Method, system, equipment and medium for detecting and judging defects of aluminum pipe by fusing eddy current phase and three-dimensional morphology

Abstract

The invention relates to an eddy current phase and three-dimensional morphology fused aluminum pipe defect detection and discrimination method, system and device and a medium, and the method comprises the steps: obtaining electromagnetic response signals of different depths of an aluminum pipe, and extracting eddy current phase distribution characteristics; matching the eddy current phase distribution characteristics with a preset defect type association rule to generate a phase classification map; extracting time-frequency response characteristics in the eddy current phase distribution characteristics, and constructing a geometric characteristic spectrum in combination with eddy current impedance change characteristics; converting the coordinates of the phase classification atlas into coordinates under the same reference as the geometric feature atlas; and generating a composite tensor by calculating phase classification probability distribution extracted from the phase classification map and a covariance matrix of a morphology parameter set extracted from the geometric feature map, and calling a defect morphology association criterion library to execute judgment so as to output a defect type. Therefore, according to the method, the objectivity and the physical interpretability of defect detection are effectively improved through multi-dimensional feature fusion and a space mapping mechanism.

Description

Technical Field

[0001] This invention relates to the field of defect detection technology, and in particular to a method, system, device and medium for detecting and identifying defects in aluminum tubes by integrating eddy current phase and three-dimensional morphology. Background Technology

[0002] In the field of defect detection for aluminum and aluminum alloy pipes, eddy current testing technology is widely used due to its non-contact and high-efficiency characteristics. Existing conventional testing methods mainly include two implementation methods: one is manual detection, where operators use a handheld eddy current probe to scan the pipe surface segment by segment, relying on changes in the amplitude or phase angle of the observed impedance plane signal to determine defects. Because this method is highly dependent on the operator's experience, visual fatigue during continuous testing can easily lead to missed defects, and its ability to distinguish between surface micro-cracks and internal inclusions is limited. The second method is machine vision detection based on neural network models, which trains deep learning networks to identify the time-frequency characteristics of eddy current signals to achieve automatic defect classification. However, the decision-making process of neural network models lacks interpretability, resulting in a "black box" effect in practical applications. Engineers find it difficult to verify the physical rationality of the defect judgment logic, thus affecting the credibility of on-site implementation. Furthermore, this type of method requires large-scale labeled defect samples for model training, and when faced with complex defect morphologies such as pipe bends or irregular joints, insufficient training data coverage can easily lead to misjudgments.

[0003] Therefore, manual detection suffers from insufficient stability due to its strong subjectivity, while automatic detection methods based on neural networks are limited by data dependence and the black box effect of the model. Ultimately, neither of them can meet the dual requirements of stability, reliability and ease of implementation in industrial scenarios. Summary of the Invention

[0004] (a) Technical problems to be solved In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method, system, device and medium for detecting and judging aluminum tube defects by integrating eddy current phase and three-dimensional morphology. It solves the problems of strong subjectivity and poor result stability of traditional manual detection, while automatic detection methods based on neural networks are limited by data dependence and model black box effect. Both are difficult to achieve stable, reliable and easy-to-implement aluminum tube defect judgment.

[0005] (II) Technical Solution To achieve the above objectives, the main technical solutions adopted by the present invention include: In a first aspect, embodiments of the present invention provide a method for detecting and identifying defects in aluminum tubes that integrates eddy current phase and three-dimensional morphology, comprising: Eddy current phase distribution characteristics are extracted from the electromagnetic response signals at different depths of the aluminum tube obtained by layering through an eddy current probe. The eddy current phase distribution characteristics are matched with preset defect type association rules to generate a phase classification map that uses color coding to differentiate defect types. Extract the time-frequency response features related to the geometric dimension of the defect from the eddy current phase distribution features, and combine them with the eddy current impedance variation characteristics to construct a geometric feature map characterizing the physical morphology of the defect; Based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and the three-dimensional space, the planar coordinates of the phase classification map are converted into three-dimensional coordinates under the same reference as the geometric feature map by using the obtained probe moving speed and dwell time parameters. Under the same benchmark, a composite tensor is generated by calculating the covariance matrix of the defect type probability extracted from the phase classification map and the morphology parameter set extracted from the geometric feature map, and then a preset defect morphology association criterion library is called to perform the judgment to output the defect type.

[0006] Optionally, the eddy current phase distribution characteristics can be extracted from the electromagnetic response signals at different depths of the aluminum tube obtained layer by layer through the eddy current probe, including: The eddy current probe is controlled to move in a stepped manner along the axial direction of the aluminum tube. After each fixed axial distance is moved, a circumferential rotation scan is performed to obtain the electromagnetic response signal of each layer section. Phase-sensitive demodulation is performed on the electromagnetic response signals of each layered section to separate the real and imaginary parts of the impedance at each detection point; For the demodulated imaginary component, a sliding window truncation and bandpass filtering are performed to obtain the filtered imaginary component; Based on the ratio of the filtered imaginary component to the corresponding real component, the arctangent angle is taken to obtain the phase angle sequence of the corresponding detection point. Combined with the standard deviation and skewness coefficient of the phase angle distribution in the corresponding layered cross section obtained by statistics, the eddy current phase distribution characteristics are generated.

