A weld full focus ultrasonic imaging inspection method and system

Abstract

The application discloses a kind of weld full focusing ultrasonic imaging detection method and system, it is related to nondestructive testing technical field, by constructing the standard geometric model of fusion material anisotropy characteristics, the microstructure parameter of weld not directly measurable is converted into the input of calculable acoustic model, fundamentally solve the problem that conventional ultrasonic imaging is misaligned due to the anisotropy of coarse grain of stainless steel weld.In this basis, using bottom echo signal as internal scale, through the inversion iteration of image quality index driven model parameter, make model from initial theoretical estimate approach real physical state, realize the adaptive calibration of specific weld acoustic characteristics.Finally, the calibration model is used for actual defect detection, can effectively suppress the interference of organization noise, so that small defect is clearly highlighted in image, thereby the reliability and resolution of stainless steel weld ultrasonic testing are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology, and in particular to a method and system for full-focus ultrasonic imaging inspection of welds. Background Technology

[0002] Nuclear-grade austenitic stainless steel, with its excellent mechanical properties and corrosion resistance, is widely used in civil nuclear fission power generation projects and nuclear fusion projects. Its structural integrity and reliability are crucial to the operation of nuclear facilities and equipment.

[0003] Ultrasonic testing is one of the most widely used methods for detecting weld cracks. However, the anisotropic and coarse grain characteristics of austenitic stainless steel welds cause significant acoustic attenuation, scattering, and deflection of ultrasonic waves during their propagation. In addition, the ultrasonic field response of vertical and bifurcated cracks is extremely low, resulting in problems such as low signal-to-noise ratio, difficulty in defect detection, or inaccurate localization and quantification in the detection of these types of welds by existing testing technologies. Summary of the Invention

[0004] The present invention aims to provide a method and system for full-focus ultrasonic imaging detection of welds. By inverting and iterating, the theoretical model is calibrated to the actual physical state, overcoming the interference of the anisotropy of stainless steel welds on ultrasonic imaging. This enables adaptive suppression of structural noise and accurate identification of defects.

[0005] In a first aspect, embodiments of the present invention provide a method for full-focus ultrasonic imaging inspection of weld seams, comprising: Ultrasonic full matrix data, material properties, and geometric features of stainless steel weld specimens are collected, and a standard geometric model of the stainless steel weld specimens is constructed based on the material properties and geometric features. Based on the material properties, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld specimen is calculated. Combined with the structural parameters of the stainless steel weld specimen, an initial anisotropic microstructure model of the weld is constructed. Based on the initial weld anisotropic microstructure model, the propagation delay time of sound waves in the stainless steel weld specimen was calculated. Based on the delay time and the ultrasonic full matrix data, a standard full-focus image is obtained by performing full-focus imaging on the stainless steel weld sample. Identify the bottom echo signal in the standard full-focus image to obtain the image quality. If the image quality is lower than a preset quality threshold, perform inversion iteration on the anisotropic partition width and grain orientation angle of the initial weld anisotropic microstructure model until the generated standard full-focus image meets the preset quality threshold, and obtain the corrected inversion actual weld anisotropic microstructure model. Adaptive full-focus imaging of defect samples was performed using the actual weld anisotropic microstructure model.

[0006] As an improvement to the above scheme, the acquisition of ultrasonic full-matrix data, material properties, and geometric features of the stainless steel weld specimen, and the construction of a standard geometric model of the stainless steel weld specimen based on the material properties and geometric features, includes: An array of ultrasonic sensors is arranged along the welding direction of the stainless steel weld sample. Each crystal element is used as an excitation source in sequence, and all crystal elements are used as receiving sources simultaneously to collect the full matrix ultrasonic data of the stainless steel weld sample. The elastic modulus, shear modulus, and Poisson's ratio of stainless steel weld specimens were collected to obtain material properties. Based on these material properties, the stiffness matrix of the stainless steel weld specimens was constructed. Geometric features of stainless steel weld specimens are collected, and a standard geometric model of the stainless steel weld specimens is constructed based on the stiffness matrix and the geometric features.

[0007] As an improvement to the above scheme, the step of calculating the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample based on the material properties, and constructing an initial weld anisotropic microstructure model by combining the structural parameters of the stainless steel weld sample, includes: Based on the aforementioned material properties, a stiffness matrix for the stainless steel weld specimen is constructed. Based on the stiffness matrix, and considering the propagation characteristics of sound waves in an elastic medium, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample is calculated. Obtain the structural parameters of the stainless steel weld specimen; the structural parameters include the width of the anisotropic microstructure partitions and the grain orientation. An initial anisotropic microstructure model of the weld is constructed based on the anisotropic sound velocity matrix and the structural parameters.

[0008] As an improvement to the above scheme, the step of calculating the propagation delay time of sound waves in the stainless steel weld specimen based on the initial weld anisotropic microstructure model includes: Based on the initial anisotropic microstructure model of the weld, and using Snell's law, the propagation path of the acoustic wave in the stainless steel weld sample is determined by forward modeling. Calculate the delay time based on the sound wave propagation path.

[0009] As an improvement to the above scheme, the step of determining the sound wave propagation path in the stainless steel weld sample using forward modeling based on the initial weld anisotropic microstructure model and Snell's law includes: An initial acoustic wave incident angle is set, and based on Snell's theorem, the initial refraction angle is calculated layer by layer according to the initial acoustic wave incident angle and the sound velocity of the current layer; wherein, the sound velocity of the current layer is obtained through the initial weld anisotropic structure model; Based on the thickness of the stainless steel weld sample and the initial refraction angle of each layer, calculate the coordinate offset of the path endpoint coordinates relative to the initial focal point coordinates. Adjust the initial sound wave incident angle and recalculate the coordinate offset value. If the coordinate offset value is less than the preset offset threshold in n consecutive iterations, then the latest initial sound wave incident angle is taken as the sound wave incident angle. Based on the incident angle of the acoustic wave and the initial anisotropic microstructure model of the weld, the propagation path of the acoustic wave in the stainless steel weld sample is determined by forward modeling based on Snell's law.

[0010] As an improvement to the above scheme, the bottom echo signal in the standard full-focus image is identified to obtain the image quality. If the image quality is lower than a preset quality threshold, the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model are inverted and iterated until the generated standard full-focus image meets the preset quality threshold, thus obtaining the corrected inverted actual weld anisotropic microstructure model, including: The bottom echo signal in the standard full-focus image is identified by a pre-trained deep learning model, and imaging quality parameters are obtained based on the bottom echo signal; the imaging quality parameters include at least one of image entropy, signal-to-noise ratio and echo integrity. Whenever a standard full-focus image is generated, it is determined whether the imaging quality parameters meet the preset quality threshold. If not, then the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model are inverted and iterated according to the preset adjustment range to regenerate the standard full-focus image; If so, the corrected and inverted anisotropic microstructure model of the actual weld can be obtained based on the current anisotropic microstructure partition width and grain orientation angle.

[0011] As an improvement to the above scheme, the step of determining whether the imaging quality parameters meet the preset quality threshold each time a standard full-focus image is generated includes: Obtain the target imaging quality parameters of the target image; If the deviation between the imaging quality parameters of the standard full-focus image and the target imaging quality parameters is less than a preset deviation threshold, then the imaging quality parameters are considered to meet the preset quality threshold; otherwise, the imaging quality parameters are considered not to meet the preset quality threshold.

