A method and system for evaluating the degree of metal pitting defects based on artificial intelligence

Abstract

The application provides a kind of metal pitting defect degree evaluation method and system based on artificial intelligence, by obtaining the micro porosity distribution data of metal plate in the forming process, and the internal fluctuation signal of metal plate under the action of directional stress wave;Based on the propagation attenuation rate of internal fluctuation signal, the geometric parameters of structure abnormal area corresponding to micro porosity distribution data are generated;Obtain the current density distribution data of structure abnormal area;Micro porosity distribution data, geometric parameters of structure abnormal area and current density distribution data are input into artificial intelligence model to associate micro porosity distribution data with current density distribution data, generate the pore current coupling characteristic matrix of metal plate;According to pore current coupling characteristic matrix, the quantitative evaluation result of internal pitting defect degree of metal plate is generated;The technical scheme provided by the application realizes the dynamic quantitative evaluation of the degree of metal internal pitting defect, and breaks through the technical bottleneck of traditional single index detection method in defect evolution mechanism analysis and quantitative precision.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based metal pitting defect degree assessment method and system. Background Art

[0002] In the sheet metal manufacturing process, early detection and quantitative assessment of internal pitting defects are core requirements for ensuring material performance and product reliability. Because pitting defects are often caused by the accumulation of microscopic pores during the forming stage and gradually spread with electrochemical corrosion, traditional methods have difficulty simultaneously capturing the dynamic correlation between material structural anomalies and changes in electrical properties during the manufacturing process, resulting in delayed and inaccurate defect prediction. Industrial scenarios urgently need a technical solution that can integrate multi-dimensional physical characteristics and analyze defect evolution trends in real time to improve the initiative and accuracy of quality control.

[0003] Currently, the typical existing solution to this demand uses a composite method that combines ultrasonic testing with porosity statistical analysis. This method uses ultrasonic flaw detection technology to obtain structural anomaly signals within the metal sheet, and uses microscopic imaging technology to calculate the porosity distribution data of the local area. Finally, an empirical formula is used to linearly weight the two to estimate the defect risk level. However, the existing solution still has significant flaws: for example, the correlation between ultrasonic signals and porosity distribution data is based only on a static empirical model and cannot adapt to dynamic changes in the material forming process; the quantification of defect geometric characteristics and evolution trends relies on manual experience, lacking the ability to perform gradient analysis in the spatial dimension and multi-parameter collaborative modeling, making it difficult to achieve early warning and accurate classification of defects, which restricts the optimization space for metal sheet manufacturing processes. Summary of the Invention

[0004] The present invention provides an artificial intelligence-based metal pitting defect degree assessment method and system to solve the problems in the prior art of being unable to adapt to dynamic changes in the material forming process and difficult to achieve early warning and accurate classification of defects.

[0005] In a first aspect, the present invention provides a method for evaluating the degree of metal pitting defects based on artificial intelligence, comprising:

[0006] Acquiring microscopic porosity distribution data of the metal sheet during the forming process and internal fluctuation signals of the metal sheet under the action of directional stress waves;

[0007] generating geometric parameters of a structural abnormality region corresponding to the microscopic porosity distribution data based on a propagation attenuation rate of the internal fluctuation signal;

[0008] Acquiring current density distribution data of the structural abnormal area;

[0009] Inputting the micro porosity distribution data, the geometric parameters of the structural abnormal region, and the current density distribution data into an artificial intelligence model to correlate the micro porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate;

[0010] A quantitative evaluation result of the internal pitting corrosion defect degree of the metal plate is generated according to the pore current coupling characteristic matrix.

[0011] Optionally, generating geometric parameters of the structural abnormality region corresponding to the microscopic porosity distribution data based on the propagation attenuation rate of the internal fluctuation signal includes:

[0012] Acquiring a starting end energy value and an ending end energy value of the internal fluctuation signal on a propagation path, and calculating a propagation attenuation rate of the internal fluctuation signal based on the starting end energy value and the ending end energy value;

[0013] determining a spatial coverage of the propagation path in the metal sheet, and mapping the spatial coverage with spatial positions of micropores in the microporosity distribution data to calculate a regional mean of the microporosity within the spatial coverage;

[0014] Selecting, from the spatial coverage range, a plurality of target spatial coverage ranges whose area averages exceed a preset density threshold and whose propagation attenuation rates exceed a preset attenuation threshold;

[0015] Based on the boundary overlap range of the target spatial coverage range, geometric parameters of the structural abnormality region corresponding to the microscopic porosity distribution data are generated, and the geometric parameters include a radial extension range and a depth distribution range of the structural abnormality region.

[0016] Optionally, determining a spatial coverage range of the propagation path in the metal sheet, and mapping the spatial coverage range with spatial positions of micropores in the microporosity distribution data to calculate a regional mean of the microporosity within the spatial coverage range includes:

[0017] defining a spatial coverage range of the propagation path in the metal plate according to the wavelength and propagation direction of the internal fluctuation signal;

[0018] Based on the spatial coverage, a local coordinate system is established, the three-dimensional coordinates of each micropore in the microporosity distribution data are converted into the local coordinate system, and target micropores whose three-dimensional coordinates are within the boundary of the spatial coverage are selected;

[0019] Divide the spatial coverage area into a plurality of subintervals along the length direction, wherein each subinterval is divided into a fixed number of strips along the width direction to generate a plurality of grid units;

[0020] The micro-porosity of the target micro-pores of each grid cell is accumulated to calculate the micro-pore density parameter of each grid cell;

[0021] The microscopic pore density parameters of all grid cells are summed up and combined with the total number of the grid cells to generate a regional mean of the microscopic porosity within the spatial coverage area.

[0022] Optionally, based on the boundary overlap range of the target spatial coverage range, geometric parameters of the structural abnormality region corresponding to the micro-porosity distribution data are generated, the geometric parameters including the radial extension range and depth distribution range of the structural abnormality region, including:

[0023] Establishing a global coordinate system based on the metal plate, mapping the coordinate range of the target space coverage range into the global coordinate system to generate a three-dimensional geometric intersection of all target space coverage ranges;

[0024] Determine a target closed area according to the three-dimensional geometric intersection, and define an outer contour boundary of the target closed area as a boundary overlap range;

[0025] Dividing the boundary overlap range into a plurality of equally spaced plane analysis units along the processing direction of the metal sheet, calculating the coverage area ratio of the boundary overlap range within each plane analysis unit, and defining the radial extension range as the difference between the first and last coordinates of the plane analysis unit in the processing direction whose coverage area ratio exceeds a preset plane threshold;

[0026] Dividing the boundary overlap range into a plurality of equally spaced thickness analysis layers along the thickness direction of the metal plate, calculating the coverage volume ratio of the boundary overlap range within each thickness analysis layer, and defining the difference between the first and last coordinates of the thickness analysis layer in the thickness direction whose coverage volume ratio exceeds a preset thickness threshold as the depth distribution range;

[0027] Geometric parameters of the structural abnormality region are generated according to the radial extension range and the depth distribution range.

[0028] Optionally, the micro-porosity distribution data, the geometric parameters of the structural abnormality region, and the current density distribution data are input into an artificial intelligence model to associate the micro-porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate, including:

[0029] Converting the geometric parameters of the structural abnormality region into spatial constraints, wherein the spatial constraints include boundary coordinates of a radial extension range and a depth distribution range of the structural abnormality region in a global coordinate system;

[0030] Selecting a plurality of target micropores whose spatial positions are within the boundaries of the spatial constraints from the microporosity distribution data, and selecting a plurality of target current density measurement points whose spatial positions are within the boundaries of the spatial constraints from the current density distribution data;

[0031] Performing proximity matching on the spatial positions of the target microscopic pores and the target current density measurement points to determine the target current density measurement point corresponding to each target microscopic pore, and generating a plurality of matching pairs;

[0032] Using an artificial intelligence model, a nonlinear correlation calculation is performed on the microscopic porosity and current density values ​​of each matching pair to generate a coupling strength value for each matching pair to construct a pore current coupling characteristic matrix of the metal sheet.