[0007] Optionally, the eddy current phase distribution characteristics are matched with preset defect type association rules to generate a phase classification map that uses color coding to differentiate defect types, including: Call the preset defect type association rule library, which is established by statistically analyzing historical defect samples. It includes the phase distribution standard deviation threshold range, skewness coefficient polarity conditions, and a mapping table between defect type and color for different defect types. The standard deviation of the eddy current phase distribution characteristics of the detection points in each layered section is matched with the threshold range of the standard deviation of the phase distribution in the defect type association rule base, and the polarity of the skewness coefficient is simultaneously checked to see if it meets the polarity condition of the skewness coefficient of the corresponding defect type. When any detection point simultaneously satisfies the condition that the membership degree of the corresponding standard deviation interval is greater than or equal to the preset membership degree threshold and the polarity condition of the skewness coefficient is met, the characteristic color value of the corresponding point is assigned according to the mapping table between defect type and color. Using the circumferential unfolding direction of the aluminum tube as the horizontal axis and the axial layer position as the vertical axis, the points with assigned colors in each layer section are mapped in a plane to generate a phase classification map.

[0008] Optionally, the time-frequency response features related to the defect's geometric dimension are extracted from the eddy current phase distribution characteristics, and combined with the eddy current impedance variation characteristics, a geometric feature map characterizing the physical morphology of the defect is constructed, including: The phase angle sequence in the eddy current phase distribution characteristics is transformed by time and frequency to extract the frequency domain energy attenuation slope and phase change time width as time and frequency response characteristics; Based on the separated real and imaginary components of the impedance, the rate of change of the real amplitude gradient and the cumulative phase lag of the imaginary component at each detection point are calculated to construct the eddy current impedance characteristic parameters. The time-frequency response characteristics and eddy current impedance characteristic parameters are fused in multiple dimensions according to spatial location to generate morphological parameters including defect depth, length and width. A three-dimensional space is established, and the morphological parameters of each detection point are projected onto the corresponding spatial coordinate nodes in the three-dimensional space to form a geometric feature map that characterizes the physical morphology of the defect.

[0009] Optionally, the frequency domain energy attenuation slope is calculated by performing a short-time Fourier transform on the phase angle sequence and extracting the attenuation slope of the high-frequency components in the energy spectrum that are greater than a preset high-frequency threshold. The phase transition time width is obtained by tracing the phase transition interval from the phase angle sequence using the zero-crossing detection method and calculating the duration span of the phase transition interval; The rate of change of the real amplitude gradient is calculated by performing a first-order difference calculation on the amplitudes of adjacent points along the probe movement direction based on the amplitude of the real component of the impedance, and taking the maximum rate of change of the absolute value of the first-order difference. The cumulative phase lag is obtained by weighting the skewness coefficient and the absolute value of the phase difference between the spatial neighborhood of the corresponding detection point; The defect depth is obtained by mapping the frequency domain energy attenuation slope, and the absolute value of the frequency domain energy attenuation slope is negatively correlated with the defect depth. The defect length is obtained by mapping the phase change time width, and is converted into physical length by the product of the probe moving speed and the time width. The defect width is characterized by the real amplitude gradient change rate as the basic value of the width, and is obtained by joint calibration after compensating and correcting the edge sharpness with the cumulative phase lag.

[0010] Optionally, based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and three-dimensional space, the planar coordinates of the phase classification map are converted into three-dimensional coordinates under the same reference as the geometric feature map by using the obtained probe moving speed and dwell time parameters, including: Based on the obtained scanning path coordinate axes, moving speed and dwell time parameters of the eddy current detection probe, a mapping relationship between the time series of the probe scanning trajectory and the axial depth layer in three-dimensional space is established. The coordinate points in the phase classification map are bound to the axial depth layer in three-dimensional space according to the layer position and the corresponding probe dwell time point through a mapping relationship; The three-dimensional coordinate nodes of the upgraded phase classification map are matched with the three-dimensional spatial nodes of the geometric feature map according to axial layering, circumferential angle and radial depth. The matching condition is that the spatial coordinate deviation between the two is less than the preset matching threshold. For successfully matched 3D spatial nodes, the proportion of each color point is extracted from the phase classification map to generate the defect type probability, and the morphology parameters of the geometric feature map are bound to the same spatial coordinates through spatial coordinates to form a data node set under the same benchmark.

[0011] Optionally, under the same benchmark, a composite tensor is generated by calculating the covariance matrix of the defect type probability extracted from the phase classification map and the morphology parameter set extracted from the geometric feature map, and a preset defect morphology association criterion library is invoked to perform the determination, so as to output the defect types including: The covariance of the phase classification probability distribution and morphological parameters at the same three-dimensional spatial node is calculated. The covariance matrix characterizes the correlation strength and direction of change between the defect type probability and the morphological parameters. The coordinate information, defect type probability, shape parameters and covariance matrix eigenvalues ​​of the same three-dimensional spatial node are superimposed to obtain a composite tensor that supports parallel analysis of multi-dimensional features. Call the criteria in the preset defect morphology association criteria library to perform joint mapping and matching on the covariance matrix eigenvalues, morphology parameters and coordinate information in the composite tensor. When the matching result satisfies that the covariance eigenvalue is greater than the preset type association threshold and the coordinate information meets the preset defect distribution prior conditions, trigger the defect type determination label. Based on the triggered judgment label, the coordinate information of the composite tensor, the probability of defect type, the shape parameters, and the covariance feature credibility index obtained by the ratio of the eigenvalue of the covariance matrix to the type association threshold are integrated to form a quantitative report.

