Alternating electromagnetic field quantitative detection system and method based on spatially distributed magnetic sensors

Abstract

The application discloses an alternating electromagnetic field quantitative detection system and method based on a spatial distributed magnetic sensor, relates to the technical field of nondestructive testing and signal processing, and comprises the following steps: a spatial distributed probe is used to scan the surface of a metal component; the probe comprises a plurality of magnetic sensors of the same type; the projection positions of the sensors in the scanning direction are known geometric parameters; time domain magnetic field signal waveform sequences of the sensors are synchronously acquired; time points and amplitudes of wave crests and troughs are extracted; a scanning motion function is linearly fitted based on the known projection positions of the sensors and the wave crest time points; and the ratio relationship of the peak-valley amplitude difference of the sensors at different lift-off heights is used to inversely calculate the defect depth based on the peak-valley amplitude difference and the power-law decay model of the magnetic field amplitude with the lift-off height.

Description

Technical Field

[0001] This invention relates to a quantitative detection system and method for AC electromagnetic fields based on spatially distributed magnetic sensors, belonging to the field of nondestructive testing and signal processing technology. Background Technology

[0002] In the detection of surface defects in metal components, defect quantification (especially depth quantification) is a key technical challenge. Alternating Current Field Measurement (ACFM) technology, as a novel electromagnetic non-destructive testing technique, induces a uniform alternating current on the surface of the workpiece through an excitation coil. When a defect is present, it causes a distortion in the spatial magnetic field. The distortion signal is picked up by a magnetic sensor to detect and assess the defect. It has advantages such as being non-contact, penetrating coatings, and requiring no coupling medium, and is widely used in marine engineering, petrochemicals, nuclear power plant operation and maintenance, bridge construction, and other fields.

[0003] Traditional ACFM inspection methods typically analyze the characteristics of a single sensor signal waveform. Specifically, the time interval between defect signal peaks (Bx or Bz components) is denoted as Δt. Assuming a constant scanning speed v, the defect length is estimated using the following formula: Then, based on empirical maps or semi-empirical models, the defect depth d is calculated from the defect length L.

[0004] For example, patent application publication number CN117074513A discloses an AC electromagnetic field detection system and a method for quantifying defects. This technical solution measures the magnetic field signal waveform of a single magnetic sensor, utilizes the time interval Δt between the Bx or Bz components of the defect signal peak, and estimates the defect length using the formula L = v × Δt, assuming a constant scanning speed v. Then, it extrapolates the defect depth from the defect length based on an empirical model or experimental calibration spectrum. The technical problems with this solution are: firstly, the defect length calculation heavily depends on the scanning speed v. During manual scanning, the speed is inevitably uneven, resulting in a large fluctuation in the calculated length L despite a constant Δt, ultimately leading to unreliable depth quantification results; secondly, to obtain an accurate speed v, an encoder or speed measuring device must be installed, increasing system complexity and cost, which is detrimental to the development of portable devices; thirdly, the empirical or semi-empirical model for estimating depth from length is heavily simplified, resulting in significant errors for complex shapes or small-sized defects; and fourthly, the quantitative determination of defect depth relies on empirical relationships established through prior experimental calibration, which itself introduces uncertainty.

[0005] Patent application publication number CN119086704A discloses a method for quantitative assessment of defects using AC electromagnetic field detection. This technical solution achieves a quantitative error of approximately 5% to 10% for defect length and depth. The technical problems with this solution are as follows: First, its depth quantification also relies on an empirical model established through experimental calibration, and the uncertainty introduced during the calibration process limits further improvement in quantitative accuracy. Second, its length quantification depends on the encoder to obtain the scanning speed, making accurate defect length quantification impossible under conditions without an encoder or with encoder signal distortion. Third, under manual variable-speed scanning conditions, its quantitative accuracy significantly decreases, making it difficult to meet the high-precision detection requirements of engineering sites.

