Inverse Gaussian resonance regulation and control distribution eddy current detection sensor and detection method

Abstract

The invention provides an eddy current detection sensor with inverse Gaussian resonance regulation and control distribution and a detection method, and relates to the technical field of eddy current detection, and the eddy current detection sensor comprises an excitation coil layer and a detection coil layer which are stacked; the excitation coil layer comprises a first PCB, the first PCB is provided with two rectangular spiral excitation coils arranged side by side, the two spiral excitation coils are connected in parallel, and the directions of currents in the two spiral excitation coils are opposite; when excitation signals are introduced into the two spiral excitation coils, a differential magnetic field is formed; the detection coil layer comprises a second PCB, a plurality of rectangular spiral detection coils and a plurality of capacitors are arranged on the second PCB, the spiral detection coils and the capacitors are connected in parallel in a one-to-one correspondence mode, and a low-pass filter is formed by connecting the capacitors in parallel; the number of turns of the plurality of spiral detection coils is in inverse Gaussian distribution according to an arrangement sequence; the spiral excitation coil and the spiral detection coil are arranged at an interval, and the spiral excitation coil and the spiral detection coil are interactively coupled in the vertical direction.

Description

Technical Field

[0001] The present invention relates to the technical field of eddy current detection, and in particular to an eddy current detection sensor and a detection method with inverse Gaussian resonance controlled distribution. Background Art

[0002] Basic components such as pipelines and aerospace will inevitably lead to aging, wear and blockage in long-term operation. In order to prevent these problems from accelerating and causing more serious social hazards, regular inspection, cleaning and maintenance are essential. Non-destructive testing technology is widely used in the industrial field due to its effectiveness in ensuring the integrity of workpieces and operational efficiency. Among them, ultrasonic testing can effectively detect and quantify internal and external corrosion defects in thick-walled pipelines. However, it requires the use of indispensable coupling agents, which limits its application. Magnetic flux leakage testing is one of the most widely used technologies in long-distance oil and gas pipeline inspection, which can effectively detect metal loss, cracks and weld defects. However, due to the weight and complex structure of permanent magnets, it faces challenges when passing through small-diameter pipelines and 1.5D elbows at low pressure. In contrast, eddy current testing is a promising method for small-diameter pipeline inspection with the advantages of small size, light weight and low cost. However, under random working conditions, the effects of lift-off and background noise hinder the induction of detectable electromotive force in the secondary coil. This effect is particularly evident when using sinusoidal excitation, as the sensitivity decreases significantly after a distance of more than 5 mm. In summary, balancing sensitivity and resolution is crucial to better identify and quantify defects. This principle applies not only to eddy current systems but also to other application scenarios.

[0003] To address the above challenges, array probe package sensors based on GMR and TMR can provide high sensitivity and a small effective sensing area for detecting eddy current signals. However, these sensors are not better than coil-based sensors when detecting surface features at high frequencies. In addition, since these sensors only directly detect the magnetic field distribution through devices such as giant magnetoresistance, their sensitivity depends entirely on the excitation magnetic field, while ignoring the relationship between sensitivity and receiving field. Balancing sensitivity and resolution remains a key issue in the current field of eddy current detection. In practice, in order to improve the spatial resolution of the sensor array while maintaining sufficient sensitivity, coils with smaller sizes and more turns are usually used. Although the inductance value of a single coil can be increased by adding an iron core to the cylindrical coil, this method requires more space. In addition, as the inductance increases, the resistance also increases significantly, resulting in higher eddy current losses. However, in the case of limited space, the printing technology of multi-layer coils is required to improve the inductance of the coil and the sensitivity of the probe, but this increases the cost. Summary of the invention

[0004] In order to solve the technical problems of the current eddy current detection technology that it is difficult to balance the sensitivity and resolution, the detection equipment requires more space and the manufacturing cost is high, the present invention provides an eddy current detection sensor and a detection method with inverse Gaussian resonance controlled distribution.

