Eddy current detection reliability evaluation method and system based on signal space distribution characteristics

Abstract

The invention discloses an eddy current testing reliability evaluation method and system based on signal space distribution characteristics, and belongs to the technical field of nondestructive testing. According to the method, eddy current signal space distribution characteristics and the detection probability (POD) and the false detection rate (PFA) in nondestructive testing reliability evaluation indexes are fused, and the method is used for comprehensively evaluating the defect detection capacity and the reliability of a detection system. By introducing the Gini coefficient to quantitatively detect the imbalance of signal space distribution, the method effectively characterizes the distinguishability of signals in a defect state and a defect-free state, and replaces traditional POD and PFA analysis methods based on single-point amplitude or phase characteristics. Compared with a traditional method, the Gini coefficient can more comprehensively represent the overall distribution characteristics of the signals, the influence of noise is small, and the method is particularly suitable for the eddy current detection environment under the condition of the low signal-to-noise ratio. Besides, a POD and PFA relation curve is constructed based on Gini coefficients, ROC analysis is performed, the defect detection rate can be effectively increased, the false detection rate can be reduced, and the reliability of the nondestructive testing system can be evaluated more efficiently and accurately.

Description

Technical Field

[0001] This invention belongs to the field of nondestructive testing (NDT) technology, specifically to a method and system for evaluating the reliability of eddy current testing based on the spatial distribution characteristics of signals. This method and system are particularly suitable for detecting and quantitatively evaluating defects in conductive materials and are widely applicable to industries with stringent requirements for structural safety and reliability, such as aerospace, nuclear power, rail transportation, and petrochemicals. Background Art

[0002] As a mature non-destructive testing technology, eddy current testing has been widely used in defect detection of various conductive engineering components. Its basic principle is that when a component is damaged (such as cracks, corrosion) and causes local conductivity changes, the eddy current probe in the scan can sense it and generate a corresponding detection signal response. The key is that the eddy current field disturbance caused by the defect and its corresponding detection signal change will show a specific spatial distribution feature on the probe scanning surface. This spatial distribution contains rich information about the geometric parameters of the defect (such as position, shape, size, orientation) and its physical properties. However, traditional eddy current testing reliability assessment methods have significant limitations. These methods mainly rely on single-point characteristic parameters extracted from the detection signal (such as signal amplitude, phase). When dealing with defects with significant spatial ductility or complex geometric shapes (such as irregular cracks, dense corrosion pit groups, and weld area defects), the shortcomings of such methods are particularly prominent:

[0003] (1) Lack of spatial information: Single-point features only reflect the instantaneous signal value at a specific location and cannot effectively characterize the spatial distribution characteristics of the signal caused by the defect in the entire scanning area (such as gradient changes, texture structure, regional intensity differences, energy concentration patterns, etc.);

[0004] (2) Insufficient characterization capability: The disturbance of the eddy current field by complex defects is essentially multidimensional in space. Single-point features are difficult to fully capture this multidimensional spatial effect, resulting in incomplete defect information extraction and limited accuracy;

[0005] (3) Reliability assessment bias: Reliability assessment models based on single-point features are unable to accurately quantify key indicators such as the probability of detection (POD) and the rate of false positives (PFA) for complex defects. This severely limits the applicability and confidence of the assessment results, ultimately impacting the accuracy of detection decisions.

[0006] The Gini coefficient is a classic statistical indicator whose core value lies in quantifying the degree of imbalance in the distribution of resources or attributes across groups. It originates from the study of income inequality in economics and is defined as the area between the Lorenz curve and the line of absolute equality. Its mathematical nature makes it a powerful, standardized tool for measuring distribution imbalance, capable of condensing complex spatial distribution differences into a single, standardized, easily comparable, and interpretable value. Despite its significant advantages and universal principles in quantifying spatial distribution imbalances, the Gini coefficient has a significant gap in its application in the field of eddy current testing technology:

[0007] (1) Lack of systematic application: The existing technology lacks research and practice on systematically applying the Gini coefficient to quantify the spatial distribution imbalance characteristics of eddy current detection signals caused by defects;

[0008] (2) Model construction gap: The method and system for constructing a non-destructive testing reliability assessment model using the Gini coefficient as the core input parameter is still a gap in the existing technology.