[0006] In summary, existing technologies do not disclose methods for eliminating the influence of scanning speed by employing a spatially distributed sensor layout and using known sensor spatial spacing as a "spatial ruler," nor do they disclose technical solutions for achieving calibration-free defect depth inversion through the peak-to-valley amplitude difference ratio of sensors at different lift heights. Existing technologies suffer from significant deficiencies in defect quantification accuracy under practical conditions such as uneven scanning speeds, lack of encoders, and absence of calibration conditions. Summary of the Invention

[0007] To address the technical problems of existing technologies, such as the dependence of defect quantification on scanning speed, the need for additional encoder configuration, the requirement for experimental calibration, and insufficient accuracy, the present invention aims to provide an AC electromagnetic field quantitative detection system and method based on spatially distributed magnetic sensors. By adopting a spatially distributed sensor layout, using known geometric spacing as a spatial scale to fit the scanning motion, and combining a power-law decay model for ratio inversion, the system achieves accurate and simple calculation of defect length and depth without the need for encoders or calibration.

[0008] To achieve the above objectives, the present invention provides the following technical solution: The quantitative detection method for AC electromagnetic fields based on spatially distributed magnetic sensors described in this invention includes the following steps: S1. A spatially distributed probe is used to scan and detect the surface of the metal component to be tested. The spatially distributed probe includes multiple magnetic sensors of the same type arranged along the scanning direction. The projection position of each sensor in the scanning direction is a known geometric parameter, and the detection direction of each sensor is located on the same scanning path. S2. The time-domain magnetic field signal waveform sequence of each sensor is acquired synchronously through the acquisition system; S3. Extract the time point and amplitude corresponding to the peak and the time point and amplitude corresponding to the trough from the waveforms of each sensor data. S4. Based on the known projection positions of each sensor in the scanning direction and their corresponding peak time points, the relationship between the scanning position and time is linearly fitted to obtain the scanning motion function X(t). S5. Based on the scanning motion function X(t) and the peak-valley time difference of each sensor, calculate the defect space length corresponding to each sensor, and then calculate the final defect length by weighted average. S6. Based on the peak-valley amplitude difference of each sensor, combined with the attenuation model of magnetic field disturbance amplitude with lift-off height, the defect depth is obtained by inverting the ratio of peak-valley amplitude difference measured by sensors at different lift-off heights.

[0009] Furthermore, the spatially distributed probe includes a first sensor, a second sensor, and a third sensor. The first sensor and the second sensor are located in a plane parallel to the detection surface and their detection directions are on the same scanning path. The third sensor is located in a plane perpendicular to the detection surface, on the perpendicular bisector of the line connecting the first sensor and the second sensor, and its detection direction is on the same scanning path as the first and second sensors. The distance between the first sensor and the second sensor is equal to the distance between the third sensor and the line connecting the first and second sensors.

[0010] Furthermore, in the linear fitting step, the projection position of the first sensor in the scanning direction is taken as 0, the second sensor as a, and the third sensor as a / 2, combined with the time point T corresponding to the peaks of the three sensors. 10 T 20 T 30 The position-time relationship was fitted using the least squares method: ; Where v is the fitted scanning speed, and X0 is the fitted intercept; for the i-th sensor, its defect space length is: ; Among them, T 11 T 21 T 31 The time corresponding to the troughs of the three sensors.

[0011] Furthermore, in the weighted average, the weighting coefficients are set according to the different lift-off heights of each sensor. Sensors with smaller lift-off heights are assigned larger weighting coefficients, and sensors with larger lift-off heights are assigned smaller weighting coefficients. The final defect length is: ; Among them, W1, W2, and W3 are the weighting coefficients corresponding to the three sensors.

[0012] Furthermore, in the defect depth inversion step: Calculate the peak-to-valley amplitude difference for each sensor: ; Where: D 10 D 20 D30 D represents the peak amplitude of the three sensors. 11 D 21 D 31 The amplitude of the valleys from the three sensors; Calculate the average peak-to-valley amplitude difference between the first and second sensors parallel to the detection surface: ; Based on a power-law model where the amplitude of the magnetic field disturbance decays with lift-off height: ; Where z is the lift-off height, C is a constant related to the defect size and excitation intensity, d is the equivalent depth of the defect, and n is the attenuation exponent. The peak-to-valley amplitude difference ratio between the first and second sensor planes and the third sensor plane is established: ; Solving the above ratio relationship, we obtain the defect depth inversion formula: ; The constant C is completely eliminated in the ratio operation.

[0013] Furthermore, the attenuation index n is a constant related to the electromagnetic properties of the material being tested, and is predetermined through simulation or experimental determination; the projection positions of each sensor are fixed geometric parameters known during probe design.

