Mark type cargo containment system FSB bonding quality automatic detection integrated system

Abstract

A Mark type cargo containment system FSB bonding quality automatic detection integrated system uses a high-resolution camera or a three-dimensional scanning device to obtain image data of an FSB surface, converts an image from a spatial domain to a frequency domain through Fourier transform, uses a low-pass wave filter to remove high-frequency noise, and retains low-frequency components to highlight corrugations, wrinkles and other defects, so that the FSB bonding quality automatic detection integrated system is obtained. Morphological optimization and curvature analysis are carried out, defects are classified and quantified, and position coordinates of the defects are determined. An electromagnetic ultrasonic transducer is utilized to excite and shear horizontal guided waves on the surface of the FSB, a guided wave reflection signal is received, characteristic quantities such as envelope information are extracted through signal processing, whether the characteristic quantities exceed the acceptance standard or not is judged, and the defect position is determined. And integrating the position, size and type of the defect into a map, and automatically generating a quality detection report for tracing and maintenance guidance. The FSB surface and lower glue layer synchronous detection can be realized, all key defects are fully covered, the detection precision is matched, and the FSB quality control requirement can be met.

Description

Technical Field

[0001] This invention belongs to the field of LNG carrier construction and design, and specifically relates to an integrated automatic detection system for FSB bonding quality of Mark-type cargo containment system. Background Technology

[0002] Liquefied natural gas (LNG), as an environmentally friendly and clean energy source, plays an increasingly important role in the global energy market. Among various LNG carrier types, the Mark III membrane LNG carrier is widely favored for its double-protected storage tanks. This design ensures extremely high safety during the transportation and handling of LNG.

[0003] The primary containment layer of this type of LNG carrier cargo containment system is made of stainless steel (SUS304L), directly contacting the liquefied natural gas under cryogenic (-163°C) conditions, effectively resisting stress caused by temperature differences and cargo sloshing. The secondary containment layer, also known as the triplet, consists of three composite layers: a 0.6-0.7 mm thick Flexible Secondary Barrier (FSB), a two-component polyurethane adhesive layer (FSB adhesive), and a rigid secondary containment layer of equal thickness (RSB). Thanks to the secondary containment layer, at least 14 days of sealing can be maintained even if the primary containment layer fails. Therefore, for Mark III membrane-type LNG carriers, the bonding of the secondary containment layer is not only a crucial step in ensuring the integrity of the cargo containment system but also an important component in maintaining LNG transportation safety. To ensure the airtightness of the secondary containment layer, quality inspection of the FSB surface and the adhesive layer beneath it is required after bonding. This inspection is typically carried out manually. The ripples and wrinkles on the FSB surface need to be judged by visual inspection, while the air bubbles in the adhesive layer need to be checked by the round bar test.

[0004] The specific procedure for the round bar test is as follows: the inspector uses a round bar with a spherical end and a smooth surface to apply a certain pressure to the FSB and scratch it. The quality is then assessed by visual inspection and measurement to determine whether it meets the standards.

[0005] Statistics show that, if working continuously, a single inspector can complete the testing and inspection of 10 round bars, each approximately 20 meters long, within one hour. This demonstrates the extremely low efficiency of manual inspection. Furthermore, prolonged and large-scale testing significantly increases the workload for inspectors, easily leading to fatigue, decreased testing efficiency, and compromised testing quality.

[0006] In addition, traditional manual inspection methods cannot perform real-time data processing and feedback, making it impossible to effectively monitor and manage the inspection process. This problem affects the continuity and reliability of inspection results and hinders the effective implementation of subsequent processes. Summary of the Invention

[0007] To address the aforementioned problems, this invention provides an integrated automatic inspection system for FSB bonding quality in Mark-type cargo containment systems. The aim is to completely resolve the issues of efficiency and reliability in manual FSB bonding quality inspection. The technical solution adopted is as follows: An integrated automatic detection system for the bonding quality of a Mark-type cargo containment system FSB is disclosed. The FSB consists of two layers of fiber fabric with an aluminum foil sandwiched between them. The fiber fabric and the aluminum foil are bonded together using a polyurethane adhesive.

[0008] The specific steps are as follows: S1: Automatic inspection of surface ripples and wrinkles in FSB. Use a high-resolution camera or 3D scanning device to acquire FSB surface image data f(x,y), and perform a two-dimensional Fourier transform on the image f(x,y) to obtain the frequency domain representation F(u,v): .

[0009] in: f(x,y) is the spatial domain signal of the image.

[0010] F(u,v) is the frequency component of the image in the frequency domain.

[0011] j is the imaginary unit.

[0012] M represents the number of pixels in the image along the x-axis (horizontal direction), in pixels.

[0013] N is the number of pixels in the image along the y-direction (vertical direction), in pixels.