[0005] The technical solution provided by the embodiment of the present invention is as follows:

[0006] An eddy current detection sensor with inverse Gaussian resonance control distribution provided by an embodiment of the present invention comprises: an excitation coil layer and a detection coil layer stacked;

[0007] The excitation coil layer comprises a first PCB board, on which two rectangular spiral excitation coils are arranged side by side, the two spiral excitation coils are connected in parallel, and the current directions of the two spiral excitation coils are opposite;

[0008] When an excitation signal is applied to the two spiral excitation coils, a differential magnetic field is formed between the two spiral excitation coils;

[0009] The detection coil layer includes a second PCB board, on which a plurality of rectangular spiral detection coils and a plurality of capacitors are arranged, the spiral detection coils and the capacitors are connected in parallel in a one-to-one correspondence, and a low-pass filter is formed by connecting the capacitors in parallel;

[0010] The number of turns of the plurality of spiral detection coils is inversely Gaussian distributed according to the arrangement order;

[0011] The spiral excitation coil and the spiral detection coil are arranged at an interval, and the spiral excitation coil and the spiral detection coil are interactively coupled in a vertical direction.

[0012] A detection method provided by an embodiment of the present invention is applied to the above-mentioned eddy current detection sensor with inverse Gaussian resonance control distribution, and the detection method specifically includes:

[0013] S1: placing the spiral detection coil between the object to be detected and the spiral excitation coil;

[0014] S2: inputting an excitation signal to the two spiral excitation coils to form a differential magnetic field between the two spiral excitation coils;

[0015] S3: acquiring eddy current signals respectively through the plurality of spiral detection coils;

[0016] S4: performing defect detection on the part to be detected according to the eddy current signal.

[0017] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0018] (1) In the present invention, the number of turns of the detection coil is in the form of an inverse Gaussian distribution, so that the magnetic field received by the detection signal in the non-uniform magnetic field has a high consistency, each detection coil has a similar sensitivity, and the detection resolution is increased by arranging multiple coils.

[0019] (2) In the present invention, the spiral excitation coil and the spiral detection coil are arranged at intervals, and the spiral excitation coil and the spiral detection coil are interactively coupled in the vertical direction. The spiral detection coil array is located between the spiral excitation coil and the test piece in space. On the one hand, the direct coupling energy of the excitation coil and the detection coil is reduced, and on the other hand, the coupling path between the test piece and the detection coil array is reduced, thereby increasing the voltage intensity of the excitation coil and the detection coil related to the test piece in the receiving coil. When the space size and the receiving inductance are limited, and, through the interactive coupling in the vertical direction, the magnetic field generated by the excitation coil can first store a part of the energy through the receiving coil, and then be inductively coupled with the test piece, thereby improving the coupling ability of small inductance weak signals, allowing the detection coil to receive more test piece information, thereby improving the detection sensitivity.

[0020] (3) In the present invention, the spiral detection coil forms a low-pass filter by connecting the capacitor in parallel, thereby improving the eddy current signal strength under defect disturbance. The adjustment and layout of different numbers of turns and capacitors help to adjust the resonant frequency of each coil and optimize the resonance effect of the array to adapt to different working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 A schematic diagram of the structure of an eddy current detection sensor with inverse Gaussian resonance control distribution provided by an embodiment of the present invention;

[0023] Figure 2 A schematic structural diagram of an excitation coil layer provided by an embodiment of the present invention;

[0024] Figure 3 A schematic diagram of the structure of a detection coil layer provided by an embodiment of the present invention;

[0025] Figure 4 A diagram of detection results of pipelines with different defects provided by an embodiment of the present invention.

[0026] [Reference Signs]

[0027] 1. First PCB board; 2. Spiral excitation coil; 3. Spiral detection coil; 4. Capacitor. DETAILED DESCRIPTION

[0028] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0029] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0030] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0031] Reference Manual Attached Figure 1 , shows a schematic structural diagram of an eddy current detection sensor with inverse Gaussian resonance control distribution provided by an embodiment of the present invention.

[0032] Reference Manual Attached Figure 2 , shows a schematic structural diagram of an excitation coil layer provided in an embodiment of the present invention.

[0033] Reference Manual Attached Figure 3 , shows a schematic structural diagram of a detection coil layer provided by an embodiment of the present invention.

[0034] An embodiment of the present invention provides an eddy current detection sensor with inverse Gaussian resonance control distribution, comprising: an excitation coil layer and a detection coil layer that are stacked.