[0009] Therefore, an innovative solution is urgently needed to address the issues of insufficient spatial information utilization and limited reliability assessment caused by traditional methods' overreliance on single-point signal characteristics. This solution should be able to deeply explore and effectively quantify the spatial distribution characteristics of eddy current testing signals, providing a computationally efficient, adaptable, and particularly adept technical approach for handling complex spatial morphological defects, thereby directly improving the accuracy and practicality of eddy current testing system reliability assessments. The proposed eddy current nondestructive testing reliability assessment method and system based on signal spatial distribution characteristics is designed to meet this urgent need. Summary of the Invention

[0010] The technical problem to be solved by the present invention is to overcome the inherent defects of the traditional eddy current detection reliability assessment method (which essentially relies on single-point characteristic parameters, such as signal amplitude and phase): namely, the serious lack of spatial information utilization, the lack of characterization ability of multi-dimensional disturbances of complex defects, and the resulting significant deviations in reliability assessment results (such as detection probability POD and false positive rate PFA). In order to solve the above technical problems, the present invention proposes an eddy current detection reliability assessment method based on signal spatial distribution characteristics. The core innovation of the method is to introduce an indicator that can quantify the imbalance of signal spatial distribution, namely the Gini coefficient, and use it as the detection signal response characteristic to construct a reliability assessment model. Compared with traditional methods, this method significantly improves the detection rate and assessment confidence of complex morphological defects by analyzing the distribution differences of signals in the spatial domain. This method can effectively capture the spatial imbalance characteristics of defect signals under low signal-to-noise ratio conditions, thereby providing a more comprehensive and accurate technical means for the reliability assessment of eddy current detection.

[0011] The present invention solves the above technical problems through the following technical solutions, which include the following steps:

[0012] S1: Data Collection

[0013] The eddy current detection system is used to scan and detect the defective specimen and the non-defective specimen within the same area to obtain the eddy current detection signal corresponding to each sampling point;

[0014] S2: Signal post-processing

[0015] A standard specimen containing the same type of defects is selected, and the eddy current detection signal of the standard specimen is used as a reference signal, and its amplitude is normalized to 1V and the phase is 0°; the eddy current detection signal collected in step S1 is corrected to be unified to the reference standard defined by the reference signal; after completing the signal correction, the eddy current detection signal is further centered and flattened;

[0016] S3: Calculate the Gini coefficient

[0017] Sorting the signal amplitude or phase data of the defective specimen and the non-defective specimen after processing in step S2 from small to large; calculating the spatial distribution characteristics of the signal data on the specified scanning line based on the sorted signal data, and quantifying the distribution difference between the defective signal and the non-defective signal in the spatial dimension using the Gini coefficient;

[0018] S4: Reliability Assessment

[0019] The signal spatial distribution characteristics quantified by the Gini coefficient are used to replace the traditional signal response characteristics based on single-point amplitude or phase. The detection probability POD and false positive rate PFA in the detection process are calculated, the receiver operating characteristic ROC curve is drawn, and the area under the curve AUC is calculated, thereby realizing the quantitative evaluation of the reliability of the eddy current detection system.

[0020] Furthermore, in step S1, the eddy current detection signal is obtained by performing a surface scan on the test piece using an eddy current detection system; before scanning, the detection signal obtained by the surface scanning of the test piece is extracted based on the eddy current detection signal of the defect-free area of ​​the test piece.

[0021] Furthermore, in step S1, the eddy current detection signal is collected by the eddy current probe and transmitted to the phase-locked amplifier for phase-sensitive detection and signal amplification processing, and a DC signal containing amplitude and phase information is output after low-pass filtering, and the corresponding real and imaginary voltage values ​​are obtained through the phasor decomposition method.

[0022] Furthermore, in step S3, the Gini coefficient is used to quantify the distribution difference between the defect signal and the defect-free signal in the spatial dimension. The closer the value is to 0, the more similar the signal distributions of the two are, and the smaller the possibility of defects. Conversely, the larger the value is, the greater the difference in the signal distributions of the two is, and the higher the possibility of defects.

