Carbon fiber inhaul cable multi-defect detection method

Abstract

The invention provides a multi-defect detection method for a carbon fiber inhaul cable. The multi-defect detection method comprises the following steps: establishing a sensor simulation model, a reference carbon fiber rod simulation model and a carbon fiber inhaul cable simulation model; calculating a sensitivity matrix according to the reference carbon fiber rod simulation model; calculating full field voltage according to the carbon fiber inhaul cable simulation model; independently setting a defect for the carbon fiber inhaul cable simulation model each time, performing voltage detection on the carbon fiber inhaul cable simulation model with the defect set each time, and performing full-field regularization to obtain a defect reference voltage matrix; detecting the voltage of the detected carbon fiber inhaul cable by using a sensor to obtain a detection voltage; calculating a voltage matched with the detection voltage by using the defect reference voltage matrix; according to the voltage matched with the detection voltage and the sensitivity matrix, the outer-layer carbon fiber rod is imaged; splicing the images of the carbon fiber rod to obtain an imaging result, and if the imaging result has a defect, determining the position of the defect according to the imaging result. The device can detect multiple defects of the carbon fiber inhaul cable.

Description

Technical Field

[0001] The invention relates to a defect detection method for a carbon fiber cable, in particular to a multi-defect detection method for a carbon fiber cable. Background Art

[0002] Electromagnetic tomography (EMT) is a non-contact, non-invasive, non-destructive testing technology that uses the principle of electromagnetic induction to visualize the conductivity distribution of the measured area. It is widely used in fields such as metal flaw detection and multiphase flow testing. However, electromagnetic tomography of defects in low-conductivity carbon fiber composites is still in the exploratory stage.

[0003] Carbon fiber composite cables overcome the heavy weight and surface corrosion challenges of traditional steel cables, meeting the load-bearing requirements of long-span cables and representing a key development direction for lightweighting ultra-large amusement rides and aerial passenger ropeways. Due to the influence of manufacturing processes and the service environment, carbon fiber composite cables inevitably experience surface damage defects such as debonding and strand breakage. This directly impacts the load-bearing capacity and service life of carbon fiber composite cables, and can even lead to cable failure and safety accidents. Therefore, conducting electromagnetic tomography of carbon fiber composite cable defects is crucial for their safe and reliable service.

[0004] Cables are primarily used at high altitudes, and some are wrapped in rubber, making surface defects difficult to detect. Electromagnetic tomography (EMT) can be used to detect defects in carbon fiber cables. Electromagnetic tomography can be divided into total variation regularization (TVR) and sparse reconstruction methods based on the sparsity of the solution. The TVR method generates globally continuous and smooth image reconstruction results, but exhibits a high number of edge artifacts, hindering accurate identification of defect locations. Given the small and concentrated defect areas of carbon fiber composite cables, which appear as locally sparse distributions, sparse reconstruction methods offer advantages for local defect imaging and are more suitable for identifying defect locations in carbon fiber composite cables.

[0005] In addition, carbon fiber composite cables have low electrical conductivity, so defects only cause minimal changes in the induced voltage of the detection coil, leading to more pronounced changes in the electromagnetic field gradient near the excitation coil and the interference effect of the superposition of the primary and secondary electric fields. Furthermore, there are positive and negative sensitivity regions in the object field being imaged. In the positive sensitivity region, as the conductivity increases, the coil induced voltage signal increases, and vice versa. Therefore, when multiple defects exist simultaneously, the voltage responses of different defects to the same detection coil will interfere with and cancel each other out. This makes it difficult to detect and identify multiple defects simultaneously in carbon fiber composite cables. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for detecting multiple defects of carbon fiber cables, so as to solve the problem that the existing electromagnetic tomography is inaccurate in detecting multiple defects on the surface of carbon fiber cables.

[0007] The purpose of the present invention is achieved like this:

[0008] A method for detecting multiple defects of a carbon fiber cable comprises the following steps:

[0009] S1. Construct a sensor and establish a sensor simulation model with the same specifications as the sensor. The sensor structure comprises a plurality of annular coils pointing toward the center of the ring carrier, arranged on the inner surface of the ring carrier. The annular coils are divided into detection coils and excitation coils. The number of detection coils is the same as the number of excitation coils, and the coils are evenly distributed in a double-layer ring around the object field. The inner coils are configured as detection coils, and the outer coils are configured as excitation coils.

[0010] S2 establishes a reference carbon fiber rod simulation model; establishes a carbon fiber cable simulation model of the same specifications as the tested carbon fiber cable; the cross-sectional area of the reference carbon fiber rod simulation model is less than the cross-sectional area of a carbon fiber rod in the carbon fiber cable simulation model 1 / 150;

[0011] S3. The reference carbon fiber rod simulation model passes through the ring of the sensor simulation model, the sensor simulation model applies current excitation to the reference carbon fiber rod simulation model, and calculates the sensitivity matrix; the carbon fiber cable simulation model passes through the center position of the sensor simulation model, the sensor simulation model applies current excitation to the carbon fiber cable simulation model, and obtains the full-field voltage;

[0012] S4. Setting a defect on the carbon fiber cable simulation model and sampling the voltage of the defective carbon fiber cable for a period of time to obtain a set of defect voltages, restoring the defect-free state of the carbon fiber cable simulation model; the defect is located near the sensor simulation model on the circumference of the outer carbon fiber rod;

[0013] S5. Changing the defect setting position, repeating step S4 until at least two defects are set for each outer carbon fiber rod of the carbon fiber cable simulation model, obtaining a defect voltage matrix corresponding to all defects; performing full-field regularization on the defect voltage matrix using the full-field voltage to obtain a defect reference voltage matrix corresponding to all defects;

[0014] S6. Using a sensor to detect the voltage of the carbon fiber cable under test, the detection voltage of the carbon fiber cable under test is obtained;

[0015] S7. Calculating a voltage matrix that matches the tested carbon fiber cable based on the defect reference voltage matrix;

[0016] S8. Imaging the outer carbon fiber rods based on a voltage matrix and a sensitivity matrix matching the carbon fiber cable being tested;

[0017] S9. splice the images of the carbon fiber rods according to the positions of the carbon fiber rods in the tested carbon fiber cable to obtain an imaging result of the tested carbon fiber cable. If the imaging result has defects, determine the defect location based on the imaging result.