[0012] As an improvement to the above scheme, the step of inverting and iterating the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model according to a preset adjustment range to regenerate a standard full-focus image includes: Within a preset first adjustment range, the width of the anisotropic microstructure partition of the initial weld anisotropic microstructure model is randomly adjusted, and the width of the anisotropic microstructure partition is updated; the first adjustment range includes the maximum deviation value and spacing compared to the original anisotropic microstructure partition width. Within a preset second adjustment range, the grain orientation angle of the initial weld anisotropic microstructure model is randomly adjusted and the grain orientation angle is updated; the second adjustment range includes the maximum deviation value and spacing compared to the original grain orientation angle. Based on the updated anisotropic structure partition width and the grain orientation angle, and combined with the anisotropic sound velocity matrix, the anisotropic sound velocity matrix is ​​recalculated. Based on the anisotropic sound velocity matrix, recalculate the sound velocity delay; A standard full-focus image is regenerated based on the recalculated sound velocity delay and the ultrasound full matrix data.

[0013] As an improvement to the above scheme, after determining whether the imaging quality parameters meet the preset quality threshold each time a standard full-focus image is generated, the weld full-focus ultrasonic imaging detection method further includes: If the imaging quality parameters of the standard full-focus images generated consecutively m times do not meet the preset quality threshold, then the adjustment range is expanded. If the adjustment range is expanded t times consecutively or the adjustment range reaches a preset limit, the elastic modulus and grain winding angle are adjusted, and the anisotropic sound velocity matrix is ​​recalculated.

[0014] Secondly, embodiments of the present invention also provide a weld seam full-focus ultrasonic imaging inspection system, comprising: The standard geometric model construction module is used to acquire ultrasonic full matrix data, material properties and geometric features of stainless steel weld specimens, and construct a standard geometric model of the stainless steel weld specimens based on the material properties and geometric features. The initial model building module is used to calculate the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample based on the material properties, and to build an initial weld anisotropic microstructure model in combination with the structural parameters of the stainless steel weld sample. The delay time calculation module is used to calculate the delay time of sound wave propagation in the stainless steel weld sample based on the initial weld anisotropic microstructure model. A standard full-focus image generation module is used to perform full-focus imaging on a stainless steel weld sample based on the delay time and the ultrasonic full matrix data to obtain a standard full-focus image. The initial model correction module is used to identify the bottom echo signal in the standard full-focus image and obtain the image quality. If the image quality is lower than the preset quality threshold, the anisotropic partition width and grain orientation angle of the initial weld anisotropic structure model are inverted and iterated until the generated standard full-focus image meets the preset quality threshold, and the corrected and inverted actual weld anisotropic structure model is obtained. The actual model application module is used to perform adaptive full-focus imaging of defect samples using the actual weld anisotropic microstructure model.

[0015] Compared with existing technologies, the present invention discloses a method and system for full-focus ultrasonic imaging inspection of welds. This method acquires full-matrix ultrasonic data, material properties, and geometric features of stainless steel weld samples. Based on the material properties and geometric features, a standard geometric model of the stainless steel weld sample is constructed. Based on the material properties, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample is calculated. Combined with the structural parameters of the stainless steel weld sample, an initial anisotropic microstructure model of the weld is constructed. Based on the initial anisotropic microstructure model, the propagation delay time of sound waves in the stainless steel weld sample is calculated. Based on the delay time and the ultrasonic full matrix data, full-focus imaging is performed on the stainless steel weld sample to obtain a standard full-focus image. The bottom echo signal in the standard full-focus image is identified to obtain the image quality. If the image quality is lower than a preset quality threshold, the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model are inverted and iterated until the generated standard full-focus image meets the preset quality threshold, resulting in a corrected and inverted actual weld anisotropic microstructure model. The actual weld anisotropic microstructure model is then used for adaptive full-focus imaging of defect samples. Using this embodiment of the invention, microstructure noise can be adaptively suppressed, achieving accurate defect identification. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of a full-focus ultrasonic imaging detection method for weld seams provided in an embodiment of the present invention; Figure 2 This is a standard geometric model of a stainless steel weld specimen constructed in one embodiment of the present invention; Figure 3 This is the anisotropic sound velocity distribution of a stainless steel weld sample calculated in one embodiment of the present invention. Figure 4 This is an initial weld anisotropic microstructure model constructed according to one embodiment of the present invention; Figure 5 This is a schematic diagram of the propagation path of sound waves in austenitic stainless steel weld, calculated according to an embodiment of the present invention. Figure 6 This is an adaptive iterative process for a standard all-focus image in one embodiment of the present invention; Figure 7 An anisotropic microstructure model of an actual weld constructed for one embodiment of the present invention; Figure 8 This is a schematic diagram comparing the fully focused images obtained before and after optimization using an actual weld anisotropic microstructure model in one embodiment of the present invention. Figure 9 This is a schematic diagram of the structure of a weld seam full-focus ultrasonic imaging inspection system provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] In the description and claims, it should be understood that the terms "first," "second," etc., used in the description and claims are only for the purpose of distinguishing the description of the same technical features, and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated, nor necessarily the order of description or chronological order. The terms are interchangeable where appropriate. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature.

[0019] This invention provides a method for full-focus ultrasonic imaging inspection of weld seams. Please refer to... Figure 1 In this embodiment, the weld full-focus ultrasonic imaging detection method is specifically executed through steps S1 to S6: S1. Collect ultrasonic full matrix data, material properties and geometric features of stainless steel weld specimens, and construct a standard geometric model of stainless steel weld specimens based on the material properties and geometric features.

[0020] Stainless steel weld specimens refer to stainless steel test blocks manufactured according to standard welding processes. Stainless steel weld specimens can be used to calibrate the physical parameters of sound wave propagation when the material has no defects or under conditions of inevitable defects. In this way, the anisotropic microstructure model of the actual weld can be used to accurately calculate the delay of the unknown defect sample at each focal point.

[0021] In ultrasonic testing, full-matrix acquisition can be achieved by alternating transmission and receiving by all personnel. The full-matrix ultrasonic data records the acoustic signals received by each possible pair of transmitter-receiver wafers. During full-focus imaging, it is necessary to traverse every signal in the full-matrix ultrasonic data, extract the amplitude at different times, and perform superposition calculations to synthesize a high-resolution image. Furthermore, the purpose of the inversion iteration in step S5 is to make the theoretical data calculated by the model approximate the full-matrix ultrasonic data.

[0022] The propagation speed and attenuation of sound waves in stainless steel are determined by the material's inherent properties. These properties reflect the material's elastic response to mechanical vibrations and its energy dissipation characteristics. Exemplarily, material properties include elastic modulus, shear modulus, and Poisson's ratio.

[0023] The reflection, refraction, and propagation path of sound waves are constrained by the workpiece boundaries, and the geometry determines the boundary conditions and propagation space of the sound waves. Geometric features reflect the physical boundaries and macroscopic morphology of the object being inspected, and are the basis for determining the start and end points of the sound wave propagation path and the reflection interface.

