A method and system for detecting the quality of an industrial profile after aging treatment

Abstract

The application discloses a kind of quality detection method and system of industrial section after aging treatment, belong to industrial metal profile nondestructive testing technical field, its method includes the following steps: S1, scan trajectory planning;S2, synchronous joint scanning;S3, structural interference decoupling;S4, residual stress imaging;S5, microfeature extraction;S6, comprehensive quality evaluation;It aims at solving the core defects existing in the quality detection technology of the existing industrial section after aging treatment: the spatiotemporal synchronism of current multi-physical field detection is poor, cannot be adapted to complex cross-section profile to realize conformal adaptive scanning;Cannot effectively decouple and filter out structural interference signals caused by profile wall thickness mutation and inner cavity edge, and the signal-to-noise ratio of effective detection signal is low;Detection dimension is single, cannot realize the integrated correlation evaluation of macro residual stress characterization, microstructure feature extraction and service performance quantitative prediction, it is difficult to meet the full-dimensional quality control needs of high-end profile.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology for industrial metal profiles, and more specifically, to a method and system for quality testing of industrial profiles after aging treatment. Background Technology

[0002] High-strength aluminum alloys, high-strength steel and other industrial profiles are core structural components in high-end equipment fields such as aerospace, rail transportation and new energy vehicles. Aging heat treatment is a key process for controlling the mechanical properties and microstructure stability of profiles. The residual stress distribution and microstructure of profiles after aging treatment directly determine the dimensional stability, resistance to intergranular corrosion and fatigue service life of structural components. Therefore, conducting full-dimensional and high-precision quality testing on industrial profiles after aging treatment is a core link to ensure the long-term service safety of high-end equipment.

[0003] Currently, for quality inspection of industrial profiles after aging treatment, the industry mostly adopts single physical field testing or destructive sampling methods, both of which have significant technical limitations. Destructive methods such as metallographic testing and tensile testing can only achieve sampling inspection and cannot complete online inspection of the entire surface and batch of profiles. Moreover, the inspection cycle is long and the cost is high, which cannot be adapted to the pace of industrial mass production. X-ray diffraction can only characterize the stress state of the extreme surface layer of the profile and cannot detect non-visible areas such as cavities and corners of complex cross-section profiles. In addition, there is a risk of radiation, making it difficult to apply in the field. Conventional ultrasonic and pulsed eddy current testing can only identify macroscopic defects or changes in surface conductivity. The quantitative characterization accuracy of residual stress and microstructure variation is insufficient. When dealing with complex profiles with variable wall thickness and internal cavities, strong interference is easily generated by interface reflection and edge effects, resulting in distorted test results.

[0004] Existing multiphysics field fusion detection technologies are mostly simple superpositions of detection methods, failing to achieve spatiotemporal synchronous acquisition with the same clock source. The scanning position and multiphysics field signals cannot be accurately matched, and there is a lack of conformal adaptive scanning schemes for complex cross-section profiles, making it impossible to effectively decouple and filter out structural interference. At the same time, existing technologies mostly remain at the level of qualitative characterization of macroscopic defects and stress, failing to establish cross-scale mapping between macroscopic stress and microstructure, and even more so failing to achieve quantitative prediction of service performance at the same time, making it difficult to meet the stringent requirements of full-dimensional quality control for high-end profiles. Summary of the Invention

[0005] To address the problems existing in the prior art, the present invention aims to provide a quality inspection method and system for industrial profiles after aging treatment, which aims to solve the core defects in the existing quality inspection technology for industrial profiles after aging treatment: the current multi-physics field detection has poor spatiotemporal synchronization, cannot adapt to complex cross-section profiles to achieve shape-adaptive scanning, resulting in inaccurate matching between detection data and spatial position; it cannot effectively decouple and filter out structural interference signals caused by sudden changes in profile wall thickness and inner cavity edges, resulting in low signal-to-noise ratio of effective detection signals; and the detection dimension is single, which cannot achieve integrated correlation evaluation of macroscopic residual stress characterization, microstructure feature extraction and quantitative prediction of service performance, making it difficult to meet the full-dimensional quality control needs of high-end profiles.

[0006] To solve the above problems, the present invention adopts the following technical solution.

[0007] A quality inspection method for industrial profiles after aging treatment includes the following steps:

[0008] S1. Scan trajectory planning: Obtain the three-dimensional geometric model of the industrial profile to be tested, import it into the control system as prior information, and plan and generate a conformal non-contact adaptive scanning trajectory covering the surface of the industrial profile based on the spatial position information of the three-dimensional geometric model.

[0009] S2. Synchronous joint scanning: Using a multi-axis scanning device equipped with a composite detection probe, the time-treated industrial profile is simultaneously and jointly scanned along the conformal non-contact adaptive scanning trajectory, and the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal of the industrial profile are collected in real time.

[0010] S3. Structural Interference Decoupling: The topography and stress decoupling algorithm is invoked to perform position matching based on the three-dimensional geometric model, compensate for and filter out the structural interference signals generated by the sudden changes in wall thickness and the reflection of the inner cavity edge of the industrial profile, and obtain pure multi-frequency pulse eddy current signals and pure nonlinear ultrasonic signals.

[0011] S4. Residual stress imaging: Feature extraction is performed on the pure nonlinear ultrasonic signal and the pure multi-frequency pulsed eddy current signal respectively to separate the sound velocity change characteristics and electromagnetic variation characteristics that characterize the residual stress of the material, and convert them into spatial stress distribution data to generate a three-dimensional residual stress cloud map.

[0012] S5. Microscopic feature extraction: Simultaneously perform cross-scale feature mapping and microscopic physicochemical property evaluation, extract high-order harmonic features characterizing dislocation density from the pure nonlinear ultrasonic signal, and extract high-frequency impedance phase features characterizing grain boundary conductivity anomalies from the pure multi-frequency pulse eddy current signal.

[0013] S6. Comprehensive Quality Assessment: The extracted high-order harmonic features and high-frequency impedance phase features are concatenated to construct a joint feature vector, which is then input into a pre-trained machine learning prediction model. This model outputs the intergranular corrosion sensitivity index and fatigue life loss rate for each scanning point of the industrial profile. The performance indicators of the points are mapped one-to-one with the stress data of the corresponding spatial coordinates in the three-dimensional residual stress cloud map. Finally, a comprehensive quality inspection report of the industrial profile is output, realizing multi-physics integrated non-contact inspection of industrial profiles after aging treatment. This model is suitable for complex cross-section profiles and simultaneously completes residual stress characterization, microscopic feature extraction, and quantitative evaluation of service performance.

[0014] As a further technical solution of the present invention, S2 specifically includes the following sub-steps:

[0015] S21. The motion control system of the multi-axis scanning device and the signal acquisition module of the composite detection probe are configured to be synchronized by the same clock source. The preset high-frequency reference clock signal provided by the same high-precision clock source is used to provide synchronous trigger clocks for the position encoder of the motion control axis, the pulse eddy current detection unit, and the nonlinear ultrasonic detection unit, respectively, to ensure that the spatiotemporal synchronization error between the scanning position and the signal acquisition does not exceed the preset spatiotemporal synchronization error threshold.

[0016] S22. Set position-triggered synchronous acquisition logic in the composite detection probe. Pre-set acquisition trigger points on the conformal non-contact adaptive scanning trajectory with equal spatial step size. When the multi-axis scanning device drives the composite detection probe to each acquisition trigger point, the position encoder synchronously outputs a trigger signal, which synchronously triggers the pulse eddy current detection unit and the nonlinear ultrasonic detection unit to complete the synchronous acquisition of the signal at that point, realizing a one-to-one correspondence between the scanning position and the multi-physics field signal.

[0017] S23. The composite detection probe adopts a non-contact coplanar integrated architecture, integrating the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit on the same detection end face. The pulsed eddy current detection unit includes a multi-frequency pulsed eddy current detection coil, and the nonlinear ultrasonic detection unit includes a nonlinear ultrasonic electromagnetic transducer. This ensures that the detection areas of the two types of detection units coincide, eliminating spatial position deviations of multi-physical field signals. Through clock homogeneous configuration, position-triggered acquisition, and coplanar integrated architecture, the spatiotemporal deviations of multi-physical field signals are eliminated, ensuring accurate correspondence between the detection signal and the scanning position.

[0018] As a further technical solution of the present invention, the spatiotemporal synchronization acquisition mechanism of S2 is constrained and calibrated through the following logic:

[0019] The signal sampling time difference between the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit is compared, and this difference is limited to being less than the period of the synchronous reference clock.