[0014] u and v are the horizontal and vertical frequencies in the frequency domain, respectively, with a unit of 1 / pixel.

[0015] e is a natural constant with a value of approximately 2.178.

[0016] S2: In the frequency domain, the low-pass filter H(u,v) is multiplied by the spectrum F(u,v) to filter out high-frequency noise, and the resulting spatial domain image is g(x,y).

[0017] .

[0018] S3: Perform morphological operations on the filtered spatial domain image, and optimize the image structure by combining structural elements to ensure the segmentability and classification accuracy of defective regions.

[0019] After completing the morphological optimization process, the sequence of contour boundary points for each suspected defect region is extracted, and the local curvature is calculated for each contour boundary point using the three-point difference method.

[0020] S4: By statistically analyzing known samples, two thresholds, a low threshold and a high threshold, are preset. Defects are compared with these preset thresholds for classification. When the curvature value is lower than the preset low threshold, it is judged as a ripple. When it exceeds the preset high threshold, it is judged as a wrinkle. Based on the curvature value, it is determined whether a defect exists.

[0021] Once the defect is identified, it needs to be determined based on the pixel coordinates (x, y) on the image. p ,y p Transform to actual physical coordinates (X,Y).

[0022] X=x p -s x , Y=y p -s y .

[0023] in: Pixel coordinates are obtained directly from the imaging device without additional calculation; the unit is pixels.

[0024] S5: Automatic inspection of FSB adhesive layer bubble defects Shear horizontal waves (SH waves) are generated on the surface of the adhesive layer by EMAT technology, and their reflected signals s(t) are collected. The envelope curve A(t) is obtained by enveloping the collected reflected signals. The unit is volts (V).

[0025] .

[0026] in (t) is the Hilbert transform of the signal s(t).

[0027] Analyze the envelope curve A(t) to identify significant peaks. Based on these peaks, set a time window of interest [t1, t2] that should cover possible defect reflection signals. Extract relevant features from the samples, including the envelope energy E (unit: V). 2 -s, spectral bandwidth B, unit: Hz, maximum envelope amplitude A max .

[0028] .

[0029] Where dt is the integral variable, representing a small time increment in the continuous time domain.

[0030] .

[0031] Where f1 represents the lower limit frequency of the spectrum bandwidth, the starting point of the -3dB bandwidth. f2 represents the upper limit frequency of the spectrum bandwidth, the ending point of the -3dB bandwidth.

[0032] S6: Construct a defect size estimation model to achieve quantitative estimation of the size of adhesive layer defects.

[0033] .

[0034] Where: D is the estimated equivalent defect diameter, in mm.

[0035] Coefficients k1, k2, k3, C: obtained by fitting several standard defect samples of known size using a multiple linear regression method.

[0036] The estimated diameter D is compared with a preset standard threshold of 8 mm to determine if a defect exists.

[0037] If D > the preset standard threshold, it is judged as an excessive defect.

[0038] If D < the preset standard threshold, then the defect is considered to be within an acceptable range.

[0039] For a single bubble, calculate the area from the equivalent diameter D output by the size estimation model: .

[0040] This can be achieved through an equivalent area threshold, i.e., 50mm. 2 Compare them to determine if they are defects.

[0041] When multiple reflection peaks are identified within the time window of interest, the time interval between adjacent peaks is calculated and converted into spatial spacing based on the waveguide propagation speed. If the number and spacing criteria of the bubble string are met, it is determined to be a bubble string.

[0042] Using scan data, the spatial distribution direction of the bubble string is determined, and the angle between the bubble string and the length direction of the FSB is calculated to determine whether it is a transverse bubble string.

[0043] For horizontal bubbles, they are directly judged as unqualified.

[0044] For non-lateral bubble strings, estimate the equivalent diameter Di of each bubble individually, and then calculate the total equivalent area: .

[0045] This value can be directly compared with the rejection criteria to determine whether it is unqualified.

[0046] S7: Send SH guided waves via the EMAT sensor and measure the time difference Δt of the reflected waves.

[0047] Calculate the distance d from the defect to the sensor.

[0048] .

[0049] in: v: Propagation speed of SH waves in the adhesive layer, unit: mm / s.

[0050] Δt=t p -t0 is the time delay between wave emission and reception / reflection, in seconds.

[0051] Based on the location and propagation direction of EMAT, the distance d is mapped to the structural coordinate system. The formula for calculating the defect location coordinates is as follows: .

[0052] .

[0053] in: (x0, y0): The current position coordinates of the EMAT sensor.

[0054] θ: The angle between the direction of guided wave propagation and the x-axis, expressed in radians or degrees; (x) d ,y d ): The location of the defect to be determined in the structural coordinate system.

[0055] S9: Integrates defect location, size, and type into a map, automatically generating quality inspection reports for traceability and repair guidance.