[0018] Furthermore, the specific method of calculating the sensitivity matrix is:

[0019] S3a-1. The cross section of the carbon fiber cable simulation model is set in the middle of the sensor simulation model when the empty position, and the cross section of the carbon fiber cable simulation model of the carbon fiber rod in the form of a grid distribution of squares approximating a circle to represent;

[0020] S3a-2 reference carbon fiber rod simulation model through the sensor simulation model, and in the same direction as the central axis of the sensor simulation model, the reference carbon fiber rod simulation model is set in the position of the grid;

[0021] S3a-3. Using the sensor simulation model, the reference carbon fiber rod simulation model is sampled for a period of voltage at the grid position to obtain a set of voltage values;

[0022] S3a-4. Axially change the square where the reference carbon fiber rod simulation model is located, repeat step S3a-3 until the detection voltage of the reference carbon fiber rod simulation model in all squares is obtained, and perform normalization processing to obtain the sensitivity matrix of the carbon fiber cable simulation model; each column of data in the sensitivity matrix is the detection voltage of the reference carbon fiber rod simulation model in each square.

[0023] Furthermore, two defects are set on each outer carbon fiber rod of the carbon fiber cable simulation model;

[0024] The two defects set on the same outer carbon fiber rod are symmetrical about the line connecting the centers of the outer carbon fiber rod and the middle carbon fiber rod, and the sizes of the defects at different positions are the same.

[0025] Furthermore, the defect reference voltage matrix U b for:

[0026] U b =(U full -U)U full

[0027] Among them, U full is the full-field voltage matrix, and U is the defect voltage matrix.

[0028] Furthermore, in step S7, the specific method for calculating the voltage matching the carbon fiber cable being tested is:

[0029] The sparsity, the preset defect number threshold and the AICC are set; the sparse coefficient vector under the sparsity is calculated, and the updated AICC is calculated based on the sparse coefficient vector, the sparsity is updated, and when the updated AICC is less than the AICC before the update, the updated AICC and the sparse coefficient vector are saved; the steps of calculating the sparse coefficient vector under the sparsity, and the updated AICC is calculated based on the sparse coefficient vector, the sparsity is updated, and when the updated AICC is less than the AICC before the update, the updated AICC and the sparse coefficient vector are saved are executed in a loop; until the sparsity is greater than the preset defect number threshold, a voltage matrix matching the detection voltage of the carbon fiber cable under test is calculated based on the defect reference voltage matrix and the saved sparse coefficient vector.

[0030] Furthermore, the updated AICC is:

[0031]

[0032] Where e is U b The number of rows, M is the number of non-zero values in the sparse coefficient vector, and θ is the residual.

[0033] Furthermore, the specific method of imaging the carbon fiber rod is as follows:

[0034] For each column voltage signal corresponding to a defect in the voltage matrix that matches the detection voltage of the tested carbon fiber cable, the corresponding sparse signal is calculated; the sparse signals corresponding to defects on the same carbon fiber rod are superimposed to obtain a reconstructed image of each carbon fiber rod.

[0035] Furthermore, the specific method of calculating the sparse signal for the column voltage signal corresponding to a defect is as follows:

[0036] S8-1-1. In the voltage matrix U that matches the detection voltage of the carbon fiber cable under test c Select the column voltage signal U corresponding to one of the defects g , select the voltage sensitivity signal A corresponding to the carbon fiber rod where the defect is located in the sensitivity matrix A g ; Make the initial value of the residual

[0037] S8-1-2. Calculate the initial value of the residual and the voltage sensitivity signal A corresponding to the carbon fiber rod g Correlation C:

[0038]

[0039] Among them, the number of C and A g The same number of columns;

[0040] S8-1-3. Select 2H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The column numbers are merged into the set T 0 In the set T 1 , the initial set T 0 is an empty set;

[0041] S8-1-4. Calculate the sparse signal X by least squares based on the voltage matrix and sensitivity matrix that match the detection voltage of the carbon fiber cable being tested. g , sparse signal X g for:

[0042]

[0043] in, is the sensitivity matrix A g corresponds to the support set T 1 Column; when the value of the formula is the smallest, the corresponding X g is the sparse signal to be sought;

[0044] S8-1-5. In the collection T 1 Select the largest H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The set of column number pairs T 1 Update and get the set T 2 ;

[0045] S8-1-6. Update the initial value of the residual according to the obtained sparse signal, and the obtained residual for:

[0046]

[0047] in, is the sensitivity matrix A g corresponds to the support set T 2 Columns;

[0048] S8-1-7. Determine the residual Is it greater than 0.1, and is the number of iterations less than or equal to 10? If the judgment result is yes, loop through steps S8-1-8-S8-1-13;

[0049] S8-1-8. Calculate the residual and the voltage sensitivity signal A corresponding to the carbon fiber rod g Correlation C:

[0050]

[0051] S8-1-9. Select 2H maximum correlation C as the correlation index, and set the A corresponding to the correlation index gMerge the columns into the set T 2i In the set T 2i+1 ;