[0020] A unique correspondence is established between the three-dimensional spatial coordinates of the acquisition point and the output signal sets of the two types of detection units; the planar geometric distance between the detection area centers of the two types of detection units in the probe coordinate system is calculated, and this distance is limited to be less than the preset center spatial deviation threshold. The synchronization accuracy is constrained from three dimensions: sampling time, spatial mapping, and detection area overlap, so as to ensure the consistency and matching of multi-physics detection data from the source.

[0021] As a further technical solution of the present invention, S3 specifically includes the following sub-steps:

[0022] S31. Establish a unified three-dimensional spatial coordinate system, convert the three-dimensional geometric model of the industrial profile to be tested into a standard digital twin model under the workpiece coordinate system, and convert the coordinates of the acquisition points under the probe coordinate system to the workpiece coordinate system through the motion forward solution algorithm of the multi-axis scanning device for the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal collected at each point during the scanning process, so as to complete the accurate matching of the spatial position of the acquired signal and the three-dimensional geometric model.

[0023] S32, based on the geometric features of the three-dimensional geometric model, pre-generates a structural interference signal template library. For the locations of sudden changes in wall thickness, inner cavity edges, and corner feature areas of industrial profiles, the pulse eddy current edge effect interference signal and nonlinear ultrasonic interface reflection interference signal at the corresponding locations are calculated by numerical simulation or experimental calibration methods, and stored as a structural interference signal template library according to spatial coordinate index.

[0024] S33 invokes the topography and stress decoupling algorithm. For each spatially matched acquisition point, it calls the corresponding coordinate interference signal template from the structural interference signal template library. Adaptive matched filtering and coherent cancellation algorithms are used to filter out structural interference signals from the measured signals, obtaining pure multi-frequency pulse eddy current signals and pure nonlinear ultrasonic signals that are only related to material properties. Through a unified coordinate system, a pre-generated interference template library, and a decoupling algorithm, the detection signal and the three-dimensional model are accurately matched, effectively filtering out interference signals brought by the profile structure.

[0025] As a further technical solution of the present invention, the coordinate transformation and structural disturbance decoupling algorithm of S3 is implemented through the following logic:

[0026] By combining the rotation angle and linear displacement of the probe, the coordinates of the acquisition points in the probe coordinate system are transformed to the workpiece coordinate system through spatial translation and rotation matrix operations.

[0027] For each acquisition trigger point, the optimal weights of the adaptive matched filter are calculated through iterative optimization to adjust the amplitude and phase of the interference signal template so that it matches the interference components in the measured signal. Then, the interference template signal after optimal weight adjustment is subtracted from the measured original signal to complete the filtering of structural interference. Precise point matching is achieved through spatial coordinate transformation. The adaptive iterative optimization filtering algorithm significantly improves the filtering accuracy of structural interference and the effective signal-to-noise ratio.

[0028] As a further technical solution of the present invention, the extraction of high-order harmonic features characterizing dislocation density in step S5 specifically includes the following sub-steps:

[0029] S51. Perform time-domain to frequency-domain conversion on the pure nonlinear ultrasound signal. Use Fast Fourier Transform to convert the time-domain ultrasound signal into a frequency-domain spectrum signal. Accurately extract the fundamental amplitude corresponding to the fundamental frequency and the second harmonic amplitude corresponding to the second harmonic frequency from the spectrum signal. Then, use window function filtering to eliminate spectrum leakage and noise interference.

[0030] S52. Based on the extracted fundamental and second harmonic amplitudes, the nonlinear core proportional characteristics of the material are extracted, and a quantitative mapping relationship between the nonlinear characteristics and the dislocation density inside the material is established. Through the calibrated mapping model, the nonlinear characteristics are converted into the dislocation density values ​​of the corresponding detection points, and the quantitative extraction of dislocation density-related higher harmonic characteristics is completed. Ultrasonic harmonic characteristics are accurately extracted through time-frequency conversion, and the mapping relationship between nonlinear characteristics and dislocation density is established to achieve accurate quantitative characterization of the microstructure of the profile.

[0031] As a further technical solution of the present invention, the extraction of the nonlinear features and the physical mapping with the dislocation density are achieved through the following logic:

[0032] The ratio of the second harmonic amplitude to the square of the fundamental amplitude is extracted as the core proportional feature characterizing the nonlinearity of the material. Based on this core feature, the relative nonlinear coefficient of the material can be derived.

[0033] Based on the inherent nonlinear coefficient of dislocation-free ideal single-crystal materials, and combined with the dislocation nonlinear contribution coefficient obtained by calibration of material elastic modulus, dislocation type, and average spacing of pinning points, the relative nonlinear coefficient is linearly mapped to the dislocation density at the detection point. The core proportional features are extracted to characterize the material nonlinearity. The dislocation density is accurately converted through the calibrated linear mapping relationship, providing a reliable quantitative basis for the microscopic performance evaluation of profiles.

[0034] As a further technical solution of the present invention, the high-frequency impedance phase feature extraction in S5, which characterizes the abnormal conductivity of grain boundaries, specifically includes the following steps:

[0035] S53. Perform multi-frequency impedance analysis on the pure multi-frequency pulsed eddy current signal. For the eddy current response signal under different excitation frequencies, use the quadrature lock-in amplification algorithm to separate the real part and imaginary part of the impedance corresponding to each excitation frequency, and calculate the impedance amplitude and impedance phase angle at the corresponding frequency.

[0036] S54. Based on multi-frequency impedance phase characteristics, a quantitative model for grain boundary conductivity anomalies is established. Taking the impedance phase angle corresponding to the selected high-frequency excitation point as the core feature, the impedance phase shift of different detection points relative to the standard sample without anomalies is calculated. Through conductivity inversion algorithm, the impedance phase shift is converted into the grain boundary conductivity change value, and the quantitative extraction of grain boundary conductivity anomaly features is completed. The phase features are accurately extracted through multi-frequency impedance analysis, and the grain boundary conductivity change is inverted based on the phase shift, so as to achieve high-sensitivity and accurate identification of profile grain boundary structure anomalies.

[0037] As a further technical solution of the present invention, the multi-frequency impedance phase analysis and the calculation of grain boundary conductivity anomalies are implemented through the following logic:

[0038] For any excitation frequency, calculate the ratio of the imaginary part to the real part of the impedance of the eddy current response signal, and then calculate the arctangent of this ratio to obtain the eddy current impedance phase angle at the corresponding excitation frequency.

[0039] The impedance phase shift is obtained by calculating the impedance phase angle difference between the test point of the industrial profile under test and the standard sample without grain boundary conductivity abnormality at the same selected specific high-frequency excitation frequency point.

[0040] Based on the phase-conductivity linear conversion relationship obtained in advance through standard sample calibration, the above impedance phase offset is converted into the change in grain boundary conductivity of the detection point relative to the standard sample. The accurate impedance phase angle is obtained through rigorous phase calculation logic, and the conductivity change is inverted based on the calibrated linear relationship, thereby improving the reliability and accuracy of grain boundary anomaly detection.

[0041] A quality inspection system for industrial profiles after aging treatment includes a model import and trajectory planning module, a scanning execution and signal acquisition module, a structural interference decoupling module, a residual stress imaging module, a microscopic feature extraction module, and a comprehensive quality assessment module.

[0042] The model import and trajectory planning module is used to acquire the three-dimensional geometric model of the industrial profile to be tested, import it into the control system as prior information, and plan and generate a conformal non-contact adaptive scanning trajectory covering the surface of the industrial profile based on the spatial position information of the three-dimensional geometric model.

[0043] The scanning execution and signal acquisition module is used to control the multi-axis scanning device equipped with a composite detection probe to perform synchronous joint scanning of the time-treated industrial profiles along the conformal non-contact adaptive scanning trajectory, and to acquire the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal of the industrial profiles in real time.

[0044] The structural interference decoupling module is used to call the morphology and stress decoupling algorithm, perform position matching based on the three-dimensional geometric model, compensate for and filter out the structural interference signals generated by the sudden changes in wall thickness and the reflection of the inner cavity edge of the industrial profile, and obtain pure multi-frequency pulse eddy current signals and pure nonlinear ultrasonic signals.

[0045] The residual stress imaging module is used to extract features from the pure nonlinear ultrasonic signal and the pure multi-frequency pulsed eddy current signal respectively, separate the sound velocity change features and electromagnetic variation features that characterize the residual stress of the material, and convert them into spatial stress distribution data to generate a three-dimensional residual stress cloud map.

