency and strong anti-interference capability, and is suitable for rapid detection and evaluation of metal component surface cracks.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing of industrial equipment, and in particular to a method mainly used for imaging and quantifying cracks on metal surfaces. Background Technology

[0002] With the development of modern manufacturing, industrial production equipment generally operates continuously for extended periods, making it prone to fatigue cracks on the surface of parts. These cracks are often difficult to detect visually due to coatings or oil stains, leading to structural failure and impacting equipment safety. Therefore, regular inspection of critical load-bearing components is necessary to mitigate safety risks. Furthermore, estimating the remaining lifespan of equipment relies on the quantitative detection of surface cracks. In mainstream electromagnetic detection methods, quantitative crack identification depends on operators using a handheld single-point probe to scan along the crack length, often requiring prior inspection of the entire surface to determine crack location and direction, thus affecting detection efficiency. Secondly, the detection signal is affected by the alignment of the crack direction with the probe trajectory, making rapid and accurate quantitative detection difficult. In high-risk industries such as energy and chemicals, where crack tolerance is extremely low, the only way to compensate for technological deficiencies is to increase the number of inspections. This necessitates technological innovation in existing detection methods to improve detection efficiency and quantitative accuracy, reducing the probability of missed detections.

[0003] Therefore, by using a linear array sensor to scan in induced current, magnetic field response data within a surface region is acquired, and a high-resolution magnetic field matrix is ​​constructed through interpolation. This allows for the identification of magnetic field disturbance signals when the crack is swept, thereby achieving quantitative inversion of crack characteristics and surface imaging of crack magnetic field response characteristics. Summary of the Invention

[0004] 1. Purpose of the invention This invention aims to address one of the technical problems in related technologies by providing a method for imaging and quantifying metal surface cracks by applying an induced current beam excitation in any direction. By applying an induced current beam excitation in any direction, the linear array of magnetic sensors is controlled to scan parallel to the induced current direction, acquiring the magnetic field response characteristics within the scanned area. This avoids the need for pre-screening of crack locations and reliance on the alignment of the scanning trajectory required by traditional single-point probes for crack quantification. A high-resolution interpolation matrix is ​​constructed from the acquired data to achieve magnetic field surface imaging of the scanned area. Combined with physical information constraints, the characteristics of the crack endpoints in the matrix are found, enabling rapid inversion of crack length and deflection angle, improving detection efficiency and reducing the risk of missed detections.

[0005] 2. Technical Solutions The method applies an induced current in any direction to the surface being tested, and scans along the direction parallel to the induced current using a magnetic sensor array. When the array passes over the crack location, it captures the induced magnetic field disturbance caused by the distortion of the induced current, constructs a surface imaging distribution of the induced magnetic field disturbance, and extracts the location of the disturbance signal to achieve imaging and quantitative identification of the crack.

[0006] Furthermore, the aforementioned method for imaging and quantifying metal surface cracks by applying an induced current beam in any direction is based on the fact that the magnetic field response of the specimen surface is a characterization of the surface crack morphology, and the extreme point of the normal magnetic flux density of the specimen surface magnetic field response is the physical endpoint of the surface crack.

