Metal pipeline defect depth eddy current detection system and method based on flexible honeycomb-shaped sensor

Abstract

The invention relates to a metal pipeline defect depth eddy current detection system and method based on a flexible honeycomb-shaped sensor, the system comprises the flexible honeycomb-shaped sensor, an excitation source, a driving mechanism, a signal acquisition module and a data processing unit, the sensor adopts a flexible substrate, and the center of the sensor is provided with a receiving coil; a honeycomb-shaped excitation coil unit composed of six regular triangle excitation coil units is arranged on the periphery of the tube wall, and reverse current is conducted in the adjacent regular triangle excitation coil units so that a multi-pole magnetic field can be formed on the tube wall and lift-off interference can be restrained. During detection, the sensor is attached to the pipe wall for scanning, and voltage signals of the receiving coil are collected; and the data processing unit extracts the impedance trajectory slope of the signal on the complex plane as a feature, inputs the feature into a pre-trained defect depth prediction model, and outputs a defect depth prediction result. According to the method, key features which are insensitive to the defect direction and sensitive to the depth are extracted through unique sensor design, and rapid and accurate nondestructive quantitative detection of the defect depth of the metal pipeline is achieved.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology, specifically to a system and method for detecting deep eddy current defects in metal pipelines based on a flexible honeycomb sensor. Background Technology

[0002] Metal pipelines are widely used in industries such as chemical, energy, and aerospace, and their structural integrity directly affects production safety and equipment lifespan. Defects such as cracks, corrosion, and perforation are the main forms of pipeline failure; therefore, efficient and accurate non-destructive testing of metal pipelines is crucial.

[0003] Eddy current testing (ECT), a non-contact, coupling agent-free electromagnetic non-destructive testing technique, is particularly suitable for detecting surface and near-surface defects in conductive metallic materials. Its basic principle is that an excitation coil carrying an alternating current induces eddy currents in a nearby conductive workpiece. Defects in the workpiece disturb the distribution of these eddy currents, thereby changing the impedance or induced voltage of the detection coil. By analyzing these changes in electrical signals, defects can be identified.

[0004] However, existing eddy current testing technology still faces the following major problems and challenges when applied to the assessment of defect depth in metal pipes:

[0005] 1) Sensor fit problem: Traditional rigid eddy current sensors are difficult to conformally fit with complex curved surfaces (such as the outer wall of a pipe). The gap between the two will introduce significant signal interference, reducing detection sensitivity and reliability.

[0006] 2) Defect orientation sensitivity: The orientation of the defect (such as the angle with the eddy current direction) significantly affects the response characteristics of the eddy current signal, resulting in huge differences in the signals presented by defects of the same depth but different orientations. This greatly increases the difficulty of inverting the key parameter of defect depth from complex signals. Existing methods usually require the defect orientation as an additional input, leading to complex inversion models and high computational costs.

[0007] 3) Insufficient signal stability and sensitivity: In complex industrial environments, interference factors such as sensor lift-off fluctuations and temperature changes can affect signal stability. Furthermore, traditional sensors exhibit weak response signals for early, minute defects, making high-precision quantitative assessment difficult.

[0008] Therefore, there is an urgent need for a new eddy current detection method and system that can adaptively fit curved surfaces, suppress lift-off interference, and have high-precision quantitative inversion capability for defect depth. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the prior art and provide a metal pipeline defect depth eddy current detection system and method based on a flexible honeycomb sensor to solve technical problems such as poor fit between traditional rigid sensors and curved surfaces, interference in defect direction depth assessment, and susceptibility of signals to lift-off effects, thereby achieving high-precision and high-efficiency quantitative detection of metal pipeline defect depth.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: a metal pipeline defect depth eddy current detection system based on a flexible honeycomb sensor, comprising a flexible honeycomb sensor, an excitation source, a sensor driving mechanism, a signal acquisition module, and a data processing unit; the flexible honeycomb sensor includes a flexible substrate, a honeycomb excitation coil unit, and a receiving coil. The honeycomb excitation coil unit is composed of six equilateral triangular excitation coil units arranged in a ring array around the center of the flexible substrate and formed on the flexible substrate to form a honeycomb conformal array. The receiving coil is located at the center of the flexible substrate, adjacent to the equilateral triangular excitation coil units. The currents are in opposite directions to form a multi-pole alternating magnetic field on the surface of the metal tube under test; the excitation source is connected to the honeycomb excitation coil unit to provide an alternating excitation current signal to the honeycomb excitation coil unit; the signal acquisition module is connected to the receiving coil and the data processing unit respectively to acquire the voltage signal output by the receiving coil and transmit it to the data processing unit; the flexible honeycomb sensor is set on the sensor driving mechanism and slides along the surface of the metal tube under test under its drive; the data processing unit extracts the defect feature slope based on the received voltage signal and calls the pre-trained defect depth prediction model to output the defect depth prediction result.