[0035] The excitation coil layer includes a first PCB (Printed Circuit Board) board 1, on which two rectangular spiral excitation coils 2 arranged side by side are arranged. The two spiral excitation coils 2 are connected in parallel, and the current directions of the two spiral excitation coils 2 are opposite.

[0036] When an excitation signal is applied to the two spiral excitation coils 2, a differential magnetic field is formed between the two spiral excitation coils 2. The differential magnetic field has a unique characteristic: due to the opposite current directions, their magnetic fields cancel each other out in some areas, while reinforcing each other in other areas. This differential magnetic field can effectively reduce external noise interference and enhance the magnetic field strength in a local area, so that the eddy current detection sensor has higher sensitivity and accuracy when performing defect detection.

[0037] The detection coil layer includes a second PCB board, on which a plurality of rectangular spiral detection coils 3 and a plurality of capacitors are arranged. The spiral detection coils 3 are connected in parallel with the capacitors in a one-to-one correspondence, and a low-pass filter is formed by connecting the capacitors in parallel.

[0038] It should be noted that the low-pass filter can effectively filter out the high-frequency noise in the eddy current signal and retain the effective signal in the low-frequency part. This can reduce the influence of environmental interference and irrelevant signals and improve the quality and stability of the signal.

[0039] Among them, the capacitance of the capacitor can be adjusted according to actual needs.

[0040] The number of turns of the plurality of spiral detection coils 3 is inversely Gaussian distributed according to the arrangement order.

[0041] Among them, the Inverse Gaussian Distribution is a continuous probability distribution, which is usually used to describe random variables with positive skewness. Its probability density function PDF presents a single peak and a long right tail.

[0042] Furthermore, the reason why the number of turns of the multiple spiral detection coils 3 is arranged in an inverse Gaussian distribution in the order of arrangement is that it was found during the research process that the eddy current distribution / magnetic flux distribution of the differential magnetic field formed between the two spiral excitation coils 2 in the entire area presents a Gaussian distribution. Therefore, it is also decided that the number of turns of the multiple spiral detection coils 3 adopts an inverse Gaussian distribution, so that the magnetic field received by the detection signal in the inhomogeneous magnetic field has a higher consistency, and each detection coil has similar sensitivity. The detection resolution is increased by arranging multiple coils.

[0043] The spiral excitation coil 2 and the spiral detection coil 3 are arranged at intervals, and the spiral excitation coil 2 and the spiral detection coil 3 are interactively coupled in a vertical direction.

[0044] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0045] (1) In the present invention, the number of turns of the detection coil is in the form of an inverse Gaussian distribution, so that the magnetic field received by the detection signal in the non-uniform magnetic field has a high consistency, each detection coil has a similar sensitivity, and the detection resolution is increased by arranging multiple coils.

[0046] (2) In the present invention, the spiral excitation coil and the spiral detection coil are arranged at intervals, and the spiral excitation coil and the spiral detection coil are interactively coupled in the vertical direction. The spiral detection coil array is located between the spiral excitation coil and the test piece in space. On the one hand, the direct coupling energy of the excitation coil and the detection coil is reduced, and on the other hand, the coupling path between the test piece and the detection coil array is reduced, thereby increasing the voltage intensity of the excitation coil and the detection coil related to the test piece in the receiving coil. When the space size and the receiving inductance are limited, and, through the interactive coupling in the vertical direction, the magnetic field generated by the excitation coil can first store a part of the energy through the receiving coil, and then be inductively coupled with the test piece, thereby improving the coupling ability of small inductance weak signals, allowing the detection coil to receive more test piece information, thereby improving the detection sensitivity.

[0047] (3) In the present invention, the spiral detection coil forms a low-pass filter by connecting the capacitor in parallel, thereby improving the eddy current signal strength under defect disturbance. The adjustment and layout of different numbers of turns and capacitors help to adjust the resonant frequency of each coil and optimize the resonance effect of the array to adapt to different working conditions.

[0048] The number of the spiral detection coils 3 can be set by those skilled in the art according to actual conditions, and the present invention does not limit this.

[0049] For example, the number of the spiral detection coils 3 is 9, and the number of turns of the 9 spiral detection coils 3 is inverse Gaussian distribution according to the arrangement order.