[0024] By constructing a standard geometric model, material properties and macroscopic geometric features can be bound together in digital space, enabling physical entities to be fully mapped in digital space. The standard geometric model contains all the boundary conditions and material parameters required for acoustic wave calculations.

[0025] S2. Based on the material properties, calculate the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample, and construct an initial weld anisotropic microstructure model by combining the structural parameters of the stainless steel weld sample.

[0026] During weld solidification, grains grow along a specific direction. The angle between this growth direction and the reference coordinate system is called the grain orientation angle, which is a key parameter describing the anisotropy of the weld. For the weld, the velocity of sound is a function of direction, and the anisotropic velocity matrix reflects the projection of the material's elastic stiffness matrix in different directions. Essentially, it is the anisotropic distribution of interatomic bonding forces in spatial directions.

[0027] Different regions on the cross-section of the weld have different grain morphologies and acoustic properties. By combining the anisotropic sound velocity matrix and structural parameters to establish an initial anisotropic weld structure model, the non-uniform and anisotropic propagation environment encountered by sound waves in a real weld can be reproduced.

[0028] S3. Calculate the propagation delay time of sound waves in the stainless steel weld specimen based on the initial weld anisotropic microstructure model.

[0029] In stainless steel welds, due to grain deflection, the path of sound waves from the emitting crystal to the focal point is tortuous, and the propagation path from the focal point to the receiving crystal is not a straight line.

[0030] The core algorithm of all-focus imaging is time-delay superposition. Sound waves from different paths must be aligned in time before they can be superimposed and enhanced. Based on the anisotropic microstructure model of the initial weld, the sound waves can be tracked to determine the delay time. The delay time reflects the cumulative effect of the sound path experienced by the sound wave as it propagates along a specific path.

[0031] S4. Based on the delay time and the ultrasonic full matrix data, perform full-focus imaging on the stainless steel weld sample to obtain a standard full-focus image.

[0032] It should be noted that the standard fully focused image described in this embodiment refers to the spatial distribution estimation of acoustic reflection intensity at various points inside the stainless steel weld sample based on the initial weld anisotropic microstructure model. The signal amplitudes at corresponding moments can be extracted from the full matrix data and superimposed using the time delay to obtain the standard fully focused image.

[0033] S5. Identify the bottom echo signal in the standard full-focus image to obtain the image quality. If the image quality is lower than a preset quality threshold, perform inversion iteration on the anisotropic partition width and grain orientation angle of the initial weld anisotropic microstructure model until the generated standard full-focus image meets the preset quality threshold, and obtain the corrected inversion actual weld anisotropic microstructure model.

[0034] Bottom surface echo is the reflection of sound waves after they penetrate the entire workpiece. Its signal clarity and signal-to-noise ratio directly reflect the accuracy of the initial weld anisotropic microstructure model. If the parameters in the initial weld anisotropic microstructure model are accurate, the generated standard fully focused image will have high quality. If the parameters are inaccurate, the sound waves cannot be properly focused, and the bottom surface echo will be messy.

[0035] In the initial anisotropic weld microstructure model, geometric features and material properties are measured values ​​and relatively certain; however, microstructural parameters, including the width of the anisotropic microstructure partitions and the grain orientation angle, are only estimated values ​​with the greatest uncertainty, which is the cause of imaging distortion. Therefore, when the image quality is substandard, the most uncertain parameters in the model are automatically and directionally adjusted until the image quality meets the standards, thereby inversely approximating the true weld microstructure.

[0036] S6. Adaptive full-focus imaging of defect samples is performed using the actual weld anisotropic microstructure model.

[0037] It should be noted that the defect sample is a stainless steel welded workpiece containing real internal defects generated in actual work, and its material grade and welding process are consistent with those of the stainless steel weld sample.

[0038] The modified anisotropic microstructure model of the actual weld accurately understands the anisotropic law of the weld. By using ultrasonic testing, abnormal echoes that deviate significantly from the normal sound field can be screened out, thereby identifying defect samples and performing full-focus imaging to automatically and accurately find the defects.

[0039] In the above scheme, by constructing a standard geometric model that integrates the anisotropic characteristics of the material, the microstructure parameters of the weld, which cannot be directly measured, are transformed into computable acoustic model inputs. This fundamentally solves the problem of inaccurate conventional ultrasonic imaging caused by the coarse-grained anisotropy of stainless steel welds. Based on this, the bottom echo signal is used as an internal scale, and the model parameters are iterated and inverted through image quality indicators. This allows the model to approximate the actual physical state from the initial theoretical estimate, achieving adaptive calibration for the acoustic characteristics of specific welds. Finally, this calibrated model is used for actual defect detection, effectively suppressing structural noise interference and clearly highlighting minute defects in the image, thus significantly improving the reliability and resolution of ultrasonic testing of stainless steel welds.

[0040] As a preferred embodiment, step S1 involves acquiring the full ultrasonic matrix data, material properties, and geometric features of the stainless steel weld specimen, and constructing a standard geometric model of the stainless steel weld specimen based on the material properties and geometric features, including: An array of ultrasonic sensors is arranged along the welding direction of the stainless steel weld sample. Each crystal element is used as an excitation source in sequence, and all crystal elements are used as receiving sources simultaneously to collect the full matrix ultrasonic data of the stainless steel weld sample. The elastic modulus, shear modulus, and Poisson's ratio of stainless steel weld specimens were collected to obtain material properties. Based on these material properties, the stiffness matrix of the stainless steel weld specimens was constructed. Geometric features of stainless steel weld specimens are collected, and a standard geometric model of the stainless steel weld specimens is constructed based on the stiffness matrix and the geometric features.

[0041] An array ultrasonic sensor consists of multiple independent piezoelectric crystals arranged in a specific pattern, with each crystal element capable of independently emitting or receiving ultrasonic waves. Arranging the array ultrasonic sensor along the welding direction of the stainless steel weld sample ensures a defined geometric relationship between the sound wave propagation path and the grain growth direction within the weld, enabling the acquired full-matrix data to reflect the actual propagation characteristics of sound waves passing through anisotropic media in different directions.

[0042] The elastic modulus is the ratio of normal stress to normal strain in a material during the elastic deformation stage, reflecting the material's ability to resist tensile or compressive deformation. The shear modulus is the ratio of shear stress to shear strain in a material during the elastic deformation stage, reflecting the material's ability to resist shear deformation. Poisson's ratio is the ratio of transverse strain to axial strain in a material under uniaxial tension or compression, reflecting the material's ability to deform laterally. Based on the elastic model, shear modulus, and Poisson's ratio, the calculated stiffness matrix can describe the elastic response of the material in various directions.

[0043] In some preferred embodiments, when the grain orientation angle of the weld is When the stiffness matrix of the stainless steel weld specimen is expressed as: ; Among them, E 11 E 22 E 33 These are the elastic moduli in three directions, respectively; v 12 v 23 v 13 Poisson's ratio in three directions; G 12 G 23 G 13 These are the shear moduli in three directions.

[0044] When the grain orientation angle of the weld is When the stiffness matrix of the stainless steel weld specimen is expressed as: , , ; Where m is the grain orientation angle The cosine value; n is the grain orientation angle. The sine value.