[0056] Furthermore, the aforementioned Mark-type cargo containment system FSB bonding quality automatic detection integrated system further features a flexible secondary screen wall thickness of 0.7mm.

[0057] Furthermore, in the aforementioned Mark-type cargo containment system FSB bonding quality automatic detection integrated system, in step S2, the low-pass filter H(u,v) is a Gaussian low-pass filter. .

[0058] The cutoff frequency D0 is determined through pre-experimentation or simulation to ensure that ripple / wrinkle information is preserved while suppressing texture interference.

[0059] Furthermore, in step S3 of the aforementioned Mark-type cargo containment system FSB bonding quality automatic detection integrated system, the curvature formula is as follows: Let three adjacent points on the boundary (unit: mm) be: P1=(x1,y1), P2=(x2,y2), P3=(x3,y3) Area of ​​the triangle formed by them (unit: mm) 2 ) is defined as: .

[0060] The discrete curvature of point P2 is estimated as follows: .

[0061] This curvature value reflects the degree of curvature of the boundary at that point.

[0062] Morphological operations include dilation, erosion, opening, and closing operations, which can be adaptively selected based on the morphological characteristics of the defect to be detected.

[0063] Erosion is used to shrink the boundaries of defect areas and remove noise; it is defined as follows: .

[0064] in: A: The set of foreground pixels in the image, i.e., the defect area.

[0065] B: Structural element.

[0066] z: Coordinates of the center point of the structuring element.

[0067] Bz : Translate the structuring element to coordinate z.

[0068] Pixels in the current image region are preserved only if B is completely contained within that region.

[0069] Expansion is used to connect fracture edges or reinforce defect boundaries, and it is defined as follows: .

[0070] That is, as long as the structuring element has any intersection with the image region, the pixel is included in the foreground.

[0071] The opening operation is used to remove small noise regions and smooth edges without destroying the main contour. It is defined as follows: .

[0072] This involves first corroding the material, then expanding the resulting material, which is suitable for removing fine textures or pseudo-defects.

[0073] The closing operation is used to fill in small black holes and repair discontinuities in defective regions; it is defined as follows: .

[0074] This involves first expanding and then corroding, which is used to make the boundaries of defective areas more coherent.

[0075] Furthermore, in step S4 of the aforementioned Mark-type cargo containment system FSB bonding quality automatic detection integrated system, the horizontal coordinate X and vertical coordinate Y in the actual physical coordinates (X,Y) are obtained in the following way: Physical horizontal coordinate acquisition: Let the width of the captured image in pixels be W. img The corresponding FSB actual physical width is W. phy。

[0076] The horizontal scaling factor is .

[0077] When the horizontal coordinate of a defective pixel in the image is x p Unit: mm, its corresponding physical horizontal coordinate is: X=x p -s x .

[0078] Physical vertical coordinate acquisition: Let the height in pixels of a single frame be H. img The corresponding physical length of the actual coverage is Lphy.

[0079] The vertical scaling factor is: .

[0080] If the vertical pixel coordinate of the defect in a single frame image is y p (Unit: mm), then its physical vertical coordinate within this frame is: Y local =y p .s y .

[0081] During the detection process, the camera or sensor moves along the length of the FSB, and its cumulative displacement is recorded as ΔL by the track encoder. Therefore, the global longitudinal physical coordinates of the defect are: Y=Y local +ΔL.

[0082] Furthermore, in the aforementioned Mark-type cargo containment system FSB bonding quality automatic detection integrated system, the FSB adhesive layer is made of two-component polyurethane cured. Taking automatic bonding as an example, the thickness of the standard adhesive layer is 0.2mm to 1.2mm.

[0083] Furthermore, in the aforementioned Mark-type cargo containment system FSB bonding quality automatic detection integrated system, the rejection criterion in step S6 is that the surface area of ​​the bubbles within the layer is less than 5% of the total area of ​​the bonding area within a span of 150mm along each side of the bonding joint.

[0084] Bubble diameters greater than 8 mm or equivalent areas greater than 50 mm² are unacceptable. Transverse bubble bands are unacceptable.

[0085] When three or more bubbles are detected, and the distance between any adjacent bubbles is less than or equal to 15 mm, they are considered as a single bubble string, regardless of the size of each bubble.

[0086] Furthermore, the aforementioned Mark-type cargo containment system FSB bonding quality automatic detection integrated system further defines a transverse bubble string as one where the angle between the bubble string and the length direction of the FSB is between 45 and 90 degrees, and one is not considered a transverse bubble string if the angle is less than 45 degrees.

[0087] The beneficial effects of this invention are: 1. Integrated testing It enables simultaneous inspection of the FSB surface and the underlying adhesive layer, comprehensively covering all critical defects. Its inspection accuracy matches the requirements of FSB quality control.