[0050] Furthermore, the numbers of turns of the nine spiral detection coils 3 are 39, 34, 29, 25, 19, 25, 29, 34, and 39 respectively.

[0051] It should be noted that the numbers of turns of 39, 34, 29, 25, 19, 25, 29, 34, and 39 are respectively a typical inverse Gaussian distribution scheme.

[0052] In a possible implementation manner, the length and width of the spiral detection coil 3 are consistent.

[0053] In the present invention, the spiral coils of uniform size can ensure that the electromagnetic field is evenly distributed in each coil area, thereby improving the sensitivity and uniformity of the detection area and facilitating the detection of small defects.

[0054] In a possible implementation manner, the symmetry center formed by the two spiral excitation coils 2 coincides with the symmetry center formed by the plurality of spiral detection coils 3 .

[0055] In the present invention, when the symmetry centers of the excitation coil and the detection coil coincide, the magnetic field generated by the excitation coil can be evenly distributed in the detection coil area, thereby improving the symmetry and stability of the magnetic field and facilitating more accurate defect detection.

[0056] In a possible implementation, the structural parameters of the eddy current detection sensor are optimized based on the Seagull optimization algorithm.

[0057] Among them, Seagull Optimization Algorithm (SOA) is a nature-inspired optimization algorithm that simulates the foraging behavior of seagulls.

[0058] Optionally, the structural parameters include: the wire diameter, number of turns, number of layers, distance between each layer of coils, distance between two coils of the spiral excitation coil 2, the number of spiral detection coils 3, wire diameter, number of turns, number of layers, distance between each layer of coils, distance between each coil, and capacitance of the capacitor.

[0059] Specifically, based on the Seagull optimization algorithm, the specific methods for optimizing the structural parameters of the eddy current detection sensor include:

[0060] The structural optimization fitness function of the Seagull optimization algorithm is set with the goal of improving the received eddy current signal strength, magnetic field uniformity, and reducing the eddy current signal-to-noise ratio:

[0061] ;

[0062] Where f represents the structural optimization fitness function, J k represents the eddy current signal strength of the kth spiral detection coil, B k represents the magnetic field strength received by the k-th spiral detection coil, Represents the average magnetic field strength received by multiple spiral detection coils, SNR k represents the signal-to-noise ratio of the eddy current signal of the kth spiral detection coil, K represents the total number of spiral detection coils, λ1 represents the weight coefficient of the eddy current signal intensity, λ2 represents the weight coefficient of the magnetic field uniformity, and λ3 represents the weight coefficient of the eddy current signal signal-to-noise ratio.

[0063] Among them, those skilled in the art can set the weight coefficient λ1 of the eddy current signal intensity, the weight coefficient λ2 of the magnetic field uniformity, and the weight coefficient λ3 of the eddy current signal signal-to-noise ratio according to actual conditions, and the present invention does not limit this.

[0064] In the present invention, by setting a structural optimization fitness function based on the Seagull optimization algorithm, the purpose is to improve the eddy current signal strength, optimize the magnetic field uniformity and reduce the signal-to-noise ratio, which can significantly improve the overall performance of the eddy current detection sensor. By considering the weighted sum of the three key indicators of eddy current signal strength, magnetic field uniformity and signal-to-noise ratio, the fitness function can comprehensively optimize the design parameters of the sensor to ensure a stronger and more stable signal, while reducing noise interference and improving detection accuracy. By balancing these three goals, the optimized sensor not only improves sensitivity and resolution, but also maintains good performance under different working environments, thereby enhancing the detection capability of small object defects.

[0065] Initialize the seagull individuals. Each seagull individual represents a feasible set of structural parameters. Each seagull individual consists of multiple dimensional components, and each component represents a structural parameter.

[0066] In the global search phase, avoid collisions and move towards the optimal individual:

[0067] ;

[0068] ;

[0069] ;

[0070] in, represents the position of the i-th seagull individual after the global search phase at the t-th iteration, represents the position of the i-th seagull after anti-collision processing at the t-th iteration, A represents the control factor, represents the position of the i-th seagull at the t-th iteration, represents the displacement of the i-th seagull individual towards the optimal individual at the t-th iteration, B represents the search balance factor, represents the optimal individual position at the tth iteration.