[0011] Furthermore, the equilateral triangular excitation coil unit is composed of several layers of equilateral triangular excitation coils arranged vertically; the current direction of all equilateral triangular excitation coils in the same equilateral triangular excitation coil unit is the same; the current direction of the equilateral triangular excitation coils located on the same layer in adjacent equilateral triangular excitation coil units is opposite; and all equilateral triangular excitation coils in the honeycomb excitation coil unit are connected in series.

[0012] Furthermore, the equilateral triangular excitation coil unit consists of two layers of equilateral triangular excitation coils; the winding method of the honeycomb excitation coil unit is as follows:

[0013] The wire is wound clockwise or counterclockwise from the outside in to the center or a predetermined number of turns to form the upper equilateral triangle excitation coil of the first equilateral triangle excitation coil unit; then it extends downwards to the lower layer, and is wound clockwise or counterclockwise from the inside out to form the lower equilateral triangle excitation coil of the first equilateral triangle excitation coil unit; after the first equilateral triangle excitation coil unit is wound, the wire extends backwards to the position of the second equilateral triangle excitation coil unit and begins to wind the second equilateral triangle excitation coil unit: the wire is wound counterclockwise or clockwise from the outside in to the center or a predetermined number of turns to form the lower equilateral triangle excitation coil of the second equilateral triangle excitation coil unit; then it extends upwards to the upper layer, and is wound counterclockwise or clockwise from the inside out to form the upper equilateral triangle excitation coil of the second equilateral triangle excitation coil unit;

[0014] After the second equilateral triangle excitation coil unit is wound, the wire continues to extend backward, and the subsequent equilateral triangle excitation coil units are wound using the same method as the first and second equilateral triangle excitation coil units.

[0015] Furthermore, the slope of the defect feature is:

[0016]

[0017] Where k represents the slope of the defect feature, and These represent the real and imaginary parts of the receiving coil impedance, respectively. This represents the maximum value of the real part when the imaginary part is greater than 0. This represents the minimum value of the real part when the imaginary part is less than 0. It represents the change in the imaginary part between the maximum and minimum values ​​of the real part.

[0018] Furthermore, a high-permeability magnetic core is embedded in the middle of the receiving coil.

[0019] Furthermore, the data processing unit is connected to the excitation source and the sensor driving mechanism respectively, so as to control the operation of the excitation source and the sensor driving mechanism respectively.

[0020] This invention also provides a method for detecting the depth of defects in metal pipelines based on a flexible honeycomb sensor, implemented using the above-mentioned system, and comprising the following steps:

[0021] S1: Select a flexible honeycomb sensor of appropriate size according to the size and specifications of the metal tube to be tested, install the flexible honeycomb sensor on the sensor driving mechanism, and then place the sensor driving mechanism next to the metal tube to be tested so that the flexible honeycomb sensor is attached to the surface of the metal tube to be tested.

[0022] S2: Start the excitation source and apply an alternating excitation current signal to the honeycomb excitation coil unit; at the same time, control the sensor drive mechanism to work, so that it drives the flexible honeycomb sensor to slide along the surface of the metal tube to be tested in order to scan the surface of the metal tube to be tested.

[0023] S3: The signal acquisition module acquires the voltage signal output by the coil during the scanning process of the flexible honeycomb sensor and transmits it to the data processing unit;

[0024] S4: The data processing unit obtains the real and imaginary parts of the impedance of the receiving coil based on the received voltage signal, maps the real and imaginary parts to the complex plane to form an impedance trajectory, and then extracts the defect feature slope k based on the impedance trajectory.