[0007] Furthermore, the method characterizes the crack endpoints by extracting positive and negative polarity feature points, including the following steps: Step 1-1. Generation of in-plane induced current beam An excitation unit consisting of a coil and a magnetic core is constructed. Alternating current is passed through the coil to generate an alternating magnetic field. The direction of the magnetic field is parallel to the surface being measured, generating an induced current that is perpendicular to the direction of the magnetic field and uniformly distributed on the surface. Step 1-2. Sensor Arrangement and Scanning Multiple magnetic sensors are arranged in a linear, equally spaced array perpendicular to the direction of the induced current and placed between the surface being measured and the excitation unit. The sensor array is controlled to scan parallel to the direction of the induced current. The scanning path has equally spaced path nodes, with the spacing between the path nodes much smaller than the spacing between the sensors. Each sensor pauses briefly when passing a path node, recording a segment of the waveform of the normal magnetic flux density. A waveform in phase with the induced current is positive, and a 180-degree phase difference is negative. The root mean square (RMS) value is calculated and assigned a sign, denoted as B. z ; Steps 1-3. Data Measurement Prepare a calibration part with the same material as the part to be tested. As described in steps 1-2, obtain the scan data of the crack-free reference state. Then, as described in steps 1-2, obtain the scan data of the current part to be tested. Steps 1-4. Data Matrix Processing Based on the baseline data and test part data obtained in steps 1-3, construct the baseline data matrix A and the test part data matrix B. The two matrices are identical, m rows and n columns, where m represents the number of sensors and n represents the number of path nodes. Subtract the calibration sample matrix A from the test part matrix B to obtain the crack feature matrix C. Generate a curve group S by row from matrix C, with a total of m curves. The resulting curve group S is used for qualitative observation and preliminary judgment of cracks. Keeping n constant, perform bicubic spline interpolation on matrix C to increase the number of rows m to m1, so that m1 / total length of sensor linear array = n / scanning path length = path node spacing, to obtain the crack feature interpolation matrix D. Use matrix D to directly generate an image, and map the element values ​​to colors. Steps 1-5. Feature point extraction Perform a spatial first-order difference operation on the crack feature interpolation matrix obtained in steps 1-4, and iterate through the matrix elements where the first-order difference is zero and the difference signs between adjacent elements are reversed as preliminary candidates for feature points. Further, use a preset protrusion force threshold H... th Calculate the relative amplitude difference Δh between each candidate feature point and its local envelope reference plane, and filter out those that do not satisfy |Δh|>H. th The feature points are then subjected to non-maximum suppression. A suppression radius R is defined. The elements in the feature point set are sorted in descending order of absolute value. The first element is regarded as the extreme value center. All elements within the radius R are removed. Then, among the remaining elements that have not been removed, the element with the largest absolute value is regarded as the extreme value center. All elements within the radius R are removed. This process is repeated in sequence to obtain the final set of remaining feature points. Positive values ​​in the set of remaining feature points are positive polarity feature points, and negative values ​​are negative polarity feature points. Furthermore, positive and negative polarity feature points characterize the crack endpoints, and the line segment connecting two feature points characterizes the basic geometric features of the crack. The feature points of any real crack satisfy a defined spatial arrangement direction, which is constrained by the physical information of the electromagnetic field spatial distribution. This physical information constraint is used to pair the feature points of all cracks within the scanning domain, specifically including the following physical constraint conditions: Constraint 1. Crack polarity characteristics When the induced current passes around the two ends of the crack, the directions of the current are opposite. A real crack corresponds to a set of feature points with opposite polarities. That is, in the set of remaining feature points described in claim 3, only one positive polarity feature point and one negative polarity feature point are allowed to be paired to represent the crack endpoint. Two feature points that are both positive polarity feature points or both negative polarity feature points are not allowed to represent a crack. Constraint 2. Orthogonal azimuth constraint A crack impedes the flow of induced current, causing it to bypass at its endpoint. This condition only holds true when the crack orientation and the direction of the induced current have a certain orthogonal component. An angular tolerance threshold θ is set. th The angle θ between the line segment connecting the two feature points and the direction of the induced current satisfies θ th ≤θ≤180°-θ th ; Constraint 3. Deterministic distribution of characteristic point polarity A Cartesian coordinate system is established on the surface being tested. The induced current is positive when it is along the positive Y direction and negative when it is along the negative Y direction. When the induced current is positive, the crack endpoint in the negative X direction generates a reverse distorted magnetic field, and the crack endpoint in the positive X direction generates a unidirectional distorted magnetic field. When the induced current is negative, the crack endpoint in the negative X direction generates a unidirectional distorted magnetic field, and the crack endpoint in the positive X direction generates a reverse distorted magnetic field. In the frequency domain, at the crack endpoint in the negative X direction, B... zThe signal and the induced current signal are 180° out of phase, and the crack is located at point B in the positive X direction. z Since the signal and the induced current signal are in phase, only negative polarity feature points are allowed to represent the endpoints in the negative X direction, and positive polarity feature points are allowed to represent the endpoints in the positive X direction. Furthermore, the method pairs positive and negative polarity feature points under physical information constraints, including the following steps: Step 2-1 Global feature point pairing Traverse the remaining feature point set obtained in steps 1-5, construct all combinations of positive and negative polarity feature point pairing lines that meet the physical information constraints described in claim 4, select the combination with the smallest total line length among all combinations, and determine it as the final feature point pairing scheme. The global distance minimization method ensures that the feature point pairing lines in space do not intersect. For isolated feature points that fail to pair successfully in the end, they are regarded as noise and discarded. Step 2-2 Crack geometric parameter calculation Explicit calculations are performed on the paired feature points. The coordinates of the paired positive polarity feature points are (x1, y1), and the coordinates of the negative polarity feature points are (x2, y2). The formulas for calculating the crack length L and the angle θ between the crack and the induced current are as follows: The above technical solution has the following advantages or beneficial effects: This invention provides a method for quantitative detection and surface imaging of metal surface cracks based on the magnetic field response to an induced current beam applied in any direction. This method solves the problems of inconvenient qualitative and quantitative crack identification, low efficiency, and difficulty in visually displaying crack contours in existing technologies. Specifically, it addresses these issues by providing a method for quantitative and qualitative crack identification based on the magnetic field response to an induced current beam applied in any direction. This invention provides a method for detecting a surface region. Compared with the single-point detection probe of traditional AC electromagnetic field detection, it provides a larger detection range and can achieve two-dimensional magnetic field imaging without subsequent point stitching. As long as the crack exists in the scanning domain, a high-resolution magnetic field matrix can be generated in one scan and the crack position, length and angle can be deduced. It can detect multiple cracks in the scanning domain at the same time and can effectively suppress the mutual interference of different cracks.