[0088] 2. Non-contact detection It can achieve non-destructive testing without relying on contact media, thus avoiding damage to the FSB.

[0089] 3. Standardized output It can automatically generate judgment results and defect reports, avoiding human error.

[0090] 4. Improve efficiency and reduce labor intensity It reduces the time required for manual inspection, greatly improves testing efficiency, and reduces the workload of inspection. Attached Figure Description

[0091] Figure 1 This is a schematic diagram of the bonding of flexible secondary screen walls.

[0092] Figure 2 Define ripples and wrinkles.

[0093] Figure 3 This is a schematic diagram of the corrugated board acceptance standard.

[0094] Figure 4 Flowchart for automated inspection of surface ripples and wrinkles in FSB.

[0095] Figure 5 This is a schematic diagram of bubbles and bubble strings.

[0096] Figure 6 This is a schematic diagram of a horizontal bubble string.

[0097] Figure 7 This is a flowchart for the automatic inspection of air bubble defects in FSB adhesive layers.

[0098] Among them, 1-fiber fabric, 2-aluminum foil, 3-polyurethane adhesive, and 4-FSB. Detailed Implementation

[0099] The present invention will be described in detail with reference to specific embodiments.

[0100] like Figure 1 and 2 The illustrated Mark-type cargo containment system FSB bonding quality automatic detection integrated system has the following specific operating steps: 1. Automatic inspection of surface ripples and wrinkles in FSB Flexible secondary screen (FSB) Definition: The flexible secondary screen wall consists of two layers of fiber fabric (such as fiberglass cloth) on the top and bottom, with an aluminum foil in the middle, bonded together by a polyurethane adhesive. Its thickness is approximately 0.7 mm. (See attached image) Figure 1 The texture of the fiber fabric on the surface of the FSB varies depending on the weaving process of different manufacturers.

[0101] Wave Definition: Ripples refer to long, narrow, and smooth-shaped irregularities on a surface, typically characterized by minute fluctuations. (See appendix) Figure 2 .

[0102] The inspection standard is as follows: taking automatic bonding as an example, a corrugation height of less than or equal to 0.8 mm and a length of less than or equal to 40 mm are considered acceptable. However, a round bar test is still required to confirm that there are no air bubbles in the underlying adhesive layer. See attached document. Figure 3 .

[0103] Fold Definition: Wrinkles and ripples look similar, but their shape is sharp and they are often accompanied by missing glue or folded aluminum foil. See appendix. Figure 2 .

[0104] Taking automatic bonding as an example, the inspection standard is: all wrinkles are considered defects and need to be repaired.

[0105] To address the aforementioned characteristics, this invention employs passive optical detection technology to inspect FSB surface defects. The inspection process is detailed in the appendix. Figure 4 The specific inspection method is as follows.

[0106] 1.1 Image Acquisition Use a high-resolution camera or 3D scanning device to acquire FSB surface image data f(x,y), with a resolution of millimeters to meet accuracy requirements.

[0107] 1.2. Frequency Domain Filtering 1.2.1 Transformation to the Frequency Domain – Fourier Transform The Fourier transform converts an image from the spatial domain to the frequency domain, revealing its components at different frequencies. Typically, low-frequency components represent smooth variations in the image, such as large shapes, structures, and overall trends. Therefore, on an FSB surface, ripples / wrinkles are low-frequency components because they represent relatively large variations. The texture of an FSB surface, on the other hand, is a high-frequency component because the texture of the fibrous fabric is a subtle and frequently repeating structure.

[0108] Performing a two-dimensional Fourier transform on the image f(x,y) yields the frequency domain representation F(u,v), which contains information about each frequency component in the image.

[0109] .

[0110] Where f(x,y) is the spatial domain signal of the image. F(u,v) is the frequency component of the image in the frequency domain. j is the imaginary unit. u and v are the horizontal and vertical frequencies in the frequency domain, respectively.

[0111] 1.2.2 Frequency Domain Processing – Low-Pass Filter To more accurately identify and quantify ripple / wrinkle defects, a low-pass filter is used, which allows low-frequency signals to pass through while removing high-frequency noise. This preserves low-frequency defects such as ripples and wrinkles in the image while removing high-frequency noise such as textures.

[0112] In the frequency domain, the low-pass filter H(u,v) is multiplied by the spectrum F(u,v) to filter out high-frequency noise: G(u,v)=H(u,v)– F(u,v) 1.2.3 Restoring to the spatial domain – Inverse Fourier Transform After filtering, the frequency domain image G(u,v) is restored to the spatial domain using inverse Fourier transform, resulting in a noise-removed image where low-frequency ripples and wrinkles are more prominent. The filtered spatial domain image is g(x,y).

[0113] .