[0033] Optionally, an artificial intelligence model is used to perform a nonlinear correlation calculation on the microscopic porosity and current density values ​​of each matching pair to generate a coupling strength value for each matching pair to construct a pore current coupling characteristic matrix of the metal plate, including:

[0034] Using the artificial intelligence model, mapping the microscopic porosity of each matching pair into a first eigenvector, and mapping the current density value of each matching pair into a second eigenvector;

[0035] Performing a product operation on the first feature vector and the second feature vector to generate an initial interaction feature vector for each matching pair;

[0036] Based on the dynamic weight parameters of the artificial intelligence model, the initial interaction feature vectors of each matching pair are weighted superimposed to obtain the superimposed feature vector of each matching pair, and the dynamic weight parameters are dynamically adjusted according to the relative distance between the spatial position of each matching pair and the geometric parameters of the structural abnormality area;

[0037] Converting the superimposed eigenvector of each matching pair into a corresponding scalar value, where the scalar value is defined as the coupling strength value of the corresponding matching pair;

[0038] According to the radial extension and depth distribution range of the structural anomaly area, the global coordinate system is divided into multiple spatial analysis areas;

[0039] The coupling strength values ​​of all matching pairs in each spatial analysis area were calculated by arithmetic averaging to obtain the mean coupling strength value of each spatial analysis area;

[0040] The coupling strength means of all spatial analysis areas are arranged to construct the pore current coupling characteristic matrix.

[0041] In a second aspect, the present invention provides a metal pitting defect degree assessment system based on artificial intelligence, comprising:

[0042] A first acquisition module is used to acquire microscopic porosity distribution data of the metal sheet during the forming process, and an internal fluctuation signal of the metal sheet under the action of a directional stress wave;

[0043] A first generating module is configured to generate geometric parameters of a structural abnormality region corresponding to the microscopic porosity distribution data based on a propagation attenuation rate of the internal fluctuation signal;

[0044] A second acquisition module is used to acquire current density distribution data of the structural abnormality area;

[0045] a correlation module, configured to input the micro-porosity distribution data, the geometric parameters of the structural abnormality region, and the current density distribution data into an artificial intelligence model, so as to correlate the micro-porosity distribution data with the current density distribution data and generate a pore current coupling characteristic matrix of the metal plate;

[0046] The second generating module is used to generate a quantitative evaluation result of the internal pitting defect degree of the metal plate according to the pore current coupling characteristic matrix.

[0047] In a third aspect, the present invention provides a computing device comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute an artificial intelligence-based metal pitting defect degree assessment method as described in any one of the first aspects.

[0048] In a fourth aspect, the present invention provides a computer storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement an artificial intelligence-based metal pitting defect degree assessment method as described in any one of the first aspects.

[0049] In the present invention, the micro-porosity distribution data of the metal plate during the forming process and the internal fluctuation signal of the metal plate under the action of a directional stress wave are obtained; based on the propagation attenuation rate of the internal fluctuation signal, the geometric parameters of the structural abnormality area corresponding to the micro-porosity distribution data are generated; the current density distribution data of the structural abnormality area is obtained; the micro-porosity distribution data, the geometric parameters of the structural abnormality area and the current density distribution data are input into an artificial intelligence model to associate the micro-porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate; based on the pore current coupling characteristic matrix, a quantitative evaluation result of the degree of internal pitting defects of the metal plate is generated. The technical solution provided by the present invention establishes the physical basis for defect detection by synchronously collecting the microscopic pore distribution and stress wave propagation characteristics during the material forming stage, breaking through the limitations of traditional single detection dimensions (such as porosity only or ultrasound only), and providing underlying support for multi-source data fusion; through the dynamic correlation between the propagation attenuation rate of the internal fluctuation signal and the microscopic porosity distribution data, the spatial range of the structural abnormality area is accurately located, overcoming the problem of fuzzy quantification of defect geometric characteristics in traditional ultrasonic detection; introducing electrical parameters (i.e. current density) to characterize the electrochemical response characteristics of the defect area, capturing the dynamic corrosion characteristics of pitting defects, and making up for the deficiency of existing technologies that ignore the correlation between electrical signals and material defects; through the artificial intelligence model, the nonlinear dynamic coupling of material properties (i.e. microporosity), physical characteristics (i.e. internal fluctuation signals) and electrical characteristics (current density) is realized, and a pore current coupling characteristic matrix is ​​constructed to solve the defect that the traditional linear weighted model cannot adapt to complex working conditions, and quantify the defect expansion rate and penetration depth, realizing the leap from static detection to dynamic prediction, and improving the timeliness and accuracy of defect assessment. Furthermore, by calculating the propagation attenuation rate of the internal fluctuation signal on the propagation path, mapping the regional average of the microporosity within its spatial coverage range in the metal plate, screening the target spatial coverage range that meets both the preset density threshold and the preset attenuation threshold, and generating the geometric parameters of the structural abnormality area, the limitation of the traditional method of relying on a single parameter to determine defects is broken through. Through the dual-threshold screening mechanism, namely the micropore density and the propagation attenuation rate, the complex defect area is accurately identified, the defect positioning accuracy and the anti-noise interference ability are improved, and reliable spatial constraints are provided for subsequent multi-source data fusion.

[0050] These and other aspects of the present invention will become more readily apparent from the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 A flowchart of a method for evaluating the degree of metal pitting defects based on artificial intelligence provided by an embodiment of the present invention;

[0053] Figure 2 A schematic structural diagram of a metal pitting defect degree assessment system based on artificial intelligence provided by an embodiment of the present invention;

[0054] Figure 3 A schematic diagram of the structure of a computing device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0056] In some of the processes described in the specification and claims of the present invention and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence, nor do they limit "first" and "second" to be different types.

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0058] Figure 1 The present invention provides a flowchart of a method for evaluating the degree of metal pitting defects based on artificial intelligence, as shown in FIG. Figure 1 As shown, the method includes:

[0059] In response to the urgent need for early detection and dynamic evaluation of internal pitting defects in the metal sheet manufacturing process, although the existing technology has achieved a certain degree of defect identification through ultrasonic detection and porosity analysis, it relies on static empirical models, ignores the dynamic changes of electrical characteristics, and lacks the ability of multi-dimensional data collaborative modeling, resulting in delayed defect prediction and insufficient accuracy. The present invention focuses on breaking through the limitations of a single detection dimension. By inputting the microscopic porosity distribution in the material forming stage, the propagation attenuation rate of the internal fluctuation signal, and the current density distribution in the structural abnormality area into the artificial intelligence model, the nonlinear correlation between porosity and current density is dynamically coupled to generate a spatial gradient-driven pore current coupling characteristic matrix. By combining the geometric parameters of the structural abnormality area, the defect area is accurately located and the evolution trend is quantified; by capturing the dynamic correlation between material structural anomalies and electrochemical responses in real time, the core problems of the existing technology such as poor adaptability of static models, lack of electrical parameters, and reliance on manual experience are solved, the real-time and accuracy of defect assessment are improved, and technical support is provided for active quality control of metal sheet manufacturing processes. Based on this, the present invention provides a metal pitting defect degree assessment method based on artificial intelligence, such as Figure 1 ,include:

[0060] Step 101: Acquire microscopic porosity distribution data of the metal sheet during the forming process, and internal fluctuation signals of the metal sheet under the action of a directional stress wave;

[0061] In this step, microporosity distribution data refers to the spatial distribution of microscopic pores within the metal sheet, obtained through microscopic imaging technology. This includes pore location, size, and pore volume per unit area, and is used to reflect the microstructural characteristics of material forming defects. Directed stress wave action refers to mechanical vibration excitation applied along a specific direction in the metal sheet, which is used to stimulate internal fluctuation signals to detect structural anomalies. The internal fluctuation signal refers to the vibration waveform data generated by the stress wave propagating through the metal sheet. Its attenuation characteristics reflect the internal structural integrity of the material.

[0062] In an embodiment of the present invention, high-resolution microscopic imaging technology is used to scan the microscopic pore distribution during the metal sheet forming process to generate microscopic porosity distribution data including the position and size of each pore; at the same time, a directional stress wave is applied along a preset direction of the metal sheet, and a piezoelectric sensor array is used to collect the original waveform data of the internal fluctuation signal of the metal sheet, wherein the microscopic porosity distribution data reflects the microstructural characteristics of the material forming defects, and the internal fluctuation signal characterizes the propagation characteristics of the stress wave in the material.

[0063] Step 102: generating geometric parameters of the structural abnormality region corresponding to the microscopic porosity distribution data based on the propagation attenuation rate of the internal fluctuation signal;

[0064] In this step, the propagation attenuation rate, which refers to the ratio of energy lost by the internal fluctuation signal along the propagation path, is calculated as the difference between the energy values ​​at the starting and ending points and is used to quantify the degree of structural anomaly in the material. The structural anomaly region, defined as a continuous spatial region that meets both the porosity density threshold and the fluctuation attenuation rate threshold, represents the core location of potential pitting defects. Geometric parameters, including the radial extension and depth distribution of the structural anomaly region, are used to quantify the defect's expansion trend and the degree of structural damage.