[0025] S5: Input the defect feature slope k into the pre-trained defect depth prediction model to obtain the defect depth prediction result of the metal pipe to be tested.

[0026] Furthermore, the method for calculating the slope of the defect feature is as follows:

[0027]

[0028] Where k represents the slope of the defect feature, and These represent the real and imaginary parts of the receiving coil impedance, respectively. This represents the maximum value of the real part when the imaginary part is greater than 0. This represents the minimum value of the real part when the imaginary part is less than 0. It represents the change in the imaginary part between the maximum and minimum values ​​of the real part.

[0029] Furthermore, the defect depth prediction model is a machine learning model trained based on the correspondence between the defect feature slope k and the defect depth and tilt angle.

[0030] Furthermore, the training method for the defect depth prediction model is as follows:

[0031] 1) Set up m metal tube samples with defects of different depths. Each metal tube sample has i prefabricated defects at different angles to the cross-section of the metal tube along the axial direction on its surface.

[0032] 2) The flexible honeycomb sensor is driven by the sensor driving mechanism to slide along the surface of each metal tube sample to scan the surface of each metal tube sample; each metal tube sample is scanned repeatedly n times; m×n×i sets of original defect sample data are obtained;

[0033] 3) Extract the defect feature slope k based on the obtained original defect sample data;

[0034] 4) Based on the extracted defect feature slope k, the natural gradient boosting algorithm is used to train the model and obtain the defect depth prediction model.

[0035] Compared with existing technologies, this invention has the following advantages: Through the design of a flexible honeycomb sensor, it adaptively fits the surface of metal pipes with various curvatures. Utilizing its unique "adjacent antiphase" excitation topology, it effectively suppresses lift-off interference and ensures the stability of signal acquisition. The eddy current signals from defects in different directions acquired by this sensor exhibit highly consistent trajectory slope characteristics on the impedance complex plane. This feature is weakly correlated with the defect direction but strongly correlated with the defect depth, thus simplifying the complex multi-parameter inversion problem into a regression problem based on a single feature, significantly reducing the analytical difficulty. Furthermore, the equilateral triangular coil design has a better magnetic field distribution and spatial sensitivity than traditional circular coils, improving the detection capability for minute defects. Based on the extracted robust slope characteristics, combined with machine learning algorithms for depth prediction, a rapid and high-precision quantitative assessment of the defect depth in metal pipelines is achieved. This invention integrates flexible sensing, feature extraction, and intelligent diagnosis into a single system, offering high practicality and providing an efficient and reliable solution for non-destructive testing and health management of industrial pipelines. Attached Figure Description

[0036] Figure 1 This is a schematic diagram (top view) of the flexible honeycomb sensor provided in an embodiment of the present invention.

[0037] Figure 2 This is a schematic diagram (side view) of the flexible honeycomb sensor provided in an embodiment of the present invention.

[0038] Figure 3 This is a schematic diagram of the metal pipeline defect depth eddy current detection system provided in the embodiment of the present invention;

[0039] Figure 4 This is a flowchart of the eddy current detection method for metal pipeline defects provided in this embodiment of the invention;

[0040] Figure 5 This is a schematic diagram of a metal tube sample in an embodiment of the present invention;

[0041] Figure 6 This is a slope diagram of the complex plane of the defect at a depth of 2mm to 0.2mm in the 0° direction in an embodiment of the present invention;

[0042] Figure 7 These are slope diagrams of the complex plane of defects in different depths from 0° to 90° in the embodiments of the present invention; where (a) is 2mm, (b) is 1.6mm, (c) is 1.2mm, and (d) is 0.8mm.