[0046] The micro-feature extraction module is used to simultaneously perform cross-scale feature mapping and micro-physicochemical property evaluation, extract high-order harmonic features characterizing dislocation density from the pure nonlinear ultrasonic signal, and extract high-frequency impedance phase features characterizing grain boundary conductivity anomalies from the pure multi-frequency pulse eddy current signal.

[0047] The comprehensive quality assessment module is used to construct a joint feature vector by concatenating the extracted high-order harmonic features and the high-frequency impedance phase features, and input it into a pre-trained machine learning prediction model. It outputs the intergranular corrosion sensitivity index and fatigue life loss rate of each scanning point of the industrial profile, and maps the performance indicators of the points to the stress data of the corresponding spatial coordinates in the three-dimensional residual stress cloud map. Finally, it outputs a comprehensive quality inspection report of the industrial profile. Through a modular collaborative architecture, it realizes the full-process automation of the time-sensitive profile inspection, ensuring the accuracy of multi-physics field inspection while significantly improving the efficiency and consistency of batch inspection of industrial profiles.

[0048] Compared with the prior art, the advantages of this invention are:

[0049] This invention plans a conformal non-contact adaptive scanning trajectory based on the three-dimensional geometric model of the industrial profile to be tested. It can perfectly adapt to complex cross-section industrial profiles with internal cavities, variable wall thickness, and irregular curved surfaces, achieving full-surface coverage of the profile without dead angles. It completely solves the problems of large fluctuations in lifting distance and incomplete detection of irregular and non-visible areas in existing fixed trajectory scanning.

[0050] By using a multi-axis scanning device equipped with an integrated composite detection probe, synchronous joint scanning of multi-frequency pulsed eddy current and nonlinear ultrasonic signals is achieved. With the synchronous triggering mechanism based on the same clock source, the precise one-to-one correspondence between the scanning spatial position and the multi-physics field detection signal is ensured. This avoids spatial misalignment and temporal deviation of signals from different detection methods, significantly improving the consistency and matching degree of the detection data. At the same time, non-contact online detection is realized, avoiding the industry pain points of long cycle, high cost and inability to cover the entire batch in destructive sampling inspection.

[0051] By employing a morphology and stress decoupling algorithm, the system accurately matches the acquired signals with the spatial position based on a three-dimensional geometric model. This algorithm can specifically compensate for and filter out structural interference signals caused by abrupt changes in profile wall thickness, reflections at the inner cavity edges, and corner structures. It precisely decouples the interference caused by the profile geometry from the effective signals generated by changes in the material's own properties, thus completely solving the core problems in existing technologies where effective detection signals are submerged by structural interference, the detection signal-to-noise ratio is low, and the results are easily distorted. This provides highly reliable raw data support for the subsequent accurate quantitative characterization of material properties.

[0052] It can simultaneously and accurately generate three-dimensional residual stress cloud maps of profiles and quantitatively extract microscopic features of dislocation density and grain boundary conductivity anomalies, realizing cross-scale correlation mapping between macroscopic stress distribution and microstructure features, breaking through the limitation of existing technologies that can only achieve qualitative identification of macroscopic defects. At the same time, it can quantitatively predict the intergranular corrosion sensitivity and fatigue life loss rate of profiles through machine learning prediction models, and complete the spatial multidimensional mapping between service performance indicators and three-dimensional residual stress cloud maps, realizing the upgrade from single qualification judgment to full-dimensional performance evaluation. The test results can directly provide data support for the optimization of aging process, fully meeting the stringent requirements of high-end equipment field for the whole chain quality control of industrial profiles. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] This invention provides an embodiment 1

[0056] Please see Figure 1 A quality inspection method for industrial profiles after aging treatment includes the following steps:

[0057] S1. Scan trajectory planning: Obtain the three-dimensional geometric model of the industrial profile to be tested, import it into the control system as prior information, and plan and generate a conformal non-contact adaptive scanning trajectory covering the surface of the industrial profile based on the spatial position information of the three-dimensional geometric model.

[0058] S2. Synchronous joint scanning: Using a multi-axis scanning device equipped with a composite detection probe, the time-treated industrial profile is simultaneously and jointly scanned along the conformal non-contact adaptive scanning trajectory, and the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal of the industrial profile are collected in real time.

[0059] S3. Structural Interference Decoupling: The topography and stress decoupling algorithm is invoked to perform position matching based on the three-dimensional geometric model, compensate for and filter out the structural interference signals generated by the sudden changes in wall thickness and the reflection of the inner cavity edge of the industrial profile, and obtain pure multi-frequency pulse eddy current signals and pure nonlinear ultrasonic signals.

[0060] S4. Residual stress imaging: Feature extraction is performed on the pure nonlinear ultrasonic signal and the pure multi-frequency pulsed eddy current signal respectively to separate the sound velocity change characteristics and electromagnetic variation characteristics that characterize the residual stress of the material, and convert them into spatial stress distribution data to generate a three-dimensional residual stress cloud map.

[0061] S5. Microscopic feature extraction: Simultaneously perform cross-scale feature mapping and microscopic physicochemical property evaluation, extract high-order harmonic features characterizing dislocation density from the pure nonlinear ultrasonic signal, and extract high-frequency impedance phase features characterizing grain boundary conductivity anomalies from the pure multi-frequency pulse eddy current signal.

[0062] S6. Comprehensive Quality Assessment: The extracted high-order harmonic features and high-frequency impedance phase features are connected in series to construct a joint feature vector, which is then input into a pre-trained machine learning prediction model. The model outputs the intergranular corrosion sensitivity index and fatigue life loss rate of each scanning point of the industrial profile. The performance indicators of the points are mapped one-to-one with the stress data of the corresponding spatial coordinates in the three-dimensional residual stress cloud map. Finally, a comprehensive quality inspection report of the industrial profile is output, realizing multi-physics field integrated non-contact inspection of industrial profiles after aging treatment. It is suitable for complex cross-section profiles and simultaneously completes residual stress characterization, micro-feature extraction and quantitative evaluation of service performance.

[0063] S2 specifically includes the following steps:

[0064] S21. The motion control system of the multi-axis scanning device and the signal acquisition module of the composite detection probe are configured to be synchronized by the same clock source. The preset high-frequency reference clock signal provided by the same high-precision clock source is used to provide synchronous trigger clocks for the position encoder of the motion control axis, the pulse eddy current detection unit, and the nonlinear ultrasonic detection unit, respectively, to ensure that the spatiotemporal synchronization error between the scanning position and the signal acquisition does not exceed the preset spatiotemporal synchronization error threshold.

[0065] S22. Set position-triggered synchronous acquisition logic in the composite detection probe. Pre-set acquisition trigger points on the conformal non-contact adaptive scanning trajectory with equal spatial step size. When the multi-axis scanning device drives the composite detection probe to each acquisition trigger point, the position encoder synchronously outputs a trigger signal, which synchronously triggers the pulse eddy current detection unit and the nonlinear ultrasonic detection unit to complete the synchronous acquisition of the signal at that point, realizing a one-to-one correspondence between the scanning position and the multi-physics field signal.

[0066] S23. The composite detection probe adopts a non-contact coplanar integrated architecture, integrating the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit on the same detection end face. The pulsed eddy current detection unit includes a multi-frequency pulsed eddy current detection coil, and the nonlinear ultrasonic detection unit includes a nonlinear ultrasonic electromagnetic transducer. This ensures that the detection areas of the two types of detection units coincide, eliminating spatial position deviations of multi-physical field signals. Through clock co-source configuration, position-triggered acquisition, and coplanar integrated architecture, the spatiotemporal deviations of multi-physical field signals are eliminated, ensuring accurate correspondence between the detection signal and the scanning position.

[0067] The spatiotemporal synchronization acquisition mechanism of S2 is constrained and calibrated through the following logic:

[0068] The signal sampling time difference between the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit is compared, and this difference is limited to being less than the period of the synchronous reference clock.

[0069] Establish a unique correspondence between the three-dimensional spatial coordinates of the acquisition point and the output signal sets of the two types of detection units; calculate the planar geometric distance between the center of the detection area of ​​the two types of detection units in the probe coordinate system, and limit this distance to be less than the preset center spatial deviation threshold. Constrain the synchronization accuracy from three dimensions: sampling time, spatial mapping, and detection area overlap, so as to ensure the consistency and matching of multi-physics detection data from the source.

[0070] S3 specifically includes the following steps:

[0071] S31. Establish a unified three-dimensional spatial coordinate system, convert the three-dimensional geometric model of the industrial profile to be tested into a standard digital twin model under the workpiece coordinate system, and convert the coordinates of the acquisition points under the probe coordinate system to the workpiece coordinate system through the motion forward solution algorithm of the multi-axis scanning device for the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal collected at each point during the scanning process, so as to complete the accurate matching of the spatial position of the acquired signal and the three-dimensional geometric model.