[0065] In an embodiment of the present invention, the difference between the energy value at the starting end and the energy value at the ending end of the internal fluctuation signal on the propagation path is calculated, and the difference is divided by the energy value at the starting end to obtain the propagation attenuation rate; the spatial coverage range of the fluctuation signal propagation path in the metal plate is determined, and the range is spatially mapped to the pore coordinates in the micro-porosity distribution data, and the regional mean of the micro-porosity within the spatial coverage range is statistically calculated; the target spatial coverage range whose regional mean exceeds a preset density threshold and whose propagation attenuation rate exceeds the preset attenuation threshold is screened out to generate the geometric parameters of the structural abnormality area according to its boundary overlap range.

[0066] Step 103: Acquire current density distribution data of the structural abnormality region;

[0067] In this step, the current density distribution data refers to the current intensity data per unit area of ​​the defect area obtained through electrochemical measurement, which reflects the electrochemical corrosion activity of the pitting defect.

[0068] In an embodiment of the present invention, a multi-channel current probe is arranged on the surface of the structural abnormality area, a microcurrent is applied by a constant potentiostat and the current density distribution in the structural abnormality area is measured, and the current density value of each measuring point is calculated according to Ohm's law (current density = current intensity / measurement area), and current density distribution data covering the structural abnormality area is generated to reflect the electrochemical corrosion activity of the defective area.

[0069] Step 104: Inputting the micro-porosity distribution data, the geometric parameters of the structural abnormal region, and the current density distribution data into an artificial intelligence model to correlate the micro-porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate;

[0070] In this step, the pore current coupling characteristic matrix refers to a three-dimensional numerical matrix generated by fusing microscopic porosity and current density distribution data through an artificial intelligence model, and each element represents the defect coupling strength of the corresponding spatial unit.

[0071] In an embodiment of the present invention, the micro-porosity distribution data, the geometric parameters of the structural anomaly area, and the current density distribution data are input into a pre-trained artificial intelligence model. The model uses a feature cross layer to calculate the nonlinear product relationship between the micro-porosity and the current density value, generates a coupling strength value for each matching pair, and finally constructs a pore current coupling feature matrix, in which the matrix dimensions match the geometric parameters of the structural anomaly area.

[0072] Step 105: generating a quantitative evaluation result of the internal pitting defect degree of the metal plate according to the pore current coupling characteristic matrix;

[0073] In this step, the quantitative evaluation results of the internal pitting defect degree refer to the defect expansion rate and penetration depth level generated based on the pore current coupling characteristic matrix gradient analysis, which are used to guide the optimization of the production process.

[0074] In an embodiment of the present invention, the coupling strength differences between adjacent matrix elements of the pore current coupling characteristic matrix along the processing direction, width direction and thickness direction of the global coordinate system are calculated respectively to generate processing direction gradient values, width direction gradient values ​​and thickness direction gradient values, and the target matrix elements whose gradients in the three directions exceed the preset thresholds are screened, and the span parameters of their target aggregation areas are counted, including the processing direction span, width direction span and thickness direction span, to confirm the quantitative evaluation parameters of the internal pitting defect degree of the metal plate, and output the quantitative evaluation results, such as high risk or medium risk, according to the preset defect level mapping table.

[0075] For example, in the continuous rolling production line of automobile steel plates, it is necessary to inspect galvanized steel plates with a thickness of 1.5 mm. First, the micro-porosity distribution data of the rolled steel plates are obtained by laser confocal microscopy, where the average micro-porosity is 8%. A 20 kHz directional stress wave is applied simultaneously and the internal fluctuation signal is collected. The propagation attenuation rate of the internal fluctuation signal on the propagation path is calculated to be 35%. The target spatial coverage range in the automobile steel plates with a micro-porosity of more than 7% and a propagation attenuation rate of more than 30% is screened out to generate the geometric parameters of the structural abnormal area, which includes a radial extension of 15 mm. , depth distribution 1.2mm; 32-channel current probes were arranged on the surface of the structural abnormal area, and the current density distribution data of the structural abnormal area was measured, where the maximum current density was 2.5A / m²; the above data were input into the artificial intelligence model to generate a pore current coupling characteristic matrix, and the processing direction gradient of 0.4 / mm and the thickness direction gradient of 0.3 / mm were calculated according to the matrix, thereby determining the defect expansion rate of 0.2mm / h and the penetration depth level of 0.15, and matching the two with the preset defect level mapping table. The output assessment result was medium risk, and process adjustment was recommended.

[0076] The embodiments of the present invention overcome the defects of existing technologies that rely on static empirical formulas, ignore electrical characteristics and lack spatial gradient analysis by integrating multi-source data such as micro-porosity, stress wave propagation attenuation rate and current density distribution; by accurately quantifying defect geometric parameters and evolution trends, the early detection capability and evaluation accuracy of pitting defects are improved, providing reliable technical support for active quality control in the metal sheet manufacturing process.

[0077] The present invention provides a specific embodiment, step 102, generating geometric parameters of the structural abnormality region corresponding to the microscopic porosity distribution data based on the propagation attenuation rate of the internal fluctuation signal, specifically includes the following steps:

[0078] Step 201: obtaining a starting end energy value and an ending end energy value of the internal fluctuation signal on a propagation path, and calculating a propagation attenuation rate of the internal fluctuation signal based on the starting end energy value and the ending end energy value;

[0079] In this step, the propagation path refers to the propagation trajectory of the internal fluctuation signal within the metal sheet, including the starting and ending locations and propagation direction, which is determined by the frequency of the mechanical excitation and the physical properties of the metal sheet. The starting energy value refers to the energy intensity of the internal fluctuation signal at the starting point of the propagation path, directly measured by a sensor. The ending energy value refers to the energy intensity of the internal fluctuation signal at the ending point of the propagation path, reflecting the remaining energy after the signal propagates.

[0080] In an embodiment of the present invention, the energy value of the starting end and the energy value of the ending end of the internal fluctuation signal on the propagation path are collected by a piezoelectric sensor array; the energy value at the starting end is subtracted from the energy value at the ending end to obtain the energy difference, and then the energy difference is divided by the energy value at the starting end to calculate the propagation attenuation rate. The propagation attenuation rate is used to quantify the energy loss ratio of the fluctuation signal on the propagation path, reflecting the degree of abnormality of the internal structure of the material.

[0081] Step 202: determining a spatial coverage of the propagation path in the metal plate, mapping the spatial coverage with the spatial positions of the micropores in the microporosity distribution data to calculate a regional mean of the microporosity within the spatial coverage;

[0082] In this step, spatial coverage refers to the physical area within the metal sheet that is actually affected by the propagation path of the fluctuating signal. It is rectangular in shape and twice the width of the wavelength. Micropores refer to the tiny gaps within the metal sheet formed during the forming process. Their distribution data is obtained using microscopic imaging technology. Microporosity refers to the volume fraction of micropores per unit area and reflects the density of the material. The regional mean refers to the average value of all microporosities within the spatial coverage area and is used to assess the local pore density.

[0083] In an embodiment of the present invention, the spatial coverage range actually affected by the propagation path in the metal plate is determined based on the propagation direction and wavelength of the internal fluctuation signal. The spatial coverage range is a rectangular area extending along the propagation path, and the width is twice the wavelength. The spatial coverage range is matched with the three-dimensional coordinates of the micropores in the microporosity distribution data, and all target micropores within the spatial coverage range are screened out. The sum of the microporosities of these pores is counted, and the regional mean of the microporosity within the spatial coverage range is calculated.

[0084] Step 203: Selecting target spatial coverages from the spatial coverages, wherein the average values ​​of the regions exceed a preset density threshold and the propagation attenuation rates exceed a preset attenuation threshold;

[0085] In this step, the preset density threshold refers to the upper limit of porosity allowed in the metal sheet manufacturing process and is used to determine whether the material is qualified. The preset attenuation threshold refers to the critical value for energy loss of the fluctuation signal and is used to identify significant structural anomalies. The target spatial coverage refers to the area where both the regional mean value exceeds the preset density threshold and the propagation attenuation rate exceeds the preset attenuation threshold.