[0008] This invention provides a complementary detection method that combines crack contour images with quantitative inversion of crack features. By combining the recognition of the original scan curve group and the two-dimensional magnetic field image, when the quantitative inversion method of crack features fails under special working conditions, manual visual qualitative detection can be performed, solving the problem of single result data in traditional detection and reducing the probability of missed detection when quantitative failure occurs.

[0009] This invention provides a novel scanning path method. Traditional AC electromagnetic field detection technology requires aligning the probe scanning path with the crack direction when quantitatively detecting cracks. This necessitates multiple probes to estimate the approximate crack direction and location, aiming to maximize the overlap between the crack direction and the scanning path. This method specifies that the scanning path of the sensor linear array is parallel to the direction of the induced current, detecting the crack's disturbance to the current. As long as the crack is within the scanning domain and the angle between it and the induced current is greater than the angle tolerance threshold θ, the method will work. th This allows for quantitative detection of cracks. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of a quantitative crack detection model in some embodiments of the present invention; Figure 2 The curve group generated row by row from the reference data matrix A in some embodiments of the present invention; Figure 3 This refers to the curve group generated row by row from the data matrix B of the part under test in some embodiments of the present invention; Figure 4 This refers to a set of curves generated row by row from the crack feature matrix C in some embodiments of the present invention; Figure 5 This refers to a set of curves generated row-wise from the crack feature interpolation matrix D in some embodiments of the present invention. Figure 6 Images generated for the crack feature matrix C in some embodiments of the present invention; Figure 7 The image generated for the crack feature interpolation matrix D in some embodiments of the present invention; Figure 8 These are the feature point pairing and crack quantification results in some embodiments of the present invention. Detailed Implementation

[0011] The present invention will be further described below with reference to the accompanying drawings and specific examples.