[0043] Figure 8This is a scatter plot of defect depth features in an embodiment of the present invention;

[0044] Figure 9 This is a structural diagram of the NGBoost model in an embodiment of the present invention;

[0045] Figure 10 This is a diagram showing the defect depth prediction results in an embodiment of the present invention. Detailed Implementation

[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0047] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0048] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0049] like Figure 1-3 As shown, this embodiment provides a metal pipe defect depth eddy current detection system based on a flexible honeycomb sensor, including a flexible honeycomb sensor, an excitation source, a sensor driving mechanism, a signal acquisition module, and a data processing unit. The flexible honeycomb sensor includes a flexible substrate, a honeycomb excitation coil unit, and a receiving coil. The honeycomb excitation coil unit consists of six equilateral triangular excitation coil units arranged in a ring array around the center of the flexible substrate and formed on the flexible substrate to form a honeycomb conformal array. The receiving coil is located at the center of the flexible substrate, and the current directions in adjacent equilateral triangular excitation coil units are opposite to form a multi-pole alternating magnetic field on the surface of the metal pipe under test. The excitation source is connected to the honeycomb excitation coil unit and is used to provide an alternating excitation current signal to the honeycomb excitation coil unit. The signal acquisition module is connected to the receiving coil and the data processing unit respectively and is used to acquire the voltage signal output by the receiving coil and transmit it to the data processing unit. The flexible honeycomb sensor is disposed on the sensor driving mechanism and slides along the surface of the metal pipe under test under its drive. The data processing unit extracts the defect feature slope based on the received voltage signal and calls a pre-trained defect depth prediction model to output the defect depth prediction result.

[0050] The equilateral triangular excitation coil unit consists of several layers of equilateral triangular excitation coils arranged vertically; the current direction of all equilateral triangular excitation coils in the same equilateral triangular excitation coil unit is the same; the current direction of equilateral triangular excitation coils located on the same layer in adjacent equilateral triangular excitation coil units is opposite; and all equilateral triangular excitation coils in the honeycomb excitation coil unit are connected in series.

[0051] In this embodiment, the equilateral triangular excitation coil unit consists of two layers of equilateral triangular excitation coils. For example... Figure 1 As shown, the winding method of the honeycomb excitation coil unit is as follows:

[0052] The wire is wound clockwise or counterclockwise from the outside in to the center or a predetermined number of turns to form the upper equilateral triangle excitation coil of the first equilateral triangle excitation coil unit. Then, it extends downwards to the lower layer and is wound clockwise or counterclockwise from the inside out to form the lower equilateral triangle excitation coil of the first equilateral triangle excitation coil unit. After the first equilateral triangle excitation coil unit is wound, the wire extends backwards to the location of the second equilateral triangle excitation coil unit and begins winding the second equilateral triangle excitation coil unit: the wire is wound counterclockwise or clockwise from the outside in to the center or a predetermined number of turns to form the lower equilateral triangle excitation coil of the second equilateral triangle excitation coil unit. Then, it extends upwards to the upper layer and is wound counterclockwise or clockwise from the inside out to form the upper equilateral triangle excitation coil of the second equilateral triangle excitation coil unit. After the second equilateral triangle excitation coil unit is wound, the wire continues to extend backwards, using the same method as the first and second equilateral triangle excitation coil units to wind subsequent equilateral triangle excitation coil units.

[0053] In this embodiment, laser etching technology is used to fabricate the excitation coil on a flexible polyimide (PI) substrate, allowing the sensor to conformally fit the tube wall and eliminate lift-off interference. The receiving coil, requiring a high number of turns to ensure a high signal-to-noise ratio, is still a rigid coil. A high-permeability graphite core is embedded in the center of the receiving coil to significantly compress the magnetic flux bundle, enhance local magnetic flux density, and improve the defect response amplitude.

[0054] In this embodiment, the sensor's geometric dimensions, number of turns, and other parameters are shown in Table 1.

[0055] Table 1 Sensor Parameters

[0056] parameter value <![CDATA[Inner side length l1 (mm) of the excitation coil]]> 1.5 <![CDATA[Outer side length l2 (mm) of the excitation coil]]> 8.5 <![CDATA[Height h1 (mm) of the excitation coil]]> 0.2 <![CDATA[Inner diameter r1 of the receiving coil (mm)]]> 1 <![CDATA[Outer diameter r2 of the receiving coil (mm)]]> 2 <![CDATA[Receiving coil height h2 (mm)]]> 1.5 <![CDATA[Number of turns N1 of the excitation coil]]> 18 <![CDATA[Number of turns N of the receiving coil 2 > 200 coil conductivity (S / m)

[0057] In this embodiment, the sensor driving mechanism is a three-axis motion mechanism that can adjust the horizontal position and height of the sensor and drive it to perform sliding motion. The signal acquisition module is an eddy current detector. The data processing unit is a computer with corresponding data processing and control programs installed. Preferably, the data processing unit is connected to both the excitation source and the sensor driving mechanism to control their operation.