[0086] In an embodiment of the present invention, the regional mean is compared with a preset density threshold, and the propagation attenuation rate is compared with a preset attenuation threshold. The spatial coverage range that satisfies both the regional mean exceeding the preset density threshold and the propagation attenuation rate exceeding the preset attenuation threshold is screened out and defined as the target spatial coverage range. The above-mentioned preset density threshold is set based on the metal sheet forming process standard, and the preset attenuation threshold is dynamically adjusted according to the material type and the fluctuation signal characteristics.

[0087] Step 204: generating geometric parameters of the structural abnormal region corresponding to the microscopic porosity distribution data based on the boundary overlap range of the target spatial coverage range, wherein the geometric parameters include a radial extension range and a depth distribution range of the structural abnormal region;

[0088] In this step, the boundary overlap range refers to the intersection of multiple target spatial coverage areas in three dimensions, representing the core location of the composite defect. The radial extension range refers to the maximum extension length of the structural anomaly area along the metal sheet's machining direction, reflecting the defect's extended length along the machining direction. The depth distribution range refers to the maximum penetration span of the structural anomaly area along the metal sheet's thickness, reflecting the depth of the defect's penetration into the sheet.

[0089] In an embodiment of the present invention, a global coordinate system is established based on the metal sheet, and the coordinates of the target space coverage range are mapped into the global coordinate system to generate a three-dimensional geometric intersection of all target space coverage ranges. The target closed area is determined based on the intersection, and the outer contour boundary of the target closed area is defined as the boundary overlap range. The boundary overlap range is divided into equally spaced plane analysis units along the processing direction of the metal sheet, and the coverage area ratio within each unit is calculated. The difference between the first and last coordinates of the unit exceeding the preset plane threshold in the processing direction is defined as the radial extension range. The thickness analysis layer is divided into equally spaced layers along the thickness direction of the metal sheet, and the coverage volume ratio within each layer is calculated. The difference between the first and last coordinates of the layer exceeding the preset thickness threshold in the thickness direction is defined as the depth distribution range. Finally, based on the radial extension range and the depth distribution range, the geometric parameters of the structural abnormality area are generated.

[0090] The embodiment of the present invention uses dual-threshold screening (including micro-porosity and propagation attenuation rate) to accurately locate complex defect areas, breaking through the limitations of traditional single-parameter detection methods; generates geometric parameters through boundary overlapping ranges to achieve objective quantification of defect spatial dimensions, avoiding reliance on manual experience; improves the accuracy and anti-interference ability of defect detection, and provides spatial constraints for subsequent multi-source data fusion.

[0091] The present invention provides a specific embodiment, step 202, determining the spatial coverage of the propagation path in the metal plate, mapping the spatial coverage with the spatial positions of the micropores in the microporosity distribution data to calculate the regional mean of the microporosity within the spatial coverage, specifically comprising the following steps:

[0092] Step 211: defining the spatial coverage of the propagation path in the metal plate according to the wavelength and propagation direction of the internal fluctuation signal;

[0093] In this step, wavelength refers to the spatial distance between adjacent peaks or troughs of the internal wave signal as it propagates through the metal sheet, reflecting the frequency characteristics of the signal. Propagation direction refers to the spatial distance between adjacent peaks or troughs of the internal wave signal as it propagates through the metal sheet, reflecting the frequency characteristics of the signal.

[0094] In an embodiment of the present invention, the spatial coverage range of the propagation path in the metal plate is defined based on the wavelength and propagation direction of the internal fluctuation signal; the spatial coverage range is defined as a rectangular area extending along the propagation direction, whose width is twice the wavelength and whose length is the actual distance of the propagation path. This range is used to limit the actual impact area of ​​the fluctuation signal on the material structure.

[0095] Step 212: establishing a local coordinate system based on the spatial coverage, converting the three-dimensional coordinates of each micropore in the microporosity distribution data into the local coordinate system, and selecting target micropores whose three-dimensional coordinates are within the boundary of the spatial coverage;

[0096] In this step, the local coordinate system refers to a temporary coordinate system established along the propagation direction (X-axis), the perpendicular propagation direction (Y-axis), and the thickness direction (Z-axis) with the starting point of the internal fluctuation signal's propagation path as the origin. This coordinate system is used to map the coordinates of microscopic pores to the area of ​​influence of the internal fluctuation signal. Target microscopic pores are those located within the spatial coverage range whose coordinates satisfy the X, Y, and Z axis boundary conditions after transformation through the local coordinate system. These pores represent potential defect pores affected by the internal fluctuation signal.

[0097] In this embodiment of the present invention, a local coordinate system is established with the starting point of the spatial coverage range as the origin, the propagation direction as the length axis (X-axis), the perpendicular propagation direction as the width axis (Y-axis), and the plate thickness direction as the depth axis (Z-axis). The three-dimensional coordinates of each micropore in the microporosity distribution data are converted to this local coordinate system. Micropores with X-axis coordinates between 0 and the propagation path length, Y-axis coordinates between -wavelength and +wavelength, and Z-axis coordinates within the plate thickness range are screened and defined as target micropores. These pores are considered valid data points within the area actually affected by the fluctuation signal.

[0098] Step 213: Divide the spatial coverage area into a plurality of sub-intervals along the length direction, wherein each sub-interval is divided into a fixed number of strips along the width direction to generate a plurality of grid units;

[0099] In this step, the length dimension refers to the dimension extending along the propagation direction of the internal fluctuation signal (the X-axis). The width dimension refers to the dimension extending perpendicular to the propagation direction (the Y-axis), and the range is limited to ± wavelength. A fixed number of strips refers to evenly dividing each subinterval into a preset number of narrow strips (e.g., 5) along the width direction to generate regularly distributed grid cells. Grid cells are regular rectangular regions within the spatial coverage area, divided by subintervals and strips, and are used for refined statistics of local pore density.

[0100] In an embodiment of the present invention, the spatial coverage range is equally divided into multiple sub-intervals along the length direction (X-axis), and the length of each sub-interval is determined according to a preset resolution (such as 1 mm / segment); within each sub-interval, it is divided into a fixed number of strips (such as 5 strips per segment) along the width direction (Y-axis) to form evenly distributed grid units. Each grid unit is a rectangular area, and its size is determined by the sub-interval length, strip width and plate thickness.

[0101] Step 214: Accumulating the micro-porosity of the target micro-pores of each grid cell to calculate the micro-pore density parameter of each grid cell;

[0102] In this step, the microscopic pore density parameter refers to the cumulative value of the microscopic porosity within the unit volume grid cell, which is used to quantify the local pore density.

[0103] In this embodiment of the present invention, the porosity values ​​of the target microscopic pores within each grid cell are accumulated and divided by the volume of the grid cell to obtain the microscopic pore density parameter for that grid cell, which reflects the density of pores within the grid cell. For example, if the volume of a grid cell is 1 mm³ and the accumulated porosity is 15%, the density parameter is 0.15 / mm³.

[0104] Step 215: Summarize the microscopic pore density parameters of all grid cells and generate a regional mean of the microscopic porosity within the spatial coverage area in combination with the total number of the grid cells;

[0105] In this embodiment of the present invention, the microscopic pore density parameters of all grid cells are aggregated, added together, and divided by the total number of grid cells to generate a regional mean of the microscopic porosity within the spatial coverage area. For example, if the sum of the density parameters for 100 grid cells is 12.0, the regional mean is 0.12 / mm³.

[0106] The embodiments of the present invention solve the problems of ambiguous porosity analysis range and insufficient statistical accuracy in traditional methods; through the refined division of grid units, the calculation accuracy of regional means is improved, a reliable data basis is provided for dual-threshold screening, and the robustness of composite defect detection is enhanced.

[0107] The present invention provides a specific embodiment, step 204, generating geometric parameters of the structural abnormal region corresponding to the micro-porosity distribution data based on the boundary overlap range of the target spatial coverage range, the geometric parameters including the radial extension range and depth distribution range of the structural abnormal region, specifically comprising the following steps:

[0108] Step 221: establishing a global coordinate system based on the metal plate, mapping the coordinate range of the target space coverage range to the global coordinate system to generate a three-dimensional geometric intersection of all target space coverage ranges;

[0109] In this step, the global coordinate system refers to a three-dimensional spatial reference system established with the sheet metal's machining direction (X-axis), width direction (Y-axis), and thickness direction (Z-axis) as reference axes. This system is used to unify the spatial positions of all inspection data. The three-dimensional geometric intersection refers to the common overlapping area of ​​multiple target spatial coverage areas, extracted through three-dimensional Boolean operations, and represents the synergistic range of composite defects.