[0014] Furthermore, the defect length calculation step also includes a step of verifying the consistency of the lengths calculated by each sensor: if the deviation of the length calculation value of a certain sensor from the length calculation values ​​of other sensors exceeds a preset threshold, then the data of that sensor is determined to be an outlier, and its weight in the weighted average is reduced or removed.

[0015] Furthermore, when the number of sensors in the spatially distributed probe is N and N≥5, N sensors parallel to the detection surface are arranged in the scanning direction. Multi-point linear regression fitting is performed using the known spatial projection position and peak time point of each sensor to improve the scanning speed fitting accuracy and defect length quantitative accuracy.

[0016] The AC electromagnetic field quantitative detection system based on a spatially distributed magnetic sensor described in this invention comprises: A spatially distributed probe, comprising multiple magnetic sensors of the same type, each sensor being arranged along the scanning direction, the projection position of each sensor in the scanning direction having known geometric parameters, and the detection direction of each sensor being located on the same scanning path; The signal acquisition module is used to synchronously acquire the time-domain magnetic field signal waveform sequences of each sensor; The feature extraction module is used to extract the time point and amplitude corresponding to the peak and the time point and amplitude corresponding to the trough from the waveforms of data from various sensors. The data processing module is used to fit the scanning motion relationship based on the known projection position of each sensor in the scanning direction and its peak time point, and to calculate the defect length based on the fitting results and the peak-valley time difference of each sensor, and to invert the defect depth based on the peak-valley amplitude difference of each sensor combined with the attenuation model of the magnetic field amplitude with the lift-off height. The output module is used to output the quantitative detection results of the defect length and defect depth.

[0017] Furthermore, the plurality of magnetic sensors are any one of tunnel magnetoresistive sensors, giant magnetoresistive sensors, anisotropic magnetoresistive sensors, or induction coil magnetic sensors; the system also includes a preprocessing submodule for performing bandpass filtering, detrending, and amplitude normalization on the time-domain signals of each sensor to eliminate background magnetic field interference and sensor inherent noise.

[0018] Compared with existing technologies, the AC electromagnetic field quantitative detection system and method based on spatially distributed magnetic sensors of the present invention exhibit the following beneficial effects in terms of technical performance and practical application: First, the invention utilizes the known spatial spacing of the sensors as a "spatial ruler" to eliminate the influence of speed. By combining the known geometric spacing 'a' of the sensors in the spatially distributed probe with multiple peak time points for least-squares linear fitting to obtain the scanning motion function, the time-domain measurement is converted into a spatial-domain measurement, and the defect length is directly calculated. Unlike traditional methods that rely on scanning speed, the defect length calculation in this invention no longer depends on external speed measurement equipment, fundamentally overcoming the quantitative errors caused by manual or variable-speed scanning, and solving the technical problem that existing technologies heavily rely on scanning speed for defect length calculation and require the installation of encoders.

[0019] Second, calibration-free depth quantification is achieved through ratio inversion based on a power-law decay model. This invention utilizes the peak-to-valley amplitude difference measured by sensors at different lift-off heights, combined with a power-law model showing the decay of magnetic field disturbance amplitude with lift-off height. By constructing a ratio relationship, the constant C is completely eliminated in the ratio calculation, thus obtaining depth quantification that depends only on the known geometric spacing a, the measured peak-to-valley amplitude difference ΔD3, and ΔD. 12 The depth inversion formula for the inherent decay index n. Compared with existing technologies that rely on experimental calibration to establish empirical models, this invention avoids the uncertainties introduced by the calibration process and improves the accuracy and adaptability of depth quantification.

[0020] Third, the encoder-free system architecture reduces cost and complexity. This invention eliminates the need for additional speed measurement encoder hardware, reducing system cost, size, and complexity. It is particularly beneficial for the development and application of portable devices, and solves the technical problem in the prior art that an additional encoder must be installed to obtain accurate scanning speed.

[0021] Fourth, the algorithm is simple and has good real-time performance. The algorithm involved in this invention has a clear principle, low computational load, and good real-time performance. It only requires the peak and trough characteristics of a small amount of sensor data to complete the quantitative calculation of defect length and depth. It does not require a complex data training process or large-scale computing resources, which is conducive to its application in real-time online detection in industrial fields.