[0012] Secondly, embodiments of the present invention provide a detection and discrimination system for aluminum tube defects that integrates eddy current phase and three-dimensional morphology, comprising: The phase distribution feature extraction module is used to extract eddy current phase distribution features from the electromagnetic response signals of aluminum tubes at different depths obtained by layering through the eddy current probe. The phase classification map output module is used to match the eddy current phase distribution characteristics with preset defect type association rules to generate a phase classification map that uses color coding to differentiate defect types. The geometric feature map output module is used to extract the time-frequency response features related to the geometric dimension of the defect from the eddy current phase distribution features, and combine them with the eddy current impedance change characteristics to construct a geometric feature map characterizing the physical morphology of the defect. The coordinate transformation module is used to convert the planar coordinates of the phase classification map into three-dimensional coordinates under the same reference as the geometric feature map, based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and the three-dimensional space, and by using the obtained probe moving speed and dwell time parameters. The defect determination module is used to generate a composite tensor by calculating the covariance matrix of the defect type probability extracted from the phase classification map and the shape parameter set extracted from the geometric feature map under the same benchmark, and then calling the preset defect shape association criterion library to perform the determination and output the defect type.

[0013] Thirdly, embodiments of the present invention provide a detection and discrimination device for aluminum tube defects that integrates eddy current phase and three-dimensional morphology, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the detection and discrimination method for aluminum tube defects that integrates eddy current phase and three-dimensional morphology as described above.

[0014] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the method described above for detecting and identifying aluminum tube defects by fusing eddy current phase and three-dimensional morphology.

[0015] (III) Beneficial Effects The beneficial effects of this invention are: through multi-dimensional feature fusion and spatial mapping mechanism, this invention effectively improves the objectivity and physical interpretability of defect detection.

[0016] First, by acquiring electromagnetic response signals in layers and extracting eddy current phase distribution features, the limitations of traditional single-layer detection in depth perception are overcome, enabling accurate separation of the electromagnetic feature differences between surface cracks and internal defects, and significantly improving the ability to analyze the depth attributes of defects.

[0017] Secondly, by combining the preset defect type association rules to generate a color-coded phase classification map, the differences in defect categories can be intuitively reflected, avoiding the subjective errors of manual interpretation and ensuring the physical rationality of the judgment logic. On this basis, the time-frequency response features are extracted simultaneously and the eddy current impedance change characteristics are fused to construct a geometric feature map, realizing the physical quantitative decoupling of three-dimensional morphology parameters such as defect depth and shape, and providing verifiable physical basis for morphology quantification.

[0018] Meanwhile, a three-dimensional spatial mapping relationship between the phase classification map and the geometric feature map is established based on the probe motion parameters, eliminating the measurement deviation caused by the lack of a unified spatial benchmark in the traditional two-dimensional analysis method.

[0019] Furthermore, by calculating the covariance matrix of the two types of map features to generate a composite tensor data structure, the statistical correlation between defect type and morphological parameters is explicitly expressed, breaking through the black box limitation of neural network models.

[0020] Finally, the judgment is performed based on the defect morphology association criterion library. When outputting the quantitative report, the covariance law of physical parameters is retained simultaneously. This allows the judgment logic to be traced back to the physical mechanism of electromagnetic response and to adapt to the detection needs of complex morphological defects. This multi-source feature collaborative analysis mechanism reduces reliance on human experience and overcomes the generalization defects of purely data-driven models. Attached Figure Description

[0021] Figure 1 A flowchart illustrating the method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the specific process of step S1 of the method provided in this embodiment of the invention; Figure 3 This is a detailed flowchart illustrating step S2 of the method provided in this embodiment of the invention; Figure 4 Phase classification maps provided in embodiments of the present invention; Figure 5 This is a detailed flowchart illustrating step S3 of the method provided in this embodiment of the invention; Figure 6 Geometric feature maps provided for embodiments of the present invention; Figure 7 This is a detailed flowchart illustrating step S4 of the method provided in this embodiment of the invention; Figure 8 A schematic diagram of the specific process of step S5 of the method provided in the embodiment of the present invention. Detailed Implementation

[0022] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] like Figure 1 As shown in the embodiment of the present invention, a method for detecting and identifying aluminum tube defects by integrating eddy current phase and three-dimensional morphology is proposed, comprising: extracting eddy current phase distribution features from electromagnetic response signals at different depths of the aluminum tube obtained layer by layer through an eddy current probe; matching the eddy current phase distribution features with preset defect type association rules to generate a phase classification map that differentiates defect types using color coding; extracting time-frequency response features related to the geometric dimension of the defect from the eddy current phase distribution features, and constructing a geometric feature map characterizing the physical morphology of the defect by combining the eddy current impedance change characteristics; based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and three-dimensional space, converting the planar coordinates of the phase classification map into three-dimensional coordinates under the same reference as the geometric feature map by using the obtained probe moving speed and dwell time parameters; under the same reference, generating a composite tensor by calculating the covariance matrix of the defect type probability extracted from the phase classification map and the morphology parameter set extracted from the geometric feature map, and calling a preset defect morphology association criterion library to perform the judgment to output the defect type.

