Ultrasonic-based weld defect detection method and system

Abstract

The invention relates to the technical field of ultrasonic detection, and discloses an ultrasonic-based weld defect detection method and system.The method comprises the steps that a historical database is established, ultrasonic echo signals are traversed in the historical database, a reuse defect strategy or an ultrasonic verification strategy is determined according to the traversal result, and when the reuse defect strategy or the ultrasonic verification strategy is determined, the ultrasonic verification strategy is determined; determining a first ultrasonic result based on the echo delay and the echo intensity, determining a second ultrasonic result based on the ultrasonic echo signal and the ultrasonic feature model, determining whether the weld joint to be detected has defects or not according to the first ultrasonic result and the second ultrasonic result, and performing Fourier transform on the ultrasonic echo signal to obtain the weld joint to be detected. The defect type of the weld joint to be detected is determined based on the ultrasonic characteristic spectrogram, the attenuation result is determined based on the three-dimensional model and the metal sample and written into the simulation database of the simulation environment, the defect type is verified based on the relation between the ultrasonic characteristic spectrogram and the simulation spectrum curve, and the reliability and stability of ultrasonic detection are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic testing, and in particular to an ultrasonic-based weld defect detection method and system. Background Art

[0002] In industrial fields such as petrochemicals, aerospace, and rail transportation, welding is often used to connect components. The quality of the weld directly determines the safety and reliability of the structure of the connected components. The weld formation process is affected by multiple factors such as welding process parameters, material properties, and operating environment, and may have certain defects. If these defects are not discovered and handled in time, they will cause serious accidents such as leakage and fracture.

[0003] At present, weld defect detection usually relies on the experience accumulation and subjective judgment of inspectors. There are differences in the technical level and practical experience of different inspectors. When performing defect detection on welds of metal materials, the obtained ultrasonic signal will deviate from the judgment of the defect type, resulting in misjudgment or omission of the defect type, thereby affecting the reliability and stability of the detection results. Moreover, with the increasing complexity of industrial products and the increasing requirements for detection accuracy, relying on human experience is difficult to meet the requirements of high-precision and intelligent quality control. Therefore, it is necessary to design an ultrasonic-based weld defect detection method and system to solve the problems existing in current technology. Summary of the Invention

[0004] In view of this, the present invention proposes an ultrasonic-based weld defect detection method and system, which aims to solve the problem that there are differences in technical levels and practical experience among different inspectors. When detecting defects in metal material welds, the obtained ultrasonic signal will deviate from the judgment of the defect type, resulting in misjudgment or omission of the defect type, thereby affecting the reliability and stability of the detection results. It is difficult to meet the high-precision and intelligent quality control requirements by relying on human experience.

[0005] In one aspect, the present invention provides a method for detecting weld defects based on ultrasound, comprising: Establishing a historical database, which includes a historical normal database and a historical defect database, obtaining an ultrasonic echo signal of the weld to be inspected, and traversing the ultrasonic echo signal in the historical database, and determining a multiplexing defect strategy or an ultrasonic verification strategy based on the traversal result; When the ultrasonic verification strategy is determined, determining the echo delay and echo intensity of the ultrasonic echo signal, determining a first ultrasonic result based on the echo delay and echo intensity, determining a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic feature model, and determining whether the weld to be inspected has a defect based on the first ultrasonic result and the second ultrasonic result; When it is determined that the weld to be inspected has a defect, the ultrasonic echo signal is Fourier transformed to determine an ultrasonic characteristic spectrum diagram, the defect type of the weld to be inspected is determined based on the ultrasonic characteristic spectrum diagram, a three-dimensional model of the weld to be inspected is established, a simulation environment is built and the three-dimensional model is imported into the simulation environment, and an attenuation result is determined based on the three-dimensional model and the metal sample and written into a simulation database of the simulation environment; A simulated ultrasonic probe array is determined based on the simulation database, a wave field propagation simulation is performed based on the simulated ultrasonic probe array, a simulated spectrum curve is determined based on the wave field propagation simulation result, and the defect type is verified based on the relationship between the ultrasonic characteristic spectrum diagram and the simulated spectrum curve.

[0006] Furthermore, when obtaining the ultrasonic echo signal of the weld to be inspected and traversing the ultrasonic echo signal in the historical database, and determining the multiplexing defect strategy or the ultrasonic verification strategy according to the traversal result, it includes: If the weld to be inspected is parallel to the detection surface, a straight probe is used to obtain the ultrasonic echo signal. If the weld to be inspected is not parallel to the detection surface, an oblique probe is used to obtain the ultrasonic echo signal. When using the straight probe or the oblique probe to obtain the ultrasonic echo signal, a coupling agent is first applied between the straight probe or the oblique probe and the detection surface. The historical database includes historical ultrasonic echo signals, historical ultrasonic sensor operation data, historical simulation spectrum curves and historical defect types, and the historical ultrasonic echo signals, historical ultrasonic sensor operation data, historical simulation spectrum curves and historical defect types correspond to each other one by one; The ultrasonic echo signal is traversed in the historical defect database to determine the multiplexing defect strategy or the ultrasonic verification strategy.

[0007] Furthermore, when traversing the ultrasonic echo signal in the historical defect database to determine the multiplexing defect strategy or the ultrasonic verification strategy, the method includes: When the same data as the ultrasonic echo signal exists in the historical defect database, the multiplexing defect strategy is determined to be used, and the historical defect type corresponding to the same data is multiplexed; When there is no data identical to the ultrasonic echo signal in the historical defect database, the ultrasonic verification strategy is determined to be used.

[0008] Furthermore, when determining the echo delay and echo intensity of the ultrasonic echo signal, determining a first ultrasonic result based on the echo delay and echo intensity, and determining a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic feature model, the method includes: Obtaining a standard echo delay interval and a standard echo intensity interval, and comparing the echo delay and echo intensity with the standard echo delay interval and the standard echo intensity interval respectively; When the echo delay is within a standard echo delay interval and the echo intensity is within a standard echo intensity interval, determining the first ultrasound result as a first output type; When the echo delay is not within the standard echo delay interval, or the echo intensity is not within the standard echo intensity interval, determining the first ultrasound result as a second output type; The sensor data set and the historical database are divided into a training set and a test set, the ultrasonic feature model is determined based on the training set and the test set, and the ultrasonic echo signal is substituted into the ultrasonic feature model to determine a second ultrasonic result, wherein the second ultrasonic result includes the first output type or the second output type.