[0045] Geometric features determine the spatial range of sound wave propagation and the location of boundary reflection, while the stiffness matrix determines the propagation speed of sound waves inside the material. By coupling the two into a standard geometric model, it is possible to calculate the propagation path of sound waves inside the material and determine the reflection conditions of sound waves at the boundary.

[0046] In the above scheme, by collecting the material properties and geometric characteristics of the specific sample, all subsequent calculations are based on the actual physical state of the weld sample, fundamentally eliminating the systematic deviation caused by material batch differences and geometric processing errors.

[0047] As a preferred embodiment, step S2 involves calculating the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample based on the material properties, and constructing an initial anisotropic microstructure model of the weld based on the structural parameters of the stainless steel weld sample, including: Based on the aforementioned material properties, a stiffness matrix for the stainless steel weld specimen is constructed. Based on the stiffness matrix, and considering the propagation characteristics of sound waves in an elastic medium, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample is calculated. Obtain the structural parameters of the stainless steel weld specimen; the structural parameters include the width of the anisotropic microstructure partitions and the grain orientation. An initial anisotropic microstructure model of the weld is constructed based on the anisotropic sound velocity matrix and the structural parameters.

[0048] When sound waves propagate in a material, the velocity distribution is different along different directions. The grain orientation angle determines the propagation direction of the sound wave relative to the grain principal axis, while the stiffness matrix describes the elastic response in the direction of the principal axis. The anisotropic sound velocity matrix calculated based on the stiffness matrix describes a set of sound velocity values ​​corresponding to each grain orientation angle.

[0049] The Christophier equations characterize the propagation properties of sound waves in elastic media. In some preferred embodiments, the anisotropic sound velocity matrix is ​​calculated by combining the stiffness matrix with the Christophier equations. The coefficient matrix of the Christophier equations is expressed as: ; in, , The angle of sound wave propagation; The grain orientation angle is The stiffness matrix of austenitic stainless steel material, i.e. (a, b = 1, 2, 3, 4, 5, 6) represents the grain orientation angles. The values ​​at the corresponding positions of the subscript rows and columns in the stiffness matrix of the weld material.

[0050] Furthermore, the anisotropic sound velocity matrix is ​​expressed as: ; in, Let be the coefficient matrix of the Christophian equation. , , , , , These are the values ​​at the corresponding index rows and columns in the coefficient matrix; Density of the weld material; Phase velocity; polarization vector , Corresponding coefficient matrices The three components of the eigenvector.

[0051] It should be noted that the structural parameters obtained in the embodiments of this invention are theoretical values ​​based on material grade, known welding process information, process qualification reports, etc., rather than precise values ​​obtained through actual measurements. More accurate structural parameters need to be obtained through subsequent inversion and iteration processes.

[0052] The anisotropic sound velocity matrix provides sound velocity data at a specific grain orientation angle. However, the grain orientation angles differ in different regions of the weld, necessitating a spatial mapping between the orientation angles of different regions and their corresponding sound velocity matrices. By combining the anisotropic sound velocity matrix and structural parameters, the resulting initial weld anisotropic microstructure model possesses the ability to simultaneously describe both direction and position dependence, providing a spatially non-uniform, direction-dependent sound velocity field input for subsequent delay time calculations.

[0053] The above scheme establishes an acoustic model that can reflect the dual characteristics of orientation dependence and spatial non-uniformity inside the weld, so that the subsequent delay time calculation can truly reflect the path bending and time difference caused by the velocity change when the sound wave passes through different grain regions. This fundamentally solves the problem of inaccurate beam deflection prediction caused by the assumption of isotropy in traditional models.

[0054] In a preferred embodiment, step S3, calculating the propagation delay time of sound waves in the stainless steel weld sample based on the initial weld anisotropic microstructure model, is performed through steps S31-S32: S31. Based on the initial weld anisotropic structure model, and using Snell's law, determine the acoustic propagation path in the stainless steel weld sample by forward modeling. S32. Calculate the delay time based on the sound wave propagation path.

[0055] Snell's law describes the physical relationship between the angle of incidence and the angle of transmission of a sound wave as it passes through the interface between two different media, expressed as: ; in, The angle of refraction of the (n-1)th layer; For the nth layer The speed of sound corresponding to the direction; The angle of refraction of the nth layer; For the (n-1)th layer The velocity of sound corresponding to the direction. Using Snell's law and the known anisotropy of the velocity of sound in stainless steel weld specimens, the refraction angles of each layer can be obtained based on any incident angle of the sound wave.

[0056] Given the parameters of the anisotropic microstructure model of the initial weld, the sound wave propagation path can be traced forward from the emission point using Snell's law. The delay time is the total time it takes for the sound wave to travel from the emission element to the target position along the sound wave propagation path.

[0057] In the above scheme, the description of the different sound velocities at different locations in the initial weld anisotropic structure model is transformed into specific sound wave propagation paths and delay times, realizing the transformation from static material description to dynamic sound wave propagation process, and providing accurate time calibration values ​​for subsequent signal superposition imaging.

[0058] Further, preferably, step S31, determining the sound wave propagation path in the stainless steel weld sample using forward modeling based on the initial weld anisotropic microstructure model and Snell's law, includes: An initial acoustic wave incident angle is set, and based on Snell's theorem, the initial refraction angle is calculated layer by layer according to the initial acoustic wave incident angle and the sound velocity of the current layer; wherein, the sound velocity of the current layer is obtained through the initial weld anisotropic structure model; Based on the thickness of the stainless steel weld sample and the initial refraction angle of each layer, calculate the coordinate offset of the path endpoint coordinates relative to the initial focal point coordinates. Adjust the initial sound wave incident angle and recalculate the coordinate offset value. If the coordinate offset value is less than the preset offset threshold in n consecutive iterations, then the latest initial sound wave incident angle is taken as the sound wave incident angle. Based on the incident angle of the acoustic wave and the initial anisotropic microstructure model of the weld, the propagation path of the acoustic wave in the stainless steel weld sample is determined by forward modeling based on Snell's law.

[0059] It should be noted that the initial incident angle of the sound wave can be set randomly or based on prior experience. The specific setting value does not affect the beneficial effects produced by the present invention. This incident angle setting is only for triggering calculations, and the accurate incident angle of the sound wave needs to be determined through subsequent iterations.

[0060] In some preferred embodiments, the thickness direction of the stainless steel weld specimen is set as the y-axis, and the initial focal point coordinates are... Based on the given initial angle of incidence Based on this, the refraction angle of each layer is calculated, and then the coordinates of the end point of the sound propagation path are obtained by combining the thickness calculation. , ,in, Let c be the thickness of the c-th layer. Further calculate the coordinate offset values. By continuously changing the initial incident angle of the first layer And repeatedly calculate the coordinate offset value. until satisfied Finally, the sound wave propagation path is calculated based on the latest initial incident angle.

[0061] In the above scheme, the path search problem is transformed into a numerical optimization problem of the incident angle through an iterative strategy of hypothesis, verification and continuous correction. This enables the computer to automatically and efficiently find the propagation path of the sound wave to the given focal point in a complex weld model, providing a physically correct spatial trajectory for the subsequent accurate calculation of the delay time.