[0023] Furthermore, in step S3, the calculation formula of the Gini coefficient is as follows:

[0024]

[0025] Where N is the number of sampling intervals along the one-dimensional scan line, b and a are the maximum and minimum values ​​of the detection signal respectively, S n-1 is the defect signal amplitude corresponding to the n-1th sorting position, x n-1 is the amplitude of the defect-free signal corresponding to the n-1th sorting position, |S n-1 -x n-1 |+|S n -x n | is to calculate the absolute difference between each pair of corresponding position points of the two sets of data in the QQ graph.

[0026] Furthermore, in step S3, the Gini coefficient is used as a signal response feature. And the relationship between it and the defect size a is transformed accordingly, so that It is linearly related to a, and the corresponding transformation includes logarithmic transformation; constructed by regression analysis The linear relationship is established, and the 95% confidence interval and 95% prediction interval of the linear regression line are calculated. Then, the maximum likelihood method or the least squares method is used to estimate the parameters β0, β1, and σ in the linear model, where the parameters β0 and β1 are the intercept and slope of the linear model, and σ is the standard deviation of the fitting error. In the signal response analysis, It follows a normal distribution with a mean and standard deviation As shown in the following formula:

[0027]

[0028] Furthermore, in step S4, the ROC curve uses PFA as the horizontal coordinate and POD as the vertical coordinate to evaluate the performance of the eddy current detection system; wherein POD is defined as the probability density integral area in the area where the defect signal characteristics are greater than the detection threshold, and PFA is defined as the probability density integral area in the area where the noise signal characteristics are greater than the detection threshold; the reliability of the eddy current detection system is quantified by calculating the area under the ROC curve AUC.

[0029] Furthermore, when the parameters β0, β1, and σ are known, for a given detection threshold to determine whether a defect is detected, Defect detection probability It obeys the standard normal distribution and the calculation formula is as follows:

[0030]

[0031] The formula for calculating the false positive rate PFA is as follows:

[0032]

[0033] in, is a noise signal, which is assumed to obey normal distribution or gamma distribution;

[0034] The calculation formula of AUC value is as follows:

[0035]

[0036] Where m is the number of points on the ROC curve, x i and y i Indicates the PFA and POD corresponding to the i-th detection threshold on the ROC curve, x i+1 and y i+1 represents the PFA and POD corresponding to the i+1th detection threshold, x i+1 -x i Indicates the amplitude of PFA change, y i+1 +y i Indicates the POD value of two adjacent points on the ROC curve.

[0037] The present invention also provides an eddy current detection reliability assessment system based on signal spatial distribution characteristics, which uses the above reliability assessment method to perform reliability assessment on the eddy current detection system, including:

[0038] A signal acquisition module is used to scan and detect defective specimens and non-defective specimens within the same area using an eddy current detection system to obtain eddy current detection signals corresponding to each sampling point;

[0039] a signal post-processing module, configured to select a standard specimen containing defects of the same type, use the eddy current detection signal of the standard specimen as a reference signal, and normalize its amplitude to 1V and phase to 0°; perform correction processing on the eddy current detection signal collected in step S1 to unify it to the reference standard defined by the reference signal; and after completing the signal correction, further perform centralization and flattening processing on the eddy current detection signal;

[0040] A Gini coefficient calculation module is used to sort the signal amplitude or phase data of the defective test piece and the non-defective test piece processed in step S2 from small to large; based on the sorted signal data, calculate its spatial distribution characteristics on the specified scanning line, and quantify the distribution difference between the defective signal and the non-defective signal in the spatial dimension using the Gini coefficient;

[0041] The reliability evaluation calculation module is used to use the signal spatial distribution characteristics quantified by the Gini coefficient to replace the traditional signal response characteristics based on single-point amplitude or phase, calculate the detection probability POD and false positive rate PFA during the detection process, draw the receiver operating characteristic ROC curve, and calculate the area under the curve AUC, thereby realizing a quantitative evaluation of the reliability of the eddy current detection system.