[0009] Furthermore, when determining whether the weld to be inspected has defects based on the first ultrasonic result and the second ultrasonic result, the method includes: When the output types of the first ultrasonic result and the second ultrasonic result are both the first output type, it is determined that the weld to be inspected does not have defects, and the ultrasonic echo signal is stored in the historical normal database; When the output types of the first ultrasonic result and the second ultrasonic result are both the second output type, or the output types are inconsistent, it is determined that the weld to be inspected has defects.

[0010] Furthermore, when determining the defect type of the weld to be detected based on the ultrasonic characteristic spectrum, it includes: Determining an amplitude jump of the ultrasonic characteristic spectrum, and determining a first amplitude jump and a second amplitude jump, wherein the first amplitude jump is greater than the second amplitude jump; When the amplitude jump is greater than the first amplitude jump, it is determined that the defect type of the weld to be inspected is a crack; When the amplitude jump is less than or equal to the first amplitude jump and greater than or equal to the second amplitude jump, it is determined that the defect type of the weld to be inspected is porosity; When the amplitude jump is smaller than the second amplitude jump, it is determined that the defect type of the weld to be inspected is slag inclusion.

[0011] Furthermore, when establishing a three-dimensional model of the weld to be inspected, building a simulation environment and importing the three-dimensional model into the simulation environment, determining the attenuation result based on the three-dimensional model and the metal sample and writing the result into the simulation database of the simulation environment, the method includes: The weld to be inspected is collected section by section to determine point cloud data, and isolated points and noise points of the point cloud data are removed based on neighborhood statistics and curvature filtering algorithms, and a three-dimensional model of the weld to be inspected is established based on ICP registration of the least squares method and surface reconstruction algorithms; The simulation environment is built using simulation software and the three-dimensional model is imported. A tetrahedral grid is generated in the three-dimensional model according to an adaptive grid division algorithm. The grid size is adjusted according to the wavelength of the ultrasonic echo signal. The metal sample made of the same material as the weld to be inspected and the detection surface is obtained. The metal sample is subjected to an attenuation test using pulse echo. The attenuation result is determined by frequency domain analysis and fitting. The attenuation result includes sound velocity, density and attenuation parameters. The sound velocity, density and attenuation parameters are imported into the software database of the simulation software and the simulation database is constructed.

[0012] Furthermore, when determining a simulated ultrasound probe array based on the simulation database, performing wave field propagation simulation according to the simulated ultrasound probe array, and determining a simulated spectrum curve according to the wave field propagation simulation result, the method includes: Determining an arrangement of phased array ultrasonic sensors according to the curvature of the three-dimensional model, and deploying array surfaces on the three-dimensional model according to the arrangement to determine the simulated ultrasonic probe array, and determining an optimal spacing of a finite element grid and an operating frequency of the phased array ultrasonic sensors based on the simulation database; A beam bandwidth and an excitation function are selected from the Gaussian pulse signal of the simulated ultrasonic probe array, and the vertical displacement component of the simulated ultrasonic probe array is obtained. A three-dimensional time domain wave field analysis is performed based on the simulation environment to output the original time domain waveform, and the signal envelope is determined using the Hilbert transform. Based on the short-time Fourier transform and multi-layer wavelet packet decomposition, the frequency drift, bandwidth expansion, and energy distribution characteristics of the signal envelope between each frequency band are extracted to determine the simulated spectrum curve.

[0013] Furthermore, when verifying the defect type based on the relationship between the ultrasonic characteristic spectrum diagram and the simulation spectrum curve, the method includes: When the simulation spectrum curve is consistent with the ultrasonic characteristic spectrum Figure 1 If the defect type is correct, the defect type and ultrasonic echo signal are stored in a historical defect database; When the simulation spectrum curve is inconsistent with the ultrasonic characteristic spectrum diagram, it is determined that the defect type is incorrect, and the ultrasonic echo signal of the weld to be inspected is reacquired.

[0014] Compared with the existing technology, the present invention has the following advantages: Traditional ultrasonic testing relies on human experience and is prone to misjudgment and omission due to differences in human skills. By establishing a historical database, a reference basis is provided for the analysis of ultrasonic echo signals, reducing the interference of human subjective bias on the detection results, and improving the reliability and stability of ultrasonic testing. When identifying defects, the echo delay, echo intensity and ultrasonic feature model are combined to obtain dual results and judge the existence of defects, realizing multi-dimensional analysis of ultrasonic echo signals, accurately capturing signal characteristics, improving the accuracy of weld defect type identification, and effectively avoiding the limitations of a single judgment method. The ultrasonic feature spectrum is obtained through Fourier transform and verified by combining the simulated spectrum curve formed by wavefield propagation simulation, constructing a closed-loop process of identification-simulation-verification, and enhancing the accuracy of defect type determination in ultrasonic testing. At the same time, the introduction of three-dimensional models and simulation environments makes ultrasonic testing no longer limited to the signal itself. Through the simulation database and simulated ultrasonic probe array, the characteristics of defects under different ultrasonic propagation conditions are comprehensively analyzed, realizing the transformation of ultrasonic testing from manual dependence to data-driven to intelligent analysis, and improving the reliability and stability of ultrasonic testing.

[0015] On the other hand, the present application also provides an ultrasonic-based weld defect detection system for applying the above-mentioned ultrasonic-based weld defect detection method, comprising: a storage and analysis unit configured to establish a historical database, the historical database including a historical normal database and a historical defect database, obtain an ultrasonic echo signal of a weld to be inspected, traverse the ultrasonic echo signal in the historical database, and determine a multiplexing defect strategy or an ultrasonic verification strategy based on the traversal result; an ultrasonic detection unit configured to, when the ultrasonic verification strategy is determined, determine an echo delay and an echo intensity of the ultrasonic echo signal, determine a first ultrasonic result based on the echo delay and the echo intensity, determine a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic characteristic model, and determine whether the weld to be inspected has a defect based on the first ultrasonic result and the second ultrasonic result; an ultrasonic simulation unit configured to, when it is determined that the weld to be inspected has a defect, perform Fourier transform on the ultrasonic echo signal to determine an ultrasonic characteristic spectrum, determine the defect type of the weld to be inspected based on the ultrasonic characteristic spectrum, establish a three-dimensional model of the weld to be inspected, build a simulation environment and import the three-dimensional model into the simulation environment, determine an attenuation result based on the three-dimensional model and the metal sample, and write the result into a simulation database of the simulation environment; The ultrasonic verification unit is configured to determine a simulated ultrasonic probe array based on the simulation database, perform wave field propagation simulation according to the simulated ultrasonic probe array, determine a simulation spectrum curve according to the wave field propagation simulation results, and verify the defect type based on the relationship between the ultrasonic characteristic spectrum diagram and the simulation spectrum curve.