[0114] 1.2.4 Morphological Optimization To address the issues of fiber optic interference, pseudo-small ripples, and boundary breaks that affect defect identification, morphological image processing methods are introduced, and structural elements are combined to optimize the image structure, ensuring the segmentability and classification accuracy of defect areas.

[0115] The basic operations used include erosion, dilation, opening, and closing operations, and their principles are as follows: Based on the spatial domain image obtained above, morphological operations can be performed on the image, including dilation, erosion, opening and closing operations, to enhance the contours of defects in the image.

[0116] corrosion Used to shrink the boundary of defect areas and remove noise, it is defined as: .

[0117] Where, A: the set of foreground pixels in the image, i.e. the defect area; B: Structural element; z: Coordinates of the center point of the structural element Bz: Translates the structuring element to coordinate z.

[0118] Pixels in the current image region are preserved only if B is completely contained within that region.

[0119] expansion Used to connect fracture edges or reinforce defect boundaries, it is defined as: .

[0120] That is, as long as the structuring element has any intersection with the image region, the pixel is included in the foreground.

[0121] Opening operation Used to remove small noise areas and smooth edges without destroying the main contour, it is defined as follows: .

[0122] This involves first corroding the material, then expanding the resulting material, which is suitable for removing fine textures or pseudo-defects.

[0123] Closing operation Used to fill in small black holes and repair discontinuities in defective regions, it is defined as: .

[0124] This involves first expanding and then corroding, which is used to make the boundaries of defective areas more coherent.

[0125] 1.2.5 Curvature Analysis After morphological optimization, the contour boundary point sequence of each suspected defect region is extracted. The local curvature is calculated for each point using the three-point difference method, with the curvature formula as follows: Let the three adjacent points on the boundary be: P1=(x1,y1), P2=(x2,y2), P3=(x3,y3) The area of ​​the triangle they form is defined as: .

[0126] The discrete curvature of point P2 is estimated as follows: .

[0127] This curvature value reflects the degree of curvature of the boundary at that point.

[0128] 1.2.6 Defect Classification and Quantification Based on inspection standards, preset thresholds are used to classify defects into ripples and wrinkles. Low curvature areas correspond to smooth ripples, while high curvature areas correspond to sharp wrinkles. The presence of a defect is determined based on the calculated curvature values.

[0129] 1.2.7 Defect Location Once a defect is identified, it needs to be converted from the pixel coordinates on the image to the actual physical coordinates (unit: mm).

[0130] Direct image coordinate positioning Once the defect is identified, its pixel coordinates (x, y) on the image are... p ,y p It can be determined directly.

[0131] Image coordinate origin: usually the top left corner (0, 0).

[0132] The coordinates of the center point of the defect can be obtained by calculating the defect area. , .

[0133] Among them, (x) i ,y i ) represents the coordinates of all pixels within the defect area, and N is the number of pixels.

[0134] These coordinates can be labeled on the image and used for subsequent transformations to the physical coordinate system.

[0135] Image coordinates to physical coordinates conversion The method for converting defect points in an image to their actual FSB location is shown below.

[0136] Assumption: Pixel size: The actual size corresponding to each pixel is s x mm / pixel (horizontal) and s y mm / pixel (vertical).

[0137] Image size: Total width W img and height H img .

[0138] FSB Physical Dimensions: Total Width W phy and length L phy .

[0139] Pixel coordinates (x) p ,y p Convert to physical coordinates (X, Y): X=x p -s x , Y=y p -s y in: , .

[0140] 2. Automatic inspection of FSB adhesive layer bubble defects FSB adhesive is a two-component polyurethane cured adhesive. Taking automatic bonding as an example, the standard adhesive layer thickness is 0.2mm to 1.2mm. The requirements for air bubbles within the adhesive layer are as follows: within a 150mm span on each side of the bond, the surface area of ​​the bubble must be less than 5% of the total bond area; bubbles with a diameter > 8mm or an equivalent area > 50mm² are unacceptable; transverse air bubbles are also unacceptable.

[0141] Bubble String: When three or more bubbles are detected, and the distance between any adjacent bubbles is less than or equal to 15 mm, they are considered a bubble string regardless of the size of each bubble. See Appendix. Figure 5 .

[0142] Horizontal bubble string: A bubble string is defined as one whose angle with the length direction of the FSB is between 45 and 90 degrees. If the angle is less than 45 degrees, it is not considered a horizontal bubble string. See appendix. Figure 6 .

[0143] To address the above issues, this invention utilizes electromagnetic acoustic transducer (EMAT) sheared horizontal guided wave (SH guided wave) technology to detect bubble defects. The detection process is detailed in the appendix. Figure 7 The specific inspection method is as follows.