[0110] In this embodiment of the present invention, a global coordinate system is established with the sheet metal's machining direction as the X-axis, its width as the Y-axis, and its thickness as the Z-axis. The three-dimensional coordinates of each target spatial coverage area, including the starting and ending coordinates, are mapped to this coordinate system. A three-dimensional Boolean operation is then performed to calculate the common overlap area of ​​all target spatial coverage areas, generating a three-dimensional geometric intersection.

[0111] Step 222: determining a target closed area based on the three-dimensional geometric intersection, and defining an outer contour boundary of the target closed area as a boundary overlap range;

[0112] In this step, the target closed region refers to the spatially continuous and closed subregion of the 3D geometric intersection, obtained through connected domain analysis and used to eliminate discrete noise interference. The outer contour boundary refers to the outermost surface boundary of the target closed region, composed of polygonal patches, which defines the geometric outline of the defect area. The boundary overlap range refers to the 3D spatial range defined by the outer contour boundary and represents the actual physical size of the structural anomaly area.

[0113] In an embodiment of the present invention, a spatially continuous target closed area is screened out from the three-dimensional geometric intersection, where the target closed area is a three-dimensional spatial range with no internal holes and closed boundaries; the outer contour boundary of the target closed area is extracted, and the outer contour boundary is composed of multiple polygonal faces, which is defined as a boundary overlapping range, which is used to limit the core position of the structural abnormality area.

[0114] Step 223: Divide the boundary overlap range into a plurality of equally spaced plane analysis units along the machining direction of the metal sheet, calculate the coverage area ratio of the boundary overlap range within each plane analysis unit, and define the radial extension range as the difference between the first and last coordinates of the plane analysis unit in the machining direction whose coverage area ratio exceeds a preset plane threshold;

[0115] In this step, the processing direction refers to the forming movement direction of the metal sheet on the production line (X-axis), and the expansion of defects along this direction reflects the stability of the process. The plane analysis unit refers to the two-dimensional plane area (XY plane) divided at equal intervals along the processing direction, which is used to count the lateral coverage ratio of the boundary overlap range. The coverage area ratio refers to the ratio of the area occupied by the boundary overlap range within the plane analysis unit to the total area of ​​the unit. The area ratio is used to screen the significant defect areas. The preset plane threshold refers to the lower limit of the area ratio (such as 60%) for determining whether the plane analysis unit belongs to the defect extension area, and is dynamically adjusted according to the material type. The difference between the first and last coordinates in the processing direction refers to the difference between the starting end coordinate and the ending end coordinate of the continuous plane analysis unit in the processing direction, which is used to quantify the radial extension range.

[0116] In an embodiment of the present invention, the boundary overlap range is divided into a plurality of equally spaced planar analysis units (e.g., each segment is 2 mm). The ratio of the coverage area of ​​the boundary overlap range within each planar analysis unit to the total unit area is calculated. Continuous planar analysis units whose coverage area ratio exceeds a preset planar threshold (e.g., 60%) are screened out, and the difference between their starting coordinates (minimum X value) and ending coordinates (maximum X value) is defined as the radial extension range.

[0117] Step 224: Divide the boundary overlap range into a plurality of equally spaced thickness analysis layers along the thickness direction of the metal plate, calculate the coverage volume ratio of the boundary overlap range in each thickness analysis layer, and define the difference between the first and last coordinates in the thickness direction of the thickness analysis layer whose coverage volume ratio exceeds a preset thickness threshold as the depth distribution range;

[0118] In this step, the thickness direction refers to the vertical direction (Z-axis) of the metal sheet. The penetration of defects along this direction reflects the structural integrity of the material. The thickness analysis layer refers to a three-dimensional hierarchical area (Z-axis interval) divided at equal intervals along the thickness direction, which is used to calculate the longitudinal coverage ratio of the boundary overlap range. The coverage volume ratio refers to the ratio of the volume occupied by the boundary overlap range within the thickness analysis layer to the total volume of the layer. The volume ratio is used to screen significant penetration areas. The preset thickness threshold refers to the lower limit of the volume ratio (such as 50%) for determining whether the thickness analysis layer belongs to the defect penetration area, and is dynamically adjusted according to the thickness of the plate. The difference between the first and last coordinates in the thickness direction refers to the difference between the starting and ending coordinates of the continuous plane analysis unit in the thickness direction, which is used to quantify the depth distribution range.

[0119] In an embodiment of the present invention, the boundary overlap range is divided into multiple equally spaced thickness analysis layers (e.g., 0.5 mm per layer) along the thickness direction (Z-axis), and the proportion of the coverage volume of the boundary overlap range in each thickness analysis layer to the total volume of the layer is counted. Continuous thickness analysis layers with a coverage volume ratio exceeding a preset thickness threshold (e.g., 50%) are screened out, and the difference between their upper end coordinates (maximum Z value) and lower end coordinates (minimum Z value) is defined as the depth distribution range.

[0120] Step 225: generating geometric parameters of the structural abnormality region according to the radial extension range and the depth distribution range;

[0121] In an embodiment of the present invention, geometric parameters of the structural abnormality area are generated based on the numerical values ​​of the radial extension range and the depth distribution range, combined with preset defect level classification rules, such as extension range > 10mm is high risk.

[0122] The embodiment of the present invention accurately extracts the boundary range of composite defects from the multi-target coverage area through three-dimensional geometric intersection and multi-layer threshold screening, solving the problems of vague defect area definition and strong dependence on manual experience in traditional methods; geometric parameters improve the standardization and operability of defect assessment, providing a direct basis for process optimization.

[0123] The present invention provides a specific embodiment, step 104, inputting the micro porosity distribution data, the geometric parameters of the structural abnormal region, and the current density distribution data into an artificial intelligence model to associate the micro porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate, specifically comprising the following steps:

[0124] Step 401: converting the geometric parameters of the structural abnormality region into spatial constraints, wherein the spatial constraints include the boundary coordinates of the radial extension range and the depth distribution range of the structural abnormality region in the global coordinate system;

[0125] In this step, spatial constraints refer to the three-dimensional coordinate boundary range (X1≤X≤X2, Z1≤Z≤Z2) converted from the geometric parameters of the structural anomaly region, which is used to define the core area for defect analysis. Boundary coordinates refer to the starting and ending coordinates of the structural anomaly region in the global coordinate system, including X1 and X2 in the machining direction (X-axis) and Z1 and Z2 in the thickness direction (Z-axis).

[0126] In an embodiment of the present invention, the geometric parameters of the structural anomaly area are converted into boundary coordinates in the global coordinate system, wherein the radial extension range has a starting coordinate X1 and an ending coordinate X2 along the processing direction (X-axis), and the depth distribution range has a starting coordinate Z1 and an ending coordinate Z2 along the thickness direction (Z-axis), thereby generating a boundary coordinate range of the spatial constraint condition: X1≤X≤X2, Y is arbitrary, Z1≤Z≤Z2.

[0127] Step 402: selecting a plurality of target micropores whose spatial locations are within the boundaries of the spatial constraints from the microporosity distribution data, and selecting a plurality of target current density measurement points whose spatial locations are within the boundaries of the spatial constraints from the current density distribution data;

[0128] In an embodiment of the present invention, all microscopic pores whose three-dimensional coordinates satisfy X1≤X≤X2 and Z1≤Z≤Z2 are selected from the microscopic porosity distribution data and defined as target microscopic pores; and all current density values ​​whose coordinates of measurement points satisfy the same conditions are selected from the current density distribution data and defined as target current density measurement points. The selection process is implemented by coordinate range comparison to ensure that only valid data within the structural abnormality area is processed.

[0129] Step 403: performing proximity matching on the spatial positions of the target microscopic pores and the target current density measurement points to determine the target current density measurement point corresponding to each target microscopic pore, and generating a plurality of matching pairs;

[0130] In this step, the matching pair refers to the one-to-one correspondence between the microscopic pores and the current density measurement points associated by the spatial proximity principle, including the microscopic porosity and current density values.

[0131] In an embodiment of the present invention, the Euclidean distance between the coordinates of each target microscopic pore and the coordinates of all target current density measurement points is calculated, and the current density measurement point with the closest distance is selected to form a matching pair with the pore; if the minimum distance is less than a preset threshold (such as 0.1 mm), the match is successful; otherwise, the pore is discarded, and finally multiple matching pairs are generated, each matching pair containing a microscopic porosity value and a current density value.

[0132] Step 404: Using an artificial intelligence model, a nonlinear correlation calculation is performed on the microscopic porosity and current density values ​​of each matching pair to generate a coupling strength value for each matching pair, thereby constructing a pore current coupling characteristic matrix of the metal plate.