[0052] S8-1-10. Calculate the sparse signal X by least squares based on the voltage matrix and sensitivity matrix that match the detection voltage of the carbon fiber cable being tested. g , sparse signal X g for:

[0053]

[0054] in, is the sensitivity matrix A g corresponds to the support set T 2i+1 Column; when the value of the formula is the smallest, the corresponding X g is the sparse signal to be sought;

[0055] S8-1-11. In the collection T 2i+1 Select the largest H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The set of column number pairs T 2i+1 Update and get the set T 2i+2 ;

[0056] S8-1-12. Update the residual according to the obtained sparse signal, and the obtained residual for:

[0057]

[0058] in, is the sensitivity matrix A g corresponds to the updated support set T 2i+2 Columns;

[0059] S8-1-13. Determine the residual Is it greater than 0.1, and is the number of iterations less than or equal to 10? If the judgment result is yes, loop through steps S8-1-8-S8-1-12 until the judgment result is yes, and output X g .

[0060] This invention provides a method for detecting multiple defects in carbon fiber cables. Based on the low electrical conductivity and geometric structure of carbon fiber composite cables, this method combines pattern matching and compressed sensing techniques to overcome interference between multiple defect detection voltages. This method enables electromagnetic tomography of multiple defects in carbon fiber composite cables, addressing the problem of offsetting positive and negative responses to detection voltages in specific defect regions.

[0061] First, a sensitivity matrix for the irregularly shaped regions of a carbon fiber cable cross section is established based on a carbon fiber rod model. This matrix is more accurate than conventional circular region sensitivity matrices for defect imaging and incorporates the sensitivity characteristics of electrical anisotropy. Second, the detection signals of 12 types of defects are used as reference signals for pattern matching to perform sparse representation of the multi-defect detection signals of carbon fiber cables. Combined with the AICC judgment method, pattern matching of multi-defect detection signals is achieved. By improving the AICC judgment method, the optimal combination is selected by weighing the number of defects and residuals, making it more suitable for finding the optimal matching signal combination. On this basis, multi-signal joint image reconstruction is performed for the optimal matching signal combination. The CoSaMP reconstruction algorithm is used within the group. Given the sparse characteristics of carbon fiber cable defects, the CoSaMP reconstruction algorithm is more conducive to concentrated imaging of defects, enabling the superposition of sparse signals for multi-signal joint image reconstruction, achieving accurate imaging of multiple defects in carbon fiber composite cables and accurately detecting the multi-defect state of carbon fiber cables. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of the present invention.

[0063] Figure 2 This is a schematic diagram of using a sensor simulation model to detect voltage on a carbon fiber rod.

[0064] Figure 3 This is a schematic diagram of the defects set in the carbon fiber cable simulation model.

[0065] Figure 4 is the voltage histogram of defects at different locations.

[0066] Figure 5 It is a flowchart of multi-defect detection signal pattern matching combining Lasso regression and improved AIC.

[0067] Figure 6 This is a flow chart of the multi-signal joint reconstruction method for carbon fiber cable defects.

[0068] Figure 7 This is a flow chart of the present invention for detecting multiple defects in a tested carbon fiber cable.

[0069] Figure 8 This is a comparison chart of the defect image reconstruction effects of the present invention and LBP. DETAILED DESCRIPTION

[0070] The present invention will be further described below in conjunction with the accompanying drawings.

[0071] like Figure 1 As shown, the present invention provides a method for detecting multiple defects of carbon fiber cables, comprising the following steps:

[0072] S1. Build a sensor; establish a sensor simulation model with the same specifications as the sensor.

[0073] The sensor structure is as follows: a number of annular coils pointing to the center of the ring are set on the inner ring surface of the ring carrier. These are divided into detection coils and excitation coils. The number of detection coils is the same as the number of excitation coils. They are evenly distributed in a double-layer ring around the object field. The inner coil is set as the detection coil, and the outer coil is set as the excitation coil.

[0074] S2. Establish a reference carbon fiber rod simulation model; establish a carbon fiber cable simulation model with the same specifications as the tested carbon fiber cable.

[0075] The cross-sectional area of the reference carbon fiber rod simulation model is smaller than 1 / 150 of the cross-sectional area of a carbon fiber rod in the carbon fiber cable simulation model.

[0076] The carbon fiber cable simulation model, i.e. the defect-free model, has the same length, width, and material properties as the tested carbon fiber cable, and the height can be set by yourself.

[0077] like Figure 2 As shown, the sensor simulation model used in the present invention is a three-dimensional sensor model of 8 coil excitation and 8 coil detection; the specifications of the reference carbon fiber rod simulation model established in the present invention are 1×1×60mm 3 .

[0078] S3. The reference carbon fiber rod simulation model is passed through the ring of the sensor simulation model. The sensor simulation model applies current excitation to the reference carbon fiber rod simulation model and calculates the sensitivity matrix. The carbon fiber cable simulation model is passed through the center position of the sensor simulation model. The sensor simulation model applies current excitation to the carbon fiber cable simulation model to obtain the full-field voltage.

[0079] The specific method of calculating the sensitivity matrix is:

[0080] S3a-1. Divide the object field area in the middle of the sensor: set the cross section of the carbon fiber cable simulation model in the empty position in the middle of the sensor simulation model, and represent the cross section of the carbon fiber rod in the carbon fiber cable simulation model in the form of a grid-like distribution of squares approximating a circle.

[0081] The cross-section of the object being tested is smaller than the hollow area of the sensor simulation model. When voltage sampling is performed on the tested carbon fiber cable and the carbon fiber cable model, the voltage is sampled in the middle of the hollow area of the sensor. Therefore, the cross-section of the carbon fiber cable simulation model is set in the middle empty position of the sensor simulation model.