[0022] Fifth, quantitative accuracy is significantly improved. By using information from multiple spatial points and performing local calibration on the scanning speed, the final defect feature length obtained is more accurate than that of traditional methods; depth calculation is performed without calibration, avoiding the uncertainties introduced by the calibration process, thereby improving the quantitative accuracy of depth. Attached Figure Description

[0023] Figure 1 This is a front view of the sensor layout in this invention; Figure 2 This is a top view of the sensor layout in this invention; Figure 3 This is a right view of the sensor layout in this invention; Figure 4 This is a structural diagram of the external housing of the spatially distributed probe in this invention; Figure 5 for Figure 4 Sectional view at DD; Figure 6 This is a schematic flowchart of the method of the present invention; Figure 7 This is a connection block diagram of the system of the present invention. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0025] Example 1: like Figures 1-5 As shown, the spatially distributed probe of the present invention includes three magnetic sensors of the same type: a first sensor, a second sensor, and a third sensor.

[0026] The first and second sensors are located in a plane parallel to the detection surface, and their detection directions are along the same scanning path. The third sensor is located in a plane perpendicular to the detection surface, on the perpendicular bisector of the line connecting the first and second sensors, and its detection direction is along the same scanning path as the first and second sensors.

[0027] The distance between the first and second sensors is equal to the distance from the third sensor to the line connecting the first and second sensors, denoted as 'a'. Taking the projected position of the first sensor in the scanning direction as 0, the projected position of the second sensor in the scanning direction is 'a', and the projected position of the third sensor in the scanning direction is 'a / 2'.

[0028] All three sensors are the same type of high-sensitivity magnetic sensors, which can be any one of tunnel magnetoresistive sensors, giant magnetoresistive sensors, anisotropic magnetoresistive sensors, or induction coil magnetic sensors.

[0029] like Figure 6 As shown, the quantitative detection method for AC electromagnetic fields based on spatially distributed magnetic sensors provided by this invention includes the following steps: Step S1: Acquire sensor data. The time-domain magnetic field signal waveform sequences of the three sensors are simultaneously acquired through the acquisition system, denoted as... , , .

[0030] Step S2: Feature Extraction. Based on the data waveforms from the three sensors, extract the time point and amplitude corresponding to the peaks and the time point and amplitude corresponding to the troughs, respectively. The time corresponding to the peaks in the waveforms from the three sensors is denoted as... , , The numerical value corresponding to the peak is denoted as , , The time corresponding to the trough is denoted as , , The value corresponding to the trough is denoted as , , .

[0031] Step S3: Fit the scanning motion relationship. The projected positions of the three sensors along the scanning direction are known: First sensor... Third sensor Second sensor Using the coordinates of three peak time points , , By performing linear fitting using the least squares method, the relationship between position and time can be obtained: ; Specifically, for N data points, the linear regression slope and intercept are calculated as follows: ; ; in, and These are the time average and the location average, respectively. Substituting the coordinates of the three peak time points, we get: ; ; ; ; Step S4: Defect length calculation. For each sensor i (i=1,2,3), using the fitted X(t) from step S3 and the peak-valley time difference of that sensor, calculate the corresponding spatial length: ; The three sensors are measuring the same defect, theoretically To improve accuracy, a weighted average is used as the final result. The weighting coefficients are set based on the lift-off height of each sensor: sensors with smaller lift-off heights are assigned larger weighting coefficients, and sensors with larger lift-off heights are assigned smaller weighting coefficients. The final defect length is: ; Step S5: Defect Depth Calculation. For AC electromagnetic field detection, the amplitude of the magnetic field disturbance caused by the defect decreases with the lift-off height, and this decrease can be approximated by a power-law model: ; Where: z is the lift-off height, C is a constant related to the defect size and excitation intensity, d is the equivalent depth of the defect, and n is the attenuation index.

[0032] Extract the peak-to-valley amplitude difference based on data from the three sensors: ; ; ; Define the average peak-to-valley amplitude difference between the first and second sensor planes: ; By substituting the peak-valley amplitude difference data and spatial location information into the power-law model equation, we can obtain: ; ; The following can be obtained by constructing the ratio relationship: ; Solving this equation yields the relationship between the defect depth d and the peak-to-valley amplitude difference, geometric spacing, and attenuation index: ; This formula depends only on the known geometric spacing 'a' and the difference between the measured peak and valley amplitude data. and The inherent decay exponent n requires no experimental calibration, and the constant C is completely eliminated in ratio calculations.