[0071] In the present invention, in the global search stage, the strategy of avoiding collision and moving towards the optimal individual helps to improve the search efficiency and stability of the seagull optimization algorithm. Through the control factor and the search balance factor, the algorithm can effectively guide each seagull individual to avoid collision with other individuals in the search space, while ensuring that the individual moves towards the optimal solution. This process not only accelerates the convergence speed, but also ensures the exploration ability of the global search stage, so that the algorithm can widely explore the solution space and avoid falling into the local optimal solution, thereby improving the quality of the final solution and the diversity of the global search.

[0072] ;

[0073] Among them, t represents the current number of iterations, T represents the maximum number of iterations, and f c Indicates a linearly decreasing frequency.

[0074] In the present invention, as the number of iterations increases, the control factor will gradually decrease. This means that in the early stages of the search, the individual search range is large and can be widely explored; as the iterations proceed, the search range gradually narrows, and more energy is concentrated near the optimal solution, which helps to accelerate convergence.

[0075] ;

[0076] Among them, r1 represents a random number between 0 and 1.

[0077] In the present invention, as iterations proceed, changes in the search balance factor can gradually guide the individuals to converge toward the global optimal solution, thereby improving the global search capability and accuracy of the optimization algorithm.

[0078] In the local search phase, a random number r2 is generated. Based on the random number r2, the spiral search strategy and the encirclement strategy are selected in parallel, and the displacement is performed in a spiral motion:

[0079] ;

[0080] ;

[0081] ;

[0082] ;

[0083] ;

[0084] in, represents the position of the i-th seagull after spiral motion at the t-th iteration, x represents the spiral flight coefficient in the x-direction, y represents the spiral flight coefficient in the y-direction, z represents the spiral flight coefficient in the z-direction, r represents the radius of the spiral flight trajectory, θ represents a random number between 0 and 2π, u and v represent spiral constants, and e represents a natural constant.

[0085] In the present invention, by generating a random number to select between the spiral search strategy and the encirclement strategy, and combining the spiral motion for displacement, the diversity and flexibility of the search process can be significantly enhanced. When , a spiral search strategy is adopted, so that the individual can make spiral movements in the solution space, thereby widely exploring the surrounding area; when The encirclement strategy is adopted when searching locally, so that individuals can focus on the optimal solution more accurately. The algorithm can maintain strong exploration ability during local search and converge to the optimal solution efficiently, thereby improving the accuracy and convergence speed of the optimization process. In addition, the introduction of spiral motion trajectory enhances the randomness and directionality of the search process, making the algorithm less likely to fall into the local optimum, ensuring a higher probability of the global optimal solution.

[0086] Perform mutation operations on each seagull individual:

[0087] ;

[0088] in, represents the position of the ith seagull individual after mutation at the tth iteration, P r represents a random individual, and ω represents an adaptive scaling factor.

[0089] In the present invention, the diversity of search and global exploration capability can be increased by performing mutation operation. The mutation operation helps individuals jump out of the local optimal solution and avoids the algorithm from converging to the suboptimal solution in the early stage by introducing the difference between a random individual and the current optimal individual.

[0090] ;

[0091] Among them, ω max represents the maximum scale factor, ω min Indicates the minimum scale factor.

[0092] In the present invention, as the number of iterations increases, the scale factor will gradually decrease, starting from a larger value and eventually tending to a smaller value, which helps to maintain a larger search range in the early stage of optimization and encourages extensive exploration of the solution space. When the optimization gradually converges in the later stage, by reducing the scale factor, the local search accuracy is enhanced, so that the algorithm can more carefully approach the global optimal solution. This method can effectively balance the ability of exploration and development, avoid premature convergence and improve the quality of the final solution.

[0093] Determine whether the fitness value of the position after mutation is greater than the fitness value of the position before mutation. If so, use the position after mutation to replace the position before mutation. Otherwise, keep the position before mutation unchanged.

[0094] Update the fitness value of each seagull individual and the global optimal individual.

[0095] Determine whether the current number of iterations has reached the maximum number of iterations. If so, output the set of structural parameters representing the seagull individual with the highest current fitness. Otherwise, return to continue iterating.