[0042] Compared with the existing technology, the present invention has the following advantages: the eddy current detection reliability assessment method based on the spatial distribution characteristics of the signal combines the spatial distribution characteristics of the eddy current signal with the detection probability POD and the false detection rate PFA in the non-destructive testing reliability assessment indicators, and can comprehensively evaluate the defect detection capability and the reliability of the detection system; the Gini coefficient can quantify the spatial distribution characteristics of the signal, judge the discrimination degree of the detection signal in the defective and non-defective states, and replace the amplitude or phase of the traditional POD and PFA analysis to perform subsequent detection reliability assessment; compared with the amplitude or phase characteristics of a single point, the Gini coefficient is calculated based on the overall signal spatial distribution characteristics and is less affected by noise; at the same time, the Gini coefficient is used to construct the POD and PFA relationship curve, and perform ROC analysis, so as to improve the defect detection rate, reduce the false detection rate, and achieve a more efficient and accurate assessment of the reliability of the non-destructive testing system. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 1 is a flow chart of a reliability assessment method for eddy current testing based on signal spatial distribution characteristics in an embodiment of the present invention;

[0044] Figure 2 1 is a schematic block diagram of the structure of an eddy current detection system (surface scanning) in an embodiment of the present invention;

[0045] Figure 3 2 is a schematic diagram of a C-scan of eddy current detection according to an embodiment of the present invention;

[0046] Figure 4 Schematic diagram of the distribution of POD and PFA in an embodiment of the present invention;

[0047] Figure 5 Schematic diagram of the ROC curve based on POD and PFA in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process. However, the protection scope of the present invention is not limited to the following embodiment.

[0049] Example 1

[0050] like Figure 1 As shown, this embodiment provides a technical solution: an eddy current detection reliability assessment method based on signal spatial distribution characteristics. This method uses an eddy current detection system to perform surface scanning detection on the test piece (defective and non-defective test pieces), obtains the amplitude, phase and other characteristics of the detection signal in the scanning area, and then performs necessary post-processing on the acquired signal. The distribution difference between the defect signal and the non-defective signal in the spatial dimension is quantified by the Gini coefficient, replacing the signal response characteristics that rely on single-point amplitude or phase in the traditional reliability assessment method (POD and PFA analysis) for subsequent detection reliability assessment. This method can more comprehensively reflect the impact of defects on the detection field, and the evaluation results are more reliable and objective, and the evaluation process is efficient and intuitive. Specifically comprising the following steps:

[0051] Step 1: Eddy current detection signal acquisition

[0052] like Figure 2 As shown in Figure 1, the eddy current testing system consists of two parts: motion control and data acquisition and processing. The motion control part includes a motion controller and a displacement motion platform. The data acquisition and processing part mainly includes a signal generator, a lock-in amplifier, an eddy current probe, a junction box, and a host computer (computer) with a built-in data acquisition card. The host computer runs software developed based on LabVIEW. Its core functions are: controlling the motion of the displacement motion platform; acquiring, storing, and displaying eddy current testing signals.

[0053] The test piece includes several defective and non-defective test pieces. This embodiment takes a defective test piece with a size of 135mm×150mm×6mm as an example. Figure 3 As shown in the figure, the specimen has six defects machined onto its surface. These defects are 1mm, 3mm, 5mm, 7mm, 10mm, and 15mm long, 0.5mm wide, and 0.5mm deep, 1mm, 2mm, 3mm, and 4mm deep. All defects were inspected using surface scanning within the same area.

[0054] The eddy current probe consists of an excitation coil and a receiving coil. During testing, the specimen is placed under the probe. Before scanning, the host computer drives the displacement motion platform through the motion controller to move the probe to the non-destructive area on the specimen surface (the starting point of the scan), and through the automatic bias adjustment function of the phase-locked amplifier, the real and imaginary voltage signals at this point are set to zero. Subsequently, the probe is Figure 3The single-pass interlaced scanning trajectory shown performs a surface scan of a specified area of ​​the specimen while simultaneously acquiring signals. The scanning process is as follows: the probe continuously moves in the positive X-axis direction (forward travel) and acquires a predetermined number of samples. Upon reaching the X-axis endpoint, it returns along the original path (return travel, no data acquisition). Subsequently, it moves a predetermined distance in the Y-axis direction. This process constitutes one scanning cycle. This scanning cycle is repeated until the specified area is scanned, and the probe finally returns to the scanning starting point.