[0058] like Figure 4 As shown, this embodiment also provides a method for detecting the depth of defects in metal pipelines based on a flexible honeycomb sensor. Implemented based on the above system, it includes the following steps:

[0059] S1: Select a flexible honeycomb sensor of appropriate size according to the size and specifications of the metal tube to be tested, install the flexible honeycomb sensor on the sensor driving mechanism, and then place the sensor driving mechanism next to the metal tube to be tested so that the flexible honeycomb sensor is attached to the surface of the metal tube to be tested.

[0060] S2: Start the excitation source and apply an alternating excitation current signal to the honeycomb excitation coil unit; at the same time, control the sensor drive mechanism to work, so that it drives the flexible honeycomb sensor to slide along the surface of the metal tube to be tested in order to scan the surface of the metal tube to be tested.

[0061] S3: The signal acquisition module acquires the voltage signal output by the coil during the scanning process of the flexible honeycomb sensor and transmits it to the data processing unit;

[0062] S4: The data processing unit obtains the real and imaginary parts of the impedance of the receiving coil based on the received voltage signal, maps the real and imaginary parts to the complex plane to form an impedance trajectory, and then extracts the defect feature slope k based on the impedance trajectory.

[0063] S5: Input the defect feature slope k into the pre-trained defect depth prediction model to obtain the defect depth prediction result of the metal pipe to be tested.

[0064] In step S4, the method for calculating the slope of the defect feature is as follows:

[0065]

[0066] Where k represents the slope of the defect feature, and These represent the real and imaginary parts of the receiving coil impedance, respectively. This represents the maximum value of the real part when the imaginary part is greater than 0. This represents the minimum value of the real part when the imaginary part is less than 0. It represents the change in the imaginary part between the maximum and minimum values ​​of the real part in a single defect signal.

[0067] In step S5, the defect depth prediction model is a machine learning model trained based on the correspondence between the defect feature slope k and the defect depth and tilt angle. The training method for the defect depth prediction model is as follows:

[0068] 1) Set up m metal tube samples with defects of different depths. Each metal tube sample has i prefabricated defects at different angles to the cross-section of the metal tube along the axial direction on its surface.

[0069] 2) The flexible honeycomb sensor is driven by the sensor driving mechanism to slide along the surface of each metal tube sample to scan the surface of each metal tube sample; each metal tube sample is scanned repeatedly n times; m×n×i sets of original defect sample data are obtained;

[0070] 3) Extract the defect feature slope k based on the obtained original defect sample data;

[0071] 4) Based on the extracted defect feature slope k, the Natural Gradient Boosting (NGBoost) algorithm is used to train the model and obtain the defect depth prediction model.

[0072] In this embodiment, the experimental sample is as follows: Figure 5 As shown, 10 6063 aluminum alloy tubes are selected, each with a length of 500 mm, an inner diameter of 16 mm, and an outer diameter of 20 mm. Seven defects are pre-cut on the surface, with an angle interval of 15°, covering the entire direction from 0° to 90°. The defect size is 10 mm in length, 0.2 mm in width, and the depth d is in the range of 0.2 to 2.0 mm, increasing in increments of 0.2 mm.

[0073] Then, the experimental setup was used to scan aluminum tubes with different defect orientations during the experiment. The excitation frequency was 30 kHz, the excitation current was 1 A, and the voltage signal of the receiving coil was acquired in real time using an eddy current detector. During acquisition, the sensor... Figure 5 The scanning direction shown was used to scan each metal tube sample 30 times, thereby obtaining 210 sets of valid data at each depth level, accumulating a total of 2100 sets of original defect sample data.