[0062] As a preferred embodiment, step S5 involves identifying the bottom echo signal in the standard full-focus image to obtain the image quality. If the image quality is lower than a preset quality threshold, the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model are inverted and iterated until the generated standard full-focus image meets the preset quality threshold, thus obtaining the corrected and inverted actual weld anisotropic microstructure model. This includes steps S51-S54. S51. Identify the bottom echo signal in the standard full-focus image using a pre-trained deep learning model, and obtain imaging quality parameters based on the bottom echo signal; the imaging quality parameters include at least one of image entropy, signal-to-noise ratio, and echo integrity. S52. Whenever a standard full-focus image is generated, determine whether the imaging quality parameters meet the preset quality threshold. If not, proceed to step S53: perform inversion iteration on the anisotropic partition width and grain orientation angle of the initial weld anisotropic microstructure model according to the preset adjustment range, and regenerate the standard full-focus image. If so, proceed to step S54: based on the current anisotropic microstructure partition width and grain orientation angle, obtain the corrected and inverted actual weld anisotropic microstructure model.

[0063] When ultrasonic waves propagate to the bottom surface of a stainless steel weld sample, they generate a reflected echo, also known as the bottom surface echo signal (which appears at a specific depth in the image). This signal is an important indicator of whether the sound wave can effectively penetrate the entire workpiece. If the signal is clear and complete, it indicates that the sound wave has not been severely disturbed during propagation; otherwise, it indicates that there is a deviation in the model parameters or an anomaly in the material itself.

[0064] It should be noted that the training samples of the deep learning model are labeled images that are highly consistent with the actual delay and imaging delay obtained from standard stainless steel weld materials. The deep learning model trained in this way has high sensitivity to bottom surface echoes and can achieve accurate identification and extraction of bottom surface echo signals.

[0065] Image entropy is a statistical measure used to describe the degree of disorder in the gray-level distribution of pixels in an image. The larger the image entropy, the more dispersed the gray-level distribution and the more cluttered the background; the smaller the entropy value, the more concentrated the gray-level distribution and the clearer the contrast.

[0066] In some preferred embodiments, image entropy is expressed as: , ; in, The grayscale value of a pixel; The grayscale value of the neighborhood; For characteristic binary pairs Frequency of occurrence; The scale of the image; For characteristic binary pairs Probability of occurrence; Let be the two-dimensional entropy of the constructed image.

[0067] In a fully focused image of a weld, the signal refers to the amplitude of the bottom surface echo, while the noise refers to the amplitude of clutter in the image background, such as from grain scattering. A higher signal-to-noise ratio indicates a more prominent effective signal and makes defects easier to identify.

[0068] Echo integrity refers to the continuity and shape integrity of the bottom surface echo in the image. A complete bottom surface echo should be a continuous, smooth bright line; if there are breaks, gaps, or deformations, it indicates that the sound waves have not propagated effectively in some areas or that the model is not in focus.

[0069] For each imaging quality parameter, a qualified judgment standard value is preset. When all imaging quality parameters meet their respective thresholds at the same time, the image quality is judged to be qualified, and the anisotropic microstructure model of the actual weld can be obtained. If any one of them is not met, it is judged to be unqualified, and then the width of the anisotropic microstructure partition and the grain orientation angle are inverted and iterated.

[0070] In this embodiment of the invention, by adjusting the range to limit the variation space of the anisotropic structure partition width and grain orientation angle during the inversion iteration, the search process is ensured to be carried out within a physically reasonable space, while preventing the optimal solution from being crossed or entering an unreasonable physical state due to excessive step size.

[0071] In the above scheme, bottom surface echo is used as an intrinsic physical benchmark. Image features are converted into quantitative indicators through a deep learning model. The model parameters are then adjusted based on whether the quantitative indicators meet the standards until the model is accurate. In this way, the model is calibrated from the initial empirical estimation state to the real physical state.

[0072] Further, preferably, step S52, each time a standard full-focus image is generated, determining whether the imaging quality parameters meet a preset quality threshold, includes: Obtain the target imaging quality parameters of the target image; If the deviation between the imaging quality parameters of the standard full-focus image and the target imaging quality parameters is less than a preset deviation threshold, then the imaging quality parameters are considered to meet the preset quality threshold; otherwise, the imaging quality parameters are considered not to meet the preset quality threshold.

[0073] It should be noted that the target image refers to a reference image that is predetermined according to the detection standards and process requirements and represents the ideal imaging state under the detection scenario.

[0074] In some preferred embodiments, if the signal-to-noise ratio and / or image entropy deviation between the standard full-focus image and the target image exceeds 30%, or the bottom echo contour integrity is less than 70%, then the imaging quality parameters are considered not to meet the preset quality threshold.

[0075] In the above scheme, the absolute signal-to-noise ratio of the bottom surface echo may naturally differ for stainless steel welds of different thicknesses and bevel types. If a fixed threshold is used, a separate threshold standard needs to be set for each type of weld, increasing the complexity of the scheme deployment. However, this scheme only requires determining a typical target image for that type of weld, and all other welds of the same type are compared with this target. The threshold standard is unified as the deviation range, eliminating the need to reset the absolute value for each type of weld.

[0076] Preferably, step S53, performing inversion iteration on the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model according to the preset adjustment range, and regenerating a standard full-focus image, includes: Within a preset first adjustment range, the width of the anisotropic microstructure partition of the initial weld anisotropic microstructure model is randomly adjusted, and the width of the anisotropic microstructure partition is updated; the first adjustment range includes the maximum deviation value and spacing compared to the original anisotropic microstructure partition width. Within a preset second adjustment range, the grain orientation angle of the initial weld anisotropic microstructure model is randomly adjusted and the grain orientation angle is updated; the second adjustment range includes the maximum deviation value and spacing compared to the original grain orientation angle. Based on the updated anisotropic structure partition width and the grain orientation angle, the anisotropic sound velocity matrix is ​​recalculated; Based on the anisotropic sound velocity matrix, recalculate the sound velocity delay; A standard full-focus image is regenerated based on the recalculated sound velocity delay and the ultrasound full matrix data.

[0077] In some preferred embodiments, the width of the original anisotropic tissue partition is randomly selected. Adjusting the width of anisotropic tissue partitions with a 5% deviation and a 1% spacing Simultaneously, the grain orientation angle is randomly adjusted within a 5% deviation and a 1% spacing of the original grain orientation angle ϕ. Then, the adjusted anisotropic microstructure partition width and grain orientation angle are combined, based on... and The stiffness matrix is ​​recalculated, and the anisotropic sound velocity matrix is ​​recalculated based on the new stiffness matrix.

[0078] It is understood that in the above embodiments of the present invention, the first adjustment range and the second adjustment range are both set with a deviation of 5% and a spacing of 1% relative to the original parameters. However, the first adjustment range and the second adjustment range can actually take different values, and the setting of this range depends on engineering requirements.

[0079] It should be noted that after step S53 of the present invention is completed, a standard full-focus image will be regenerated. At this time, step S52 is triggered again to determine whether the imaging quality parameters meet the preset quality threshold. This triggering loop will terminate when the imaging quality parameters meet the preset quality threshold, and then step S54 will be executed to obtain the actual weld anisotropic structure model.