[0016] It is understandable that the above-mentioned ultrasonic-based weld defect detection method and system have the same beneficial effects and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings: Figure 1 A flow chart of a weld defect detection method based on ultrasound provided in an embodiment of the present invention; Figure 2 This is a functional block diagram of an ultrasonic-based weld defect detection system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, unless there is a conflict, the embodiments of the present disclosure and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0019] See Figure 1 As shown, in some embodiments of the present application, a method for detecting weld defects based on ultrasound includes: S100: Establish a historical database, which includes a historical normal database and a historical defect database, obtain the ultrasonic echo signal of the weld to be inspected, and traverse the ultrasonic echo signal in the historical database, and determine the multiplexing defect strategy or ultrasonic verification strategy based on the traversal result.

[0020] S200: When the ultrasonic verification strategy is determined, the echo delay and echo intensity of the ultrasonic echo signal are determined, and a first ultrasonic result is determined based on the echo delay and echo intensity, and a second ultrasonic result is determined based on the ultrasonic echo signal and the ultrasonic feature model, and whether there is a defect in the weld to be inspected is determined based on the first ultrasonic result and the second ultrasonic result.

[0021] S300: When it is determined that the weld to be inspected has defects, the ultrasonic echo signal is Fourier transformed to determine the ultrasonic characteristic spectrum diagram, the defect type of the weld to be inspected is determined based on the ultrasonic characteristic spectrum diagram, a three-dimensional model of the weld to be inspected is established, a simulation environment is built and the three-dimensional model is imported into the simulation environment, and the attenuation result is determined based on the three-dimensional model and the metal sample and written into the simulation database of the simulation environment.

[0022] S400: Determine a simulated ultrasonic probe array based on a simulation database, perform wave field propagation simulation based on the simulated ultrasonic probe array, determine a simulated spectrum curve based on the wave field propagation simulation result, and verify the defect type based on the relationship between the ultrasonic characteristic spectrum diagram and the simulated spectrum curve.

[0023] Currently, weld defect detection technologies primarily include radiographic testing, ultrasonic testing, magnetic particle testing, and penetrant testing. Ultrasonic testing, with its advantages of strong penetration, high sensitivity, and real-time detection, has become a mainstream technology for detecting defects in metal welds. Its basic principle is to transmit ultrasonic waves through the ultrasonic sensor's probe toward the weld area. When the ultrasonic waves encounter the defect interface during propagation, they are reflected, refracted, or scattered. The reflected signal (i.e., the ultrasonic echo signal) is captured by the receiving probe, and the presence of defects and related information are analyzed based on the signal characteristics.

[0024] Specifically, a historical database including a historical normal database and a historical defect database is established, a benchmark database covering various normal states and defect characteristics is constructed by accumulating a large amount of past weld detection data, these data include ultrasonic echo signals and corresponding defect data under different welding processes, materials and environments, in the case of experience difference of different detection personnel leading to subjective judgment deviation, the historical database can provide objective and unified comparison standard, the ultrasonic sensor is used to form a probe array around the weld to be detected or direct detection is performed, when the ultrasonic echo signal of the weld to be detected is obtained, then the historical database is traversed, it is quickly judged whether the signal matches the signal of some known defects, so as to determine the reuse defect strategy or the ultrasonic verification strategy, under the reuse defect strategy, the defect condition is determined based on the historical database, and under the ultrasonic verification strategy, it is indicated that the signal characteristics are not very clear, and further verification is needed, thereby reducing the dependence on human experience, the detection direction is preliminarily screened through data comparison, the efficiency of ultrasonic detection is improved, and a foundation is laid for subsequent detection. When it is determined as the ultrasonic verification strategy, it is necessary to determine whether the defect exists through double verification, the echo delay and echo intensity of the ultrasonic echo signal are obtained to obtain a first ultrasonic result, the principle is that the echo delay is related to the defect position, (the sound propagation time reflects the distance), the echo intensity is related to the defect size and nature, (the reflection ability of defects to sound waves is different), a second ultrasonic result is determined based on the ultrasonic echo signal and an ultrasonic characteristic model, the ultrasonic characteristic model is trained by a large amount of data, can extract the subtle features in the signal, and identify the possible complex defect mode, the echo delay and echo intensity may be disturbed in the propagation process, and the artificial intelligence model is combined with the echo delay and echo intensity, which can be verified from different dimensions, reduces the risk of misjudgment, improves the accuracy of defect identification, and avoids missed judgment or misjudgment due to single index deviation.

[0025] It can be understood that when it is determined that there is a defect, the ultrasonic echo signal is subjected to Fourier transform to obtain an ultrasonic characteristic spectrum, the Fourier transform converts the time domain signal into the frequency domain, the spectral characteristics of different defect types (such as cracks and pores) are different, and the defect type can be preliminarily determined accordingly. A three-dimensional model is established and a simulation environment is built. The simulation environment supports multi-physical field coupling and single-physical field simulation, and completely covers the whole physical process of ultrasonic propagation. Since the welds of different metal materials and the material properties have different effects on ultrasonic propagation, after the three-dimensional model is imported, the attenuation result is determined based on the three-dimensional model and the metal sample and written into the simulation database, which provides material property parameters for subsequent wave field propagation simulation. Based on the simulation database, a simulated ultrasonic probe array is determined to perform wave field propagation simulation, which can generate a corresponding simulation spectrum curve. The simulation spectrum curve is compared with the ultrasonic characteristic spectrum to verify the defect type. The defects existing in the welds of the metal material are sensed based on the ultrasonic, and the whole process is simulated and verified to ensure the reliability of the determination of the defect type. Through the synergistic effect of historical data comparison, double verification, spectrum analysis and simulation verification, the reliability of ultrasonic detection is improved, the subjectivity and uncertainty of manual detection are avoided, and the intelligentization and precision of weld defect detection are realized.