[0133] In this step, the current density value refers to the current intensity per unit area obtained through electrochemical measurements, reflecting the local electrochemical corrosion activity of the defect area, and the unit is A / m². The coupling strength value refers to the nonlinear correlation strength between microporosity and current density, calculated by the artificial intelligence model, and is used to quantify the defect coupling effect. It is dimensionless.

[0134] In an embodiment of the present invention, the micro-porosity and current density values ​​of each matching pair are input into an artificial intelligence model, which maps the micro-porosity of each matching pair into a first eigenvector and the current density value into a second eigenvector, and generates an initial interactive eigenvector by vector product; the initial interactive eigenvectors are weightedly superimposed based on a dynamic weight parameter to obtain a superimposed eigenvector, wherein the dynamic weight parameter is adjusted according to the relative distance between the spatial position of the matching pair and the geometric parameters of the structural anomaly area; the superimposed eigenvector is converted into a scalar value as a coupling strength value; the global coordinate system is divided into spatial analysis areas according to the geometric parameters of the structural anomaly area, the coupling strength values ​​in each area are arithmetic averaged, and finally the coupling strength average values ​​of all spatial analysis areas are arranged to obtain a pore current coupling characteristic matrix.

[0135] The embodiments of the present invention solve the problems of data dimensionality fragmentation and large noise interference in the existing technology; through the dynamic nonlinear calculation of the artificial intelligence model, a coupling feature matrix of multi-source data collaboration is constructed to improve the comprehensiveness of defect characterization and evaluation accuracy, providing a reliable data basis for quantitative analysis.

[0136] The present invention provides a specific embodiment, step 404, using an artificial intelligence model to perform a nonlinear correlation calculation on the microscopic porosity and current density values ​​of each matching pair to generate a coupling strength value for each matching pair to construct a pore current coupling characteristic matrix of the metal plate, specifically comprising the following steps:

[0137] Step 411: using the artificial intelligence model, mapping the microscopic porosity of each matching pair into a first eigenvector, and mapping the current density value of each matching pair into a second eigenvector;

[0138] In this step, the first eigenvector is the vector representation of the microporosity value mapped to a high-dimensional feature space by the artificial intelligence model, which is used to capture the nonlinear effect of microporosity on defects. The second eigenvector is the vector representation of the current density value mapped to a high-dimensional feature space by the artificial intelligence model, which is used to characterize the correlation between the current density value and corrosion activity.

[0139] In an embodiment of the present invention, the microscopic porosity of each matching pair is input into the feature mapping module of the artificial intelligence model, and the porosity value is converted into a first eigenvector of dimension N through a fully connected layer; at the same time, the current density value of the matching pair is input into another fully connected layer and converted into a second eigenvector of dimension N.

[0140] Step 412: performing a product operation on the first feature vector and the second feature vector to generate an initial interaction feature vector for each matching pair;

[0141] In this step, the initial interaction eigenvector refers to a vector generated by element-by-element multiplication of the first eigenvector and the second eigenvector, reflecting the local interaction characteristics of the microscopic porosity and current density values.

[0142] In this embodiment of the present invention, the first eigenvector and the second eigenvector of each matching pair are element-wise multiplied, that is, the elements at corresponding positions are multiplied to generate an initial interaction eigenvector of the same dimension. For example, the first eigenvector is [a1, a2, ..., a n ], the second eigenvector is [b1,b2,…,b n ], then the initial interaction feature vector is [a1×b1,a2×b2,…,a n ×b n ].

[0143] Step 413: Based on the dynamic weight parameters of the artificial intelligence model, weighted superposition is performed on the initial interaction feature vectors of each matching pair to obtain a superimposed feature vector of each matching pair, wherein the dynamic weight parameters are dynamically adjusted according to the relative distance between the spatial position of each matching pair and the geometric parameters of the structural abnormality region;

[0144] In this step, the dynamic weight parameter refers to a weighting coefficient that is dynamically adjusted based on the distance between the matching pair and the geometric center of the structural anomaly. The closer the distance, the greater the weight, which is used to enhance the contribution of the core defect area. The superimposed feature vector is a one-dimensional vector generated by superimposing the initial interaction feature vectors with the dynamic weight. It is used for dimensionality reduction and feature fusion.

[0145] In the embodiment of the present invention, the weight parameter is dynamically adjusted according to the Euclidean distance between the spatial position of the matching pair and the geometric center of the structural abnormality area: the smaller the distance, the larger the weight parameter. Each element of the initial interaction feature vector is multiplied by the corresponding dynamic weight parameter, and then all weighted elements are added together to obtain the superimposed feature vector. For example, the initial interaction feature vector is [c1, c2, ..., c n ], the weight parameters are [w1,w2,…,w n ], then the superimposed eigenvector is Σ(c i ×w i ).

[0146] Step 414: converting the superimposed eigenvector of each matching pair into a corresponding scalar value, where the scalar value is defined as the coupling strength value of the corresponding matching pair;

[0147] In this step, the scalar value refers to a single numerical value generated by linear transformation of the superimposed eigenvectors, which is defined as the coupling strength value of the matching pair and quantifies the correlation strength between the microporosity and the current density value.

[0148] In the embodiment of the present invention, the superimposed feature vector is input into the output layer of the artificial intelligence model and compressed into a single scalar value through linear transformation, which is defined as the coupling strength value of the matching pair. For example, the superimposed feature vector is [d1, d2, ..., d n ], the output layer weights are [k1,k2,…,k n ], then the scalar value is Σ(d i ×k i ).

[0149] Step 415: Divide the global coordinate system into multiple spatial analysis regions according to the radial extension range and depth distribution range of the structural anomaly region;

[0150] In this step, the spatial analysis area refers to the three-dimensional space unit divided according to the geometric parameters of the structural abnormality area, which is used to aggregate the local statistical characteristics of the coupling strength value.

[0151] In an embodiment of the present invention, according to the radial extension range and depth distribution range of the structural abnormality area, the global coordinate system is divided into equal length intervals along the processing direction, such as 2 mm per segment, and divided into equal thickness intervals along the thickness direction, such as 0.5 mm per layer, to form multiple rectangular spatial analysis areas.

[0152] Step 416: performing arithmetic mean calculation on the coupling strength values ​​of all matching pairs in each spatial analysis region to obtain the coupling strength mean value of each spatial analysis region;

[0153] In this step, the coupling strength mean refers to the arithmetic mean of the coupling strength values ​​of all matching pairs in the spatial analysis area, reflecting the overall defect coupling level in the area.

[0154] In this embodiment of the present invention, the coupling strength values ​​of all matching pairs within each spatial analysis region are calculated as the arithmetic average. For example, if a spatial analysis region contains five matching pairs with coupling strength values ​​of 0.3, 0.5, 0.4, 0.6, and 0.2, the average coupling strength is (0.3 + 0.5 + 0.4 + 0.6 + 0.2) / 5 = 0.4.

[0155] Step 417: Arrange the coupling strength means of all spatial analysis regions to construct a pore current coupling characteristic matrix;

[0156] In this embodiment, the coupling strength averages for each spatial analysis region are arranged according to their spatial positions in the global coordinate system to form a pore current coupling feature matrix. For example, if the machining direction is divided into 10 segments and the thickness direction is divided into 5 layers, a 10×5 pore current coupling feature matrix is ​​constructed, with full coverage in the width direction by default.

[0157] The embodiment of the present invention strengthens the feature contribution of the core defect area through a dynamic weight mechanism, combines the mean statistics of the spatial analysis area, suppresses noise interference, and improves the robustness of the feature matrix; through nonlinear interaction and spatial aggregation, it realizes efficient collaboration of multi-source data, breaks through the expressive power limitations of traditional linear models, and improves the accuracy and reliability of defect quantitative assessment.

[0158] The present invention provides a specific embodiment, step 105, generating a quantitative evaluation result of the internal pitting defect degree of the metal plate according to the pore current coupling characteristic matrix, specifically comprising the following steps:

[0159] Step 501: Calculating the coupling strength differences between adjacent matrix elements in the pore current coupling characteristic matrix along the machining direction, width direction, and thickness direction of the global coordinate system to generate a machining direction gradient value, a width direction gradient value, and a thickness direction gradient value;

[0160] In this step, the coupling strength difference refers to the numerical difference in coupling strength between adjacent matrix elements. It is calculated by subtracting the previous element from the next element and is used to quantify the local rate of change of the defect characteristic. The machining direction gradient value, width direction gradient value, and thickness direction gradient value refer to the difference in coupling strength between adjacent matrix elements calculated along the X, Y, and Z axes, respectively, reflecting the propagation or penetration rate of the defect in the corresponding direction.