[0080] In the above scheme, a random search strategy is used to adjust the width of the anisotropic tissue partition and the grain orientation angle within a reasonable range. This can avoid the problem of stagnation in the local parameter space due to fixed step size or deterministic search path, and improve the possibility of finding the parameter combination that meets the image quality standard.

[0081] In a preferred embodiment, after performing step S52, the weld full-focus ultrasonic imaging detection method further includes: If the imaging quality parameters of the standard full-focus images generated consecutively m times do not meet the preset quality threshold, then the adjustment range is expanded. If the adjustment range is expanded t times consecutively or the adjustment range reaches a preset limit, the elastic modulus and grain winding angle are adjusted, and the anisotropic sound velocity matrix is ​​recalculated.

[0082] If the parameter combinations generated consecutively m times fail to achieve the image quality standard, it indicates that the current search space may not contain parameter combinations that enable the model to be accurate. In other words, the true value may fall outside the preset adjustment range. In this case, it is necessary to expand the search range to avoid repeated invalid iterations in the same invalid region.

[0083] When a satisfactory solution cannot be found even within the expanded parameter space, it indicates that the problem may not lie in the values ​​of the partition width and grain orientation angle, but in the deviation of the underlying physical input or the description of the higher-dimensional microstructure. Therefore, the elastic modulus and grain winding angle are adjusted, and the iteration is repeated.

[0084] It is understandable that the values ​​of m and t can be set based on the actual engineering situation, and the specific values ​​do not affect the beneficial effects produced by the embodiments of the present invention.

[0085] The above scheme avoids infinite stagnation at any single parameter level, thus improving the convergence success rate and computational efficiency of the entire inversion iteration process.

[0086] In one specific embodiment of the present invention, two 25 mm thick austenitic stainless steel welds are used as the test objects. The material is 316L stainless steel with a density of 7980 kg / m³. One weld is used as a standard weld sample, and the other has an internally machined hole defect with a hole diameter of 8 mm and a depth of 12.5 mm.

[0087] Ultrasonic data acquisition was performed using a 64-channel ultrasonic full-matrix acquisition system (64:64PR Gekko, Eddyfi Technologies, Canada). The probe employed a 64-element ultrasonic phased array sensor with a center frequency of 0.5 MHz and an element spacing of 1 mm. The array ultrasonic sensor was placed on the surface of the stainless steel weld sample along the welding direction, with the probe and weld surface well coupled using a couplant. Ultrasonic full-matrix data was acquired by sequentially using each element as the excitation source and simultaneously using all elements as the receiving source.

[0088] A stiffness matrix was constructed based on the parameters of elastic modulus, shear modulus, and Poisson's ratio of austenitic stainless steel. Geometric features of standard austenitic stainless steel weld samples were extracted to construct a standard geometric model of the austenitic stainless steel weld specimen. Table 1 shows the elastic modulus, shear modulus, and Poisson's ratio required to calculate the stiffness matrix. Figure 2 This is the standard geometric model for stainless steel weld specimens.

[0089] Table 1 Based on the stiffness matrix and Christoph's equations, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the austenitic stainless steel sample was calculated, and the results are as follows: Figure 3 The anisotropic sound velocity distribution is shown. Then, based on the anisotropic sound velocity matrix and the structural parameters of the austenitic stainless steel weld, the following is constructed: Figure 4 The initial anisotropic microstructure model of the austenitic stainless steel weld is shown, and the propagation path of the sound wave in the austenitic stainless steel weld specimen is determined by forward modeling based on Snell's law. The propagation path is as follows: Figure 5 As shown, the delay time is calculated based on the sound wave propagation path and the anisotropic sound velocity matrix. Based on the delay time and the full matrix data, full-focus imaging is performed to obtain a standard full-focus image.

[0090] The bottom echo signal in standard full-focus images is identified and extracted using a deep learning model, and the full-focus imaging quality is quantitatively judged by two-dimensional image entropy, imaging signal-to-noise ratio, and bottom echo integrity.

[0091] When the signal-to-noise ratio and / or image entropy deviation of the standard fully focused image compared to the target image exceeds 30% or the echo integrity is less than 70%, the convergence target is set at a signal-to-noise ratio and image entropy deviation of less than 10% or contour integrity greater than 90% compared to the target image. The maximum deviation value is set at 5% of the width of the anisotropic tissue partition and the grain orientation angle, with a spacing of 1%. Random arrangements are then made to recalculate the anisotropic sound velocity matrix that takes into account the deviation of sound propagation delay.

[0092] Based on the anisotropic sound velocity matrix, the above full-matrix focusing steps are performed on the full matrix data to obtain a standard full-focus image with the best imaging effect. Figure 6 This is an adaptive iterative process for standard full-focus images. The image highlighted in the figure represents the image with the best imaging effect after optimization, while the others are images from the optimization process. Based on the anisotropic tissue partition width and grain orientation angle corresponding to the image with the best imaging effect, we can obtain... Figure 7 The model shown is the actual weld anisotropic microstructure after the corrected inversion.

[0093] Defect detection can then be performed based on the actual weld anisotropic microstructure model. Figure 8 The images show the full focus of defects before and after optimization based on the anisotropic microstructure model of the actual weld. Figure 8 Image (a) is the unoptimized image. Figure 8 (b) is a fully focused image optimized based on the actual weld anisotropic microstructure model. It can be seen that the signal-to-noise ratio and quantitative accuracy of defect imaging have been significantly improved.

[0094] The weld full-focus ultrasonic imaging detection method provided in this embodiment of the invention calibrates the theoretical model to the real physical state through inversion iteration, overcomes the interference of stainless steel weld anisotropy on ultrasonic imaging, and can adaptively suppress structural noise to achieve accurate defect identification.

[0095] This invention provides a full-focus ultrasonic imaging inspection system for weld seams. Please refer to [link to relevant documentation]. Figure 9 The weld full-focus ultrasonic imaging inspection system includes a standard geometric model construction module 11, an initial model construction module 12, a delay time calculation module 13, a standard full-focus image generation module 14, an initial model correction module 15, and an actual model application module 16, wherein: The standard geometric model construction module 11 is used to collect ultrasonic full matrix data, material properties and geometric features of stainless steel weld specimens, and construct a standard geometric model of stainless steel weld specimens based on the material properties and geometric features. The initial model construction module 12 is used to calculate the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample based on the material properties, and to construct an initial weld anisotropic microstructure model in combination with the structural parameters of the stainless steel weld sample. The delay time calculation module 13 is used to calculate the delay time of sound wave propagation in the stainless steel weld sample based on the initial weld anisotropic microstructure model. The standard full-focus image generation module 14 is used to perform full-focus imaging on the stainless steel weld sample according to the delay time and the ultrasonic full matrix data to obtain a standard full-focus image. The initial model correction module 15 is used to identify the bottom echo signal in the standard full-focus image and obtain the image quality. If the image quality is lower than the preset quality threshold, the anisotropic partition width and grain orientation angle of the initial weld anisotropic structure model are inverted and iterated until the generated standard full-focus image meets the preset quality threshold, and the corrected inverted actual weld anisotropic structure model is obtained. The actual model application module 16 is used to perform adaptive full-focus imaging of defect samples using the actual weld anisotropic microstructure model.