[0033] The technical effects of the above method are experimentally verified using data from specific embodiments.

[0034] Experiment setup and shared input data: Specimen material: Q235 steel; Sensor spacing: a=6mm; Excitation frequency: 10kHz; Scanning method: Manual, with possible natural speed fluctuations within ±15%; Lifting conditions: Tested at 1mm, 2mm, 3mm, and 4mm respectively; Attenuation index: n=2.5 (Q235 steel).

[0035] (I) Complete calculation process at a lift height of 1mm The original collected data is shown in Table 1 below: Table 11 shows the original data collected at 1mm.

[0036] Length quantitative calculation process: Establish the fitted coordinate points: , , .

[0037] Calculate the time mean and the location mean: ; ; The deviations at each point are calculated, as shown in Table 2: Table 21mm deviation data for each point

[0038] Calculate the scanning speed v: ; Calculate the intercept : ; Obtain the position-time function: ; Convert the peak-valley time difference of each sensor into spatial length: ; ; ; The final length is obtained by weighted average (taking...) , ): ; Quantitative relative error in length (actual defect length is 20mm): ; Depth Quantitative Calculation Process: Calculate the peak-to-valley amplitude difference for each sensor: ; ; ; Calculate the average peak-to-valley amplitude difference between the first and second sensors at the same lift-off height: ; Calculate the 2.5th root of n (i.e., raise it to the power of 0.4): ; ; Substitute into the depth inversion formula: ; Quantitative relative error of depth (actual defect depth is 4mm): ; (II) Complete calculation process at a lifting height of 2mm: The original collected data is shown in Table 3 below: Table 32mm Original Data Collection

[0039] Length quantitative calculation process: Establish the fitted coordinate points: , , .

[0040] Calculate the time mean and the location mean: ; ; The deviations at each point are calculated, as shown in Table 4 below: Table 42mm deviation data for each point

[0041] Calculate the scanning speed v: ; Calculate the intercept : ; Obtain the position-time function: ; Convert the peak-valley time difference of each sensor into spatial length: ; ; ; The final length is obtained by weighted average (taking...) , ): ; Quantitative relative error of length: ; Depth Quantitative Calculation Process: Calculate the peak-to-valley amplitude difference for each sensor: ; ; ; Calculate the average peak-to-valley amplitude difference at the same lift height: ; Calculate the root of n = 2.5: ; ; Substitute into the depth inversion formula: ; Relative error in depth quantification: ; (III) Complete calculation process at a lifting height of 3mm The original collected data is shown in Table 5 below: Table 53mm Original Data Collection

[0042] Length quantitative calculation process: Establish the fitted coordinate points: , , .

[0043] Calculate the time mean and the location mean: ; ; The deviations at each point are calculated, as shown in Table 6 below: Table 63mm deviation data for each point

[0044] Calculate the scanning speed v: ; Calculate the intercept : ; Obtain the position-time function: ; Convert the peak-valley time difference of each sensor into spatial length: ; ; ; The final length is obtained by weighted average (taking...) , ): ; Quantitative relative error of length: ; Depth Quantitative Calculation Process: Calculate the peak-to-valley amplitude difference for each sensor: ; ; ; Calculate the average peak-to-valley amplitude difference at the same lift height: ; Calculate the root of n = 2.5: ; ; Substitute into the depth inversion formula: ; Relative error in depth quantification: ; (iv) Complete calculation process at a lifting height of 4mm The original collected data is shown in Table 7 below: Table 74mm Original Data Acquired

[0045] Length quantitative calculation process: Establish the fitted coordinate points: , , .

[0046] Calculate the time mean and the location mean: ; ; The deviations at each point are calculated, as shown in Table 8 below: Table 84mm deviation at various points

[0047] Calculate the scanning speed v: ; Calculate the intercept : ; Obtain the position-time function: ; Convert the peak-valley time difference of each sensor into spatial length: ; ; ; The final length is obtained by weighted average (taking...) , ): ; Quantitative relative error of length: ; Depth Quantitative Calculation Process: Calculate the peak-to-valley amplitude difference for each sensor: ; ; ; Calculate the average peak-to-valley amplitude difference at the same lift height: ; Calculate the root of n = 2.5: ; ; Substitute into the depth inversion formula: ; Relative error in depth quantification: ; (v) Summary of all lift height results The test results at each lift-off height are summarized in Table 9 below: Table 9. Detection results at various lift heights

[0048] (vi) Comparison with existing technologies The detection effects of this invention and existing technologies are compared in Table 10 below: Table 10 Comparison of detection methods between the present invention and prior art.