[0012] During implementation, a sinusoidal alternating current with a frequency of 5kHz is applied to the excitation unit. The excitation unit is placed 2mm from the surface of the part, inducing a current on the surface. The direction of the induced current is denoted as Y, the direction perpendicular to the induced current is denoted as X, and the direction perpendicular to the measured surface is denoted as Z. Multiple sensors are arrayed at equal intervals along the X direction, with the array range being the projection range of the excitation unit onto the part surface in the X direction. The sensor array scans along the Y direction, with the scanning range also being the projection range of the excitation unit onto the part surface in the Y direction. The excitation unit is 28mm in the X direction and 40mm in the Y direction. The path node spacing on the scanning path is specified as 0.1mm, and the sensor array spacing is 2mm. Each sensor needs to scan 401 nodes in the Y direction. The crack on the surface of the part under test is 8mm long, 0.2mm wide, and 1mm deep, with an angle of 70° to the Y direction. The calibration part is a crack-free part of the same material as the part under test.

[0013] The calibration part without cracks was scanned, and the resulting matrix A is shown in Table 1. The matrix size is 15×401. Curve groups were generated from matrix A row by row, and each curve represents the value scanned by one of the sensors. Figure 2 As shown, the curve trend within the scanned domain is smooth, with small numerical changes and no abrupt changes, but the magnetic field gradient is larger at the edge.

[0014] Table 1. Baseline Data Matrix A The part to be tested, which may have cracks, is scanned, and the resulting matrix B is shown in Table 2. The matrix size is 15×401. Curve groups are generated from matrix B row by row, and each curve represents the value scanned by one of the sensors. Figure 3 As shown, the curve has peaks and troughs in many places, which initially form positive and negative polarity characteristic points.

[0015] Table 2 Data Matrix of Parts to be Tested (B) To reduce interference from the background magnetic field, lower the strong background gradient at the edges, and make the magnetic field disturbance caused by the crack features clearer, the reference data matrix A is subtracted from the data matrix B of the part under test to obtain the crack feature matrix C, as shown in Table 3. Curve sets are generated row-wise from matrix C, and the resulting curves represent the magnetic field of the pure crack disturbance after removing the background magnetic field. Figure 4 As shown, the curve group retains complete feature point information, and the values ​​tend to be 0 at non-crack locations, eliminating interference from the background magnetic field. An image is generated by mapping the values ​​of matrix C to colors, as shown... Figure 6 As shown, the crack feature image at the original resolution is obtained.

[0016] Table 3 Crack Feature Matrix C Interpolating the crack feature matrix C with a path node spacing of 0.1 mm, the interpolated matrix D corresponds to a row spacing of 0.1 mm in the actual scanning domain. Therefore, the interpolated matrix D has 281 rows and retains 401 columns. The resulting crack feature interpolation matrix D is shown in Table 4. High-density curve groups are then generated from matrix D row by row, as shown below. Figure 5 As shown, an image is generated by mapping the values ​​of matrix D to colors, as follows. Figure 7 As shown, the interpolated crack feature image is obtained.

[0017] Table 4 Crack Feature Interpolation Matrix D Based on the interpolated matrix data, feature points are searched and paired. First-order difference operations are used to find matrix elements with zero first-order difference and inverted sign between adjacent elements, serving as preliminary candidate feature points. This yields 8388 maxima and 7758 minima, as shown in Table 5. A protrusion force threshold H is set. th =0.4, calculate the relative amplitude difference Δh between each point in the candidate feature points and its local envelope reference plane.

[0018] Table 5 Results of First-Order Difference Operations Candidate points that do not satisfy |Δh|>H are removed from the list. th The feature points were obtained, and among the remaining feature points, there were 46 maxima and 45 minima, as shown in Table 6.

[0019] Table 6. Results of Protrusion Force Threshold Screening The remaining feature points are those located near the crack endpoint. Multiple feature points may exist at a single endpoint, therefore non-maximum suppression is applied to the remaining feature points. A suppression radius of R = 3 mm is defined. The elements in the feature point set are arranged in descending order of absolute value. The first element is considered the extreme center, and all elements within radius R are removed. Then, among the remaining elements that were not removed, the element with the largest absolute value is considered the extreme center, and all elements within radius R are removed. This process is repeated sequentially. In this embodiment, after 6 rounds of removal, the resulting set of remaining feature points contains 3 positive polarity feature points and 3 negative polarity feature points, as shown in Table 7.