[0096] In a preferred embodiment, the standard geometric model construction module 11 is specifically used for: An array of ultrasonic sensors is arranged along the welding direction of the stainless steel weld sample. Each crystal element is used as an excitation source in sequence, and all crystal elements are used as receiving sources simultaneously to collect the full matrix ultrasonic data of the stainless steel weld sample. The elastic modulus, shear modulus, and Poisson's ratio of stainless steel weld specimens were collected to obtain material properties. Based on these material properties, the stiffness matrix of the stainless steel weld specimens was constructed. Geometric features of stainless steel weld specimens are collected, and a standard geometric model of the stainless steel weld specimens is constructed based on the stiffness matrix and the geometric features.

[0097] In a preferred embodiment, the initial model construction module 12 is specifically used for: Based on the aforementioned material properties, a stiffness matrix for the stainless steel weld specimen is constructed. Based on the stiffness matrix, and considering the propagation characteristics of sound waves in an elastic medium, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample is calculated. Obtain the structural parameters of the stainless steel weld specimen; the structural parameters include the width of the anisotropic microstructure partitions and the grain orientation. An initial anisotropic microstructure model of the weld is constructed based on the anisotropic sound velocity matrix and the structural parameters.

[0098] In a preferred embodiment, the delay time calculation module 13 includes: The acoustic wave propagation path calculation unit is used to determine the acoustic wave propagation path in the stainless steel weld sample based on the initial weld anisotropic structure model and Snell's law using the forward modeling method. The delay time calculation unit is used to calculate the delay time based on the sound wave propagation path.

[0099] Further, preferably, the sound wave propagation path calculation unit is specifically used for: An initial acoustic wave incident angle is set, and based on Snell's theorem, the initial refraction angle is calculated layer by layer according to the initial acoustic wave incident angle and the sound velocity of the current layer; wherein, the sound velocity of the current layer is obtained through the initial weld anisotropic structure model; Based on the thickness of the stainless steel weld sample and the initial refraction angle of each layer, calculate the coordinate offset of the path endpoint coordinates relative to the initial focal point coordinates. Adjust the initial sound wave incident angle and recalculate the coordinate offset value. If the coordinate offset value is less than the preset offset threshold in n consecutive iterations, then the latest initial sound wave incident angle is taken as the sound wave incident angle. Based on the incident angle of the acoustic wave and the initial anisotropic microstructure model of the weld, the propagation path of the acoustic wave in the stainless steel weld sample is determined by forward modeling based on Snell's law.

[0100] In a preferred embodiment, the initial model correction module 15 includes: An imaging quality parameter extraction unit is used to identify the bottom echo signal in the standard full-focus image through a pre-trained deep learning model, and obtain imaging quality parameters based on the bottom echo signal; the imaging quality parameters include at least one of image entropy, signal-to-noise ratio, and echo integrity; The quality judgment unit is used to determine whether the imaging quality parameters meet the preset quality threshold whenever a standard full-focus image is generated. The inversion iteration unit is used to perform inversion iteration according to the anisotropic partition width and grain orientation angle of the initial weld anisotropic microstructure model within the preset adjustment range when the quality judgment unit outputs "no", and regenerate the standard full-focus image. The actual model generation unit is used to obtain the corrected and inverted actual weld anisotropic microstructure model based on the current anisotropic microstructure partition width and grain orientation angle when the quality judgment unit outputs.

[0101] Further, preferably, the quality judgment unit is specifically used for: Obtain the target imaging quality parameters of the target image; If the deviation between the imaging quality parameters of the standard full-focus image and the target imaging quality parameters is less than a preset deviation threshold, then the imaging quality parameters are considered to meet the preset quality threshold; otherwise, the imaging quality parameters are considered not to meet the preset quality threshold.

[0102] Preferably, the inversion iteration unit is specifically used for: Within a preset first adjustment range, the width of the anisotropic microstructure partition of the initial weld anisotropic microstructure model is randomly adjusted, and the width of the anisotropic microstructure partition is updated; the first adjustment range includes the maximum deviation value and spacing compared to the original anisotropic microstructure partition width. Within a preset second adjustment range, the grain orientation angle of the initial weld anisotropic microstructure model is randomly adjusted and the grain orientation angle is updated; the second adjustment range includes the maximum deviation value and spacing compared to the original grain orientation angle. Based on the updated anisotropic structure partition width and the grain orientation angle, and combined with the anisotropic sound velocity matrix, the anisotropic sound velocity matrix is ​​recalculated. Based on the anisotropic sound velocity matrix, recalculate the sound velocity delay; A standard full-focus image is regenerated based on the recalculated sound velocity delay and the ultrasound full matrix data.

[0103] In a preferred embodiment, the initial model correction module 15 further includes an iterative parameter adjustment unit, used for: If the imaging quality parameters of the standard full-focus images generated consecutively m times do not meet the preset quality threshold, then the adjustment range is expanded. If the adjustment range is expanded t times consecutively or the adjustment range reaches a preset limit, the elastic modulus and grain winding angle are adjusted, and the anisotropic sound velocity matrix is ​​recalculated.

[0104] The weld full-focus ultrasonic imaging inspection system provided in this invention constructs a standard geometric model that integrates the anisotropic characteristics of the material. This transforms the microstructural parameters of the weld, which cannot be directly measured, into calculable acoustic model inputs, fundamentally solving the problem of inaccurate conventional ultrasonic imaging caused by the coarse-grained anisotropy of stainless steel welds. Based on this, the system utilizes the bottom echo signal as an internal scale and drives the inversion iteration of model parameters through image quality indicators. This allows the model to approximate the actual physical state from the initial theoretical estimate, achieving adaptive calibration for specific weld acoustic characteristics. Finally, this calibrated model is used for actual defect detection, effectively suppressing structural noise interference and clearly highlighting minute defects in the image, thereby significantly improving the reliability and resolution of ultrasonic inspection of stainless steel welds.

[0105] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0106] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for full-focus ultrasonic imaging inspection of weld seams, characterized in that, include: Ultrasonic full matrix data, material properties, and geometric features of stainless steel weld specimens are collected, and a standard geometric model of the stainless steel weld specimens is constructed based on the material properties and geometric features. Based on the material properties, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld specimen is calculated. Combined with the structural parameters of the stainless steel weld specimen, an initial anisotropic microstructure model of the weld is constructed. Based on the initial weld anisotropic microstructure model, the propagation delay time of sound waves in the stainless steel weld specimen was calculated. Based on the delay time and the ultrasonic full matrix data, a standard full-focus image is obtained by performing full-focus imaging on the stainless steel weld sample. Identify the bottom echo signal in the standard full-focus image to obtain the image quality. If the image quality is lower than a preset quality threshold, perform inversion iteration on the anisotropic partition width and grain orientation angle of the initial weld anisotropic microstructure model until the generated standard full-focus image meets the preset quality threshold, and obtain the corrected inversion actual weld anisotropic microstructure model. Adaptive full-focus imaging of defect samples was performed using the actual weld anisotropic microstructure model.