[0082] Draw a cross section of the carbon fiber cable simulation model at the center of a plane parallel to the ring of the sensor simulation model. The cross section includes the cross section of each carbon fiber rod. Each carbon fiber rod cross section is evenly divided into at least 150 squares.

[0083] The cross-sectional area of the reference carbon fiber rod simulation model is the same as the cross-sectional area of each square.

[0084] S3a-2. The reference carbon fiber rod simulation model passes through the sensor simulation model and is aligned with the central axis of the sensor simulation model. The reference carbon fiber rod simulation model is set at a grid position.

[0085] S3a-3. Use the sensor simulation model to perform voltage sampling of a reference carbon fiber rod simulation model at a grid position for one period to obtain a set of voltage values.

[0086] The reference carbon fiber rod simulation model performs a cycle of voltage sampling at each grid position. The excitation coils are excited in turn, and the detection coils are detected in turn (each time the excitation coil is excited, the detection coil directly opposite the excitation coil is not detected). The present invention uses 8 excitation coils to be excited in turn and 7 detection coils to be detected in turn, and 56 detection voltage signals are obtained in one cycle of sampling.

[0087] S3a-4. Move the reference carbon fiber rod simulation model in a direction perpendicular to the sensor simulation model, move the reference carbon fiber rod simulation model to the position of other squares, repeat step S3a-3 until the detection voltage of the reference carbon fiber rod simulation model in all squares is obtained, and perform normalization processing to obtain the sensitivity matrix of the carbon fiber cable simulation model.

[0088] The position of the sensor simulation model remains unchanged, and the reference carbon fiber rod simulation model is moved to obtain a set of voltages in each square. According to the correlation between the position of the carbon fiber rod and the voltage response of the detection coil, a matrix is obtained. The matrix is normalized to obtain the sensitivity matrix A of the cross-sectional area of the carbon fiber cable simulation model.

[0089] Each column of the sensitivity matrix A represents the sensitivity vector of the carbon fiber rod at each grid position. In this case, sensitivity A is a matrix of m × d columns and h × (h-1) rows. Here, m is the number of grids divided by each carbon fiber rod, d is the number of carbon fiber rods in the carbon fiber cable being tested, and h is the number of detection coils on the sensor.

[0090] Among them, each column of the sensitivity matrix is the detection voltage of the reference carbon fiber rod simulation model in each square.

[0091] It is only necessary to detect the detection voltage of one cross section of the reference carbon fiber rod at all squares.

[0092] The central axis directions of the reference carbon fiber rod simulation model and the carbon fiber cable simulation model are consistent with the central axis direction of the sensor simulation model. When exciting the simulation model, appropriate excitation current and excitation frequency are selected, for example, an excitation current of 1A and an excitation frequency of 1MHz.

[0093] The present invention divides the cross-section of each carbon fiber rod of the carbon fiber cable simulation model into 177 squares, sets its axial conductivity to 10060S / m and radial conductivity to 138S / m, and moves the reference carbon fiber rod simulation model to the square positions of the object field area in turn. By alternately exciting and detecting 8 coils, 56 full-field voltage signals U are obtained. full , as the detection voltage data of the non-defective type, that is, the full-field voltage matrix.

[0094] S4. Set a defect for the carbon fiber cable simulation model, and perform voltage sampling for a period of time on the defective carbon fiber cable to obtain a set of defect voltages, thereby restoring the defect-free state of the carbon fiber cable simulation model.

[0095] The defect is located on the circumference of the cross section of the outer carbon fiber rod, close to the sensor simulation model, and in contact with the outside world.

[0096] When setting a defect in a carbon fiber cable simulation model with a defect, it is only necessary to apply current excitation to the cross section of the carbon fiber cable simulation model where the defect is located, perform a cycle of voltage sampling, and the obtained set of voltages is the defect voltage group corresponding to the defect.

[0097] S5. Change the setting position of the defect and repeat step S4 until at least two defects are set for each outer carbon fiber rod of the carbon fiber cable simulation model to obtain the defect voltage matrix corresponding to all defects; use the full-field voltage to full-field regularize the defect voltage matrix to obtain the defect reference voltage matrix corresponding to all defects.

[0098] like Figure 3 As shown, two defects are set on each outer carbon fiber rod of the carbon fiber cable simulation model. The two defects set on the same outer carbon fiber rod are symmetrical about the line connecting the center of the outer carbon fiber rod and the center carbon fiber rod, and the defects at different locations are of the same size. The defects are located on the side surface of the carbon fiber rod of the carbon fiber cable simulation model and close to the location of the sensor simulation model, that is, the outer surface of the entire carbon fiber cable simulation model. The 2×(d-1) defects set represent 2×(d-1) defect types, which can represent the identification and imaging of typical defects in carbon fiber cables.

[0099] Figure 3 Taking 6 outer carbon fiber rods as an example, 12 defects are set.

[0100] To avoid the problem of defects canceling out and interfering with each other's positive and negative responses to the detection voltage, two defects were placed on the outer carbon fiber rods of each cable, with the total number of defects being twice that of the outer carbon fiber rods. With the line connecting the centers of the outer and middle carbon fiber rods as the axis of symmetry, defects were placed symmetrically on either side of the axis of the outer carbon fiber rods, located at the end closest to the coil and connected to the outside world. Defects were obtained by subtracting the carbon fiber rods from rectangular structures with a length, width, and height less than 4 mm. The defect sizes of each defect type were identical.

[0101] like Figure 4 As shown, each defect constructed appears separately in the carbon fiber cable simulation model, so that each carbon fiber cable simulation model includes a type of defect. The coils are excited and detected in turn, and each defect obtains h×(h-1) detection voltage signals, and the defect voltage matrix U corresponding to all defects is obtained.