[0020] Table 7 Nonmaximum suppression results In the remaining feature point set, based on physical information constraints, combinations of lines connecting all positive and negative polarity feature points that meet the constraints are constructed. In each group of lines, each feature point is used by only one line. The total length of each group of lines is calculated, and the shortest group is taken as the final feature point pairing scheme. The information of the three pairs of feature points is shown in Table 8. The row direction of the matrix is ​​mapped to the actual X coordinate, and the column direction is mapped to the Y coordinate. The length and angle are calculated according to the formula.

[0021] Table 8 Quantitative Results of Cracks In summary, the three cracks in the part were quantitatively analyzed using this method, and the obtained lengths were 8.1 mm, 6.71 mm, and 6.74 mm, with angles of 69.78°, 69.94°, and 69.15° to the current direction, respectively. The errors between the quantitative results and the actual crack sizes are within acceptable limits, with an average length error of 11.04% and an average angle error of 0.54%.

Claims

1. A method for imaging and quantifying metal surface cracks by applying an induced current beam in any direction, characterized in that, This method applies an induced current in any direction to the surface being tested, and scans along the direction of the induced current synchronously with the magnetic sensor array. When the array passes over the crack location, it captures the disturbance of the induced magnetic field caused by the distortion of the induced current, constructs a surface imaging distribution of the induced magnetic field disturbance, and extracts the location of the disturbance signal to achieve imaging and quantitative identification of the crack.

2. The method for imaging and quantifying metal surface cracks by applying an induced current beam in any direction as described in claim 1, characterized in that, The method is based on the fact that the magnetic field response of the specimen surface is a characterization of the surface crack morphology, and the extreme point of the normal magnetic flux density of the magnetic field response of the specimen surface is the physical endpoint of the surface crack.

3. The method for imaging and quantifying metal surface cracks by applying an induced current beam in any direction as described in claim 2, characterized in that, The method characterizes the crack endpoints by extracting positive and negative polarity feature points, and includes the following steps: Step 1-1. Generation of in-plane induced current beam An excitation unit consisting of a coil and a magnetic core is constructed. Alternating current is passed through the coil to generate an alternating magnetic field. The direction of the magnetic field is parallel to the surface being measured, generating an induced current beam that is perpendicular to the direction of the magnetic field and uniformly distributed on the surface. Step 1-2. Sensor Arrangement and Scanning Multiple magnetic sensors are arranged in a linear, equally spaced array perpendicular to the direction of the induced current and placed between the surface being measured and the excitation unit. The sensor array is controlled to scan parallel to the direction of the induced current. The scanning path has equally spaced path nodes, with the spacing between the path nodes much smaller than the spacing between the sensors. Each sensor pauses briefly when passing a path node, recording a segment of the waveform of the normal magnetic flux density. A waveform in phase with the induced current is positive, and a 180-degree phase difference is negative. The root mean square (RMS) value is calculated and assigned a sign, denoted as B. z ; Steps 1-3. Data Measurement Prepare a calibration part with the same material as the part to be tested. As described in steps 1-2, obtain the scan data of the crack-free reference state. Then, as described in steps 1-2, obtain the scan data of the current part to be tested. Steps 1-4. Data Matrix Processing Based on the baseline data and test part data obtained in steps 1-3, construct the baseline data matrix A and the test part data matrix B. The two matrices are identical, m rows and n columns, where m represents the number of sensors and n represents the number of path nodes. Subtract the calibration sample matrix A from the test part matrix B to obtain the crack feature matrix C. Generate a curve group S by row from matrix C, with a total of m curves. The resulting curve group S is used for qualitative observation and preliminary judgment of cracks. Keeping n constant, perform bicubic spline interpolation on matrix C to increase the number of rows m to m1, so that m1 / total length of sensor linear array = n / scanning path length = path node spacing, to obtain the crack feature interpolation matrix D. Use matrix D to directly generate an image, and map the element values ​​to colors. Steps 1-5. Feature point extraction Perform a spatial first-order difference operation on the crack feature interpolation matrix obtained in steps 1-4, and iterate through the matrix elements where the first-order difference is zero and the difference signs between adjacent elements are reversed as preliminary candidates for feature points. Further, use a preset protrusion force threshold H... th Calculate the relative amplitude difference Δh between each candidate feature point and its local envelope reference plane, and filter out those that do not satisfy |Δh|>H. th The feature points are then subjected to non-maximum suppression. A suppression radius R is defined. The elements in the feature point set are arranged in descending order of absolute value. The first element is regarded as the extreme value center. All elements within the radius R are removed. Then, among the remaining elements that have not been removed, the element with the largest absolute value is regarded as the extreme value center. All elements within the radius R are removed. This process is repeated in sequence to obtain the final set of remaining feature points. Positive values ​​in the set of remaining feature points are positive polarity feature points, and negative values ​​are negative polarity feature points.