[0049] Overall conclusion: Firstly, in terms of depth quantification: This invention performs excellently at all lift heights, with a maximum depth error of only 0.15% and an average error of only 0.081%, which is significantly better than the error level of the prior art of about 5% to 10%.

[0050] Secondly, regarding length measurement: When the speed uniformity is good, the error of this invention is controlled within 2%; when the speed fluctuation is large (2mm lift), the error increases, but the depth measurement remains unaffected. Under normal scanning conditions with small speed fluctuations, the length error can still be stably controlled within 2.1%.

[0051] Thirdly, the core advantage is that the deep quantitative method requires no calibration throughout the entire process (the constant C is completely eliminated), which is a fundamental technological breakthrough of this invention compared to existing technologies.

[0052] Example 2: like Figure 7 As shown, the AC electromagnetic field quantitative detection system based on a spatially distributed magnetic sensor provided by this invention includes the following modules: The spatially distributed probe includes multiple identical magnetic sensors, each arranged along the scanning direction. The projected positions of each sensor along the scanning direction have known geometric parameters, and the detection directions of each sensor are located on the same scanning path. In a preferred embodiment, the multiple magnetic sensors are any one of a tunnel magnetoresistive sensor, a giant magnetoresistive sensor, an anisotropic magnetoresistive sensor, or an induction coil magnetic sensor.

[0053] The signal acquisition module is electrically connected to the spatially distributed probe and is used to synchronously acquire the time-domain magnetic field signal waveform sequence of each sensor.

[0054] The feature extraction module, connected to the signal acquisition module, is used to extract the time point and amplitude corresponding to the peak and the time point and amplitude corresponding to the trough from the waveforms of data from each sensor. In a preferred embodiment, the feature extraction module further includes a preprocessing submodule, used to perform bandpass filtering, detrending, and amplitude normalization on the time-domain signals of each sensor to eliminate background magnetic field interference and inherent sensor noise.

[0055] The data processing module, connected to the feature extraction module, is used to fit the scanning motion relationship based on the known projection position of each sensor in the scanning direction and its peak time point, calculate the defect length based on the fitting results and the peak-valley time difference of each sensor, and invert the defect depth based on the peak-valley amplitude difference of each sensor combined with the attenuation model of the magnetic field amplitude with the lift-off height.

[0056] The output module, connected to the data processing module, is used to output the quantitative detection results of the defect length and defect depth.

[0057] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors, characterized in that, Includes the following steps: S1. A spatially distributed probe is used to scan and detect the surface of the metal component to be tested. The spatially distributed probe includes multiple magnetic sensors of the same type arranged along the scanning direction. The projection position of each sensor in the scanning direction is a known geometric parameter, and the detection direction of each sensor is located on the same scanning path. S2. The time-domain magnetic field signal waveform sequence of each sensor is acquired synchronously through the acquisition system; S3. Extract the time point and amplitude corresponding to the peak and the time point and amplitude corresponding to the trough from the waveforms of each sensor data. S4. Based on the known projection positions of each sensor in the scanning direction and their corresponding peak time points, the relationship between the scanning position and time is linearly fitted to obtain the scanning motion function X(t). S5. Based on the scanning motion function X(t) and the peak-valley time difference of each sensor, calculate the defect space length corresponding to each sensor, and then calculate the final defect length by weighted average. S6. Based on the peak-valley amplitude difference of each sensor, combined with the attenuation model of magnetic field disturbance amplitude with lift-off height, the defect depth is obtained by inverting the ratio of peak-valley amplitude difference measured by sensors at different lift-off heights.

2. The method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors according to claim 1, characterized in that, The spatially distributed probe includes a first sensor, a second sensor, and a third sensor. The first sensor and the second sensor are located in a plane parallel to the detection surface and their detection directions are on the same scanning path. The third sensor is located in a plane perpendicular to the detection surface, on the perpendicular bisector of the line connecting the first sensor and the second sensor, and its detection direction is on the same scanning path as the first and second sensors. The distance between the first sensor and the second sensor is equal to the distance between the third sensor and the line connecting the first and second sensors.