[0102] Figure 4 Taking 12 defects as an example, the defect voltages corresponding to the 12 defects are obtained. Figure 4 The voltage histogram of each defect type in is Figure 3 The defect voltage corresponding to the defect in is, for example, Figure 4 The voltage histogram of defect class 1 is Figure 3 A set of defect voltages corresponding to the defect “1” in .

[0103] By comparing the reference voltage, a single defect type was established that can represent the identification and imaging of typical defects in carbon fiber cables. A subtraction was performed between a rectangular structure with a length of 2 mm, a width of 3 mm, and a height of 2 mm and the geometric structure of the carbon fiber cable, preserving the internal boundaries to establish the defective carbon fiber cable structure.

[0104] The defect reference voltage matrix U corresponding to all defects after full field regularization b for:

[0105] U b =(U full -U)U full

[0106] Among them, U full is the full-field voltage matrix.

[0107] U b ={U1,U2…U g …U n}

[0108] Where n is the total number of defect types.

[0109] S6. Use a sensor to perform voltage detection on the carbon fiber cable to obtain a detection voltage of the carbon fiber cable.

[0110] The sensor's outer and inner coils correspond one to another, and the corresponding detection and excitation coils are connected using an insulator. The sensor generates a cross-sectional image of the carbon fiber cable once per test. To achieve real-time imaging during the test process, the sensor traverses the axial length of the carbon fiber cable at a 2mm interval, completing a voltage test each time it stops, i.e., one cycle of voltage sampling. An excitation current of 1A and an excitation frequency of 1MHz can be applied to the carbon fiber cable under test.

[0111] S7. Calculate a voltage matrix that matches the carbon fiber cable under test based on the defect reference voltage matrix.

[0112] In order to solve the phenomenon of mutual cancellation between positive and negative voltage responses between multiple defects, the present invention adopts the idea of pattern matching to decompose the detection voltage signal, selects the optimal matching combination and the optimal weight through the AICC (Akaike Information Criterion Corrected) method, and finally obtains a voltage matrix that matches the detection voltage of the carbon fiber cable under test.

[0113] like Figure 5 As shown, the sparsity and the preset defect number threshold are set. The sparse coefficient vector under the sparsity is calculated, and the updated AICC is calculated based on the sparse coefficient vector. The sparsity is updated. When the updated AICC is less than the AICC before the update, the updated AICC and the sparse coefficient vector are saved. The steps of calculating the sparse coefficient vector under the sparsity, calculating the updated AICC based on the sparse coefficient vector, updating the sparsity, and saving the updated AICC and the sparse coefficient vector when the updated AICC is less than the AICC before the update are executed in a loop until the sparsity is greater than the preset defect number threshold. A voltage matrix that matches the detection voltage of the carbon fiber cable under test is calculated based on the defect reference voltage matrix and the saved sparse coefficient vector.

[0114] S7-1. Set the initial AICC (ie, AICC0), preset defect count threshold, and sparsity.

[0115] AICC0=+∞, the preset defect number threshold is greater than or equal to 2, and the sparsity k is 1.

[0116] S7-2. Calculate the sparse coefficient vector under sparsity for:

[0117]

[0118] Among them, U de is the detection voltage of the tested carbon fiber cable.

[0119] Sparsity refers to the ratio of the number of non-zero elements in the sparse coefficient vector to the total number of elements. Iterate until the value of the entire formula is minimum. is the desired sparse coefficient vector. S7-3. Calculate the updated AICC based on the sparse coefficient vector. The updated AICC is:

[0120]

[0121] Among them, e is the length of the signal, that is, U b The number of rows, M is the sparse coefficient vector The number of non-zero values in , that is, the sparse coefficient vector The corresponding number of defects, θ is the residual.

[0122] The residual θ is:

[0123]

[0124] S7-4. Update the sparsity. The updated sparsity is:

[0125] k1=k0+1

[0126] S7-5. Determine whether the updated AICC is greater than the AICC0 before the update. When the judgment result is yes, save the updated AICC and save the sparse coefficient vector; determine whether the updated sparsity is less than or equal to m. When the judgment result is no, calculate the voltage matrix that matches the detection voltage of the carbon fiber cable under test based on the saved sparse coefficient vector; when the judgment result is yes, execute steps S7-6-S7-9.

[0127] The voltage matrix U that matches the detection voltage of the carbon fiber cable being tested c for:

[0128]

[0129] S7-6. Calculate the updated sparsity k i The sparse coefficient vector under for:

[0130]

[0131] S7-7. Calculate the updated AICC based on the sparse coefficient vector. The updated AICC is:

[0132]

[0133] Among them, e is the length of the signal, that is, U bThe number of rows, M is the sparse coefficient vector The number of non-zero values in , that is, the sparse coefficient vector The corresponding number of defects, θ is the residual.

[0134] The calculation formula of residual θ is:

[0135]

[0136] S7-8. Update the sparsity. The updated sparsity is:

[0137] k i+1 =k i +1

[0138] S7-9. Determine whether the updated AICC is greater than the AICC before the update. If the judgment result is no, and the updated sparsity k i+1 When the number of defects is less than or equal to the preset threshold, steps S4-6-S4-9 are executed repeatedly; when the judgment result is no, and the sparsity k i+1 When the number of defects is greater than the preset threshold, the voltage matrix matching the detection voltage of the tested carbon fiber cable is calculated according to the saved sparse coefficient vector; when the judgment result is yes, the updated AICC and sparse coefficient vector are saved, and when the updated sparsity k i+1 When the number of defects is less than or equal to the preset defect threshold m, execute steps S7-6-S7-8. When the updated sparsity k i+1 When the defect number is greater than a preset defect number threshold m, a voltage matrix matching the detection voltage of the tested carbon fiber cable is calculated based on the saved sparse coefficient vector.