[0074] Then, sample defect feature extraction can be performed. After adaptive noise reduction and separation, the real part of the receiving coil impedance is obtained. With the imaginary part Mapped to the complex plane, typical defect loops appear in a figure-eight shape (e.g. Figure 6-7 (As shown). The trajectory was The zero axis divides the region into two approximately circular domains, one above the other. Experiments show that defects at the same depth but in different directions... The standard deviation of the value is <0.3; when the depth decreases from 2.0 mm to 0.2 mm, the characteristic slope... It monotonically increases from 3.240 to 13.864, exhibiting deep monotonicity and quantization resolution. Figure 8 The distribution of depth features in the defect dataset is shown. A line drawn from the mean slope of the feature for each defect class shows that the slope monotonically increases as the defect depth decreases.

[0075] Using the feature slope extracted earlier The defect depth quantization is performed using the Natural Gradient Boosting (NGBoost) algorithm. The training and test sets are divided in a 7:3 ratio. The NGBoost model structure is as follows: Figure 9 As shown, it comprises three components: the base learner, which is a weak learner based on the ensemble model, and the distribution parameters, which are stably updated in the direction that allows the scoring rule to decrease the fastest in the probability distribution space, through regression tree fitting. Simultaneously, the scoring rule... Maximum likelihood estimation (MLE) of the scoring function is used to measure the degree of matching between the current predicted distribution and the true label, driving the entire boosting process. This process is repeated until the scoring rule converges, ultimately outputting the complete conditional probability distribution. The prediction accuracy of the NGBoost model is shown in Table 2, with an average accuracy of 98.6% at different depths. Specific prediction results are as follows: Figure 10 As shown.

[0076] Table 2 NGBoost Defect Depth Prediction Results

[0077] Depth (mm) 2.0 1.8 1.6 1.4 1.2 ACC 98.4% 96.8% 98.4% 96.8% 96.8% Depth (mm) 1.0 0.8 0.6 0.4 0.2 ACC 100% 100% 98.4% 100% 100%

[0078] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0079] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0080] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0082] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A flexible honeycomb sensor based metal pipeline defect depth eddy current testing system, characterized in that, The system includes a flexible honeycomb sensor, an excitation source, a sensor driving mechanism, a signal acquisition module, and a data processing unit. The flexible honeycomb sensor comprises a flexible substrate, honeycomb excitation coil units, and a receiving coil. The honeycomb excitation coil units consist of six equilateral triangular excitation coil units arranged in a ring array around the center of the flexible substrate, forming a honeycomb conformal array. The receiving coil is positioned at the center of the flexible substrate, with current directions opposite in adjacent equilateral triangular excitation coil units to create a multi-pole alternating magnetic field on the surface of the metal tube under test. The excitation source is connected to the honeycomb excitation coil units to provide alternating excitation current signals. The signal acquisition module is connected to both the receiving coil and the data processing unit to acquire the voltage signal output by the receiving coil and transmit it to the data processing unit. The flexible honeycomb sensor is mounted on the sensor driving mechanism and slides along the surface of the metal tube under test under its drive. The data processing unit extracts the defect feature slope based on the received voltage signal and calls a pre-trained defect depth prediction model to output the defect depth prediction result.

2. The flexible cellular sensor based metal pipe defect depth eddy current testing system according to claim 1, wherein, The equilateral triangular excitation coil unit consists of several layers of equilateral triangular excitation coils arranged vertically; the current direction of all equilateral triangular excitation coils in the same equilateral triangular excitation coil unit is the same; the current direction of equilateral triangular excitation coils located on the same layer in adjacent equilateral triangular excitation coil units is opposite; all equilateral triangular excitation coils in the honeycomb excitation coil unit are connected in series as one unit.

3. The flexible cellular sensor based metal pipe defect depth eddy current testing system according to claim 2, characterized in that, The equilateral triangular excitation coil unit consists of two layers of equilateral triangular excitation coils; the winding method of the honeycomb excitation coil unit is as follows: The wire is wound clockwise or counterclockwise from the outside in to the center or a predetermined number of turns to form the upper equilateral triangle excitation coil of the first equilateral triangle excitation coil unit; then it extends downwards to the lower layer, and is wound clockwise or counterclockwise from the inside out to form the lower equilateral triangle excitation coil of the first equilateral triangle excitation coil unit; after the first equilateral triangle excitation coil unit is wound, the wire extends backwards to the position of the second equilateral triangle excitation coil unit and begins to wind the second equilateral triangle excitation coil unit: the wire is wound counterclockwise or clockwise from the outside in to the center or a predetermined number of turns to form the lower equilateral triangle excitation coil of the second equilateral triangle excitation coil unit; then it extends upwards to the upper layer, and is wound counterclockwise or clockwise from the inside out to form the upper equilateral triangle excitation coil of the second equilateral triangle excitation coil unit; After the second equilateral triangle excitation coil unit is wound, the wire continues to extend backward, and the subsequent equilateral triangle excitation coil units are wound using the same method as the first and second equilateral triangle excitation coil units.