[0055] During this process, the detection signal collected by the probe is input into a lock-in amplifier for phase-sensitive detection and amplification. After low-pass filtering, it outputs a quasi-DC signal containing the detection signal's amplitude and phase information. Based on the principle of phasor decomposition, the lock-in amplifier outputs real (X) and imaginary (Y) voltage components corresponding to the changes in the detection coil's resistance and inductive reactance, respectively. These are then transmitted by a signal acquisition card to a host computer for relevant post-processing and defect imaging.

[0056] Step 2: Signal post-processing

[0057] In order to adjust the detection signal to a unified reference standard, signal calibration is required: select a standard specimen containing the same type of defects, collect the signal of the standard specimen as the reference signal, normalize the amplitude of the reference signal to 1V and the phase to 0°, and use this reference signal to calibrate all eddy current detection signals collected in step 1. After the calibration is completed, the signal is further processed to improve consistency and comparability. Centering processing: The origin of the signal data is translated to the origin of the Cartesian coordinate system (0,0) to eliminate the DC bias of the acquisition equipment; Flattening processing: Eliminate the linear trend term in the signal (mainly caused by factors such as probe tilt).

[0058] Step 3: Calculate the Gini coefficient

[0059] After signal post-processing in step 2, signal amplitude or phase data is obtained for the scanned areas of defective and non-defective specimens. The corresponding feature data for each area is then sorted from smallest to largest. The Gini coefficient is used to quantify the spatial distribution imbalance reflected by these sorted feature data. This is done by calculating the cumulative deviation between the detection signal distribution of defective specimens and the detection signal distribution of non-defective specimens on a given scan line.

[0060]

[0061] Where N is the number of sampling intervals along the one-dimensional scan line, b and a are the maximum and minimum values ​​of the detection signal respectively, S n-1 is the defect signal amplitude corresponding to the n-1th sorting position, x n-1 is the amplitude of the defect-free signal corresponding to the n-1th sorting position, |S n-1 -x n-1 |+|Sn -x n This calculates the absolute difference between each pair of corresponding points in the QQ plot. It should be noted that the above calculation method is also applicable to other signal features, such as the phase feature of the detection signal.

[0062] The Gini coefficient is used to quantify the distribution difference between defective signals and non-defective signals in the spatial dimension. The closer its value is to 0, the more similar the signal distributions of the two are and the smaller the defect possibility is. Conversely, the larger its value is, the greater the difference in the signal distributions of the two is and the higher the defect possibility is.

[0063] Step 4: Reliability Assessment

[0064] The core of this embodiment is to use the Gini coefficient to quantify the spatial distribution characteristics of the eddy current signal, and to replace the traditional eddy current detection reliability assessment based on single-point signal (amplitude or phase) analysis method to evaluate the probability of defect detection POD and false positive rate PFA, and then comprehensively evaluate the defect detection capability and the reliability of the detection system. The Gini coefficient calculated in step 3 is used as the response quantity to characterize the spatial distribution characteristics of the signal. Select a single defect parameter a (such as length or depth) as the characteristic quantity to characterize the defect size, and establish A linear regression model is built between α and α, and the parameters β0, β1, and σ in the linear model are estimated using the maximum likelihood method or the least squares method. The 95% confidence interval and 95% prediction interval of the regression line are calculated. The interval estimation method can use the Wald method, but is not limited to this. The Bootstrap method or a method based on the t distribution can also be used.

[0065] Due to the randomness of the detection process, the signal response Probability distribution around the regression line (usually normal distribution, see Figure 4 ). Therefore, a detection threshold is set Used to determine whether a defect is detected. Based on this, the probability of detection (POD) is defined as the response value of the real defect signal. Greater than the detection threshold The probability of . Figure 4 In the probability density function diagram, POD corresponds to the detection threshold under the defect signal distribution curve. The area of ​​the right region (grey shaded area in the figure). False positive rate (PFA): defined as the response value of the noise signal Greater than the detection threshold The probability of . Figure 4 In the above example, PFA corresponds to the detection threshold under the noise distribution curve. The area of ​​the right region (the red shaded area in the figure).