[0139] When saving the AICC for the second time, the newly saved AICC replaces the previously saved AICC. The sparse coefficient vector is saved in the same way as the AICC. Each calculation of the AICC and sparse coefficient vector is performed on the most recently saved data.

[0140] The voltage matrix U that matches the detection voltage of the carbon fiber cable being tested c for:

[0141]

[0142] U c ={U1,U2…U g …U n}

[0143] in, is the saved sparse coefficient vector.

[0144] The voltage U of each column of the voltage matrix matching the carbon fiber cable under test gCorresponds to a defect.

[0145] S8. Image the outer carbon fiber rod according to the voltage matrix and sensitivity matrix that match the carbon fiber cable being tested.

[0146] The small, concentrated defects of carbon fiber composite cables align with the sparse nature of the reconstructed image signal, making it suitable for reconstructing carbon fiber cable defect images using compressed sensing technology. Therefore, combining the concept of joint reconstruction with the Compressed Sampling Matching Pursuit (CoSaMP) algorithm, we achieve image reconstruction of multiple defects in carbon fiber cables.

[0147] like Figure 6 As shown in the figure, the image of each carbon fiber rod on the outside of the tested carbon fiber cable is reconstructed. Based on the detection signal pattern matching, the carbon fiber cable defect image reconstruction problem is transformed into a multi-signal joint reconstruction problem. With the support of the common sensitivity matrix, the voltage vector U g Recover sparse signal X g The cross-sectional area of the outer carbon fiber rods of the carbon fiber cable is divided into sections for imaging. Since the influence of the intermediate carbon fiber rods on the detection voltage is negligible, defect detection of the intermediate rods is not considered. The detection signals are assigned to 2(d-1) defect types through a pattern matching process. Finally, the sparse solution of the 2(d-1) signals is mapped to d-1 imaging regions to complete defect image reconstruction.

[0148] The voltage matrix U that matches the detection voltage of the carbon fiber cable being tested c Each column of data represents the voltage corresponding to a defect. The sparse signal is calculated for each defect until the sparse signals corresponding to all defects are obtained, that is, the voltage matrix U that matches the detection voltage of the tested carbon fiber cable is obtained. c The corresponding sparse signals are calculated for all columns of , and the image of the cross section of the same carbon fiber rod is reconstructed according to the sparse signals of the carbon fiber rod.

[0149] Image reconstruction is performed on all outer carbon fiber rods of the tested carbon fiber cable to generate a cross-sectional image of the tested carbon fiber cable.

[0150] The specific methods for calculating the sparse signal for the column voltage signal corresponding to a defect are:

[0151] S8-1. In the voltage matrix U that matches the detection voltage of the carbon fiber cable under test c Select the column voltage signal U corresponding to one of the defects g , select the voltage sensitivity signal A corresponding to the carbon fiber rod where the defect is located in the sensitivity matrix A g ; Make the initial value of the residual

[0152] Each column of data in the sensitivity matrix A is the data corresponding to the square on the carbon fiber rod. The column corresponding to each square in the sensitivity matrix is determined according to the position of the square. When the number of squares of each carbon fiber rod is m, A g There are m columns of data in it. Since there are two defects on a carbon fiber rod, the defects on the same fiber rod have the same sensitivity signal, A1=A n , A2=A n-1 …A g =A n-g+1 .

[0153] S8-2. Calculate the initial value of the residual and the voltage sensitivity signal A corresponding to the carbon fiber rod g Correlation C:

[0154]

[0155] Calculate A g Each column of The correlation of the sensitivity signal is obtained by multiplying the correlations C. Each column of the sensitivity signal corresponds to a correlation C.

[0156] S8-3. Select 2H maximum correlation C as the correlation index, and the A corresponding to the correlation index g The column numbers are merged into the set T 0 In the set T 1 , the initial set T 0 Is an empty set.

[0157] The column number represents the column of the sensitivity matrix. For example, the column number is 1, which represents the first column of the sensitivity matrix.

[0158] 2H is the generation of X g The sparsity of H can be defined by yourself, and 2H is less than X g The number of pixels.

[0159] S8-4. Calculate the sparse signal X by least squares based on the voltage matrix and sensitivity matrix that match the detection voltage of the carbon fiber cable being tested. g , sparse signal X g for:

[0160]

[0161] in, is the sensitivity matrix A g corresponds to the support set T 1 Column; constantly changing X g When the value of the formula is the smallest, the corresponding X gis the sparse signal to be sought.

[0162] S8-5. In set T 1 Select the largest H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The set of column number pairs T 1 Update and get the set T 2 .

[0163] The update of set T is to replace the original data with the new data.

[0164] S8-6. Update the initial value of the residual according to the obtained sparse signal, and the obtained residual for:

[0165]

[0166] in, is the sensitivity matrix A g corresponds to the support set T 2 Column.

[0167] S8-7. Determine the residual Is it greater than 0.1, and is the number of iterations less than or equal to 10? When the judgment result is yes, loop through steps S8-8-S8-13.

[0168] S8-8. Calculate the residual and the voltage sensitivity signal A corresponding to the carbon fiber rod g Correlation C:

[0169]

[0170] Calculate A g Each column of The correlations are calculated to obtain multiple correlations C.

[0171] S8-9. Select 2H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The column numbers are merged into the set T 2i In the set T 2i+1 .

[0172] S8-10. Calculate the sparse signal X by least squares based on the voltage matrix and sensitivity matrix that match the detection voltage of the carbon fiber cable being tested. g , sparse signal X g for:

[0173]

[0174] in, is the sensitivity matrix A g corresponds to the support set T2i+1 Column; when the value of the formula is the smallest, the corresponding X g is the sparse signal to be sought.