[0072] S32, based on the geometric features of the three-dimensional geometric model, pre-generates a structural interference signal template library. For the locations of sudden changes in wall thickness, inner cavity edges, and corner feature areas of industrial profiles, the pulse eddy current edge effect interference signal and nonlinear ultrasonic interface reflection interference signal at the corresponding locations are calculated by numerical simulation or experimental calibration methods, and stored as a structural interference signal template library according to spatial coordinate index.

[0073] S33 invokes the topography and stress decoupling algorithm. For each spatially matched acquisition point, it calls the corresponding coordinate interference signal template from the structural interference signal template library. Adaptive matched filtering and coherent cancellation algorithm are used to filter out structural interference signals from the measured signal, obtaining pure multi-frequency pulse eddy current signal and pure nonlinear ultrasonic signal that are only related to material properties. Through a unified coordinate system, pre-generated interference template library and decoupling algorithm, the detection signal and three-dimensional model are accurately matched, effectively filtering out interference signals brought by the profile structure.

[0074] The coordinate transformation and structural disturbance decoupling algorithm of S3 is implemented through the following logic:

[0075] By combining the rotation angle and linear displacement of the probe, the coordinates of the acquisition points in the probe coordinate system are transformed to the workpiece coordinate system through spatial translation and rotation matrix operations.

[0076] For each acquisition trigger point, the optimal weight of the adaptive matched filter is calculated through iterative optimization to adjust the amplitude and phase of the interference signal template so that it matches the interference components in the measured signal. Then, the interference template signal after optimal weight adjustment is subtracted from the measured original signal to complete the filtering of structural interference. The point matching is achieved through spatial coordinate transformation. The adaptive iterative optimization filtering algorithm significantly improves the filtering accuracy of structural interference and the effective signal-to-noise ratio.

[0077] The extraction of high-order harmonic features characterizing dislocation density in S5 specifically includes the following steps:

[0078] S51. Perform time-domain to frequency-domain conversion on the pure nonlinear ultrasound signal. Use Fast Fourier Transform to convert the time-domain ultrasound signal into a frequency-domain spectrum signal. Accurately extract the fundamental amplitude corresponding to the fundamental frequency and the second harmonic amplitude corresponding to the second harmonic frequency from the spectrum signal. Then, use window function filtering to eliminate spectrum leakage and noise interference.

[0079] S52. Based on the extracted fundamental amplitude and second harmonic amplitude, the nonlinear core proportional characteristics of the material are extracted, and a quantitative mapping relationship between the nonlinear characteristics and the dislocation density inside the material is established. Through the calibrated mapping model, the nonlinear characteristics are converted into the dislocation density values ​​of the corresponding detection points, and the quantitative extraction of dislocation density-related higher harmonic characteristics is completed. Ultrasonic harmonic characteristics are accurately extracted through time-frequency conversion, and a mapping relationship between nonlinear characteristics and dislocation density is established to achieve accurate quantitative characterization of the microstructure of the profile.

[0080] The extraction of the nonlinear features and their physical mapping to dislocation density are achieved through the following logic:

[0081] The ratio of the second harmonic amplitude to the square of the fundamental amplitude is extracted as the core proportional feature characterizing the nonlinearity of the material. Based on this core feature, the relative nonlinear coefficient of the material can be derived.

[0082] Based on the inherent nonlinear coefficient of dislocation-free ideal single crystal material, and combined with the dislocation nonlinear contribution coefficient obtained by calibration of material elastic modulus, dislocation type, and average spacing of pinning points, the relative nonlinear coefficient is linearly mapped to the dislocation density at the detection point. The core proportional features are extracted to characterize the material nonlinearity. The dislocation density is accurately converted through the calibrated linear mapping relationship, providing a reliable quantitative basis for the microscopic performance evaluation of profiles.

[0083] The high-frequency impedance phase feature extraction in S5, which characterizes the abnormal conductivity of grain boundaries, specifically includes the following steps:

[0084] S53. Perform multi-frequency impedance analysis on the pure multi-frequency pulsed eddy current signal. For the eddy current response signal under different excitation frequencies, use the quadrature lock-in amplification algorithm to separate the real part and imaginary part of the impedance corresponding to each excitation frequency, and calculate the impedance amplitude and impedance phase angle at the corresponding frequency.

[0085] S54. Based on multi-frequency impedance phase characteristics, a quantitative model for grain boundary conductivity anomaly is established. Taking the impedance phase angle corresponding to the selected high-frequency excitation point as the core feature, the impedance phase shift of different detection points relative to the standard sample without anomalies is calculated. Through conductivity inversion algorithm, the impedance phase shift is converted into the grain boundary conductivity change value, and the quantitative extraction of grain boundary conductivity anomaly features is completed. The phase features are accurately extracted through multi-frequency impedance analysis, and the grain boundary conductivity change is inverted based on the phase shift, so as to achieve high sensitivity and accurate identification of profile grain boundary structure anomalies.

[0086] The multi-frequency impedance phase analysis and the calculation of grain boundary conductivity anomalies are implemented through the following logic:

[0087] For any excitation frequency, calculate the ratio of the imaginary part to the real part of the impedance of the eddy current response signal, and then calculate the arctangent of this ratio to obtain the eddy current impedance phase angle at the corresponding excitation frequency.

[0088] The impedance phase shift is obtained by calculating the impedance phase angle difference between the test point of the industrial profile under test and the standard sample without grain boundary conductivity abnormality at the same selected specific high-frequency excitation frequency point.

[0089] Based on the phase-conductivity linear conversion relationship obtained in advance through standard sample calibration, the above impedance phase offset is converted into the grain boundary conductivity change of the detection point relative to the standard sample. The accurate impedance phase angle is obtained through rigorous phase calculation logic, and the conductivity change is inverted based on the calibrated linear relationship, thereby improving the reliability and detection accuracy of grain boundary anomaly detection.

[0090] This invention provides an embodiment 2

[0091] A quality inspection system for industrial profiles after aging treatment includes a model import and trajectory planning module, a scanning execution and signal acquisition module, a structural interference decoupling module, a residual stress imaging module, a microscopic feature extraction module, and a comprehensive quality assessment module.

[0092] The model import and trajectory planning module is used to acquire the three-dimensional geometric model of the industrial profile to be tested, import it into the control system as prior information, and plan and generate a conformal non-contact adaptive scanning trajectory covering the surface of the industrial profile based on the spatial position information of the three-dimensional geometric model.

[0093] The scanning execution and signal acquisition module is used to control the multi-axis scanning device equipped with a composite detection probe to perform synchronous joint scanning of the time-treated industrial profiles along the conformal non-contact adaptive scanning trajectory, and to acquire the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal of the industrial profiles in real time.

[0094] The structural interference decoupling module is used to call the morphology and stress decoupling algorithm, perform position matching based on the three-dimensional geometric model, compensate for and filter out the structural interference signals generated by the sudden changes in wall thickness and the reflection of the inner cavity edge of the industrial profile, and obtain pure multi-frequency pulse eddy current signals and pure nonlinear ultrasonic signals.

[0095] The residual stress imaging module is used to extract features from the pure nonlinear ultrasonic signal and the pure multi-frequency pulsed eddy current signal respectively, separate the sound velocity change features and electromagnetic variation features that characterize the residual stress of the material, and convert them into spatial stress distribution data to generate a three-dimensional residual stress cloud map.

[0096] The micro-feature extraction module is used to simultaneously perform cross-scale feature mapping and micro-physicochemical property evaluation, extract high-order harmonic features characterizing dislocation density from the pure nonlinear ultrasonic signal, and extract high-frequency impedance phase features characterizing grain boundary conductivity anomalies from the pure multi-frequency pulse eddy current signal.

[0097] The comprehensive quality assessment module is used to construct a joint feature vector by concatenating the extracted high-order harmonic features and the high-frequency impedance phase features, and input it into a pre-trained machine learning prediction model. It outputs the intergranular corrosion sensitivity index and fatigue life loss rate of each scanning point of the industrial profile, and maps the performance indicators of the points to the stress data of the corresponding spatial coordinates in the three-dimensional residual stress cloud map. Finally, it outputs a comprehensive quality inspection report of the industrial profile. Through a modular collaborative architecture, it realizes the full-process automation of the time-sensitive profile inspection, ensuring the accuracy of multi-physics field inspection while significantly improving the efficiency and consistency of batch inspection of industrial profiles.