2. The weld seam full-focus ultrasonic imaging inspection method as described in claim 1, characterized in that, The process involves acquiring full-matrix ultrasonic data, material properties, and geometric features of the stainless steel weld specimen, and constructing a standard geometric model of the stainless steel weld specimen based on the material properties and geometric features, including: An array of ultrasonic sensors is arranged along the welding direction of the stainless steel weld sample. Each crystal element is used as an excitation source in sequence, and all crystal elements are used as receiving sources simultaneously to collect the full matrix ultrasonic data of the stainless steel weld sample. The elastic modulus, shear modulus, and Poisson's ratio of stainless steel weld specimens were collected to obtain material properties. Based on these material properties, the stiffness matrix of the stainless steel weld specimens was constructed. Geometric features of stainless steel weld specimens are collected, and a standard geometric model of the stainless steel weld specimens is constructed based on the stiffness matrix and the geometric features.

3. The method for full-focus ultrasonic imaging inspection of welds as described in claim 1, characterized in that, The process involves calculating the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample based on the material properties, and constructing an initial anisotropic microstructure model of the weld based on the structural parameters of the stainless steel weld sample, including: Based on the material properties, the stiffness matrix of the stainless steel weld specimen is constructed. Based on the stiffness matrix, and considering the propagation characteristics of sound waves in an elastic medium, the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample is calculated. Obtain the structural parameters of the stainless steel weld specimen; the structural parameters include the width of the anisotropic microstructure partitions and the grain orientation. An initial anisotropic microstructure model of the weld is constructed based on the anisotropic sound velocity matrix and the structural parameters.

4. The method for full-focus ultrasonic imaging inspection of welds as described in claim 1, characterized in that, The step of calculating the propagation delay time of sound waves in the stainless steel weld specimen based on the initial weld anisotropic microstructure model includes: Based on the initial anisotropic microstructure model of the weld, and using Snell's law, the propagation path of the acoustic wave in the stainless steel weld specimen is determined by forward modeling. Calculate the delay time based on the sound wave propagation path.

5. The weld seam full-focus ultrasonic imaging inspection method as described in claim 4, characterized in that, The step of determining the sound wave propagation path in the stainless steel weld sample using forward modeling based on the initial weld anisotropic microstructure model and Snell's law includes: An initial acoustic wave incident angle is set, and based on Snell's theorem, the initial refraction angle is calculated layer by layer according to the initial acoustic wave incident angle and the sound velocity of the current layer; wherein, the sound velocity of the current layer is obtained through the initial weld anisotropic structure model; Based on the thickness of the stainless steel weld sample and the initial refraction angle of each layer, calculate the coordinate offset of the path endpoint coordinates relative to the initial focal point coordinates. Adjust the initial sound wave incident angle and recalculate the coordinate offset value. If the coordinate offset value is less than the preset offset threshold in n consecutive iterations, then the latest initial sound wave incident angle is taken as the sound wave incident angle. Based on the incident angle of the acoustic wave and the initial anisotropic microstructure model of the weld, the propagation path of the acoustic wave in the stainless steel weld sample is determined by forward modeling based on Snell's law.

6. The method for full-focus ultrasonic imaging inspection of welds as described in claim 1, characterized in that, The process involves identifying the bottom echo signal in the standard full-focus image to obtain image quality. If the image quality is lower than a preset quality threshold, the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model are inverted and iterated until the generated standard full-focus image meets the preset quality threshold, resulting in a corrected and inverted actual weld anisotropic microstructure model, including: The bottom echo signal in the standard full-focus image is identified by a pre-trained deep learning model, and imaging quality parameters are obtained based on the bottom echo signal; the imaging quality parameters include at least one of image entropy, signal-to-noise ratio and echo integrity. Whenever a standard full-focus image is generated, it is determined whether the imaging quality parameters meet the preset quality threshold. If not, then the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model are inverted and iterated according to the preset adjustment range to regenerate the standard full-focus image; If so, the corrected and inverted anisotropic microstructure model of the actual weld can be obtained based on the current anisotropic microstructure partition width and grain orientation angle.

7. The method for full-focus ultrasonic imaging inspection of welds as described in claim 6, characterized in that, The step of determining whether the imaging quality parameters meet the preset quality threshold each time a standard full-focus image is generated includes: Obtain the target imaging quality parameters of the target image; If the deviation between the imaging quality parameters of the standard full-focus image and the target imaging quality parameters is less than a preset deviation threshold, then the imaging quality parameters are considered to meet the preset quality threshold; otherwise, the imaging quality parameters are considered not to meet the preset quality threshold.

8. The method for full-focus ultrasonic imaging inspection of welds as described in claim 6, characterized in that, The process involves inverting and iterating the anisotropic microstructure partition width and grain orientation angle of the initial weld anisotropic microstructure model within a preset adjustment range to regenerate a standard full-focus image, including: Within a preset first adjustment range, the width of the anisotropic microstructure partition of the initial weld anisotropic microstructure model is randomly adjusted, and the width of the anisotropic microstructure partition is updated; the first adjustment range includes the maximum deviation value and spacing compared to the original anisotropic microstructure partition width. Within a preset second adjustment range, the grain orientation angle of the initial weld anisotropic microstructure model is randomly adjusted and the grain orientation angle is updated; the second adjustment range includes the maximum deviation value and spacing compared to the original grain orientation angle. Based on the updated anisotropic structure partition width and the grain orientation angle, and combined with the anisotropic sound velocity matrix, the anisotropic sound velocity matrix is ​​recalculated. Based on the anisotropic sound velocity matrix, recalculate the sound velocity delay; A standard full-focus image is regenerated based on the recalculated sound velocity delay and the ultrasound full matrix data.

9. The method for full-focus ultrasonic imaging inspection of welds as described in claim 6, characterized in that, After determining whether the imaging quality parameters meet a preset quality threshold each time a standard full-focus image is generated, the weld full-focus ultrasonic imaging detection method further includes: If the imaging quality parameters of the standard full-focus images generated consecutively m times do not meet the preset quality threshold, then the adjustment range is expanded. If the adjustment range is expanded t times consecutively or the adjustment range reaches a preset limit, the elastic modulus and grain winding angle are adjusted, and the anisotropic sound velocity matrix is ​​recalculated.

10. A weld seam full-focus ultrasonic imaging inspection system, characterized in that, include: The standard geometric model construction module is used to acquire ultrasonic full matrix data, material properties and geometric features of stainless steel weld specimens, and construct a standard geometric model of stainless steel weld specimens based on the material properties and geometric features. The initial model building module is used to calculate the anisotropic sound velocity matrix corresponding to different grain orientation angles in the stainless steel weld sample based on the material properties, and to build an initial weld anisotropic microstructure model in combination with the structural parameters of the stainless steel weld sample. The delay time calculation module is used to calculate the delay time of sound wave propagation in the stainless steel weld sample based on the initial weld anisotropic microstructure model. A standard full-focus image generation module is used to perform full-focus imaging on a stainless steel weld sample based on the delay time and the ultrasonic full matrix data to obtain a standard full-focus image. The initial model correction module is used to identify the bottom echo signal in the standard full-focus image and obtain the image quality. If the image quality is lower than the preset quality threshold, the anisotropic partition width and grain orientation angle of the initial weld anisotropic structure model are inverted and iterated until the generated standard full-focus image meets the preset quality threshold, and the corrected and inverted actual weld anisotropic structure model is obtained. The actual model application module is used to perform adaptive full-focus imaging of defect samples using the actual weld anisotropic microstructure model.

Images