[0175] S8-11. In set T 2i+1 Select the largest H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The set of column number pairs T 2i+1 Update and get the set T 2i+2 .

[0176] S8-12. Update the residual according to the obtained sparse signal, and the obtained residual for:

[0177]

[0178] in, is the sensitivity matrix A g corresponds to the updated support set T 2i+2 Column.

[0179] S8-13. Determine the residual Is it greater than 0.1, and is the number of iterations less than or equal to 10? If the judgment result is yes, loop through steps S8-8-S8-12 until the judgment result is yes, and output X g .

[0180] Solve the sparse signals of all outer carbon fiber rods and get X1, X2…X g …X n .

[0181] like Figure 7 As shown, X1 and X n The corresponding data are superimposed to obtain the corresponding reconstructed image G1 of the carbon fiber rod, and X2 and X n-1 The corresponding data are superimposed to obtain the corresponding reconstructed image G2 of the outer carbon fiber rod. g and X n-g+1 The corresponding data are superimposed to obtain the corresponding reconstructed image Gg of the carbon fiber rod, thereby obtaining a reconstructed image of each carbon fiber rod.

[0182] S9. splice the images of the carbon fiber rods according to the positions of the carbon fiber rods in the tested carbon fiber cable to obtain an imaging result of the tested carbon fiber cable. If the imaging result has defects, determine the defect location based on the imaging result.

[0183] The reconstructed images of different fiber rods are placed in a matrix to obtain a cross-sectional image of the tested carbon fiber cable.

[0184] When imaging all the outer carbon fiber rods, the sparse signals corresponding to all the carbon fiber rods form a matrix, with different columns corresponding to carbon fiber rods at different positions. Since the middle carbon fiber rods were not inspected, the sparse signal of the middle carbon fiber rods is 0, indicating no defects.

[0185] When the imaging result has defects, the imaging result is a single defect or a combination of multiple defects. At this time, the actual number of defects and the defect position are more accurate. The defect position is recorded. After the recording is completed, the sensor is used to image the cross section along the carbon fiber cable being tested. When the imaging result has no defects, cross-sectional imaging is continued along the carbon fiber cable being tested.

[0186] S10. Method evaluation.

[0187] like Figure 8 As shown, the imaging effects of the LBP algorithm and the joint reconstruction algorithm on multiple defects of carbon fiber cables are compared when two defects or three defects exist at the same time. The results show that only the LBP algorithm reconstructed images at positions 1-3 and 1-4 can reflect the specific positions of the two types of defects, and the reconstructed images at positions 1-2, 1-2-3, 1-2-4 and 1-3-5 can only clearly show the position of one defect. This is because the simultaneous existence of multiple defects causes interference and offset effects on the detection coil voltage, resulting in the submerging of the voltage responses of other defects. However, the reconstructed images using the joint reconstruction algorithm proposed in the present invention can reflect the specific positions of the two types of defects and the three types of defects in the carbon fiber cable. This shows that the present invention can clearly image multiple defects on the same cross section.

Claims

1. A method for detecting multiple defects in carbon fiber cables, characterized in that: The steps include: S1. Construct a sensor and establish a sensor simulation model with the same specifications as the sensor. The sensor structure comprises a plurality of annular coils pointing toward the center of the ring carrier, arranged on the inner surface of the ring carrier. The annular coils are divided into detection coils and excitation coils. The number of detection coils is the same as the number of excitation coils, and the coils are evenly distributed in a double-layer ring around the object field. The inner coils are configured as detection coils, and the outer coils are configured as excitation coils. S2 establishes a reference carbon fiber rod simulation model; establishes a carbon fiber cable simulation model of the same specifications as the tested carbon fiber cable; the cross-sectional area of the reference carbon fiber rod simulation model is less than the cross-sectional area of a carbon fiber rod in the carbon fiber cable simulation model 1 / 150; S3. The reference carbon fiber rod simulation model passes through the ring of the sensor simulation model, the sensor simulation model applies current excitation to the reference carbon fiber rod simulation model, and calculates the sensitivity matrix; the carbon fiber cable simulation model passes through the center position of the sensor simulation model, the sensor simulation model applies current excitation to the carbon fiber cable simulation model, and obtains the full-field voltage; S4. Setting a defect on the carbon fiber cable simulation model and sampling the voltage of the defective carbon fiber cable for a period of time to obtain a set of defect voltages, restoring the defect-free state of the carbon fiber cable simulation model; the defect is located near the sensor simulation model on the circumference of the outer carbon fiber rod; S5. Changing the defect setting position, repeating step S4 until at least two defects are set for each outer carbon fiber rod of the carbon fiber cable simulation model, obtaining a defect voltage matrix corresponding to all defects; performing full-field regularization on the defect voltage matrix using the full-field voltage to obtain a defect reference voltage matrix corresponding to all defects; S6. Using a sensor to detect the voltage of the carbon fiber cable under test, the detection voltage of the carbon fiber cable under test is obtained; S7. Calculating a voltage matrix that matches the tested carbon fiber cable based on the defect reference voltage matrix; S8. Imaging the outer carbon fiber rods based on a voltage matrix and a sensitivity matrix matching the carbon fiber cable being tested; S9. splice the images of the carbon fiber rods according to the positions of the carbon fiber rods in the tested carbon fiber cable to obtain an imaging result of the tested carbon fiber cable. If the imaging result has defects, determine the defect location based on the imaging result.