[0026] In some embodiments of the present application, when the ultrasonic echo signal of the weld to be detected is obtained, and the ultrasonic echo signal is traversed in the historical database, the multiplexing defect strategy or the ultrasonic verification strategy is determined according to the traversal result, including: if the weld to be detected and the detection surface are parallel, a straight probe is used to obtain the ultrasonic echo signal, if the weld to be detected and the detection surface are not parallel, an inclined probe is used to obtain the ultrasonic echo signal, and when the straight probe or the inclined probe is used to obtain the ultrasonic echo signal, a coupling agent is first applied between the straight probe or the inclined probe and the detection surface. The historical database includes historical ultrasonic echo signals, historical ultrasonic sensor operation data, historical simulation spectrum curves and historical defect types, and the historical ultrasonic echo signals, the historical ultrasonic sensor operation data, the historical simulation spectrum curves and the historical defect types correspond one by one. The ultrasonic echo signal is traversed in the historical defect database to determine the multiplexing defect strategy or the ultrasonic verification strategy.

[0027] Specifically, the detection surface is the surface of the workpiece that the ultrasonic probe directly contacts and transmits and receives ultrasonic waves. It is usually the surface of the component to be inspected (such as a pipe, plate, container, etc.) that contacts the probe. It is the entrance for ultrasonic waves to enter the interior of the workpiece from the probe. For example, when inspecting pipe welds, the outer or inner wall of the pipe may serve as the detection surface. When inspecting flat plate butt welds, the outer surface of the plate may serve as the detection surface. The parallelism or non-parallelism between the detection surface and the weld to be inspected directly affects the incident angle and propagation path of the ultrasonic wave. If the two are parallel, the vertical sound wave of the straight probe can efficiently reach the weld. If they are not parallel, the angled sound wave of the oblique probe can adjust the angle to adapt to the position of the weld to be inspected, ensuring that the ultrasonic wave enters the weld to be inspected at the appropriate angle, thereby effectively covering the inspection area. When ultrasonic waves propagate in different media, the greater the difference in acoustic impedance, the more severe the reflection loss. The acoustic impedance of air differs significantly from that of workpiece materials such as metal. If there's an air gap between the ultrasonic probe and the detection surface, the ultrasonic wave will be strongly reflected from the detection surface, with a significant amount of energy reflected back to the probe, preventing the ultrasonic echo signal from being acquired. Couplants (such as engine oil and glycerin) have an acoustic impedance intermediate between the probe (piezoelectric material) and the detection surface. They fill the tiny gap between the probe and the detection surface, displacing air and allowing the ultrasonic wave to travel smoothly through the couplant. This reduces reflection loss, ensures the strength and integrity of the ultrasonic echo signal, and avoids missed defects due to signal attenuation or loss. When acquiring the ultrasonic echo signal, the collected signal undergoes noise reduction and other processing to ensure its stability and reliability. The multi-dimensional data in the historical database comprehensively reflects the characteristics of the ultrasonic echo signal over time. Historical sensor operating data reflects detection conditions, historical simulation spectrum curves provide theoretical reference, and historical defect types provide the final conclusion determined during each historical period. By traversing the historical defect database, it is possible to accurately determine whether the signal matches a known defect, improving the stability and reliability of ultrasonic testing.

[0028] In some embodiments of the present application, when traversing the ultrasonic echo signal in the historical defect database to determine the multiplexing defect strategy or the ultrasonic verification strategy, it includes: when there is data identical to the ultrasonic echo signal in the historical defect database, it is determined to be a multiplexing defect strategy, and the historical defect type corresponding to the identical data is multiplexed; when there is no data identical to the ultrasonic echo signal in the historical defect database, it is determined to be an ultrasonic verification strategy.

[0029] Specifically, when conducting ultrasonic weld inspection, the inspection scenario is somewhat complex. The data in the historical defect database has been verified. When there are historical ultrasonic echo signals with the same ultrasonic echo signal, the corresponding historical defect type indicates that the defect has been fully identified. At this time, the reused defect strategy can directly call the known results to avoid repeated analysis of similar defects, thereby improving detection efficiency. When there is no identical historical ultrasonic echo signal, it indicates that the ultrasonic echo signal may be unknown data and corresponds to a new defect type. If it is directly judged, it is easy to cause misjudgment due to lack of reference basis. The ultrasonic verification strategy uses double verification of echo parameter analysis and feature model to deeply analyze the potential information in the signal and make up for the judgment blind spots caused by data missing. It avoids redundant operations and reduces the risk of missed judgments and misjudgments, while ensuring ultrasonic detection and improving the reliability of the results.

[0030] In some embodiments of the present application, when determining the echo delay and echo intensity of an ultrasonic echo signal, and determining a first ultrasonic result based on the echo delay and echo intensity, and determining a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic feature model, it includes: obtaining a standard echo delay interval and a standard echo intensity interval, comparing the echo delay and echo intensity with the standard echo delay interval and the standard echo intensity interval respectively; when the echo delay is within the standard echo delay interval and the echo intensity is within the standard echo intensity interval, the first ultrasonic result is determined as a first output type; when the echo delay is not within the standard echo delay interval, or the echo intensity is not within the standard echo intensity interval, the first ultrasonic result is determined as a second output type; the sensor data set and the historical database are divided into a training set and a test set; the ultrasonic feature model is determined based on the training set and the test set; and the ultrasonic echo signal is substituted into the ultrasonic feature model to determine the second ultrasonic result, and the second ultrasonic result includes the first output type or the second output type.

[0031] Specifically, due to the different parameters such as density, elastic modulus and acoustic impedance of different metal materials, which directly affect the propagation speed of ultrasound, for example, the longitudinal wave velocity in steel is about 5900m / s, and in aluminum it is about 6300m / s. Therefore, the standard echo delay interval and the standard echo intensity interval are dynamically determined according to the material properties of the detection surface and the weld to be inspected (such as aluminum, iron, etc.). When the echo delay is within the standard echo delay interval and the echo intensity is within the standard echo intensity interval, the first ultrasonic result is determined as the first output type. The first output type reflects that there are no defects in the weld to be inspected, while the second output type reflects that there are defects in the weld to be inspected. Regardless of whether the first ultrasonic result is the first output type or the second output type, it needs to be further verified. The double judgment mechanism can improve the accuracy and reliability of defect identification. The second ultrasonic result relies on an ultrasonic feature model. The sensor dataset contains parameters such as the operating frequency, reflection time, and accuracy of different types of ultrasonic sensors. The sensor dataset and historical database are divided into training and test sets, typically with a 4:1 ratio, to ensure the model's generalization ability and its ability to capture and identify relationships between different data. The training set is used to train the model to learn the characteristic patterns of ultrasonic waves, while the test set is used to verify the model's generalization ability, ensuring that the model can stably identify possible defects and thus uncover deep features that cannot be reflected, thereby compensating for the limitations of standard interval comparisons. Double results cross-validation not only enables rapid judgments within the standard interval, but also improves accuracy through model analysis, reducing the risk of missed or misjudgment of single judgments. At the same time, the model is continuously optimized based on historical data to adapt to diverse defect scenarios.