[0098] This invention provides an application example 1

[0099] The S2 spatiotemporal synchronization acquisition mechanism is constrained and calibrated using the following mathematical formula: Spatiotemporal synchronization error constraint formula:

[0100]

[0101] in, This represents the maximum acquisition time difference between the pulsed eddy current signal and the nonlinear ultrasonic signal. This refers to the signal sampling time of the pulse eddy current detection unit. The signal sampling time of the nonlinear ultrasonic detection unit. The frequency of the synchronous reference clock;

[0102] The one-to-one mapping relationship between data collection points and spatial locations is defined by the mapping function F:

[0103]

[0104] in, To scan the trajectory of the first The three-dimensional spatial coordinates of each acquisition trigger point These are the X, Y, and Z coordinates of the trigger point in the workpiece coordinate system. For the first The signal sequence output by the pulse eddy current detection unit at each acquisition trigger point. For the first The signal sequence output by the nonlinear ultrasonic detection unit acquired at each acquisition trigger point. The data set includes the output signals of the two types of detection units at that point; the formula for the overlap constraint of the detection region center is:

[0105]

[0106] in, The spatial deviation between the detection area centers of the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit is given. The coordinates of the center of the eddy current detection area in the probe coordinate system are: The coordinates of the center of the nonlinear ultrasonic testing region in the probe coordinate system are: The preset center space deviation threshold;

[0107] The coordinate transformation and structural disturbance decoupling algorithm of S3 is implemented through the following mathematical formula:

[0108] Formula for spatial coordinate transformation from probe coordinate system to workpiece coordinate system:

[0109]

[0110] in, To acquire the three-dimensional coordinate vectors of the acquisition points in the workpiece coordinate system, To obtain the three-dimensional coordinate vector of the acquisition point in the probe coordinate system, The 3×3 rotation matrix is ​​the ratio of the probe coordinate system to the workpiece coordinate system, calculated in real time from the rotation angles of each axis of the multi-axis scanning device. The 3×1 translation vector of the probe coordinate system origin relative to the workpiece coordinate system origin is calculated in real time from the linear displacement of each axis of the multi-axis scanning device.

[0111] Formula for coherent cancellation of structural interference signals:

[0112]

[0113] in, This is the filtered, clean signal sequence. For the first The time-domain sampling sequence of the measured original multi-frequency pulse eddy current signal or nonlinear ultrasonic signal at each acquisition trigger point. This refers to the sampling point number within the time-domain sequence. A template sequence of structural interference signals that matches the current acquisition point;

[0114] Optimal weight calculation and filtering formula for adaptive matched filtering:

[0115]

[0116] in, These are the weight variables of the adaptive filter, used to adjust the amplitude and phase of the interference signal template to match the interference components in the measured signal. The optimal weights for adaptive filtering are... This represents the total number of time-domain sampling points corresponding to a single acquisition trigger point. For the corresponding template in the template library A standard sequence of structural interference signals for each acquisition trigger point;

[0117] The extraction of the nonlinear features and their physical mapping to dislocation density are achieved through the following mathematical formula:

[0118] Formula for extracting the proportional characteristic parameters of nonlinear ultrasound core:

[0119]

[0120] in, The core proportional characteristic parameter characterizing material nonlinearity can be used to derive the material's relative nonlinearity coefficient. The amplitude of the fundamental signal. The amplitude of the second harmonic signal;

[0121] The quantitative mapping formula between the relative nonlinear coefficient and dislocation density is as follows:

[0122]

[0123] in, The relative nonlinear coefficient is derived based on the core proportional characteristic parameter. Let be the intrinsic nonlinear coefficient of an ideal single-crystal material without dislocations, and be an intrinsic constant of the material. The dislocation nonlinearity contribution coefficient is obtained by calibrating the material's elastic modulus, dislocation type, and average pinning point spacing. The dislocation density at the test points of the industrial profile to be tested;

[0124] The multi-frequency impedance phase analysis and the calculation of grain boundary conductivity anomalies are achieved through the following mathematical formulas:

[0125] Formula for calculating the phase angle of pulsed eddy current impedance:

[0126]

[0127] in, Excitation frequency The corresponding eddy current impedance phase angle, Excitation frequency The imaginary part of the impedance of the eddy current response signal. Excitation frequency The real part of the impedance of the lower eddy current response signal;

[0128] Formula for calculating impedance phase shift corresponding to grain boundary conductivity anomaly:

[0129]

[0130] in, For a specific high-frequency excitation point The corresponding impedance phase offset, The testing point for the industrial profile under test is located at this high-frequency excitation point. The impedance phase angle below, For standard samples without grain boundary conductivity anomalies at the same high-frequency excitation frequency point Standard impedance phase angle below;

[0131] Quantitative inversion formula for impedance phase shift versus grain boundary conductivity change:

[0132]

[0133] in, This represents the change in grain boundary conductivity at the test point of the industrial profile under test relative to the standard sample. The phase-conductivity calibration coefficient is obtained by linearly fitting the conductivity and impedance phase test data of the standard sample. For calibration constants;

[0134] This invention provides an application example 2

[0135] The specific algorithm in S1 is as follows:

[0136] Scan trajectory planning algorithm formula:

[0137]

[0138]

[0139]

[0140] in:

[0141] This is a set of conformal non-contact adaptive scanning trajectories; To scan the trajectory of the first One data collection trigger point; For the first The three-dimensional spatial coordinates of each trajectory point in the workpiece coordinate system; For the first The unit normal vector of the profile surface at each trajectory point; This represents the total number of points on the scanned trajectory. This represents the straight-line distance between two adjacent trajectory points. The preset equal spatial step size; For the first The lifting distance between the composite detection probe and the profile surface at each trajectory point; For the first Coordinates of the center point of the end face of the composite detection probe at each trajectory point; The preset standard lift-off distance;

[0142] Execution logic steps:

[0143] The processor discretizes the imported three-dimensional geometric model of the industrial profile under test into a surface point cloud with normal vector information, thus completing the calibration of the workpiece coordinate system.

[0144] According to the preset fixed scanning step size Generate initial trajectory points with equal step size on the point cloud of the profile surface to ensure that the trajectory covers the entire surface of the profile to be inspected;

[0145] For each initial trajectory point, offset outwards by a standard lift-off distance along the surface normal vector of that point. This allows us to obtain the coordinates of the center point of the probe end face, ensuring a constant lift-off distance during the scanning process.

[0146] All trajectory points are smoothly interpolated to generate the final conformal non-contact adaptive scanning trajectory, which is then output to the motion control system of the multi-axis scanning device.

[0147] The specific algorithms for S21-S23 are as follows:

[0148] Clock synchronization error constraint formula:

[0149]

[0150] in: The maximum acquisition time difference between the multi-frequency pulsed eddy current signal and the nonlinear ultrasonic signal; This refers to the signal sampling time of the pulse eddy current detection unit. The signal sampling time of the nonlinear ultrasonic detection unit; The frequency of the synchronous reference clock;

[0151] Execution logic steps:

[0152] The high-precision clock source outputs a preset high-frequency reference clock signal, which is connected to the clock trigger ports of the position encoder, pulse eddy current detection unit, and nonlinear ultrasonic detection unit of the motion control system to complete the clock homogeneity configuration.

[0153] The processor acquires the sampling times of the pulse eddy current detection unit and the nonlinear ultrasonic detection unit in real time and calculates the time deviation between them.

[0154] The calculated maximum sampling time deviation is compared with the reference clock period. If the deviation exceeds the constraint range, the sampling trigger delay of the two detection units is recalibrated until the time synchronization constraint is met.

[0155] Formula for the mapping relationship between spatial location and signal:

[0156]

[0157]

[0158] in: This is a one-to-one mapping function between the acquisition points and the detection signals; To scan the trajectory of the first The three-dimensional spatial coordinates of each acquisition trigger point; For the first The multi-physics joint signal vector corresponding to each acquisition point; For the first A sequence of multi-frequency pulsed eddy current signals collected at each sampling point; For the first Nonlinear ultrasound signal sequences acquired at each acquisition point;

[0159] Execution logic steps:

[0160] The motion control system drives the composite detection probe to move along the scanning trajectory, and the position encoder collects the probe's current spatial coordinates in real time.

[0161] When the probe moves to the... When a single acquisition trigger point is reached, the position encoder synchronously outputs a trigger signal to both detection units.