[0144] 2.1 Signal Excitation and Reception The shear horizontal wave (SH wave) is generated on the surface of the adhesive layer by EMAT technology, and its reflected signal s(t) is received.

[0145] 2.2 Signal Processing and Defect Identification 2.2.1 Signal Processing The collected reflected signals are processed by envelope processing to obtain the envelope curve A(t).

[0146] .

[0147] Where (t) is the Hilbert transform of signal s(t).

[0148] 2.2.2 Peak Identification By analyzing the envelope curve, significant peaks were identified, and the time periods in which defects might exist were preliminarily determined.

[0149] 2.2.3 Time Window Setting Based on the identified peak values, a time window of interest [t1, t2] is set, which should cover possible defect reflection signals. As test data accumulates, the time window of interest can be continuously optimized to improve the accuracy and reliability of defect detection.

[0150] 2.2.4 Feature Extraction Extract relevant features from the samples, including envelope energy E, spectral bandwidth B, and maximum envelope amplitude Amax.

[0151] Envelope energy E refers to the envelope energy of the defect reflection wave within the window [t1,t2], and is used to estimate the defect size.

[0152] .

[0153] Where dt is the integral variable, representing a small time increment in the continuous time domain.

[0154] The spectral bandwidth is determined by analyzing the spectrum of the reflected signal using Fast Fourier Transform (FFT) and extracting the -3dB bandwidth of the main frequency portion for auxiliary identification of defect types.

[0155] .

[0156] Where f1 represents the lower limit frequency of the spectrum bandwidth, the starting point of the -3dB bandwidth; and f2 represents the upper limit frequency of the spectrum bandwidth, the ending point of the -3dB bandwidth.

[0157] Amax is the maximum value of the reflected wave envelope curve A(t) in the current detected signal within the time window of interest [t1,t2], which helps to enhance the model's sensitivity to shallow or highly reflective defects.

[0158] 2.2.5 Model Fitting Based on the above-mentioned characteristic quantities, the present invention constructs a defect size estimation model, which can realize the quantitative estimation of the size of adhesive layer defects.

[0159] .

[0160] in: D: Estimated equivalent defect diameter (unit: mm); Coefficients k1, k2, k3, C: obtained by fitting several standard defect samples of known size using a multiple linear regression method.

[0161] 2.2.6 Detection Applications and Defect Judgment By substituting the parameters E, B, and Amax from the actual detection signal into the above model, the equivalent diameter D of the defect is calculated.

[0162] The estimated diameter D is compared with a preset standard threshold (e.g., 8 mm) to determine if a defect exists.

[0163] If D>8mm, it is judged as an excessive defect; If D < 8mm, the defect is considered to be within an acceptable range.

[0164] For a single bubble, calculate the area from the equivalent diameter D output by the size estimation model: .

[0165] Whether it is a defect can be determined by comparing the equivalent area threshold.

[0166] When multiple reflection peaks are identified within the time window of interest, the time interval between adjacent peaks is calculated and converted into spatial spacing based on the guided wave propagation velocity. If the criteria for the number and spacing of bubble strings are met, it is determined to be a bubble string.

[0167] Using scan data, the spatial distribution direction of the bubble string is determined, and the angle between the bubble string and the length direction of the FSB is calculated to determine whether it is a transverse bubble string.

[0168] For horizontal bubbles, they are directly judged as unqualified.

[0169] For non-lateral bubble strings, estimate the equivalent diameter Di of each bubble individually, and then calculate the total equivalent area: .

[0170] This value can be directly compared with the rejection criteria (such as A). total >50mm 2 The comparison is performed to determine whether it is a defective or non-conforming defect.

[0171] 2.3 Defect Location During guided wave signal acquisition, the aforementioned method allows for real-time assessment of any defects. Based on the assessment results, defects can be visually marked by printing different colors onto a nearby insulating board for on-site identification or subsequent re-inspection.

[0172] For inspection reports that require the physical location of defects, this invention uses the Time of Flight (TOF) method to calculate the position of the defect along its length.

[0173] Measuring the time delay t of the guided wave signal The SH guided wave is sent through the EMAT sensor, and the time difference Δt of the reflected wave is measured.

[0174] Calculate the distance d from the defect to the sensor.

[0175] .

[0176] in: v: Propagation speed of SH waves in the adhesive layer (unit: mm / μs).

[0177] Δt = tp - t0, which is the time delay between wave emission and reception / reflection; Because guided waves propagate in two paths, they need to be divided by 2.

[0178] Spatial mapping and result annotation Based on the location and propagation direction of the EMAT, the distance d is mapped to the structural coordinate system. The formula for calculating the defect location coordinates is: , .

[0179] in: (x0, y0): Current position coordinates of the EMAT sensor; θ: The angle between the direction of guided wave propagation and the x-axis (in radians or degrees); (x) d ,y d ): The location of the defect to be determined in the structural coordinate system.