[0032] In some embodiments of the present application, when determining whether a weld to be inspected has a defect based on a first ultrasonic result and a second ultrasonic result, it includes: when the output types of the first ultrasonic result and the second ultrasonic result are both the first output type, it is determined that the weld to be inspected does not have a defect, and the ultrasonic echo signal is stored in a historical normal database; when the output types of the first ultrasonic result and the second ultrasonic result are both the second output type, or the output types are inconsistent, it is determined that the weld to be inspected has a defect.

[0033] Specifically, the first ultrasonic result performs a rapid screening based on a standard interval, while the second ultrasonic result relies on a model for deep feature analysis. These two types reflect the defect status from two dimensions: intuitive parameters and complex patterns. When both output the first type, it indicates that the ultrasonic echo signal is within the normal parameter range and has not been identified as an abnormal feature by the model. This double confirmation reduces the risk of misjudgment. At this time, the historical normal database is stored for data-driven enrichment of the baseline data. However, the generalization ability of the ultrasonic feature model is limited by the training data. If the training set lacks samples of a certain type of defect, such as rare, tiny delamination defects, or if the defect characteristics are highly similar to normal signals, such as a surface defect whose reflection pattern is close to the base material echo, the model may not be able to learn distinguishing features. Furthermore, the standard interval may not cover all defect scenarios. For example, the reflection signal of an extremely small point defect is weak, and its echo intensity may fall within the normal range. Or, the echo delay of an inclined crack, due to the unique acoustic wave propagation path, may coincide with the reflection time of a normal weld. In addition, material inhomogeneities, such as loose areas in castings, may cause normal signal fluctuations. Therefore, when the outputs of the two are inconsistent, defects are still determined to exist. In essence, through the complementarity of double verification, even if one mechanism may miss a detection due to limitations, the results of the other mechanism can still trigger the defect, thereby improving the reliability of ultrasonic testing.

[0034] In some embodiments of the present application, when determining the defect type of the weld to be detected based on the ultrasonic characteristic spectrum diagram, it includes: determining the amplitude jump of the ultrasonic characteristic spectrum diagram, and determining the first amplitude jump and the second amplitude jump, the first amplitude jump is greater than the second amplitude jump, when the amplitude jump is greater than the first amplitude jump, it is determined that the defect type of the weld to be detected is a crack, when the amplitude jump is less than or equal to the first amplitude jump, and greater than or equal to the second amplitude jump, it is determined that the defect type of the weld to be detected is a porosity, and when the amplitude jump is less than the second amplitude jump, it is determined that the defect type of the weld to be detected is a slag inclusion.

[0035] Specifically, the physical structures of different defects exhibit significant differences in their ultrasonic reflection characteristics. Cracks are often sharp, continuous linear defects with large reflection areas and smooth interfaces, resulting in dramatic energy fluctuations in the reflected signal, manifested as the largest amplitude jumps. Pores are closed cavities with a more regular shape but a medium reflection area, resulting in weaker amplitude jumps than cracks. Slag inclusions, on the other hand, are mostly loose impurities with irregular reflection interfaces and severe energy scattering, resulting in the smallest amplitude jumps. The peak value of a crack echo often exceeds the reference peak by more than half, meaning that ΔA / A0 is greater than approximately 0.5. Pores both reflect and scatter ultrasonic waves, resulting in amplitude variations between those of cracks and slag inclusions. ΔA / A0 typically falls between 0.1 and 0.5, meaning the peak echo can exceed the reference by 10% to 50%. Slag inclusions, with their irregular internal interfaces, absorb and scatter ultrasonic waves less effectively, resulting in the smallest amplitude jumps. In most cases, ΔA / A0 is less than 0.1, meaning the peak echo value does not differ from the reference by more than 10%. By setting the first amplitude jump to 0.5 and the second amplitude jump to 0.1, the subjective bias due to reliance on human experience is avoided, thus ensuring the reliability and stability of ultrasonic detection.

[0036] In some embodiments of the present application, when establishing a three-dimensional model of the weld to be inspected, building a simulation environment and importing the three-dimensional model into the simulation environment, determining the attenuation result based on the three-dimensional model and the metal sample and writing it into the simulation database of the simulation environment, it includes: collecting the weld to be inspected section by section to determine the point cloud data, and removing isolated points and noise points of the point cloud data based on neighborhood statistics and curvature filtering algorithms, and establishing a three-dimensional model of the weld to be inspected based on the ICP registration and surface reconstruction algorithm of the least squares method, using simulation software to build a simulation environment and import the three-dimensional model, generating a tetrahedral mesh in the three-dimensional model according to an adaptive meshing algorithm, adjusting the mesh size according to the wavelength of the ultrasonic echo signal, obtaining a metal sample of the same material as the weld to be inspected and the detection surface, using pulse echo to perform attenuation test on the metal sample, and determining the attenuation result through frequency domain analysis fitting, the attenuation result includes sound velocity, density and attenuation parameters, importing the sound velocity, density and attenuation parameters into the software database of the simulation software and constructing a simulation database.