[0066] In signal response analysis, the effective range of the response value needs to be set according to the actual situation. If the signal response is lower than the detection capability of the system (such as the defect signal is completely submerged in the background noise), it is considered to be below the minimum threshold. If the signal response exceeds the maximum value that the detection system can record (saturation), these signals may not be effectively recorded or require special processing. Used to determine whether defects of a specific size are detected. Its setting requires a balance between POD and PFA: increasing the threshold can reduce PFA (reduce false positives), but usually also reduces POD (increases false negatives); and vice versa.

[0067] It should be noted that this embodiment uses the analysis of a single defect parameter (length or depth) as an example. However, the method of the present invention is not limited to this. By introducing a multivariate normal distribution model, it can be extended to POD and PFA analysis based on multiple defect parameters (e.g., considering both length and depth).

[0068] When the parameters β0, β1, σ ε After knowing the detection threshold for determining whether a defect is detected, Defect detection probability It obeys the normal distribution and the calculation formula is:

[0069]

[0070] Among them, the parameters β0 and β1 are the intercept and slope of the regression model, and σ is the standard deviation of the fitting error.

[0071] The formula for calculating the false positive rate PFA is:

[0072]

[0073] in, is a noise signal, which is assumed to obey normal distribution or gamma distribution.

[0074] The ROC curve is drawn with the false positive rate PFA as the horizontal axis and the probability of detection POD as the vertical axis. This curve not only intuitively shows the relationship between POD and PFA under different detection thresholds, but also the area under the curve AUC can quantitatively evaluate the overall reliability of the detection system. Figure 4 Detection threshold shown By calculating the corresponding POD and PFA values, a ROC curve can be generated, such as Figure 5 As shown. The closer the ROC curve is to the upper left corner of the coordinate system (i.e. the higher the POD value and the lower the PFA value), the higher the reliability of the detection system. Figure 5Of the five ROC curves shown, curve 5 (closest to the upper left corner) represents the detection system with the highest reliability.

[0075] After drawing the ROC curve, the area under the ROC curve AUC can be calculated. The calculation formula is:

[0076]

[0077] Where m is the number of points on the ROC curve, x i and y i Indicates the PFA and POD corresponding to the i-th detection threshold on the ROC curve, x i+1 and y i+1 represents the PFA and POD corresponding to the i+1th detection threshold, x i+1 -x i Indicates the amplitude of PFA change, y i+1 +y i Indicates the POD value of two adjacent points on the ROC curve.

[0078] When the AUC value is close to 1, it indicates that the detection system has high discrimination ability and high reliability; when the AUC value is close to 0.5, it indicates that the system has basically no discrimination ability and its performance is equivalent to random guessing.

[0079] Example 2

[0080] This embodiment provides an eddy current testing reliability assessment system based on signal spatial distribution characteristics established according to the above-mentioned assessment method, including:

[0081] The signal acquisition module is used to detect conductive materials (test pieces) such as carbon fiber composite materials and 316L stainless steel using an eddy current detection system, and obtain characteristic parameters such as the amplitude and phase of the eddy current detection signal;

[0082] a signal post-processing module, configured to select a standard specimen containing defects of the same type, use the eddy current detection signal of the standard specimen as a reference signal, and normalize its amplitude to 1V and phase to 0°; perform correction processing on the eddy current detection signal collected in step S1 to unify it to the reference standard defined by the reference signal; and after completing the signal correction, further perform centralization and flattening processing on the eddy current detection signal;

[0083] A Gini coefficient calculation module is used to sort the signal amplitude or phase data of the defective test piece and the non-defective test piece processed in step S2 from small to large; based on the sorted signal data, calculate its spatial distribution characteristics on the specified scanning line, and quantify the distribution difference between the defective signal and the non-defective signal in the spatial dimension using the Gini coefficient;

[0084] A reliability evaluation calculation module is used to use the signal spatial distribution characteristics quantified by the Gini coefficient to replace the traditional signal response characteristics based on single-point amplitude or phase, calculate the probability of detection (POD) and the false positive rate (PFA) during the detection process, draw the receiver operating characteristic (ROC) curve, and calculate the area under the curve (AUC), thereby achieving a quantitative evaluation of the reliability of the eddy current detection system;

[0085] The Adaptive Optimization module analyzes the impact of DUT defect parameters (such as length and depth) and eddy current testing system parameters (such as probe type, lift-off, and scanning direction) on eddy current testing reliability based on reliability assessment results (such as POD, PFA, AUC, and ROC curve shape). Based on this analysis, it adaptively optimizes testing parameters to improve defect detection capabilities and enhance overall system reliability.