[0161] In an embodiment of the present invention, the coupling strength differences between adjacent matrix elements in the pore current coupling characteristic matrix are calculated along the machining direction (X-axis), width direction (Y-axis), and thickness direction (Z-axis) of the global coordinate system. For example, for matrix element [i][j][k], its machining direction gradient value is the coupling strength of [i+1][j][k] minus the coupling strength of [i][j][k]; its width direction gradient value is the coupling strength of [i][j+1][k] minus the coupling strength of [i][j][k]; and its thickness direction gradient value is the coupling strength of [i][j][k+1] minus the coupling strength of [i][j][k].

[0162] Step 502: Selecting target matrix elements from the pore current coupling characteristic matrix, whose processing direction gradient value exceeds a preset processing gradient threshold, whose width direction gradient value exceeds a preset width gradient threshold, and whose thickness direction gradient value exceeds a preset thickness gradient threshold, and using the target matrix elements as gradient abnormality matrix elements;

[0163] In this step, the preset machining gradient threshold, preset width gradient threshold, and preset thickness gradient threshold refer to critical gradient values ​​set based on material type and process requirements, used to determine the significance of defect changes. The gradient anomaly matrix element refers to the target matrix element where the gradient value in all three directions exceeds the threshold, representing the core area of ​​​​drastic defect changes.

[0164] In an embodiment of the present invention, the processing direction gradient value, the width direction gradient value and the thickness direction gradient value are compared with the preset processing gradient threshold, width gradient threshold and thickness gradient threshold respectively, and the target matrix elements that simultaneously meet the requirements that the gradient values ​​of the three directions exceed the threshold are screened out and defined as gradient abnormality matrix elements. The above three thresholds are dynamically set according to the material type and process standards. For example, the processing gradient threshold of carbon steel is 0.3 / mm, and that of aluminum alloy is 0.2 / mm.

[0165] Step 503: determining a target aggregation region according to the spatial distribution density of the gradient anomaly matrix elements in the local coordinate system, wherein the target aggregation region is a continuous spatial region where the number of gradient anomaly matrix elements exceeds a preset density threshold;

[0166] In this step, spatial distribution density refers to the number of gradient anomaly matrix elements per unit volume. It is calculated by counting the total number of gradient anomaly matrix elements within the target cluster region and dividing it by the region's volume. The target cluster region is defined as a region where the spatial distribution density exceeds a preset density threshold and the elements are continuous, representing a high-risk location for defect expansion or penetration. The preset density threshold is the lower limit for determining the target cluster region. For example, 5 / mm³ indicates that at least 5 anomaly elements per cubic millimeter are required to be considered clustered.

[0167] In an embodiment of the present invention, the spatial distribution density of the gradient anomaly matrix elements in the local coordinate system is statistically analyzed, that is, the number of abnormal elements within a unit volume (such as 1 mm³), and a set of abnormal elements with a density exceeding a preset density threshold (such as 5 / mm³) and spatial continuity is screened out and defined as a target aggregation area. The area must meet the requirement that all adjacent elements are gradient anomaly matrix elements.

[0168] Step 504: generating quantitative evaluation parameters of the internal pitting corrosion defect degree of the metal plate based on the machining direction span, width direction span, and thickness direction span of the target concentration area, wherein the quantitative evaluation parameters include defect propagation rate and defect penetration depth level;

[0169] In this step, the process direction span, width direction span, and thickness direction span refer to the maximum coordinate difference of the target cluster area in the corresponding directions, respectively reflecting the defect's expansion length, lateral diffusion width, and penetration depth. Quantitative assessment parameters include the defect expansion rate (i.e., the expansion length per unit time) and the penetration depth level (i.e., the ratio of penetration depth to the plate thickness) to objectively quantify the degree of defect damage. The defect expansion rate is calculated by dividing the process direction span by a preset time parameter and reflects the dynamic expansion speed of the defect along the production line. The defect penetration depth level is calculated by dividing the thickness direction span by the total plate thickness and is expressed as a percentage or graded, such as 1-5, indicating the degree of longitudinal penetration of the defect.

[0170] In this embodiment of the present invention, the maximum span of the target concentration area along the machining direction, the maximum span along the width direction, and the maximum span along the thickness direction are calculated. The machining direction span is divided by a preset time parameter, such as the inspection cycle corresponding to the sheet forming speed, to generate the defect growth rate. The thickness direction span is divided by the total sheet thickness to generate the defect penetration depth level. For example, a machining direction span of 20mm / 24h = 0.83mm / h, and a thickness direction span of 1.5mm / 5mm = 0.3, which is level 3.

[0171] Step 505: matching the quantitative evaluation parameters with a preset defect level mapping table to generate a quantitative evaluation result of the internal pitting defect degree of the metal plate;

[0172] In this step, a preset defect level mapping table is a classification rule table formulated according to the defect growth rate and the defect penetration depth level. For example, a defect growth rate greater than 0.5 mm / h and a defect penetration level greater than level 3 are defined as high risk.

[0173] In this embodiment of the present invention, a preset defect level mapping table is queried based on the defect growth rate and penetration depth level. This mapping table is developed based on historical process data and defect failure cases. For example, a defect growth rate ≥ 0.5 mm / h and a level ≥ 3 is defined as high risk, and a quantitative assessment result is output.

[0174] The embodiments of the present invention accurately capture the dynamic change characteristics of defects through multi-directional gradient analysis, and combine spatial density screening with quantitative parameter mapping to solve the problems of single defect assessment dimension and strong subjectivity in traditional methods; through dynamic threshold setting and objective parameter calculation, the accuracy and consistency of defect classification are improved, providing direct and reliable data support for process optimization, and effectively reducing the risks of missed detection and misjudgment.

[0175] Figure 2 The present invention provides a schematic diagram of the structure of a metal pitting defect degree assessment system based on artificial intelligence, as shown in FIG. Figure 2 As shown, the system includes:

[0176] A first acquisition module 21 is used to acquire microscopic porosity distribution data of the metal sheet during the forming process, and internal fluctuation signals of the metal sheet under the action of directional stress waves;

[0177] A first generating module 22 is configured to generate geometric parameters of a structural abnormality region corresponding to the microscopic porosity distribution data based on a propagation attenuation rate of the internal fluctuation signal;

[0178] A second acquisition module 23 is used to acquire current density distribution data of the structural abnormality area;

[0179] a correlation module 24 for inputting the micro-porosity distribution data, the geometric parameters of the structural abnormality region, and the current density distribution data into an artificial intelligence model to correlate the micro-porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate;

[0180] The second generating module 25 is configured to generate a quantitative evaluation result of the internal pitting defect degree of the metal plate according to the pore current coupling characteristic matrix.

[0181] Figure 2 The metal pitting defect degree assessment system based on artificial intelligence can be performed Figure 1 The implementation principle and technical effects of the artificial intelligence-based metal pitting defect assessment method described in the illustrated embodiment will not be elaborated on here. The specific manner in which each module and unit performs operations in the artificial intelligence-based metal pitting defect assessment system in the above embodiment has been described in detail in the embodiments of the method and will not be elaborated on here.

[0182] In one possible design, Figure 2 The metal pitting defect degree assessment system based on artificial intelligence of the embodiment shown can be implemented as a computing device, such as Figure 3As shown, the computing device may include a storage component 31 and a processing component 32;

[0183] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0184] The processing component 32 is used to obtain microscopic porosity distribution data of the metal sheet during the forming process, and internal fluctuation signals of the metal sheet under the action of directional stress waves.

[0185] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0186] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0187] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0188] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0189] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0190] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0191] The embodiment of the present invention further provides a computer storage medium storing a computer program, which can achieve the above-mentioned Figure 1 The embodiment shown is an artificial intelligence-based method for evaluating the degree of metal pitting defects.

[0192] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0193] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0194] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0195] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for evaluating the degree of metal pitting defects based on artificial intelligence, characterized in that: include: Acquiring microscopic porosity distribution data of the metal sheet during the forming process and internal fluctuation signals of the metal sheet under the action of directional stress waves; generating geometric parameters of a structural abnormality region corresponding to the microscopic porosity distribution data based on a propagation attenuation rate of the internal fluctuation signal; Acquiring current density distribution data of the structural abnormal area; Inputting the micro porosity distribution data, the geometric parameters of the structural abnormal region, and the current density distribution data into an artificial intelligence model to correlate the micro porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate; A quantitative evaluation result of the internal pitting corrosion defect degree of the metal plate is generated according to the pore current coupling characteristic matrix.