[0037] Specifically, by accurately restoring the geometric features and material physical properties of the weld to be detected, a high-fidelity digital foundation is provided for subsequent wave field simulation, ensuring that the simulation results are highly consistent with the actual detection scene. After the point cloud data is collected in segments, neighborhood statistics and curvature filtering are used to remove isolated points and noise. The original collected point cloud data is prone to errors due to environmental interference. Filtering can avoid abnormal points distorting the geometry of the weld to be detected. ICP registration based on the least squares method can achieve accurate splicing of multiple point clouds. Surface reconstruction converts discrete points into a continuous three-dimensional model, ensuring accurate restoration of the geometric structure of the weld to be detected and providing an accurate spatial carrier for the simulation environment. COMSOL Multiphysics simulation software is usually used to build the simulation environment, and tetrahedral meshes are adaptively divided. The mesh size is adjusted according to the wavelength of the ultrasonic echo signal. The mesh accuracy directly affects the accuracy of subsequent wave field propagation simulation. Too coarse meshing will miss the details of the sound wave, while too fine meshing will increase the computational load. Mesh matching the wavelength can balance between accuracy and efficiency to ensure that the physical processes such as ultrasonic reflection and refraction are accurately simulated. The metal sample is the material property of the detection surface and the weld to be detected (such as aluminum, iron, etc.). Because the sound speed, density, and attenuation characteristics of different materials differ, they determine the propagation law of ultrasonic waves. Pulse echo combined with frequency domain analysis can accurately extract these parameters, making the material properties of the simulation environment consistent with the actual workpiece and avoiding simulation distortion due to parameter deviation. The sound speed, density, and attenuation parameters are imported into the software database of the simulation software and a simulation database is constructed. Generally, the simulation software provides some basic software databases as simulation data. The results of the pulse echo attenuation test on the metal sample are integrated into the software database, which can provide comprehensive and accurate physical parameter benchmarks for subsequent wave field simulation, ensuring the reliability of the simulation results. Wave field propagation simulation can truly reproduce the action process of ultrasonic waves, providing a theoretical basis for defect type verification and improving the objectivity and accuracy of defect verification, providing a solid foundation for the reliability of ultrasonic detection.

[0038] In some embodiments of the present application, when determining a simulated ultrasonic probe array based on a simulation database, performing wave field propagation simulation based on the simulated ultrasonic probe array, and determining a simulation spectrum curve based on the wave field propagation simulation results, it includes: determining an arrangement of phased array ultrasonic sensors based on the curvature of a three-dimensional model, and deploying the array surface on the three-dimensional model according to the arrangement, determining the simulated ultrasonic probe array based on the simulation database, determining the optimal spacing of the finite element grid and the operating frequency of the phased array ultrasonic sensor, selecting the beam bandwidth and excitation function in the Gaussian pulse signal of the simulated ultrasonic probe array, and obtaining the vertical displacement component of the simulated ultrasonic probe array, performing three-dimensional time domain wave field analysis based on the simulation environment to output the original time domain waveform, and using Hilbert transform to determine the signal envelope, and extracting the frequency drift, bandwidth expansion and energy distribution characteristics of the signal envelope between frequency bands based on short-time Fourier transform and multi-layer wavelet packet decomposition to determine the simulation spectrum curve.

[0039] Specifically, by accurately determining the operating state and wavefield propagation patterns of the simulated ultrasonic probe array, a spectrum curve that truly reflects defect characteristics is extracted, providing a reliable reference for defect type verification. The phased array arrangement and array surface are determined based on the curvature of the 3D model. This is because the surface curvature of the weld to be inspected affects the probe coupling effect and the angle of incidence of the sound wave. Adapting the arrangement to the curvature ensures that the sound wave evenly covers the inspection area and avoids signal distortion caused by poor probe fit. Generally, a linear array arrangement is used for planes or near-planes with large curvature (such as flat plate butt welds and the outer surfaces of large containers). The elements of the linear array probe are arranged along a straight line, and the sound wave is incident vertically, making it suitable for welds parallel to the inspection surface. For example, the curvature of flat plate welds is close to zero, so a linear array can achieve full coverage and avoid signal distortion caused by probe tilt. Cylindrical or annular structures (such as pipelines and pressure vessel girth welds) use a circular array arrangement. The elements of the circular array are distributed along the circumference, and the sound wave is incident radially, adapting to the circumferential inspection requirements of cylindrical surfaces. For example, the curvature of a pipe girth weld has a constant radius, so a ring array can cover the entire weld with full circumferential scanning. For complex, irregularly curved surfaces (such as aircraft engine blades and welds on special-shaped structures) or areas with dramatic curvature variations, a matrix array arrangement is used. The array elements are arranged in a two-dimensional rectangular pattern, and dynamic focusing and beam deflection enable three-dimensional spatial scanning. For example, the curvature of a blade tenon weld is complex and varies in multiple directions. A matrix array can electronically adjust the acoustic incident angle to ensure that reflected signals from defects are captured regardless of their location. The optimal finite element grid spacing and operating frequency of the phased array ultrasonic sensor are determined because the grid spacing must match the acoustic wavelength to balance simulation accuracy and efficiency. The operating frequency determines the acoustic resolution (high frequencies are suitable for small defects). The combination of these two allows for accurate reproduction of detailed characteristics of wavefield propagation. The beamwidth, excitation function, and acquisition of the vertical displacement component of the Gaussian pulse are selected to simulate the transmission characteristics of a real ultrasonic signal. The vertical displacement component focuses energy variations in the main propagation direction of the acoustic wave, thereby improving the signal-to-noise ratio. The three-dimensional time-domain wave field analysis, Hilbert transform and short-time Fourier transform are processed because the original time-domain waveform contains a large amount of redundant information. The Hilbert transform can extract the signal envelope to simplify the analysis. The short-time Fourier transform and wavelet packet decomposition can capture dynamic characteristics such as frequency drift and bandwidth expansion. These characteristics are directly related to the defect type (such as the rich high-frequency components of cracks and the energy dispersion of slag inclusions) and can be converted into simulated spectrum curves to distinguish the defect types. The simulated spectrum curves can accurately map the interaction between the defect type and the ultrasonic wave, thereby improving the stability and reliability of ultrasonic testing.

[0040] In some embodiments of the present application, when verifying the defect type based on the relationship between the ultrasonic characteristic spectrum and the simulation spectrum curve, it includes: when the simulation spectrum curve is consistent with the ultrasonic characteristic spectrum Figure 1When the simulated spectrum curve is inconsistent with the ultrasonic characteristic spectrum diagram, the defect type is determined to be incorrect, and the ultrasonic echo signal of the weld to be inspected is re-acquired.

[0041] Specifically, the ultrasonic characteristic spectrum diagram is the signal characteristic of the actual detection. The consistency between the simulation spectrum curve and the ultrasonic characteristic spectrum diagram indicates that the actual defect characteristics match the simulation results, verifying the correctness of the previous judgment of the defect type. Inconsistency indicates that there may be detection interference (such as poor coupling), model errors or unknown defects, and the signal needs to be re-collected to eliminate interference. The comparison of the dual spectra reduces the risk of misjudgment of defect types, ensures the accuracy of the judgment results, and improves the reliability and stability of ultrasonic detection.