[0024] This invention effectively improves the objectivity and physical interpretability of defect detection through multi-dimensional feature fusion and spatial mapping mechanisms.

[0025] First, by acquiring electromagnetic response signals in layers and extracting eddy current phase distribution features, the limitations of traditional single-layer detection in depth perception are overcome, enabling accurate separation of the electromagnetic feature differences between surface cracks and internal defects, and significantly improving the ability to analyze the depth attributes of defects.

[0026] Secondly, by combining the preset defect type association rules to generate a color-coded phase classification map, the differences in defect categories can be intuitively reflected, avoiding the subjective errors of manual interpretation and ensuring the physical rationality of the judgment logic. On this basis, the time-frequency response features are extracted simultaneously and the eddy current impedance change characteristics are fused to construct a geometric feature map, realizing the physical quantitative decoupling of three-dimensional morphology parameters such as defect depth and shape, and providing verifiable physical basis for morphology quantification.

[0027] Meanwhile, a three-dimensional spatial mapping relationship between the phase classification map and the geometric feature map is established based on the probe motion parameters, eliminating the measurement deviation caused by the lack of a unified spatial benchmark in the traditional two-dimensional analysis method.

[0028] Furthermore, by calculating the covariance matrix of the two types of map features to generate a composite tensor data structure, the statistical correlation between defect type and morphological parameters is explicitly expressed, breaking through the black box limitation of neural network models.

[0029] Finally, the judgment is performed based on the defect morphology association criterion library. When outputting the quantitative report, the covariance law of physical parameters is retained simultaneously. This allows the judgment logic to be traced back to the physical mechanism of electromagnetic response and to adapt to the detection needs of complex morphological defects. This multi-source feature collaborative analysis mechanism reduces reliance on human experience and overcomes the generalization defects of purely data-driven models.

[0030] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0031] Specifically, the present invention provides a method for detecting and identifying defects in aluminum tubes by integrating eddy current phase and three-dimensional morphology, comprising: S1. Extract the eddy current phase distribution characteristics from the electromagnetic response signals at different depths of the aluminum tube obtained by layering through the eddy current probe.

[0032] Furthermore, such as Figure 2 As shown, step S1 includes: S11. Control the eddy current probe to advance in a stepped manner along the axial direction of the aluminum tube. After each fixed axial distance, perform a circumferential rotation scan to acquire the electromagnetic response signals of each layered section. This invention controls the eddy current probe to advance in a stepped manner along the axial direction of the aluminum tube with a preset step size (e.g., 0.5 mm). After each axial step is completed, drive the probe to rotate and scan around the circumference of the aluminum tube (e.g., 360° equal intervals). By adopting the principle of multi-circle layered coverage, the original electromagnetic response signals of different depth sections of the aluminum tube are acquired, ensuring that the layered scan covers the full thickness of the tube wall.

[0033] S12. Perform phase-sensitive demodulation on the electromagnetic response signals of each layered section to separate the real and imaginary components of the impedance at each detection point. Use a quadrature lock-in amplifier to separate the real and imaginary components of the impedance at each detection point. The reason for choosing to separate the dual-channel signals in this step is that the real component reflects resistive loss (which is susceptible to lift-off interference), while the imaginary component characterizes reactive storage (which is directly related to the geometric phase characteristics of the defect).

[0034] S13. For the demodulated imaginary component, perform sliding window truncation and bandpass filtering to obtain the filtered imaginary component. High-frequency circuit noise and low-frequency environmental interference are filtered out, while retaining the defect characteristic response frequency band (e.g., 10-50kHz). This step further improves the signal-to-noise ratio of the defect signal by focusing on the effective frequency band of the imaginary component.

[0035] It is important to emphasize that the imaginary component is chosen as the core processing object here because: 1. Sensitive to deep defects: The imaginary component reflects the eddy current phase delay effect, and the detection depth of subsurface defects (such as pores and cracks in the tube wall) of aluminum tubes can reach more than 3 times that of the real component.

[0036] 2. Anti-interference advantages: The imaginary component is less affected by the lift-off due to the phase-locking mechanism, while the real component is easily affected by the fluctuation of the probe lift-off distance. Although the real component is not filtered, its sensitivity to absolute amplitude can be suppressed through subsequent ratio calculations.

[0037] S14. Based on the ratio of the filtered imaginary component to the corresponding real component, the arctangent angle is taken to obtain the phase angle sequence of the corresponding detection point. Combined with the standard deviation and skewness coefficient of the phase angle distribution in the corresponding layered cross section obtained by statistics, the eddy current phase distribution characteristics are generated.

[0038] S2. Match the eddy current phase distribution characteristics with the preset defect type association rules to generate a phase classification map that uses color coding to differentiate defect types.

[0039] Furthermore, such as Figure 3 As shown, step S2 includes: S21. Call the preset defect type association rule library, which is established by statistically analyzing historical defect samples. It includes the phase distribution standard deviation threshold range, skewness coefficient polarity conditions, and a mapping table between defect type and color for different defect types.

[0040] S22. Match the standard deviation of the eddy current phase distribution characteristics of the detection points within each layered section with the threshold range of the standard deviation of the phase distribution in the defect type association rule base, and simultaneously verify whether the polarity of the skewness coefficient meets the skewness coefficient polarity condition of the corresponding defect type. The skewness coefficient is a statistical indicator used to quantify the asymmetry of data distribution. The skewness coefficient polarity conditions include: raw material defects (negative skewness): Skewness < -0.5; foreign inclusion defects (positive skewness): Skewness > 0.6; perforation and leakage defects (weak skewness): -0.3 ≤ Skewness ≤ 0.3.