[0180] 3. Defect Output and Report Generation The location, size, and type of defects are integrated into a map, and a quality inspection report is automatically generated for traceability and repair guidance.

Claims

1. An integrated automatic detection system for FSB bonding quality of Mark-type cargo containment systems, characterized in that, FSB has two layers of fiber fabric with an aluminum foil in between. The fiber fabric and aluminum foil are bonded together with a polyurethane adhesive. The specific steps are as follows: S1: Automatic inspection of surface ripples and wrinkles in FSB. Use a high-resolution camera or 3D scanning device to acquire FSB surface image data f(x,y), and perform a two-dimensional Fourier transform on the image f(x,y) to obtain the frequency domain representation F(u,v): ； in: f(x,y) is the spatial domain signal of the image; F(u,v) is the frequency component of the image in the frequency domain; j is the imaginary unit; M is the number of pixels in the image along the x-direction (horizontal direction), in pixels; N is the number of pixels in the image along the y-direction (vertical direction), in pixels; u and v are the horizontal and vertical frequencies in the frequency domain, respectively, with a unit of 1 / pixel. e is the natural constant, with a value of approximately 2.178; S2: In the frequency domain, the low-pass filter H(u,v) is multiplied by the spectrum F(u,v) to filter out high-frequency noise, and the spatial domain image after filtering is g(x,y); ； S3: Perform morphological operations on the filtered spatial domain image, and combine structural elements to optimize the image structure to ensure the segmentability and classification accuracy of defective regions. After completing the morphological optimization process, the contour boundary point sequence of each suspected defect region is extracted, and the local curvature is calculated by applying the three-point difference method to each contour boundary point. S4: By statistically analyzing known samples, two thresholds, a low threshold and a high threshold, are preset. Defects are compared with the preset thresholds for classification. When the curvature value is lower than the preset low threshold, it is judged as a ripple; when it exceeds the preset high threshold, it is judged as a wrinkle. Based on the curvature value, it is determined whether a defect exists. Once the defect is identified, it needs to be determined based on the pixel coordinates (x, y) on the image. p ,y p Transform to actual physical coordinates (X,Y); X=x p –s x , Y=y p –s y ; S5: Automatic inspection of FSB adhesive layer bubble defects Shear horizontal waves (SH waves) are generated on the surface of the adhesive layer using EMAT technology, and their reflected signals s(t) are collected. The envelope curve A(t) is obtained by envelope processing of the collected reflected signals, with units of volts (V). ； in (t) is the Hilbert transform result of signal s(t); Analyze the envelope curve A(t) to identify significant peaks. Based on these peaks, set a time window of interest [t1, t2] that should cover possible defect reflection signals. Extract relevant features from the samples, including the envelope energy E (unit: V). 2 -s, spectral bandwidth B, unit: Hz, maximum envelope amplitude A max ; ； Where dt is the integral variable, representing a small time increment in the continuous time domain; ； Where f1 represents the lower limit frequency of the spectrum bandwidth, the starting point of the -3dB bandwidth; and f2 represents the upper limit frequency of the spectrum bandwidth, the ending point of the -3dB bandwidth. S6: Construct a defect size estimation model to achieve quantitative estimation of the size of adhesive layer defects; ； Where: D is the estimated equivalent defect diameter, in mm; Coefficients k1, k2, k3, C: obtained by fitting several standard defect samples of known size using a multiple linear regression method; The estimated diameter D is compared with a preset standard threshold of 8 mm to determine whether a defect exists. If D > the preset standard threshold, it is judged as an excessive defect; If D < the preset standard threshold, then the defect is considered to be within an acceptable range; For a single bubble, calculate the area from the equivalent diameter D output by the size estimation model: ; This can be achieved through an equivalent area threshold, i.e., 50mm. 2 Compare and determine if it is a defect; When multiple reflection peaks are identified within the time window of interest, the time interval between adjacent peaks is calculated and converted into spatial spacing based on the waveguide propagation speed. If the number and spacing criteria of the bubble string are met, it is determined to be a bubble string. Using scan data, the spatial distribution direction of the bubble string is determined, and the angle between the bubble string and the length direction of the FSB is calculated to determine whether it is a transverse bubble string. For horizontal bubble clusters, they are directly judged as unqualified; For non-lateral bubble strings, estimate the equivalent diameter Di of each bubble individually, and then calculate the total equivalent area: ; This value can be directly compared with the rejection criteria to determine whether it is unqualified; S7: Send SH guided waves via the EMAT sensor and measure the time difference Δt of the reflected waves; Calculate the distance d from the defect to the sensor; ； in: v: Propagation speed of SH waves in the adhesive layer, unit: mm / s; Δt=t p -t0 is the time delay between wave emission and reception / reflection, in seconds; Based on the location and propagation direction of EMAT, the distance d is mapped to the structural coordinate system. The formula for calculating the defect location coordinates is as follows: ； ； in: (x0, y0): Current position coordinates of the EMAT sensor; θ: The angle between the direction of guided wave propagation and the x-axis, expressed in radians or degrees; (x) d ,y d ): The location of the defect to be determined in the structural coordinate system; S9: Integrates defect location, size, and type into a map, automatically generating quality inspection reports for traceability and repair guidance.