[0162] The two detection units synchronously complete the signal acquisition at this point. The processor combines the acquired eddy current signal sequence and ultrasonic signal sequence into a joint signal vector, establishes a unique mapping relationship between the coordinates of this point and the joint signal vector, and stores it.

[0163] Formula for constraining the center overlap of the detection area:

[0164]

[0165] in: The spatial deviation of the detection area center between the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit; The coordinates of the center of the pulsed eddy current detection area in the probe coordinate system are: The coordinates of the center of the nonlinear ultrasonic testing region in the probe coordinate system are: The preset center space deviation threshold;

[0166] Execution logic steps:

[0167] Before the scan begins, the processor obtains the planar coordinates of the center of the detection area of ​​the two detection units in the probe coordinate system by calibrating the sample;

[0168] Calculate the planar geometric distance between the two detection centers to obtain the detection center deviation;

[0169] The detection center deviation is compared with the preset center deviation allowable threshold. If the deviation exceeds the threshold, the installation position of the two detection units on the probe end face is adjusted until the overlap constraint is met.

[0170] The specific algorithms for S31-S33 are as follows:

[0171] Formula for spatial coordinate transformation from probe coordinate system to workpiece coordinate system:

[0172]

[0173] in: To collect the three-dimensional coordinate vector of the acquisition point in the workpiece coordinate system; This refers to the three-dimensional coordinate vector of the acquisition point in the probe coordinate system. This is a 3×3 rotation matrix of the probe coordinate system relative to the workpiece coordinate system; This is a 3×1 translation vector relative to the origin of the probe coordinate system and the origin of the workpiece coordinate system.

[0174] Execution logic steps:

[0175] The processor uses the forward motion algorithm of the multi-axis scanning device to calculate the coordinate rotation transformation matrix in real time based on the rotation angle of each axis. The coordinate translation transformation vector is calculated in real time based on the linear displacement of each axis. ;

[0176] Substitute the coordinates of the acquisition point in the probe coordinate system into the coordinate transformation formula to calculate the coordinates of the point in the workpiece coordinate system.

[0177] The transformed workpiece coordinate system point coordinates are matched with the coordinates of the three-dimensional geometric model to complete the spatial binding of the acquired signal and the three-dimensional model.

[0178] Formula for calculating the optimal weights of adaptive matched filtering:

[0179]

[0180] in: These are the weight variables for the adaptive filter, used to adjust the amplitude and phase of the interference signal template; The optimal weights for adaptive filtering; This represents the total number of time-domain sampling points corresponding to a single acquisition trigger point. The sampling point number within a single-point time-domain signal sequence; For the first Time-domain sampling sequence of the measured original multi-frequency pulse eddy current signal or nonlinear ultrasonic signal at each acquisition trigger point; For the corresponding template in the template library A standard sequence of structural interference signals for each acquisition trigger point;

[0181] Execution logic steps:

[0182] The processor is for the first For each sampling point, the corresponding interference template sequence is retrieved from the structural interference signal template library;

[0183] With the goal of minimizing the sum of squared errors between the original measured signal and the weighted interference template, the optimal filtering weights are calculated through an iterative optimization algorithm.

[0184] The amplitude and phase of the interference template sequence are adjusted by using the optimal filtering weights to obtain an adjusted interference template sequence that matches the interference components of the measured signal.

[0185] Formula for calculating coherent cancellation of structural interference:

[0186]

[0187]

[0188] in: For the first The purified signal sequence after filtering at each acquisition point; This is the structural interference signal sequence after optimal weight adjustment;

[0189] Execution logic steps:

[0190] The processor substitutes the optimal filter weights into the formula to calculate the matched interference signal sequence;

[0191] The pure detection signal sequence is obtained by subtracting the matched interference signal sequence point by point from the original measured signal sequence;

[0192] Repeat the above calculations for all acquisition points to complete the structural interference filtering of all detection signals;

[0193] The specific calculation logic in S4 is as follows:

[0194] Formula for residual stress-related feature extraction algorithm:

[0195]

[0196]

[0197] in: The change in ultrasonic velocity; The velocity of the longitudinal wave propagating at the point to be measured is the ultrasonic velocity. The ultrasonic longitudinal wave propagation velocity of a stress-free standard specimen; The ultrasonic velocity-stress calibration coefficient; The residual stress value at the test point; This represents the change in eddy current impedance. The amplitude of the multi-frequency pulsed eddy current impedance at the point to be measured; The amplitude of the multi-frequency pulsed eddy current impedance of the stress-free standard specimen; This refers to the eddy current impedance-stress calibration factor.

[0198] Execution logic steps:

[0199] The processor calculates the measured ultrasonic velocity at the point to be measured from the pure nonlinear ultrasonic signal, and compares it with the standard stress-free ultrasonic velocity to obtain the change in sound velocity.

[0200] The measured eddy current impedance amplitude at the test point is calculated from the pure multi-frequency pulsed eddy current signal and compared with the standard stress-free eddy current impedance amplitude to obtain the impedance change.

[0201] Based on the pre-calibrated ultrasonic stress sensitivity coefficient and eddy current stress sensitivity coefficient, the residual stress value of the test point is calculated by the change in sound velocity and the change in impedance, respectively, and the average value of the two is taken as the final residual stress value of the point.

[0202] Formula for generating 3D residual stress cloud map:

[0203]

[0204] in: This is a dataset of the spatial distribution of residual stress across the entire detection area. For the first The final residual stress value at each sampling point;

[0205] Execution logic steps:

[0206] The processor combines the three-dimensional coordinates of all the collected points with the corresponding residual stress values ​​to construct a spatial distribution set of residual stress.

[0207] A triangular meshing interpolation algorithm is used to spatially interpolate the residual stress values ​​at discrete points to obtain continuous residual stress distribution data across the entire surface of the profile.

[0208] Based on the interpolated continuous distribution data, a pseudo-color mapping method is used to generate and store a three-dimensional residual stress cloud map of the industrial profile under test.

[0209] The specific calculation logic for S51-S52 is as follows:

[0210] Nonlinear ultrasound core proportion feature extraction:

[0211]

[0212]

[0213]

[0214] in: The core proportional characteristic parameter for characterizing material nonlinearity; The fundamental amplitude of the nonlinear ultrasonic signal; The amplitude of the second harmonic of the nonlinear ultrasonic signal; The wavenumber is the fundamental frequency of the ultrasound. Pi; The fundamental frequency of the ultrasonic excitation signal; The propagation speed of ultrasonic surface waves in the industrial profile under test; The fixed center-to-center distance between the ultrasonic transmitting transducer and the receiving transducer; The relative nonlinear ultrasonic coefficient of the material;

[0215] Execution logic steps:

[0216] The processor performs a fast Fourier transform on the pure nonlinear ultrasonic time-domain signal, converting it into a frequency-domain spectral signal;

[0217] Extract the fundamental frequency amplitude from the spectral signal. The second harmonic amplitude corresponding to the second harmonic frequency ;

[0218] Substituting into the formula, the nonlinear core proportional characteristics are calculated. By combining the fundamental wavenumber and the transducer center spacing, the relative nonlinear coefficient was calculated. ;

[0219] Quantitative mapping formula for dislocation density:

[0220]

[0221]

[0222] in: is the intrinsic nonlinear coefficient of the dislocation-free ideal single-crystal material, and is the intrinsic constant of the material; The nonlinear contribution coefficient of dislocation is determined by the material's elastic modulus, dislocation type, and average spacing of pinning points. The dislocation density at the test points of the industrial profile to be tested;

[0223] Execution logic steps:

[0224] The processor retrieves the pre-calibrated material inherent nonlinearity coefficients and dislocation contribution coefficients;

[0225] Substitute the calculated relative nonlinear coefficients into the formula to calculate the dislocation density at the test point.

[0226] Establish the mapping relationship between the three-dimensional coordinates of the test point and the dislocation density, and complete the quantitative extraction of the dislocation density-related higher harmonic features;

[0227] The specific calculation logic for S53-S54 is as follows:

[0228] Formula for calculating the phase angle of pulsed eddy current impedance:

[0229]

[0230] in: Excitation frequency The corresponding eddy current impedance phase angle; The excitation frequency of the pulsed eddy current; Excitation frequency The imaginary part of the impedance of the lower eddy current response signal; Excitation frequency The real part of the impedance of the lower eddy current response signal;

[0231] Execution logic steps: For the selected high-frequency excitation frequency point, the processor uses an orthogonal lock-in amplification algorithm to separate the real part and imaginary part of the impedance at the corresponding frequency from the pure multi-frequency pulse eddy current signal;

[0232] Calculate the ratio of the imaginary part of the impedance to the real part of the impedance, and then calculate the arctangent of this ratio to obtain the eddy current impedance phase angle at the corresponding high-frequency excitation frequency.