[0096] In the present invention, the structural parameters of the eddy current detection sensor (such as the number of turns, size, spacing, etc. of the coil) are optimized by the Seagull optimization algorithm, which can improve the sensitivity, resolution and signal strength of the sensor, thereby achieving higher-precision defect detection, especially in the detection of tiny defects, the optimized sensor can provide better performance.

[0097] The embodiment of the present invention provides a detection method, which is applied to the above-mentioned eddy current detection sensor with inverse Gaussian resonance control distribution. The detection method specifically includes:

[0098] S1: The spiral detection coil 3 is arranged between the object to be detected and the spiral excitation coil 2.

[0099] S2: Input an excitation signal to the two spiral excitation coils 2 to form a differential magnetic field between the two spiral excitation coils 2.

[0100] S3: acquiring eddy current signals respectively through the plurality of spiral detection coils 3 .

[0101] S4: performing defect detection on the part to be detected according to the eddy current signal.

[0102] In the present invention, by setting the spiral detection coil between the part to be detected and the spiral excitation coil, the sensitivity and accuracy of small object defects can be effectively improved by using the differential magnetic field. The differential magnetic field reduces the interference of external noise, making the detection signal clearer. The parallel operation of multiple spiral detection coils can obtain eddy current signals from different positions, further enhancing the comprehensiveness and reliability of the detection. In this way, tiny defects in the part to be detected can be accurately detected, improving the performance and accuracy of eddy current detection.

[0103] In a possible implementation manner, the S4 specifically includes sub-steps S401 to S410:

[0104] S401: performing registration processing on the eddy current signals acquired by the plurality of spiral detection coils 3.

[0105] Specifically, the registration can be performed by identifying features in the signal such as peaks, local extrema, etc., and then aligning these feature points.

[0106] It should be noted that, through the registration process, the data of different detection coils can be correctly aligned in space, which can enhance the detection capability of defects, especially for defects with tiny or complex shapes, and improve the overall detection accuracy and sensitivity.

[0107] S402: Extracting signal characteristic graphs of each eddy current signal through continuous wavelet transform.

[0108] Among them, Continuous Wavelet Transform (CWT) is a mathematical tool for signal analysis, which can decompose the signal into components with different frequency and time resolution.

[0109] S403: For the eddy current signal of the single spiral detection coil 3 and the corresponding signal characteristic graph, extract the characteristics of the one-dimensional eddy current signal through a first convolution channel to obtain a first characteristic graph.

[0110] The first convolution channel contains 10 consecutive one-dimensional convolution layers, and the output features of each one-dimensional convolution layer are specifically: FConv i , FConv i represents the output of the i-th one-dimensional convolutional layer, .

[0111] S404: Perform jump connections on the output features of each one-dimensional convolutional layer to form multiple one-dimensional intermediate features:

[0112] ;

[0113] ;

[0114] ;

[0115] ;

[0116] in, represents the j-th one-dimensional intermediate feature of the eddy current data of the u-th spiral detection coil, , up represents the upsampling operation, which is used to ensure the consistency of the feature map size. Represents an element-by-element addition operation.

[0117] In the present invention, a jump connection mechanism is innovatively adopted. The jump connection combines feature information at different levels, so that the network can retain more details and context information. Especially when processing deep networks, it can avoid information loss and improve the representation ability of the model, so that the network can better capture the eddy current signal characteristics of different scales, thereby enhancing the accuracy and sensitivity of defect detection.

[0118] S405: Perform feature extraction on the two-dimensional signal feature map through a second convolution channel to obtain a second feature map.

[0119] The second convolution channel contains 10 consecutive two-dimensional convolutional layers, and the output features of each two-dimensional convolutional layer are specifically: SConv i , SConv i represents the output of the i-th one-dimensional convolutional layer, .

[0120] S406: Perform jump connections on the output features of each two-dimensional convolutional layer to form multiple two-dimensional intermediate features:

[0121] ;

[0122] ;

[0123] ;

[0124] ;

[0125] in, represents the j-th two-dimensional intermediate feature of the eddy current data of the u-th spiral detection coil, and up represents the upsampling operation, which is used to ensure the consistency of the feature map size. Represents an element-wise addition operation.