2. The method according to claim 1, characterized in that Generating geometric parameters of the structural abnormality region corresponding to the microscopic porosity distribution data based on the propagation attenuation rate of the internal fluctuation signal includes: Acquiring a starting end energy value and an ending end energy value of the internal fluctuation signal on a propagation path, and calculating a propagation attenuation rate of the internal fluctuation signal based on the starting end energy value and the ending end energy value; determining a spatial coverage of the propagation path in the metal sheet, and mapping the spatial coverage with spatial positions of micropores in the microporosity distribution data to calculate a regional mean of the microporosity within the spatial coverage; Selecting, from the spatial coverage range, a plurality of target spatial coverage ranges whose area averages exceed a preset density threshold and whose propagation attenuation rates exceed a preset attenuation threshold; Based on the boundary overlap range of the target spatial coverage range, geometric parameters of the structural abnormality region corresponding to the microscopic porosity distribution data are generated, and the geometric parameters include a radial extension range and a depth distribution range of the structural abnormality region.

3. The method according to claim 2, characterized in that Determining a spatial coverage of the propagation path in the metal sheet, and mapping the spatial coverage with spatial positions of micropores in the microporosity distribution data to calculate a regional mean of the microporosity within the spatial coverage, including: defining a spatial coverage range of the propagation path in the metal plate according to the wavelength and propagation direction of the internal fluctuation signal; Based on the spatial coverage, a local coordinate system is established, the three-dimensional coordinates of each micropore in the microporosity distribution data are converted into the local coordinate system, and target micropores whose three-dimensional coordinates are within the boundary of the spatial coverage are selected; Divide the spatial coverage area into a plurality of subintervals along the length direction, wherein each subinterval is divided into a fixed number of strips along the width direction to generate a plurality of grid units; The micro-porosity of the target micro-pores of each grid cell is accumulated to calculate the micro-pore density parameter of each grid cell; The microscopic pore density parameters of all grid cells are summed up and combined with the total number of the grid cells to generate a regional mean of the microscopic porosity within the spatial coverage area.

4. The method according to claim 2, characterized in that Based on the boundary overlap range of the target spatial coverage range, geometric parameters of the structural abnormality region corresponding to the microscopic porosity distribution data are generated, wherein the geometric parameters include a radial extension range and a depth distribution range of the structural abnormality region, including: Establishing a global coordinate system based on the metal plate, mapping the coordinate range of the target space coverage range into the global coordinate system to generate a three-dimensional geometric intersection of all target space coverage ranges; Determine a target closed area according to the three-dimensional geometric intersection, and define an outer contour boundary of the target closed area as a boundary overlap range; Dividing the boundary overlap range into a plurality of equally spaced plane analysis units along the processing direction of the metal sheet, calculating the coverage area ratio of the boundary overlap range within each plane analysis unit, and defining the radial extension range as the difference between the first and last coordinates of the plane analysis unit in the processing direction whose coverage area ratio exceeds a preset plane threshold; Dividing the boundary overlap range into a plurality of equally spaced thickness analysis layers along the thickness direction of the metal plate, calculating the coverage volume ratio of the boundary overlap range within each thickness analysis layer, and defining the difference between the first and last coordinates of the thickness analysis layer in the thickness direction whose coverage volume ratio exceeds a preset thickness threshold as the depth distribution range; Geometric parameters of the structural abnormality region are generated according to the radial extension range and the depth distribution range.

5. The method according to claim 1, wherein Inputting the micro porosity distribution data, the geometric parameters of the structural abnormality region, and the current density distribution data into an artificial intelligence model to correlate the micro porosity distribution data with the current density distribution data to generate a pore current coupling characteristic matrix of the metal plate, including: Converting the geometric parameters of the structural abnormality region into spatial constraints, wherein the spatial constraints include boundary coordinates of a radial extension range and a depth distribution range of the structural abnormality region in a global coordinate system; Selecting a plurality of target micropores whose spatial positions are within the boundaries of the spatial constraints from the microporosity distribution data, and selecting a plurality of target current density measurement points whose spatial positions are within the boundaries of the spatial constraints from the current density distribution data; Performing proximity matching on the spatial positions of the target microscopic pores and the target current density measurement points to determine the target current density measurement point corresponding to each target microscopic pore, and generating a plurality of matching pairs; Using an artificial intelligence model, a nonlinear correlation calculation is performed on the microscopic porosity and current density values ​​of each matching pair to generate a coupling strength value for each matching pair to construct a pore current coupling characteristic matrix of the metal sheet.

6. The method according to claim 5, characterized in that Using an artificial intelligence model, a nonlinear correlation calculation is performed on the microscopic porosity and current density values ​​of each matching pair to generate a coupling strength value for each matching pair to construct a pore current coupling characteristic matrix of the metal sheet, including: Using the artificial intelligence model, mapping the microscopic porosity of each matching pair into a first eigenvector, and mapping the current density value of each matching pair into a second eigenvector; Performing a product operation on the first feature vector and the second feature vector to generate an initial interaction feature vector for each matching pair; Based on the dynamic weight parameters of the artificial intelligence model, the initial interaction feature vectors of each matching pair are weighted superimposed to obtain the superimposed feature vector of each matching pair, and the dynamic weight parameters are dynamically adjusted according to the relative distance between the spatial position of each matching pair and the geometric parameters of the structural abnormality area; Converting the superimposed eigenvector of each matching pair into a corresponding scalar value, where the scalar value is defined as the coupling strength value of the corresponding matching pair; According to the radial extension and depth distribution range of the structural anomaly area, the global coordinate system is divided into multiple spatial analysis areas; The coupling strength values ​​of all matching pairs in each spatial analysis area were calculated by arithmetic averaging to obtain the mean coupling strength value of each spatial analysis area; The coupling strength means of all spatial analysis areas are arranged to construct the pore current coupling characteristic matrix.

7. The method according to claim 1, characterized in that Generating a quantitative evaluation result of the degree of internal pitting defects of the metal plate according to the pore current coupling characteristic matrix, including: Calculating coupling strength differences between adjacent matrix elements in the pore current coupling characteristic matrix along the machining direction, width direction, and thickness direction of the global coordinate system to generate a machining direction gradient value, a width direction gradient value, and a thickness direction gradient value; Selecting a target matrix element from the pore current coupling characteristic matrix, whose processing direction gradient value exceeds a preset processing gradient threshold, whose width direction gradient value exceeds a preset width gradient threshold, and whose thickness direction gradient value exceeds a preset thickness gradient threshold, and using the target matrix element as a gradient abnormality matrix element; Determining a target aggregation region according to the spatial distribution density of the gradient anomaly matrix elements in the local coordinate system, wherein the target aggregation region is a continuous spatial region where the number of gradient anomaly matrix elements exceeds a preset density threshold; Generating quantitative evaluation parameters of the internal pitting corrosion defect degree of the metal plate based on the machining direction span, the width direction span, and the thickness direction span of the target concentration area, wherein the quantitative evaluation parameters include the defect propagation rate and the defect penetration depth level; The quantitative evaluation parameters are matched with a preset defect level mapping table to generate a quantitative evaluation result of the internal pitting defect degree of the metal plate.

8. A metal pitting defect degree assessment system based on artificial intelligence, characterized in that: include: A first acquisition module is used to acquire microscopic porosity distribution data of the metal sheet during the forming process, and an internal fluctuation signal of the metal sheet under the action of a directional stress wave; A first generating module is configured to generate geometric parameters of a structural abnormality region corresponding to the microscopic porosity distribution data based on a propagation attenuation rate of the internal fluctuation signal; A second acquisition module is used to acquire current density distribution data of the structural abnormality area; a correlation module, configured to input the micro-porosity distribution data, the geometric parameters of the structural abnormality region, and the current density distribution data into an artificial intelligence model, so as to correlate the micro-porosity distribution data with the current density distribution data and generate a pore current coupling characteristic matrix of the metal plate; The second generating module is used to generate a quantitative evaluation result of the internal pitting defect degree of the metal plate according to the pore current coupling characteristic matrix.

9. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an artificial intelligence-based metal pitting defect degree assessment method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, an artificial intelligence-based metal pitting defect degree assessment method as described in any one of claims 1 to 7 is implemented.

Images