[0042] In summary, the beneficial effects of the present invention are as follows: traditional ultrasonic testing relies on human experience and is prone to misjudgment and missed judgment due to differences in personnel skills. By establishing a historical database, a reference basis is provided for the analysis of ultrasonic echo signals, reducing the interference of human subjective bias on the detection results, and improving the reliability and stability of ultrasonic testing. When identifying defects, the echo delay, echo intensity and ultrasonic feature model are combined to obtain dual results and judge the existence of defects, realizing multi-dimensional analysis of ultrasonic echo signals, accurately capturing signal characteristics, improving the accuracy of weld defect type identification, and effectively avoiding the limitations of a single judgment method. The ultrasonic feature spectrum is obtained by Fourier transform and verified by combining the simulated spectrum curve formed by wave field propagation simulation, constructing a closed-loop process of identification-simulation-verification, and enhancing the accuracy of defect type determination in ultrasonic testing. At the same time, the introduction of three-dimensional models and simulation environments makes ultrasonic testing no longer limited to the signal itself. By using the simulation database and simulated ultrasonic probe array, the characteristics of defects under different ultrasonic propagation conditions are comprehensively analyzed, realizing the transformation of ultrasonic testing from manual dependence to data-driven to intelligent analysis, and improving the reliability and stability of ultrasonic testing.

[0043] In another preferred embodiment based on the above embodiment, refer to Figure 2 As shown, this embodiment provides an ultrasonic-based weld defect detection system for applying the above-mentioned ultrasonic-based weld defect detection method, including: The storage and analysis unit is configured to establish a historical database, which includes a historical normal database and a historical defect database, obtain the ultrasonic echo signal of the weld to be inspected, traverse the ultrasonic echo signal in the historical database, and determine a multiplexing defect strategy or an ultrasonic verification strategy based on the traversal result; an ultrasonic detection unit configured to, when the ultrasonic verification strategy is determined, determine an echo delay and an echo intensity of the ultrasonic echo signal, determine a first ultrasonic result based on the echo delay and the echo intensity, determine a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic characteristic model, and determine whether the weld to be inspected has a defect based on the first ultrasonic result and the second ultrasonic result; an ultrasonic simulation unit configured to, when it is determined that a defect exists in the weld to be inspected, perform Fourier transform on the ultrasonic echo signal to determine an ultrasonic characteristic spectrum, determine the defect type of the weld to be inspected based on the ultrasonic characteristic spectrum, establish a three-dimensional model of the weld to be inspected, build a simulation environment and import the three-dimensional model into the simulation environment, determine an attenuation result based on the three-dimensional model and the metal sample, and write the result into a simulation database of the simulation environment; The ultrasonic verification unit is configured to determine a simulated ultrasonic probe array based on a simulation database, perform wave field propagation simulation based on the simulated ultrasonic probe array, determine a simulated spectrum curve based on the wave field propagation simulation results, and verify the defect type based on the relationship between the ultrasonic characteristic spectrum diagram and the simulated spectrum curve.

[0044] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0045] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0046] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1The function specified in one or more boxes.

[0047] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A weld defect detection method based on ultrasound, characterized in that: include: Establishing a historical database, which includes a historical normal database and a historical defect database, obtaining an ultrasonic echo signal of the weld to be inspected, and traversing the ultrasonic echo signal in the historical database, and determining a multiplexing defect strategy or an ultrasonic verification strategy based on the traversal result; When the ultrasonic verification strategy is determined, determining the echo delay and echo intensity of the ultrasonic echo signal, determining a first ultrasonic result based on the echo delay and echo intensity, determining a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic feature model, and determining whether the weld to be inspected has a defect based on the first ultrasonic result and the second ultrasonic result; When it is determined that the weld to be inspected has a defect, the ultrasonic echo signal is Fourier transformed to determine an ultrasonic characteristic spectrum diagram, the defect type of the weld to be inspected is determined based on the ultrasonic characteristic spectrum diagram, a three-dimensional model of the weld to be inspected is established, a simulation environment is built and the three-dimensional model is imported into the simulation environment, and an attenuation result is determined based on the three-dimensional model and the metal sample and written into a simulation database of the simulation environment; A simulated ultrasonic probe array is determined based on the simulation database, a wave field propagation simulation is performed based on the simulated ultrasonic probe array, a simulated spectrum curve is determined based on the wave field propagation simulation result, and the defect type is verified based on the relationship between the ultrasonic characteristic spectrum diagram and the simulated spectrum curve.

2. The ultrasonic weld defect detection method according to claim 1, characterized in that: When obtaining the ultrasonic echo signal of the weld to be inspected, traversing the ultrasonic echo signal in the historical database, and determining the multiplexing defect strategy or the ultrasonic verification strategy according to the traversal result, the method includes: If the weld to be inspected is parallel to the detection surface, a straight probe is used to obtain the ultrasonic echo signal. If the weld to be inspected is not parallel to the detection surface, an oblique probe is used to obtain the ultrasonic echo signal. When using the straight probe or the oblique probe to obtain the ultrasonic echo signal, a coupling agent is first applied between the straight probe or the oblique probe and the detection surface. The historical database includes historical ultrasonic echo signals, historical ultrasonic sensor operation data, historical simulation spectrum curves and historical defect types, and the historical ultrasonic echo signals, historical ultrasonic sensor operation data, historical simulation spectrum curves and historical defect types correspond to each other one by one; The ultrasonic echo signal is traversed in the historical defect database to determine the multiplexing defect strategy or the ultrasonic verification strategy.

3. The ultrasonic weld defect detection method according to claim 2, characterized in that: When the ultrasonic echo signal is traversed in the historical defect database to determine the multiplexing defect strategy or the ultrasonic verification strategy, the method includes: When the same data as the ultrasonic echo signal exists in the historical defect database, the multiplexing defect strategy is determined to be used, and the historical defect type corresponding to the same data is multiplexed; When there is no data identical to the ultrasonic echo signal in the historical defect database, the ultrasonic verification strategy is determined to be used.