2. The integrated automatic detection system for FSB bonding quality of a Mark-type cargo containment system according to claim 1, characterized in that, The thickness of the flexible secondary screen is 0.7mm.

3. The integrated automatic detection system for FSB bonding quality of a Mark-type cargo containment system according to claim 1, characterized in that, In step S2, the low-pass filter H(u,v) is a Gaussian low-pass filter. ； The cutoff frequency D0 is determined through pre-experimentation or simulation to ensure that ripple / wrinkle information is preserved while suppressing texture interference.

4. The integrated automatic detection system for FSB bonding quality of Mark-type cargo containment system according to claim 1, characterized in that, In step S3, the curvature formula is as follows: Let three adjacent points on the boundary (unit: mm) be: P1=(x1,y1), P2=(x2,y2), P3=(x3,y3) Area of ​​the triangle formed by them (unit: mm) 2 ) is defined as: ； This curvature value reflects the degree of curvature of the boundary at that point; Morphological operations include dilation, erosion, opening, and closing operations, which can be adaptively selected according to the morphological characteristics of the defect to be detected. Erosion is used to shrink the boundaries of defect areas and remove noise; it is defined as follows: ； in: A: The set of foreground pixels in the image, i.e., the defect area; B: Structural element; z: Coordinates of the center point of the structuring element; Bz Translate the structuring element to coordinate z; Pixels in the current image region are preserved only if B is completely contained within that region. Expansion is used to connect fracture edges or reinforce defect boundaries, and it is defined as follows: ； That is, as long as the structuring element has any intersection with the image region, the pixel is included in the foreground; The opening operation is used to remove small noise regions and smooth edges without destroying the main contour. It is defined as follows: ； This involves first corroding the material, and then expanding the resulting material. This method is suitable for removing fine textures or pseudo-defects. The closing operation is used to fill in small black holes and repair discontinuities in defective regions; it is defined as follows: ; This involves first expanding and then corroding, which is used to make the boundaries of defective areas more coherent.

5. The integrated automatic detection system for FSB bonding quality of a Mark-type cargo containment system according to claim 1, characterized in that, In step S4, the horizontal coordinate X and vertical coordinate Y in the actual physical coordinates (X,Y) are obtained in the following ways: Physical horizontal coordinate acquisition: Let the width of the captured image in pixels be W. img The corresponding FSB actual physical width is W. phy ; The horizontal scaling factor is ; When the horizontal coordinate of a defective pixel in the image is x p Unit: mm, its corresponding physical horizontal coordinate is: X=x p –s x ; Physical vertical coordinate acquisition: Let the height in pixels of a single frame be H. img The corresponding physical length of the actual coverage is Lphy; The vertical scaling factor is: ; If the vertical pixel coordinate of the defect in a single frame image is y p (Unit: mm), then its physical vertical coordinate within this frame is: Y local =y p .s y ; During the detection process, the camera or sensor moves along the length of the FSB, and its cumulative displacement is recorded as ΔL by the track encoder. Therefore, the global longitudinal physical coordinates of the defect are: Y=Y local +ΔL。 6. The integrated automatic detection system for FSB bonding quality of a Mark-type cargo containment system according to claim 1, characterized in that, The FSB adhesive layer is a two-component polyurethane cured adhesive. Taking automatic bonding as an example, the standard adhesive layer thickness is 0.2mm to 1.2mm.

7. The integrated automatic detection system for FSB bonding quality of a Mark-type cargo containment system according to claim 1, characterized in that, The rejection criterion in step S6 is that, for air bubbles within the layer, the surface area of ​​the air bubble must be less than 5% of the total area of ​​the adhesive joint within a 150mm span along each side of the joint. Bubble diameter > 8mm or equivalent area > 50mm² is unacceptable; transverse bubble bands are unacceptable. When three or more bubbles are detected, and the distance between any adjacent bubbles is less than or equal to 15 mm, they are considered as a single bubble string, regardless of the size of each bubble.

8. The integrated automatic detection system for FSB bonding quality of a Mark-type cargo containment system according to claim 1, characterized in that, If the angle between the bubble string and the length direction of the FSB is between 45 degrees and 90 degrees, it is defined as a transverse bubble string. If the angle is less than 45 degrees, it is not considered a transverse bubble string.

Images