[0086] In summary, the eddy current detection reliability assessment method and system based on the spatial distribution characteristics of the signal in the above-mentioned embodiment combines the spatial distribution characteristics of the eddy current signal with the detection probability POD and false positive rate PFA in the non-destructive testing reliability assessment indicators, and can comprehensively evaluate the defect detection capability and the reliability of the detection system during the eddy current detection process; the Gini coefficient can quantify the spatial distribution characteristics of the signal, judge the discrimination degree of the detection signal in the defective and non-defective states, and replace the amplitude or phase of the traditional POD and PFA analysis for subsequent detection reliability assessment; compared with the amplitude or phase characteristics of a single point, the Gini coefficient is calculated based on the overall signal spatial distribution characteristics and is less affected by noise; at the same time, the Gini coefficient can be used to calculate POD and PFA to draw the ROC curve for analysis, effectively improving the defect detection rate and reducing the false detection rate.

[0087] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A reliability assessment method for eddy current testing based on signal spatial distribution characteristics, characterized in that: The following steps are involved: S1: Data Collection The eddy current detection system is used to scan and detect the defective specimen and the non-defective specimen within the same area to obtain the eddy current detection signal corresponding to each sampling point; S2: Signal post-processing A standard specimen containing the same type of defects is selected, and the eddy current detection signal of the standard specimen is used as a reference signal, and its amplitude is normalized to 1V and the phase is 0°; the eddy current detection signal collected in step S1 is corrected to be unified to the reference standard defined by the reference signal; after completing the signal correction, the eddy current detection signal is further centered and flattened; S3: Calculate the Gini coefficient Sorting the signal amplitude or phase data of the defective specimen and the non-defective specimen after processing in step S2 from small to large; calculating the spatial distribution characteristics of the signal data on the specified scanning line based on the sorted signal data, and quantifying the distribution difference between the defective signal and the non-defective signal in the spatial dimension using the Gini coefficient; S4: Reliability Assessment The signal spatial distribution characteristics quantified by the Gini coefficient are used to replace the traditional signal response characteristics based on single-point amplitude or phase. The detection probability POD and false positive rate PFA in the detection process are calculated, the receiver operating characteristic ROC curve is drawn, and the area under the curve AUC is calculated, thereby realizing the quantitative evaluation of the reliability of the eddy current detection system.

2. The eddy current detection reliability assessment method based on signal spatial distribution characteristics according to claim 1 is characterized in that: In step S1, the eddy current detection signal is obtained by performing surface scanning on the test piece using an eddy current detection system; before scanning, the detection signal obtained by surface scanning of the test piece is extracted based on the eddy current detection signal of the defect-free area of ​​the test piece.

3. The eddy current detection reliability assessment method based on signal spatial distribution characteristics according to claim 1 is characterized in that: In step S1, the eddy current detection signal is collected by the eddy current probe and transmitted to the phase-locked amplifier for phase-sensitive detection and signal amplification processing. After low-pass filtering, a DC signal containing amplitude and phase information is output, and the corresponding real and imaginary voltage values ​​are obtained through the phasor decomposition method.

4. The eddy current detection reliability assessment method based on signal spatial distribution characteristics according to claim 1 is characterized in that: In step S3, the Gini coefficient is used to quantify the distribution difference between the defect signal and the non-defective signal in the spatial dimension. The closer the value is to 0, the more similar the signal distributions of the two are, and the smaller the defect possibility is; conversely, the larger the value is, the greater the difference in the signal distributions of the two is, and the higher the defect possibility is.