4. The flexible cellular sensor based metal pipe defect depth eddy current testing system of claim 1, wherein, The slope of the defect feature is: Where k represents the slope of the defect feature, and These represent the real and imaginary parts of the receiving coil impedance, respectively. This represents the maximum value of the real part when the imaginary part is greater than 0. This represents the minimum value of the real part when the imaginary part is less than 0. It represents the change in the imaginary part between the maximum and minimum values ​​of the real part.

5. The eddy current detection system for metal pipeline defects based on a flexible honeycomb sensor according to claim 1, characterized in that, A high-permeability magnetic core is embedded in the middle of the receiving coil.

6. The eddy current detection system for metal pipeline defects based on a flexible honeycomb sensor according to claim 1, characterized in that, The data processing unit is connected to the excitation source and the sensor driving mechanism respectively, so as to control the operation of the excitation source and the sensor driving mechanism respectively.

7. A method for detecting the depth of defects in metal pipelines based on a flexible honeycomb sensor, characterized in that, Based on the system implementation as described in any one of claims 1-6, the system includes the following steps: S1: Select a flexible honeycomb sensor of appropriate size according to the size and specifications of the metal tube to be tested, install the flexible honeycomb sensor on the sensor driving mechanism, and then place the sensor driving mechanism next to the metal tube to be tested so that the flexible honeycomb sensor is attached to the surface of the metal tube to be tested. S2: Start the excitation source and apply an alternating excitation current signal to the honeycomb excitation coil unit; at the same time, control the sensor drive mechanism to work, so that it drives the flexible honeycomb sensor to slide along the surface of the metal tube to be tested in order to scan the surface of the metal tube to be tested. S3: The signal acquisition module acquires the voltage signal output by the coil during the scanning process of the flexible honeycomb sensor and transmits it to the data processing unit; S4: The data processing unit obtains the real and imaginary parts of the impedance of the receiving coil based on the received voltage signal, maps the real and imaginary parts to the complex plane to form an impedance trajectory, and then extracts the defect feature slope k based on the impedance trajectory. S5: Input the defect feature slope k into the pre-trained defect depth prediction model to obtain the defect depth prediction result of the metal pipe to be tested.

8. The method for detecting the depth of defects in metal pipelines based on a flexible honeycomb sensor according to claim 7, characterized in that, The method for calculating the slope of the defect feature is as follows: Where k represents the slope of the defect feature, and These represent the real and imaginary parts of the receiving coil impedance, respectively. This represents the maximum value of the real part when the imaginary part is greater than 0. This represents the minimum value of the real part when the imaginary part is less than 0. It represents the change in the imaginary part between the maximum and minimum values ​​of the real part.

9. The method for detecting the depth of defects in metal pipelines based on a flexible honeycomb sensor according to claim 7, characterized in that, The defect depth prediction model is a machine learning model trained based on the correspondence between the defect feature slope k and the defect depth and tilt angle.

10. The method for detecting the depth of defects in metal pipelines based on a flexible honeycomb sensor according to claim 9, characterized in that, The training method for the defect depth prediction model is as follows: 1) Set up m metal tube samples with defects of different depths. Each metal tube sample has i prefabricated defects at different angles to the cross-section of the metal tube along the axial direction on its surface. 2) The flexible honeycomb sensor is driven by the sensor driving mechanism to slide along the surface of each metal tube sample to scan the surface of each metal tube sample; each metal tube sample is scanned repeatedly n times; m×n×i sets of original defect sample data are obtained; 3) Extract the defect feature slope k based on the obtained original defect sample data; 4) Based on the extracted defect feature slope k, the natural gradient boosting algorithm is used to train the model and obtain the defect depth prediction model.

Images