[0126] S407: Concatenate and fuse the one-dimensional intermediate features with the two-dimensional intermediate features to obtain a fused feature map:

[0127] ;

[0128] in, represents the j-th fused feature map of the eddy current data of the u-th spiral detection coil, and flatten represents the flattening operation.

[0129] In the present invention, one-dimensional features and two-dimensional features represent the information of eddy current signals in different dimensions. By splicing and fusing one-dimensional intermediate features and two-dimensional intermediate features to obtain a fused feature map, feature information of different scales can be effectively combined to enhance the performance of the model, thereby improving the accuracy and robustness of defect detection.

[0130] S408: Perform defect detection on the part to be inspected according to the fused feature map:

[0131] ;

[0132] in, represents the defect detection probability set of the jth fusion feature map for the uth spiral detection coil eddy current data, represents the probability value that the jth fusion feature map of the uth spiral detection coil eddy current data belongs to the kth defect category, K represents the total number of defect categories, Softmax represents the Softmax activation function, and W j represents the weight matrix for the jth fusion feature map, b j Represents the bias value for the jth fused feature map.

[0133] In the present invention, by using the Softmax function to perform defect detection on the fused feature map, the extracted features can be effectively mapped to different defect categories. The Softmax function converts the output of each feature map into a probability value for each defect category, which helps to accurately classify the eddy current signal and ensures more reliable identification of different types of defects.

[0134] S409: Summarize the defect detection results of the eddy current data of each spiral detection coil:

[0135] ;

[0136] Among them, p k represents the probability value of belonging to the kth defect category, and U represents the total number of spiral detection coils.

[0137] In the present invention, by summarizing the defect detection results of the eddy current data of each spiral detection coil, the output of multiple detection coils can be comprehensively considered to improve the accuracy and stability of defect classification. By averaging the defect detection probability of each spiral detection coil, the balanced contribution of different coil data is ensured, making the final defect category probability more reliable. Such a method can reduce the noise impact that a single coil may bring, and through the information fusion of multiple coils, enhance the overall sensitivity and robustness of the system, thereby improving the accuracy of eddy current detection, especially when dealing with complex or small defects.

[0138] S410: The defect category with the largest probability value is taken as the final detection result of the eddy current detection.

[0139] In the present invention, one-dimensional and two-dimensional convolution channels are used to extract deep features of the signal respectively, and multi-level features are fused through jump connections and upsampling operations, which helps to enhance signal information at different scales. By fusing multi-scale features and combining data from multiple spiral detection coils, the accuracy and reliability of detection are improved.

[0140] Reference Manual Attached Figure 4 , showing a test result diagram of a pipeline with different defects provided by an embodiment of the present invention. Test results of a pulling tube with different defects. By arranging and imaging the time series signals of multiple channels, it can be clearly seen that each channel responds to the circumferential weld and spiral weld areas, so that the two can be clearly distinguished. In addition, abnormal signals detected between welds are identified as defects. After comparison with the defect list, it is confirmed that all defects in the pipeline are effectively detected.

[0141] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0142] There are a few points to note:

[0143] (1) The drawings of the embodiments of the present invention only involve structures related to the embodiments of the present invention. Other structures may refer to conventional designs.

[0144] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of the film or region is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a film, a region or a substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0145] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0146] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An eddy current detection sensor with inverse Gaussian resonance control distribution, characterized in that: include: An excitation coil layer and a detection coil layer are stacked; The excitation coil layer comprises a first PCB board (1), on which two rectangular spiral excitation coils (2) arranged side by side are arranged, the two spiral excitation coils (2) being connected in parallel, and the current directions in the two spiral excitation coils (2) being opposite; When an excitation signal is introduced into the two spiral excitation coils (2), a differential magnetic field is formed between the two spiral excitation coils (2); The detection coil layer comprises a second PCB board, on which a plurality of rectangular spiral detection coils (3) and a plurality of capacitors (4) are arranged, the spiral detection coils (3) and the capacitors (4) being connected in parallel in a one-to-one correspondence, and a low-pass filter is formed by connecting the capacitors (4) in parallel; The number of turns of the plurality of spiral detection coils (3) is inversely Gaussian distributed according to the arrangement order; The spiral excitation coil (2) and the spiral detection coil (3) are arranged at intervals, and the spiral excitation coil (2) and the spiral detection coil (3) are interactively coupled in a vertical direction.