5. The eddy current detection reliability assessment method based on signal spatial distribution characteristics according to claim 1 is characterized in that: In step S3, the calculation formula of the Gini coefficient is as follows: Where N is the number of sampling intervals along the one-dimensional scan line, b and a are the maximum and minimum values ​​of the detection signal respectively, S n-1 is the defect signal amplitude corresponding to the n-1th sorting position, x n-1 is the amplitude of the defect-free signal corresponding to the n-1th sorting position, |S n-1 -x n-1 |+|S n -x n | is to calculate the absolute difference between each pair of corresponding position points of the two sets of data in the QQ graph.

6. The eddy current detection reliability assessment method based on signal spatial distribution characteristics according to claim 5 is characterized in that: In step S3, the Gini coefficient is used as the signal response feature And the relationship between it and the defect size a is transformed accordingly, so that It is linearly related to a, and the corresponding transformation includes logarithmic transformation; constructed through regression analysis" vs a” and calculate the 95% confidence interval and 95% prediction interval of the linear regression line. Then, the maximum likelihood method or the least squares method is used to estimate the parameters β0, β1, and σ in the linear model, where the parameters β0 and β1 are the intercept and slope of the linear model, and σ is the standard deviation of the fitting error. In the signal response analysis, It follows a normal distribution with a mean and standard deviation As shown in the following formula:

7. The eddy current detection reliability assessment method based on signal spatial distribution characteristics according to claim 6 is characterized in that: In step S4, the ROC curve uses PFA as the horizontal coordinate and POD as the vertical coordinate to evaluate the performance of the eddy current detection system; wherein POD is defined as the probability density integral area in the region where the defect signal characteristics are greater than the detection threshold, and PFA is defined as the probability density integral area in the region where the noise signal characteristics are greater than the detection threshold; the reliability of the eddy current detection system is quantified by calculating the area under the ROC curve AUC.

8. The eddy current detection reliability assessment method based on signal spatial distribution characteristics according to claim 7 is characterized in that: When the parameters β0, β1, and σ are known, the detection threshold for determining whether a defect is detected is given. Defect detection probability It obeys the standard normal distribution and the calculation formula is as follows: The formula for calculating the false positive rate PFA is as follows: in, is a noise signal, which is assumed to obey normal distribution or gamma distribution; The calculation formula of AUC value is as follows: Where m is the number of points on the ROC curve, x i and y i Indicates the PFA and POD corresponding to the i-th detection threshold on the ROC curve, x i+1 and y i+1 represents the PFA and POD corresponding to the i+1th detection threshold, x i+1 -x i Indicates the amplitude of PFA change, y i+1 +y i Indicates the POD value of two adjacent points on the ROC curve.

9. The eddy current detection reliability assessment system based on signal spatial distribution characteristics is characterized by: The reliability evaluation method according to any one of claims 1 to 8 is used to perform reliability evaluation on an eddy current testing system, comprising: A signal acquisition module is used to scan and detect defective specimens and non-defective specimens within the same area using an eddy current detection system to obtain eddy current detection signals corresponding to each sampling point; a signal post-processing module, configured to select a standard specimen containing defects of the same type, use the eddy current detection signal of the standard specimen as a reference signal, and normalize its amplitude to 1V and phase to 0°; perform correction processing on the eddy current detection signal collected in step S1 to unify it to the reference standard defined by the reference signal; and after completing the signal correction, further perform centralization and flattening processing on the eddy current detection signal; A Gini coefficient calculation module is used to sort the signal amplitude or phase data of the defective test piece and the non-defective test piece processed in step S2 from small to large; based on the sorted signal data, calculate its spatial distribution characteristics on the specified scanning line, and quantify the distribution difference between the defective signal and the non-defective signal in the spatial dimension using the Gini coefficient; The reliability evaluation calculation module is used to use the signal spatial distribution characteristics quantified by the Gini coefficient to replace the traditional signal response characteristics based on single-point amplitude or phase, calculate the detection probability POD and false positive rate PFA during the detection process, draw the receiver operating characteristic ROC curve, and calculate the area under the curve AUC, thereby realizing a quantitative evaluation of the reliability of the eddy current detection system.

Images