4. The method for imaging and quantifying metal surface cracks by applying an induced current beam in any direction as described in claim 3, characterized in that, Positive and negative polarity feature points characterize the crack endpoints, and the line segment connecting two feature points characterizes the basic geometric features of the crack. The feature points of any real crack satisfy a defined spatial arrangement direction, which is constrained by the physical information of the electromagnetic field spatial distribution. This physical information constraint is used to pair the feature points of all cracks within the scanning domain, specifically including the following physical constraint conditions: Constraint 1. Crack polarity characteristics When the induced current passes around the two ends of the crack, the directions of the current are opposite. A real crack corresponds to a set of feature points with opposite polarities. That is, in the set of remaining feature points described in claim 3, only one positive polarity feature point and one negative polarity feature point are allowed to be paired to represent the crack endpoint. Two feature points that are both positive polarity feature points or both negative polarity feature points are not allowed to represent a crack. Constraint 2. Orthogonal azimuth constraint A crack impedes the flow of induced current, causing it to bypass at its endpoint. This condition only holds true when the crack orientation and the direction of the induced current have a certain orthogonal component. An angular tolerance threshold θ is set. th The angle θ between the line segment connecting the two feature points and the direction of the induced current satisfies θ th ≤θ≤180°-θ th ; Constraint 3. Deterministic distribution of characteristic point polarity A Cartesian coordinate system is established on the surface being tested. The induced current is positive when it is along the positive Y direction and negative when it is along the negative Y direction. When the induced current is positive, the crack endpoint in the negative X direction generates a reverse distorted magnetic field, and the crack endpoint in the positive X direction generates a unidirectional distorted magnetic field. When the induced current is negative, the crack endpoint in the negative X direction generates a unidirectional distorted magnetic field, and the crack endpoint in the positive X direction generates a reverse distorted magnetic field. In the frequency domain, at the crack endpoint in the negative X direction, B... z The signal and the induced current signal are 180° out of phase, and the crack is located at point B in the positive X direction. z Since the signal is in phase with the induced current signal, only negative polarity feature points are allowed to represent the endpoints in the negative X direction, and positive polarity feature points are allowed to represent the endpoints in the positive X direction.

5. The method for imaging and quantifying metal surface cracks by applying an induced current beam excitation in any direction according to claim 4, characterized in that, The method pairs positive and negative polarity feature points under physical information constraints, including the following steps: Step 2-1 Global feature point pairing Traverse the remaining feature point set obtained in steps 1-5, construct all combinations of positive and negative polarity feature point pairing lines that meet the physical information constraints described in claim 4, select the combination with the smallest total line length among all combinations, and determine it as the final feature point pairing scheme. The global distance minimization method ensures that the feature point pairing lines in space do not intersect. For isolated feature points that fail to pair successfully in the end, they are regarded as noise and discarded. Step 2-2 Crack geometric parameter calculation Explicit calculations are performed on the paired feature points. The coordinates of the paired positive polarity feature points are (x1, y1), and the coordinates of the negative polarity feature points are (x2, y2). The formulas for calculating the crack length L and the angle θ between the crack and the induced current are as follows: 。

Images