[0233] Impedance phase shift calculation and grain boundary conductivity inversion formula:

[0234]

[0235]

[0236] in: For a specific high-frequency excitation point; This refers to the impedance phase shift at the high-frequency excitation frequency. The impedance phase angle of the test point of the industrial profile under test at the high-frequency excitation frequency; The standard impedance phase angle of a standard sample without grain boundary conductivity anomalies at the same high-frequency excitation frequency; This represents the change in grain boundary conductivity at the test site relative to the standard sample. The phase-conductivity calibration coefficient is obtained by linearly fitting the conductivity and impedance phase test data of the standard sample; For calibration constants;

[0237] Execution logic steps:

[0238] The processor retrieves the standard impedance phase angle of a standard sample without grain boundary conductivity anomalies at the same high-frequency excitation frequency.

[0239] Calculate the difference between the impedance phase angle at the point to be measured and the standard impedance phase angle to obtain the impedance phase offset;

[0240] Based on the pre-calibrated phase-conductivity conversion coefficient and calibration constant, the impedance phase offset is substituted into the formula to calculate the change in grain boundary conductivity at the test point, thus completing the quantitative extraction of grain boundary conductivity anomaly characteristics.

[0241] The specific calculation logic in S6 is as follows:

[0242] Formula for constructing the function from joint eigenvectors:

[0243]

[0244] in: For the first The joint feature vector of each collection point is constructed by concatenating and splicing them together. For the first Dislocation density at each sampling point; For the first The change in grain boundary conductivity at each sampling point;

[0245] Execution logic steps:

[0246] The processor is for the first Each sampling point is used to retrieve the changes in dislocation density and grain boundary conductivity at that point.

[0247] The two features are combined into a one-dimensional joint feature vector by concatenating them.

[0248] Repeat the above operation for all acquisition points to generate a joint feature vector set for the entire detection area;

[0249] Quantitative prediction formula for service performance:

[0250]

[0251] in: For the first Intergranular corrosion sensitivity index at each sampling point; For the first Fatigue life loss rate at each sampling point; For pre-trained machine learning prediction models;

[0252] Execution logic steps:

[0253] The processor retrieves a pre-trained machine learning prediction model;

[0254] The first The joint feature vector of each collection point is input into the model, and the model outputs the intergranular corrosion sensitivity index and fatigue life loss rate of that point.

[0255] Repeat the above prediction operation for all collection points to obtain the service performance index matrix data of the entire detection area;

[0256] In summary, this invention plans a conformal non-contact adaptive scanning trajectory based on the three-dimensional geometric model of the industrial profile to be tested. It can perfectly adapt to complex cross-section industrial profiles with internal cavities, variable wall thickness, and irregular curved surfaces, achieving full-surface coverage of the profile without dead angles. It completely solves the problems of large fluctuations in lifting distance and inability to completely detect irregular and non-visible areas in existing fixed trajectory scanning methods.

[0257] By using a multi-axis scanning device equipped with an integrated composite detection probe, synchronous joint scanning of multi-frequency pulsed eddy current and nonlinear ultrasonic signals is achieved. With the synchronous triggering mechanism based on the same clock source, the precise one-to-one correspondence between the scanning spatial position and the multi-physics field detection signal is ensured. This avoids spatial misalignment and temporal deviation of signals from different detection methods, significantly improving the consistency and matching degree of the detection data. At the same time, non-contact online detection is realized, avoiding the industry pain points of long cycle, high cost and inability to cover the entire batch in destructive sampling inspection.

[0258] By employing a morphology and stress decoupling algorithm, the system accurately matches the acquired signals with the spatial position based on a three-dimensional geometric model. This algorithm can specifically compensate for and filter out structural interference signals caused by abrupt changes in profile wall thickness, reflections at the inner cavity edges, and corner structures. It precisely decouples the interference caused by the profile geometry from the effective signals generated by changes in the material's own properties, thus completely solving the core problems in existing technologies where effective detection signals are submerged by structural interference, the detection signal-to-noise ratio is low, and the results are easily distorted. This provides highly reliable raw data support for the subsequent accurate quantitative characterization of material properties.

[0259] It can simultaneously and accurately generate three-dimensional residual stress cloud maps of profiles and quantitatively extract microscopic features of dislocation density and grain boundary conductivity anomalies, realizing cross-scale correlation mapping between macroscopic stress distribution and microstructure features, breaking through the limitation of existing technologies that can only achieve qualitative identification of macroscopic defects. At the same time, it can quantitatively predict the intergranular corrosion sensitivity and fatigue life loss rate of profiles through machine learning prediction models, and complete the spatial multidimensional mapping between service performance indicators and three-dimensional residual stress cloud maps, realizing the upgrade from single qualification judgment to full-dimensional performance evaluation. The test results can directly provide data support for the optimization of aging process, fully meeting the stringent requirements of high-end equipment field for the whole chain quality control of industrial profiles.

[0260] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A method for quality inspection of industrial profiles after aging treatment, characterized in that: Includes the following steps: S1. Scan trajectory planning: Obtain the three-dimensional geometric model of the industrial profile to be tested, import it into the control system as prior information, and plan and generate a conformal non-contact adaptive scanning trajectory covering the surface of the industrial profile based on the spatial position information of the three-dimensional geometric model. S2. Synchronous joint scanning: Using a multi-axis scanning device equipped with a composite detection probe, the time-treated industrial profile is simultaneously and jointly scanned along the conformal non-contact adaptive scanning trajectory, and the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal of the industrial profile are collected in real time. S3. Structural Interference Decoupling: The topography and stress decoupling algorithm is invoked to perform position matching based on the three-dimensional geometric model, compensate for and filter out the structural interference signals generated by the sudden changes in wall thickness and the reflection of the inner cavity edge of the industrial profile, and obtain pure multi-frequency pulse eddy current signals and pure nonlinear ultrasonic signals. S4. Residual stress imaging: Feature extraction is performed on the pure nonlinear ultrasonic signal and the pure multi-frequency pulsed eddy current signal respectively to separate the sound velocity change characteristics and electromagnetic variation characteristics that characterize the residual stress of the material, and convert them into spatial stress distribution data to generate a three-dimensional residual stress cloud map. S5. Microscopic feature extraction: Simultaneously perform cross-scale feature mapping and microscopic physicochemical property evaluation, extract high-order harmonic features characterizing dislocation density from the pure nonlinear ultrasonic signal, and extract high-frequency impedance phase features characterizing grain boundary conductivity anomalies from the pure multi-frequency pulse eddy current signal. S6. Comprehensive Quality Assessment: The extracted high-order harmonic features and high-frequency impedance phase features are connected in series to construct a joint feature vector, which is then input into a pre-trained machine learning prediction model. The model outputs the intergranular corrosion sensitivity index and fatigue life loss rate of each scanning point of the industrial profile. The performance indicators of the points are mapped one-to-one with the stress data of the corresponding spatial coordinates in the three-dimensional residual stress cloud map. Finally, a comprehensive quality inspection report of the industrial profile is output.

2. The quality inspection method for industrial profiles after aging treatment according to claim 1, characterized in that: S2 specifically includes the following steps: S21. The motion control system of the multi-axis scanning device and the signal acquisition module of the composite detection probe are configured to be synchronized by the same clock source. The preset high-frequency reference clock signal provided by the same high-precision clock source is used to provide synchronous trigger clocks for the position encoder of the motion control axis, the pulse eddy current detection unit, and the nonlinear ultrasonic detection unit, respectively, to ensure that the spatiotemporal synchronization error between the scanning position and the signal acquisition does not exceed the preset spatiotemporal synchronization error threshold. S22. Set position-triggered synchronous acquisition logic in the composite detection probe. Pre-set acquisition trigger points on the conformal non-contact adaptive scanning trajectory with equal spatial step size. When the multi-axis scanning device drives the composite detection probe to each acquisition trigger point, the position encoder synchronously outputs a trigger signal, which synchronously triggers the pulse eddy current detection unit and the nonlinear ultrasonic detection unit to complete the synchronous acquisition of the signal at that point, realizing a one-to-one correspondence between the scanning position and the multi-physics field signal. S23. The composite detection probe adopts a non-contact coplanar integrated architecture, integrating the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit on the same detection end face. The pulsed eddy current detection unit includes a multi-frequency pulsed eddy current detection coil, and the nonlinear ultrasonic detection unit includes a nonlinear ultrasonic electromagnetic transducer, ensuring that the detection areas of the two types of detection units coincide and eliminating the spatial position deviation of multi-physics field signals.