[0041] S23. When any detection point simultaneously satisfies the condition that the membership degree of the corresponding standard deviation interval is greater than or equal to the preset membership degree threshold and the polarity condition of the skewness coefficient is met, the characteristic color value of the corresponding point is assigned according to the mapping table between defect type and color; wherein, the mapping table between defect type and color is specifically referred to in Table 1: Table 1. Mapping Table of Defect Types and Colors

[0042] S24. Using the circumferential unfolding direction of the aluminum tube as the horizontal axis and the axial layering position as the vertical axis, perform planar mapping on the points with assigned colors within each layered section, ultimately generating... Figure 4 The phase classification map is shown.

[0043] S3. Extract the time-frequency response features related to the geometric dimension of the defect from the eddy current phase distribution features, and combine them with the eddy current impedance variation characteristics to construct a geometric feature map characterizing the physical morphology of the defect.

[0044] Furthermore, such as Figure 5 As shown, step S3 includes: S31. Perform time-frequency transformation on the phase angle sequence in the eddy current phase distribution characteristics, and extract the frequency domain energy attenuation slope and phase change time width as time-frequency response characteristics.

[0045] The frequency domain energy attenuation slope is calculated by performing a short-time Fourier transform on the phase angle sequence, extracting high-frequency components in the energy spectrum that are above a preset high-frequency threshold, analyzing the attenuation trend of the selected high-frequency components as the frequency increases, and calculating their attenuation rate through linear fitting. The larger the absolute value of the attenuation slope, the faster the high-frequency energy attenuates, and the deeper the hidden defect (the two are negatively correlated).

[0046] The phase transition time width is obtained by tracing the phase transition interval from the phase angle sequence using the zero-crossing detection method and calculating the duration span of the phase transition interval. Specifically, it detects abrupt regions in the phase angle sequence that cross zero (such as a phase flip from -180° to +180°), and then records the timestamps of the start and end points of the phase transition. This parameter reflects the length of the defect along the axial direction of the aluminum tube (the larger the width, the longer the longitudinal dimension of the defect).

[0047] S32. Based on the separated real and imaginary components of the impedance, calculate the rate of change of the real amplitude gradient and the cumulative amount of the imaginary phase lag at each detection point to construct the eddy current impedance characteristic parameters.

[0048] The real amplitude gradient change rate is calculated by performing a first-order difference operation (amplitude of the later point minus amplitude of the earlier point) on the real amplitude of adjacent detection points along the probe's movement direction (axial direction) based on the amplitude of the real component of the impedance. This quantifies the rate of change of the amplitude along the axial direction, and the maximum rate of change of the absolute value of the first-order difference is taken. This parameter reflects the severity of the defect's circumferential opening (the greater the rate of change, the steeper the opening edge). For example, when the probe sweeps over a sharp-edge defect, the real amplitude changes abruptly, and the gradient change rate increases sharply.

[0049] The cumulative phase lag is obtained by weighting the skewness coefficient and the absolute value of the phase difference between the corresponding detection point and its spatial neighborhood. Specifically, taking the current detection point as the center, all points within a spatial neighborhood (e.g., a radius of 3 mm) are selected, and the absolute value of their phase difference from the center point is calculated to measure the degree of accumulation of local phase distortion. The skewness coefficient and the absolute value of the phase difference between the neighborhood are then weighted and summed (e.g., cumulative phase lag = 0.7 × skewness coefficient + 0.3 × average neighborhood phase difference) to comprehensively reflect the phase lag effect and spatial consistency of the defect edge.

[0050] S33. The time-frequency response characteristics and eddy current impedance characteristic parameters are fused in multiple dimensions according to spatial location to generate morphological parameters including defect depth, length and width.

[0051] S34. Establish a three-dimensional space, project the morphological parameters of each detection point onto the corresponding spatial coordinate nodes in the three-dimensional space, and form a geometric feature map characterizing the physical morphology of the defect. (Reference) Figure 6 Finally obtained Figure 6-1 The standard damage / defect diagram shown Figure 6-2 The diagram showing the bubble defect in the aluminum tube is shown. Figure 6-3 The diagram of the unperforated defect shown Figure 6-4 The diagram of the perforation and leakage defect shown is as follows: Figure 6-5 The diagram showing the slag inclusion defect in the inner hole of the aluminum tube and Figure 6-6 The oxide scale defect diagrams shown cover the geometric feature maps of typical defect types.

[0052] It is important to understand that the defect depth is obtained by mapping the frequency domain energy attenuation slope, and the absolute value of the frequency domain energy attenuation slope is negatively correlated with the defect depth; the defect length is obtained by mapping the phase change time width, and is converted into physical length by the product relationship between the probe moving speed and the time width; the defect width is characterized by the real part amplitude gradient change rate, and is obtained by joint calibration after compensating and correcting the edge sharpness with the cumulative phase lag.