2. The method for detecting multiple defects of carbon fiber cables according to claim 1, wherein: The specific method of calculating the sensitivity matrix is: S3a-1. The cross section of the carbon fiber cable simulation model is set in the middle of the sensor simulation model when the empty position, and the cross section of the carbon fiber cable simulation model of the carbon fiber rod in the form of a grid distribution of squares approximating a circle to represent; S3a-2 reference carbon fiber rod simulation model through the sensor simulation model, and in the same direction as the central axis of the sensor simulation model, the reference carbon fiber rod simulation model is set in the position of the grid; S3a-3. Using the sensor simulation model, the reference carbon fiber rod simulation model is sampled for a period of voltage at the grid position to obtain a set of voltage values; S3a-4. Axially change the square where the reference carbon fiber rod simulation model is located, repeat step S3a-3 until the detection voltage of the reference carbon fiber rod simulation model in all squares is obtained, and perform normalization processing to obtain the sensitivity matrix of the carbon fiber cable simulation model; each column of data in the sensitivity matrix is the detection voltage of the reference carbon fiber rod simulation model in each square.

3. The method for detecting multiple defects of carbon fiber cables according to claim 1, wherein: Two defects are set on each outer carbon fiber rod of the carbon fiber cable simulation model; The two defects set on the same outer carbon fiber rod are symmetrical about the line connecting the centers of the outer carbon fiber rod and the middle carbon fiber rod, and the sizes of the defects at different positions are the same.

4. The method for detecting multiple defects of carbon fiber cables according to claim 1 is characterized in that the defects Reference voltage matrix U b for: IN b =(U full -U)U full Among them, U full is the full-field voltage matrix, and U is the defect voltage matrix.

5. The method for detecting multiple defects of carbon fiber cables according to claim 1, wherein: The specific method of calculating the voltage matching the carbon fiber cable being tested in step S7 is: The sparsity, the preset defect number threshold and the AICC are set; the sparse coefficient vector under the sparsity is calculated, and the updated AICC is calculated based on the sparse coefficient vector, the sparsity is updated, and when the updated AICC is less than the AICC before the update, the updated AICC and the sparse coefficient vector are saved; the steps of calculating the sparse coefficient vector under the sparsity, and the updated AICC is calculated based on the sparse coefficient vector, the sparsity is updated, and when the updated AICC is less than the AICC before the update, the updated AICC and the sparse coefficient vector are saved are executed in a loop; until the sparsity is greater than the preset defect number threshold, a voltage matrix matching the detection voltage of the carbon fiber cable under test is calculated based on the defect reference voltage matrix and the saved sparse coefficient vector.

6. The method for detecting multiple defects of carbon fiber cables according to claim 5, wherein: The updated AICC is: Where e is U b The number of rows, M is the number of non-zero values in the sparse coefficient vector, and θ is the residual.

7. The method for detecting multiple defects of carbon fiber cables according to claim 1, wherein: The specific method for imaging the carbon fiber rod is: For each column voltage signal corresponding to a defect in the voltage matrix that matches the detection voltage of the tested carbon fiber cable, the corresponding sparse signal is calculated; the sparse signals corresponding to defects on the same carbon fiber rod are superimposed to obtain a reconstructed image of each carbon fiber rod.

8. The method for detecting multiple defects of carbon fiber cables according to claim 7, wherein: The specific method of calculating the sparse signal for the column voltage signal corresponding to a defect is: S8-1-1. In the voltage matrix U that matches the detection voltage of the carbon fiber cable under test c Select the column voltage signal U corresponding to one of the defects g , select the voltage sensitivity signal A corresponding to the carbon fiber rod where the defect is located in the sensitivity matrix A g ; Make the initial value of the residual S8-1-2. Calculate the initial value of the residual and the voltage sensitivity signal A corresponding to the carbon fiber rod g Correlation C: Among them, the number of C and A g The number of columns is the same; S8-1-3. Select 2H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The column numbers are merged into the set T 0 In the set T 1 , the initial set T 0 is an empty set; S8-1-4. Calculate the sparse signal X by least squares based on the voltage matrix and sensitivity matrix that match the detection voltage of the carbon fiber cable being tested. g , sparse signal X g for: in, is the sensitivity matrix A g corresponds to the support set T 1 Column; when the value of the formula is the smallest, the corresponding X g is the sparse signal to be sought; S8-1-5. In the collection T 1 Select the largest H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The set of column number pairs T 1 Update and get the set T 2 ; S8-1-6. Update the initial value of the residual according to the obtained sparse signal, and the obtained residual for: in, is the sensitivity matrix A g corresponds to the support set T 2 Columns; S8-1-7. Determine the residual Is it greater than 0.1, and is the number of iterations less than or equal to 10? If the judgment result is yes, loop through steps S8-1-8-S8-1-13; S8-1-8. Calculate the residual and the voltage sensitivity signal A corresponding to the carbon fiber rod g Correlation C: S8-1-9. Select 2H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g Merge the columns into the set T 2i In the set T 2i+1 ; S8-1-10. Calculate the sparse signal X by least squares based on the voltage matrix and sensitivity matrix that match the detection voltage of the carbon fiber cable being tested. g , sparse signal X g for: in, is the sensitivity matrix A g corresponds to the support set T 2i+1 Column; when the value of the formula is the smallest, the corresponding X g is the sparse signal to be sought; S8-1-11. In the collection T 2i+1 Select the largest H maximum correlation C as the correlation index, and set the A corresponding to the correlation index g The set of column number pairs T 2i+1 Update and get the set T 2i+2 ; S8-1-12. Update the residual according to the obtained sparse signal, and the obtained residual for: in, is the sensitivity matrix A g corresponds to the updated support set T 2i+2 Columns; S8-1-13. Determine the residual Is it greater than 0.1, and is the number of iterations less than or equal to 10? If the judgment result is yes, loop through steps S8-1-8-S8-1-12 until the judgment result is yes, and output X g .

Images