3. The quality inspection method for industrial profiles after aging treatment according to claim 2, characterized in that: The spatiotemporal synchronization acquisition mechanism of S2 is constrained and calibrated through the following logic: The signal sampling time difference between the pulsed eddy current detection unit and the nonlinear ultrasonic detection unit is compared, and this difference is limited to being less than the period of the synchronous reference clock. Establish a unique correspondence between the three-dimensional spatial coordinates of the acquisition points and the output signal sets of the two types of detection units; Calculate the planar geometric distance between the center of the detection area of ​​the two types of detection units in the probe coordinate system, and limit this distance to be less than a preset center spatial deviation threshold.

4. The quality inspection method for industrial profiles after aging treatment according to claim 1, characterized in that: S3 specifically includes the following steps: S31. Establish a unified three-dimensional spatial coordinate system, convert the three-dimensional geometric model of the industrial profile to be tested into a standard digital twin model under the workpiece coordinate system, and convert the coordinates of the acquisition points under the probe coordinate system to the workpiece coordinate system through the motion forward solution algorithm of the multi-axis scanning device for the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal collected at each point during the scanning process, so as to complete the accurate matching of the spatial position of the acquired signal and the three-dimensional geometric model. S32, based on the geometric features of the three-dimensional geometric model, pre-generates a structural interference signal template library. For the locations of sudden changes in wall thickness, inner cavity edges, and corner feature areas of industrial profiles, the pulse eddy current edge effect interference signal and nonlinear ultrasonic interface reflection interference signal at the corresponding locations are calculated by numerical simulation or experimental calibration methods, and stored as a structural interference signal template library according to spatial coordinate index. S33 invokes the topography and stress decoupling algorithm. For each spatially matched acquisition point, it calls the interference signal template corresponding to the coordinates from the structural interference signal template library. It then uses an adaptive matched filtering and coherent cancellation algorithm to filter out the structural interference signal from the measured signal, obtaining a pure multi-frequency pulsed eddy current signal and a pure nonlinear ultrasonic signal that are only related to the material properties.

5. The quality inspection method for industrial profiles after aging treatment according to claim 4, characterized in that: The coordinate transformation and structural disturbance decoupling algorithm of S3 is implemented through the following logic: By combining the rotation angle and linear displacement of the probe, the coordinates of the acquisition points in the probe coordinate system are transformed to the workpiece coordinate system through spatial translation and rotation matrix operations. For each acquisition trigger point, the optimal weight of the adaptive matched filter is calculated through iterative optimization to adjust the amplitude and phase of the interference signal template so that it matches the interference components in the measured signal. Then, the interference template signal after optimal weight adjustment is subtracted from the measured original signal to complete the filtering of structural interference.

6. The quality inspection method for industrial profiles after aging treatment according to claim 1, characterized in that: The extraction of high-order harmonic features characterizing dislocation density in S5 specifically includes the following steps: S51. Perform time-domain to frequency-domain conversion on the pure nonlinear ultrasound signal. Use Fast Fourier Transform to convert the time-domain ultrasound signal into a frequency-domain spectrum signal. Accurately extract the fundamental amplitude corresponding to the fundamental frequency and the second harmonic amplitude corresponding to the second harmonic frequency from the spectrum signal. Then, use window function filtering to eliminate spectrum leakage and noise interference. S52. Based on the extracted fundamental amplitude and second harmonic amplitude, extract the nonlinear core proportional features of the material, establish a quantitative mapping relationship between the nonlinear features and the dislocation density inside the material, and convert the nonlinear features into the dislocation density values ​​of the corresponding detection points through the calibrated mapping model, thus completing the quantitative extraction of dislocation density-related higher harmonic features.

7. The quality inspection method for industrial profiles after aging treatment according to claim 6, characterized in that: The extraction of the nonlinear features and their physical mapping to dislocation density are achieved through the following logic: The ratio of the second harmonic amplitude to the square of the fundamental amplitude is extracted as the core proportional feature characterizing the nonlinearity of the material. Based on this core feature, the relative nonlinear coefficient of the material can be derived. Based on the inherent nonlinear coefficient of a dislocation-free ideal single-crystal material, and combined with the dislocation nonlinear contribution coefficient obtained by calibration of the material's elastic modulus, dislocation type, and average spacing of pinning points, the relative nonlinear coefficient is linearly mapped to the dislocation density at the detection point.

8. The quality inspection method for industrial profiles after aging treatment according to claim 1, characterized in that: The high-frequency impedance phase feature extraction in S5, which characterizes the abnormal conductivity of grain boundaries, specifically includes the following steps: S53. Perform multi-frequency impedance analysis on the pure multi-frequency pulsed eddy current signal. For the eddy current response signal under different excitation frequencies, use the quadrature lock-in amplification algorithm to separate the real part and imaginary part of the impedance corresponding to each excitation frequency, and calculate the impedance amplitude and impedance phase angle at the corresponding frequency. S54. Based on the multi-frequency impedance phase characteristics, a quantitative model of grain boundary conductivity anomaly is established. Taking the impedance phase angle corresponding to the selected high-frequency excitation point as the core feature, the impedance phase shift of different detection points relative to the standard sample without anomalies is calculated. Through the conductivity inversion algorithm, the impedance phase shift is converted into the grain boundary conductivity change value, thus completing the quantitative extraction of grain boundary conductivity anomaly characteristics.

9. A method for quality inspection of industrial profiles after aging treatment according to claim 8, characterized in that: The multi-frequency impedance phase analysis and the calculation of grain boundary conductivity anomalies are implemented through the following logic: For any excitation frequency, calculate the ratio of the imaginary part to the real part of the impedance of the eddy current response signal, and then calculate the arctangent of this ratio to obtain the eddy current impedance phase angle at the corresponding excitation frequency. The impedance phase shift is obtained by calculating the impedance phase angle difference between the test point of the industrial profile under test and the standard sample without grain boundary conductivity abnormality at the same selected specific high-frequency excitation frequency point. Based on the phase-conductivity linear conversion relationship obtained in advance through standard sample calibration, the above impedance phase offset is converted into the change in grain boundary conductivity of the detection point relative to the standard sample.

10. A quality inspection system for industrial profiles after aging treatment, characterized in that: It includes a model import and trajectory planning module, a scan execution and signal acquisition module, a structural interference decoupling module, a residual stress imaging module, a microscopic feature extraction module, and a comprehensive quality assessment module; The model import and trajectory planning module is used to acquire the three-dimensional geometric model of the industrial profile to be tested, import it into the control system as prior information, and plan and generate a conformal non-contact adaptive scanning trajectory covering the surface of the industrial profile based on the spatial position information of the three-dimensional geometric model. The scanning execution and signal acquisition module is used to control the multi-axis scanning device equipped with a composite detection probe to perform synchronous joint scanning of the time-treated industrial profiles along the conformal non-contact adaptive scanning trajectory, and to acquire the multi-frequency pulse eddy current signal and nonlinear ultrasonic signal of the industrial profiles in real time. The structural interference decoupling module is used to call the morphology and stress decoupling algorithm, perform position matching based on the three-dimensional geometric model, compensate for and filter out the structural interference signals generated by the sudden changes in wall thickness and the reflection of the inner cavity edge of the industrial profile, and obtain pure multi-frequency pulse eddy current signals and pure nonlinear ultrasonic signals. The residual stress imaging module is used to extract features from the pure nonlinear ultrasonic signal and the pure multi-frequency pulsed eddy current signal respectively, separate the sound velocity change features and electromagnetic variation features that characterize the residual stress of the material, and convert them into spatial stress distribution data to generate a three-dimensional residual stress cloud map. The micro-feature extraction module is used to simultaneously perform cross-scale feature mapping and micro-physicochemical property evaluation, extract high-order harmonic features characterizing dislocation density from the pure nonlinear ultrasonic signal, and extract high-frequency impedance phase features characterizing grain boundary conductivity anomalies from the pure multi-frequency pulse eddy current signal. The comprehensive quality assessment module is used to construct a joint feature vector by concatenating the extracted high-order harmonic features and the high-frequency impedance phase features in series. This vector is then input into a pre-trained machine learning prediction model, which outputs the intergranular corrosion sensitivity index and fatigue life loss rate for each scanning point of the industrial profile. The module also maps the performance indicators of the points to the stress data of the corresponding spatial coordinates in the three-dimensional residual stress cloud map, and finally outputs a comprehensive quality inspection report for the industrial profile.

Images