[0053] Specifically, the defect width is determined through the following steps: Local maxima of the impedance real part gradient rate of change curve are detected using the sliding window method. The region where the amplitude fluctuation exceeds three times the baseline noise level is defined as the effective peak range, and this effective peak range is mapped to the base value of the defect width. This value reflects the macroscopic physical size of the opening but does not consider the local disturbance effect of the edge morphology on the eddy current field. The base width value is dynamically adjusted based on the cumulative phase lag. At sharp edges (where the defect opening cross-section has a near-right angle or abrupt geometric transition, such as cracks or mechanically cut defects), the base width is positively compensated to expand the width estimate and offset the measurement shrinkage effect caused by edge distortion. The sharp edge correction width = base width × [1 + 0.3 × (cumulative amount - 0.85)], with a compensation coefficient of 0.3 determined through calibration experiments. At gentle edges (where the defect opening cross-section has a sloped or arc-shaped transition, such as corrosion pits or wear defects), the compensation amplitude is reduced to maintain the base width as the dominant factor. This satisfies the following formula: Correction width = base width × [1 + 0.05 × (cumulative amount / 0.85)]. Meanwhile, the compensation for smooth edges must meet the following principles: compensation amount ≤ 5% to avoid over-correction; compensation amplitude decreases as the cumulative phase lag decreases; and only positive adjustment or maintenance of the base width is performed. Finally, the base width value and the compensation correction value are merged according to a preset weight (e.g., base width accounts for 70%, edge compensation accounts for 30%) to generate the final defect width calibration result.

[0054] S4. Based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and the three-dimensional space, the planar coordinates of the phase classification map are converted into three-dimensional coordinates under the same reference as the geometric feature map by using the obtained probe moving speed and dwell time parameters.

[0055] Furthermore, such as Figure 7 As shown, step S4 includes: S41. Based on the obtained scanning path coordinate axes, moving speed, and dwell time parameters of the eddy current detection probe, establish a mapping relationship between the time series of the probe scanning trajectory and the axial depth layer in three-dimensional space. By using the probe moving speed and dwell time, the time axis (probe detection time sequence) is converted into a spatial depth axis, solving the problem of the lack of depth dimension in two-dimensional phase maps.

[0056] S42. The coordinate points in the phase classification map are bound to the axial depth layer in three-dimensional space according to the layer position and the corresponding probe dwell time point through the mapping relationship.

[0057] S43. Spatial matching is performed between the three-dimensional coordinate nodes of the upgraded phase classification map and the three-dimensional spatial nodes of the geometric feature map according to axial layering, circumferential angle, and radial depth. The matching condition is that the spatial coordinate deviation between the two is less than a preset matching threshold. This step expands the two-dimensional planar points (circumferential × axial) of the phase map into three-dimensional coordinate points (circumferential × axial × depth), aligning them with the three-dimensional space of the geometric map.

[0058] S44. For successfully matched 3D spatial nodes, extract the proportion of each color point from the phase classification map to generate the defect type probability, and bind it to the same spatial coordinate with the morphology parameters of the geometric feature map through spatial coordinates to form a data node set under the same benchmark.

[0059] S5. Under the same benchmark, a composite tensor is generated by calculating the covariance matrix of the defect type probability extracted from the phase classification map and the shape parameter set extracted from the geometric feature map, and the preset defect shape association criterion library is called to perform the judgment to output the defect type.

[0060] Furthermore, such as Figure 8 As shown, step S5 includes: S51. Calculate the covariance between the phase classification probability distribution and the morphological parameters at the same three-dimensional spatial node. The covariance matrix represents the correlation strength and direction of change between the defect type probability and the morphological parameters.

[0061] S52. The coordinate information, defect type probability, shape parameters, and covariance matrix eigenvalues ​​of the same 3D spatial node are superimposed to obtain a composite tensor that supports parallel parsing of multi-dimensional features. This is adapted to GPU parallel computing architecture to achieve efficient parsing and joint inference of multi-dimensional features. The multi-dimensional composite tensor is a regularized hierarchical stacking based on the spatial dimension (3D coordinate XYZ encoding), parameter dimension (defect type probability + shape parameters), and covariant feature dimension (covariance matrix eigenvalues ​​and signs).

[0062] S53. Call the criteria in the preset defect morphology association criteria library to perform joint mapping and matching on the covariance matrix eigenvalues, morphology parameters and coordinate information in the composite tensor. When the matching result satisfies that the covariance eigenvalue is greater than the preset type association threshold and the coordinate information meets the preset defect distribution prior conditions, trigger the defect type determination label.

[0063] S54. Based on the triggered judgment label, integrate the coordinate information of the composite tensor, the probability of the defect type, the shape parameters, and the covariance feature credibility index obtained by calculating the ratio of the eigenvalue of the covariance matrix to the type association threshold, and form a quantitative report.

[0064] Additionally, this invention provides a system for detecting and identifying aluminum tube defects by integrating eddy current phase and three-dimensional morphology, comprising: a phase distribution feature extraction module for extracting eddy current phase distribution features from electromagnetic response signals at different depths of the aluminum tube obtained layer by layer through an eddy current probe; a phase classification map output module for matching the eddy current phase distribution features with preset defect type association rules to generate a phase classification map that uses color coding to differentiate defect types; and a geometric feature map output module for extracting time-frequency response features related to the defect's geometric dimension from the eddy current phase distribution features, combined with eddy current impedance. The system employs several methods: a geometric feature map to characterize the physical morphology of defects; a coordinate transformation module to convert the planar coordinates of the phase classification map into three-dimensional coordinates under the same reference as the geometric feature map, based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and three-dimensional space, using the obtained probe moving speed and dwell time parameters; and a defect determination module to generate a composite tensor by calculating the covariance matrix of the defect type probability extracted from the phase classification map and the morphology parameter set extracted from the geometric feature map, under the same reference, and then calling a preset defect morphology association criterion library to perform the determination and output the defect type.