4. The ultrasonic weld defect detection method according to claim 3, characterized in that: When determining the echo delay and echo intensity of the ultrasonic echo signal, determining a first ultrasonic result based on the echo delay and echo intensity, and determining a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic feature model, the method includes: Obtaining a standard echo delay interval and a standard echo intensity interval, and comparing the echo delay and echo intensity with the standard echo delay interval and the standard echo intensity interval respectively; When the echo delay is within a standard echo delay interval and the echo intensity is within a standard echo intensity interval, determining the first ultrasound result as a first output type; When the echo delay is not within the standard echo delay interval, or the echo intensity is not within the standard echo intensity interval, determining the first ultrasound result as a second output type; The sensor data set and the historical database are divided into a training set and a test set, the ultrasonic feature model is determined based on the training set and the test set, and the ultrasonic echo signal is substituted into the ultrasonic feature model to determine a second ultrasonic result, wherein the second ultrasonic result includes the first output type or the second output type.

5. The ultrasonic weld defect detection method according to claim 4, characterized in that: When determining whether the weld to be inspected has a defect according to the first ultrasonic result and the second ultrasonic result, the method includes: When the output types of the first ultrasonic result and the second ultrasonic result are both the first output type, it is determined that the weld to be inspected does not have defects, and the ultrasonic echo signal is stored in the historical normal database; When the output types of the first ultrasonic result and the second ultrasonic result are both the second output type, or the output types are inconsistent, it is determined that the weld to be inspected has defects.

6. The ultrasonic weld defect detection method according to claim 5, characterized in that: When determining the defect type of the weld to be detected based on the ultrasonic characteristic spectrum, it includes: Determining an amplitude jump of the ultrasonic characteristic spectrum, and determining a first amplitude jump and a second amplitude jump, wherein the first amplitude jump is greater than the second amplitude jump; When the amplitude jump is greater than the first amplitude jump, it is determined that the defect type of the weld to be inspected is a crack; When the amplitude jump is less than or equal to the first amplitude jump and greater than or equal to the second amplitude jump, it is determined that the defect type of the weld to be inspected is porosity; When the amplitude jump is smaller than the second amplitude jump, it is determined that the defect type of the weld to be inspected is slag inclusion.

7. The ultrasonic weld defect detection method according to claim 6, characterized in that: When establishing a three-dimensional model of the weld to be inspected, building a simulation environment and importing the three-dimensional model into the simulation environment, determining an attenuation result based on the three-dimensional model and a metal sample and writing the result into a simulation database of the simulation environment, the method includes: The weld to be inspected is collected section by section to determine point cloud data, and isolated points and noise points of the point cloud data are removed based on neighborhood statistics and curvature filtering algorithms, and a three-dimensional model of the weld to be inspected is established based on ICP registration of the least squares method and surface reconstruction algorithms; The simulation environment is built using simulation software and the three-dimensional model is imported. A tetrahedral grid is generated in the three-dimensional model according to an adaptive grid division algorithm. The grid size is adjusted according to the wavelength of the ultrasonic echo signal. The metal sample made of the same material as the weld to be inspected and the detection surface is obtained. The metal sample is subjected to an attenuation test using pulse echo. The attenuation result is determined by frequency domain analysis and fitting. The attenuation result includes sound velocity, density and attenuation parameters. The sound velocity, density and attenuation parameters are imported into the software database of the simulation software and the simulation database is constructed.

8. The ultrasonic weld defect detection method according to claim 7, characterized in that: When determining a simulated ultrasonic probe array based on the simulation database, performing wave field propagation simulation according to the simulated ultrasonic probe array, and determining a simulated spectrum curve according to the wave field propagation simulation result, the method includes: Determining an arrangement of phased array ultrasonic sensors according to the curvature of the three-dimensional model, and deploying array surfaces on the three-dimensional model according to the arrangement to determine the simulated ultrasonic probe array, and determining an optimal spacing of a finite element grid and an operating frequency of the phased array ultrasonic sensors based on the simulation database; A beam bandwidth and an excitation function are selected from the Gaussian pulse signal of the simulated ultrasonic probe array, and the vertical displacement component of the simulated ultrasonic probe array is obtained. A three-dimensional time domain wave field analysis is performed based on the simulation environment to output the original time domain waveform, and the signal envelope is determined using the Hilbert transform. Based on the short-time Fourier transform and multi-layer wavelet packet decomposition, the frequency drift, bandwidth expansion, and energy distribution characteristics of the signal envelope between each frequency band are extracted to determine the simulated spectrum curve.

9. The ultrasonic weld defect detection method according to claim 8, characterized in that: When verifying the defect type based on the relationship between the ultrasonic characteristic spectrum diagram and the simulation spectrum curve, it includes: When the simulated spectrum curve is consistent with the ultrasonic characteristic spectrum graph, it is determined that the defect type is correct, and the defect type and ultrasonic echo signal are stored in a historical defect database; When the simulation spectrum curve is inconsistent with the ultrasonic characteristic spectrum diagram, it is determined that the defect type is incorrect, and the ultrasonic echo signal of the weld to be inspected is reacquired.

10. An ultrasonic-based weld defect detection system, used for applying the ultrasonic-based weld defect detection method according to any one of claims 1 to 9, characterized in that: include: a storage and analysis unit configured to establish a historical database, the historical database including a historical normal database and a historical defect database, obtain an ultrasonic echo signal of a weld to be inspected, traverse the ultrasonic echo signal in the historical database, and determine a multiplexing defect strategy or an ultrasonic verification strategy based on the traversal result; an ultrasonic detection unit configured to, when the ultrasonic verification strategy is determined, determine an echo delay and an echo intensity of the ultrasonic echo signal, determine a first ultrasonic result based on the echo delay and the echo intensity, determine a second ultrasonic result based on the ultrasonic echo signal and an ultrasonic characteristic model, and determine whether the weld to be inspected has a defect based on the first ultrasonic result and the second ultrasonic result; an ultrasonic simulation unit configured to, when it is determined that the weld to be inspected has a defect, perform Fourier transform on the ultrasonic echo signal to determine an ultrasonic characteristic spectrum, determine the defect type of the weld to be inspected based on the ultrasonic characteristic spectrum, establish a three-dimensional model of the weld to be inspected, build a simulation environment and import the three-dimensional model into the simulation environment, determine an attenuation result based on the three-dimensional model and the metal sample, and write the result into a simulation database of the simulation environment; The ultrasonic verification unit is configured to determine a simulated ultrasonic probe array based on the simulation database, perform wave field propagation simulation according to the simulated ultrasonic probe array, determine a simulation spectrum curve according to the wave field propagation simulation results, and verify the defect type based on the relationship between the ultrasonic characteristic spectrum diagram and the simulation spectrum curve.

Images