2. The eddy current detection sensor with inverse Gaussian resonance control distribution according to claim 1 is characterized in that: The number of the spiral detection coils (3) is 9, and the number of turns of the 9 spiral detection coils (3) is inverse Gaussian distribution according to the arrangement order.

3. The eddy current detection sensor with inverse Gaussian resonance control distribution according to claim 2 is characterized in that: The numbers of turns of the nine spiral detection coils (3) are 39, 34, 29, 25, 19, 25, 29, 34, and 39 respectively.

4. The eddy current detection sensor with inverse Gaussian resonance control distribution according to claim 1, characterized in that: The length and width of the spiral detection coil (3) are consistent.

5. The eddy current detection sensor with inverse Gaussian resonance control distribution according to claim 1, characterized in that: The symmetry center formed by the two spiral excitation coils (2) coincides with the symmetry center formed by the plurality of spiral detection coils (3).

6. The eddy current detection sensor with inverse Gaussian resonance control distribution according to claim 1, characterized in that: Based on the Seagull optimization algorithm, the structural parameters of the eddy current detection sensor are optimized.

7. The eddy current detection sensor with inverse Gaussian resonance control distribution according to claim 6, characterized in that: The structural parameters include: The wire diameter, number of turns, number of layers, distance between each layer of coils, and distance between two coils of the spiral excitation coil (2); The number, wire diameter, number of turns, number of layers, distance between each layer of coils, and distance between each coil of the spiral detection coil (3); Capacitance of capacitor (4).

8. The eddy current detection sensor with inverse Gaussian resonance control distribution according to claim 6, characterized in that: The structural optimization fitness function of the Seagull optimization algorithm is set with the goal of improving the received eddy current signal strength, magnetic field uniformity, and reducing the eddy current signal-to-noise ratio: ； Where f represents the structural optimization fitness function, J k represents the eddy current signal strength of the kth spiral detection coil, B k represents the magnetic field strength received by the k-th spiral detection coil, Represents the average magnetic field strength received by multiple spiral detection coils, SNR k represents the signal-to-noise ratio of the eddy current signal of the kth spiral detection coil, K represents the total number of spiral detection coils, λ1 represents the weight coefficient of the eddy current signal intensity, λ2 represents the weight coefficient of the magnetic field uniformity, and λ3 represents the weight coefficient of the eddy current signal signal-to-noise ratio.

9. A detection method, characterized in that: The eddy current detection sensor with inverse Gaussian resonance control distribution applied to any one of claims 1 to 8, wherein the detection method specifically comprises: S1: arranging the spiral detection coil (3) between the object to be detected and the spiral excitation coil (2); S2: inputting an excitation signal into the two spiral excitation coils (2) to form a differential magnetic field between the two spiral excitation coils (2); S3: acquiring eddy current signals respectively through the plurality of spiral detection coils (3); S4: performing defect detection on the part to be detected according to the eddy current signal.

10. The detection method according to claim 9, characterized in that: The S4 specifically includes: S401: performing registration processing on eddy current signals acquired by the plurality of spiral detection coils (3); S402: extracting a signal characteristic graph of each eddy current signal through continuous wavelet transform; S403: for the eddy current signal of the single spiral detection coil (3) and the corresponding signal characteristic graph, extract the characteristics of the one-dimensional eddy current signal through a first convolution channel to obtain a first characteristic graph; S404: Perform jump connections on the output features of each one-dimensional convolutional layer to form multiple one-dimensional intermediate features; S405: extracting features from the two-dimensional signal feature map through a second convolution channel to obtain a second feature map; S406: Perform jump connections on the output features of each two-dimensional convolutional layer to form multiple two-dimensional intermediate features; S407: Concatenate and fuse the one-dimensional intermediate feature and the two-dimensional intermediate feature to obtain a fused feature map; S408: performing defect detection on the part to be inspected according to the fused feature map; S409: Summarizing defect detection results of eddy current data of each spiral detection coil; S410: The defect category with the largest probability value is taken as the final detection result of the eddy current detection.

Images