[0065] Furthermore, this embodiment of the invention also provides a device for detecting and identifying aluminum tube defects by integrating eddy current phase and three-dimensional morphology, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method for detecting and identifying aluminum tube defects by integrating eddy current phase and three-dimensional morphology as described above.

[0066] Furthermore, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the method described above for detecting and identifying aluminum tube defects by fusing eddy current phase and three-dimensional morphology.

[0067] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0069] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0070] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0071] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0072] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A method for detecting and discriminating defects of an aluminum tube by fusing eddy current phase and three-dimensional topography, characterized by, The method comprises the following steps: Extracting eddy current phase distribution features from electromagnetic response signals of different depths of the aluminum pipe obtained by the eddy current probe layer by layer; Matching the eddy current phase distribution features with the preset defect type association rules to generate a phase classification atlas that differentiates the defect types by color coding; Extracting time-frequency response features related to the geometric dimensions of the defects from the eddy current phase distribution features, and combining the eddy current impedance change characteristics to construct a geometric feature atlas representing the physical morphology of the defects; Based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and the three-dimensional space, the planar coordinates of the phase classification atlas are converted into three-dimensional coordinates under the same reference as the geometric feature atlas by obtaining the probe moving speed and residence time parameters; Under the same reference, a composite tensor is generated by calculating the covariance matrix of the defect type probability extracted from the phase classification atlas and the morphology parameter set extracted from the geometric feature atlas, and a preset defect morphology association criterion library is called to perform judgment to output the defect type.

2. The method of claim 1, wherein the method of detecting and discriminating defects in aluminum tubes by fusion of eddy current phase and three-dimensional topography is characterized by, The method comprises the following steps: Controlling the eddy current probe to move in steps along the axial direction of the aluminum pipe, and performing circumferential rotation scanning after moving a fixed axial interval to obtain electromagnetic response signals of each layer section; Phase-sensitive demodulation is performed on the electromagnetic response signals of each layer section to separate the real and imaginary parts of the impedance of each detection point; For the demodulated imaginary part, a sliding window is intercepted and band-pass filtering is performed to obtain the filtered imaginary part; Based on the ratio of the filtered imaginary part to the corresponding real part, the phase angle sequence of the corresponding detection point is obtained by taking the inverse tangent angle, and the eddy current phase distribution features are generated by combining the standard deviation and skewness coefficient of the phase angle distribution of the corresponding layer section.

3. The method of claim 1, wherein the method further comprises: The method comprises the following steps: A preset defect type association rule library is called, wherein the defect type association rule library is established by statistical historical defect samples, and contains the phase distribution standard deviation threshold interval, skewness coefficient polarity condition, and mapping table of defect type and color corresponding to different defect types; The phase distribution standard deviation of the eddy current phase distribution features of each detection point in the layer section is matched with the phase distribution standard deviation threshold interval of the defect type association rule library, and the skewness coefficient polarity is simultaneously checked to see if it meets the skewness coefficient polarity condition of the corresponding defect type; When any detection point simultaneously meets the condition that the corresponding standard deviation interval membership degree is greater than or equal to a preset membership threshold and the skewness coefficient polarity condition is met, the corresponding point value is assigned according to the mapping table of defect type and color; The points with assigned colors in each layer section are mapped in the plane with the circumferential expansion direction of the aluminum pipe as the horizontal axis and the axial layer position as the vertical axis to generate a phase classification atlas.

4. The method of claim 2, wherein the method further comprises: The method comprises the following steps: Extracting time-frequency response features related to the geometric dimensions of the defects from the eddy current phase distribution features, and combining the eddy current impedance change characteristics to construct a geometric feature atlas representing the physical morphology of the defects; The phase angle sequence in the eddy current phase distribution characteristics is subjected to time-frequency transformation, and the frequency domain energy decay slope and phase mutation time width are extracted as time-frequency response characteristics; According to the separated impedance real part and imaginary part, the real part amplitude gradient change rate and the imaginary part phase lag cumulative amount of each detection point are calculated, and the eddy current impedance characteristic parameters are constructed; The time-frequency response characteristics and the eddy current impedance characteristic parameters are fused in multiple dimensions according to the spatial position to generate the topographic parameters including the defect depth, length and width; A three-dimensional space is established, and the topographic parameters of each detection point are projected to the corresponding spatial coordinate nodes in the three-dimensional space to form a geometric feature map representing the physical topography of the defect.