3. The method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors according to claim 2, characterized in that, In the linear fitting step S4, the projection position of the first sensor in the scanning direction is taken as 0, the second sensor as a, and the third sensor as a / 2, combined with the time point T corresponding to the peaks of the three sensors. 10 T 20 T 30 The position-time relationship was fitted using the least squares method: ； Where v is the fitted scanning speed, and X0 is the fitted intercept; for the i-th sensor, its defect space length is: ； Among them, T 11 T 21 T 31 The time corresponding to the troughs of the three sensors.

4. The method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors according to claim 3, characterized in that, In step S5, the weighted average is calculated using weighting coefficients set according to the different lift-off heights of each sensor. Sensors with smaller lift-off heights are assigned larger weighting coefficients, and sensors with larger lift-off heights are assigned smaller weighting coefficients. The final defect length is: ； Among them, W1, W2, and W3 are the weighting coefficients corresponding to the three sensors.

5. The method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors according to claim 2, characterized in that, In step S6, defect depth inversion: Calculate the peak-to-valley amplitude difference for each sensor: ； Where: D 10 D 20 D 30 D represents the peak amplitude of the three sensors. 11 D 21 D 31 The amplitude of the valleys from the three sensors; Calculate the average peak-to-valley amplitude difference between the first and second sensors parallel to the detection surface: ； Based on a power-law model where the amplitude of the magnetic field disturbance decays with lift-off height: ； Where z is the lift-off height, C is a constant related to the defect size and excitation intensity, d is the equivalent depth of the defect, and n is the attenuation exponent. The peak-to-valley amplitude difference ratio relationship between the first sensor, the second sensor plane, and the third sensor plane is established: ； Solving the above ratio relationship, we obtain the defect depth inversion formula: ； In this case, the constant C is completely eliminated in the ratio operation.

6. The method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors according to claim 5, characterized in that, The attenuation index n is a constant related to the electromagnetic properties of the material being tested, and is predetermined through simulation or experimental determination; the projection positions of each sensor are fixed geometric parameters known during probe design.

7. The method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors according to claim 1, characterized in that, Step S5 further includes a step of verifying the consistency of the lengths calculated by each sensor: if the length calculation value of a certain sensor deviates from the length calculation values ​​of other sensors by more than a preset threshold, then the data of that sensor is determined to be an outlier, and its weight in the weighted average is reduced or removed.

8. The method for quantitative detection of AC electromagnetic fields based on spatially distributed magnetic sensors according to claim 1, characterized in that, When the number of sensors in the spatially distributed probe is N and N≥5, N sensors are arranged parallel to the detection surface in the scanning direction. Multi-point linear regression fitting is performed using the known spatial projection position and peak time point of each sensor to improve the accuracy of scanning speed fitting and defect length quantification.

9. A quantitative detection system for AC electromagnetic fields based on spatially distributed magnetic sensors, characterized in that, include: A spatially distributed probe, comprising multiple magnetic sensors of the same type, each sensor being arranged along the scanning direction, the projection position of each sensor in the scanning direction having known geometric parameters, and the detection direction of each sensor being located on the same scanning path; The signal acquisition module is used to synchronously acquire the time-domain magnetic field signal waveform sequences of each sensor; The feature extraction module is used to extract the time point and amplitude corresponding to the peak and the time point and amplitude corresponding to the trough from the waveforms of data from various sensors. The data processing module is used to fit the scanning motion relationship based on the known projection position of each sensor in the scanning direction and its peak time point, and to calculate the defect length based on the fitting results and the peak-valley time difference of each sensor, and to invert the defect depth based on the peak-valley amplitude difference of each sensor combined with the attenuation model of the magnetic field amplitude with the lift-off height. The output module is used to output the quantitative detection results of the defect length and defect depth.

10. The AC electromagnetic field quantitative detection system based on a spatially distributed magnetic sensor according to claim 9, characterized in that, The magnetic sensor is any one of a tunnel magnetoresistive sensor, a giant magnetoresistive sensor, an anisotropic magnetoresistive sensor, or an induction coil magnetic sensor; the system also includes a preprocessing submodule, which performs bandpass filtering, detrending, and amplitude normalization on the time-domain signals of each sensor to eliminate background magnetic field interference and sensor inherent noise.

Images