5. The method for detecting and identifying aluminum pipe defects by fusing eddy current phase and three-dimensional topography according to claim 4, wherein The frequency domain energy decay slope is calculated by performing short-time Fourier transform on the phase angle sequence, and extracting the decay slope of the high-frequency component greater than the preset high-frequency threshold in the energy spectrum; The phase mutation time width is obtained by tracking the phase jump interval from the phase angle sequence using the zero-crossing point detection method, and calculating the duration span of the phase jump interval; The real part amplitude gradient change rate is calculated by first-order difference of the amplitude of the impedance real part along the probe moving direction, and taking the maximum change rate of the absolute value of the first-order difference; The phase lag cumulative amount is obtained by the skewness coefficient and the weighted sum of the phase difference absolute values of the spatial neighborhood of the corresponding detection point; The defect depth is mapped from the frequency domain energy decay slope, and the absolute value of the frequency domain energy decay slope is negatively correlated with the defect depth; The defect length is mapped from the phase mutation time width, and is converted into physical length by the product relationship of the obtained probe moving speed and time width; The defect width is represented by the real part amplitude gradient change rate, and is jointly calibrated after compensating and correcting the edge sharpness by the phase lag cumulative amount.

6. The method of claim 1, wherein the method further comprises: Based on the mapping relationship between the scanning path coordinates of the eddy current detection probe and the three-dimensional space, the plane coordinates of the phase classification map are converted into three-dimensional coordinates under the same reference as the geometric feature map by the obtained probe moving speed and dwell time parameters, including: Based on the obtained scanning path coordinate axis, moving speed and dwell time parameters of the eddy current detection probe, the mapping relationship between the time sequence of the probe scanning trajectory and the axial depth layer of the three-dimensional space is established; The coordinate points in the phase classification map are bound to the axial depth layer of the three-dimensional space according to the layered position and the corresponding probe dwell time point through the mapping relationship; The three-dimensional coordinate nodes of the upgraded phase classification map and the three-dimensional space nodes of the geometric feature map are matched in spatial position according to the axial layering, circumferential angle and radial depth, and the matching condition is that the spatial coordinate deviation of both is less than the preset matching threshold; For the three-dimensional space nodes that match successfully, the proportion of each color point in the phase classification map is extracted to generate the defect type probability, and the topographic parameters of the geometric feature map are bound to the same space coordinate through the spatial coordinate to form a data node set under the same reference.

7. The method for detecting and judging aluminum tube defects by integrating eddy current phase and three-dimensional morphology as described in claim 2, characterized in that, The composite tensor is generated by calculating the covariance matrix of the defect type probability extracted from the phase classification atlas and the set of topographic parameters extracted from the geometric feature atlas under the same reference, and a preset defect topography correlation criterion library is called to perform determination to output the defect type, including: The phase classification probability distribution and the topographic parameters on the same three-dimensional space node are subjected to covariance calculation, and the covariance matrix represents the correlation strength and change direction of the defect type probability and the topographic parameters; The coordinate information, the defect type probability, the topographic parameters and the covariance matrix eigenvalues of the same three-dimensional space node are subjected to dimension stacking to obtain a composite tensor supporting multi-dimensional feature parallel analysis; A criterion in a preset defect topography correlation criterion library is called to jointly map and match the covariance matrix eigenvalues, the topographic parameters and the coordinate information in the composite tensor, and when the matching result satisfies the condition that the covariance eigenvalue is greater than a preset type correlation threshold and the coordinate information meets a preset defect distribution prior condition, a defect type determination label is triggered; According to the triggered determination label, the coordinate information, the defect type probability, the topographic parameters and the covariance eigenvalue-based covariance feature confidence index obtained by ratio calculation of the type correlation threshold are integrated to form a quantitative report.

8. A system for detecting and discriminating defects in aluminum tubes by fusion of eddy current phase and three-dimensional topography, characterized by, It includes: A phase distribution feature obtaining module is configured to extract eddy current phase distribution features from electromagnetic response signals of different depths of an aluminum pipe obtained by layering of an eddy current probe; A phase classification atlas output module is configured to match the eddy current phase distribution features with a preset defect type correlation rule to generate a phase classification atlas that differentiates defect types by color coding differences; A geometric feature atlas output module is configured to extract time-frequency response features related to defect geometric dimensions from the eddy current phase distribution features, and construct a geometric feature atlas representing defect physical topography in combination with eddy current impedance change characteristics; A coordinate conversion module is configured to convert planar coordinates of the phase classification atlas into three-dimensional coordinates under the same reference as the geometric feature atlas based on a mapping relationship between scanning path coordinates of the eddy current detection probe and a three-dimensional space, and by obtaining probe moving speed and dwell time parameters; A defect determination module is configured to generate a composite tensor by calculating a covariance matrix of defect type probability extracted from the phase classification atlas and a set of topographic parameters extracted from the geometric feature atlas under the same reference, and calling a preset defect topography correlation criterion library to perform determination to output the defect type.

9. A device for detecting and discriminating defects of an aluminum tube by fusion of eddy current phase and three-dimensional topography, characterized by, At least one processor; and a memory connected in communication with the at least one processor; The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the detection and discrimination method of aluminum pipe defects integrating eddy current phase and three-dimensional topography as claimed in any one of claims 1-7. Executable instructions are executed by the processor to implement the detection and discrimination method of aluminum pipe defects integrating eddy current phase and three-dimensional topography as claimed in any one of claims 1-7.

10. A computer-readable storage medium having stored thereon computer- executable instructions, the computer-executable instructions comprising instructions for: receiving a request for a resource; determining whether the request is for a resource that is subject to a policy; and if the request is for a resource that is subject to a policy, then determining whether the request is from a client that is subject to the policy. ​

Images