Non-destructive evaluation method and apparatus for densification of selective laser melting additively manufactured components

Abstract

This invention discloses a non-destructive evaluation method and apparatus for the density of selective laser melting additive manufacturing components. The method includes: exciting an ultrasonic guided wave at a specific frequency in the workpiece under test using an air-coupled ultrasonic transducer; receiving the ultrasonic signal using the air-coupled ultrasonic transducer; analyzing the phase change of the ultrasonic guided wave at the specific frequency; and using a convolutional neural network model to obtain the mapping relationship between the phase change and the density of the test piece, thereby predicting the density of the selective laser melting sample. This invention overcomes the problems of low detection efficiency, high cost, and the need for destructive testing in conventional selective laser melting component density detection methods, achieving non-contact, high-precision, non-destructive testing of the density of selective laser melting components, thus improving detection efficiency, accuracy, and flexibility.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing technology and relates to a technique for non-destructive evaluation and characterization of material properties using ultrasonic waves. Specifically, it is a non-destructive evaluation method and apparatus for the density of selective laser melting additive manufacturing components. Background Technology

[0002] Additive manufacturing is a technology that uses layer-by-layer material accumulation based on sliced ​​data from a three-dimensional digital model to create solid parts. Since its inception, this technology has developed rapidly and has been widely applied in various fields, such as automotive, aerospace, medical, and military industries, enabling the rapid prototyping of complex samples. Selective laser melting (SLM) is an additive manufacturing technology based on the layer-by-layer melting, deposition, and shaping of a metal powder bed. Its advantages include the ability to process samples with complex geometries, high dimensional accuracy, and a wide variety of selectable materials, using either single-material or multi-component powders. Furthermore, because the metal powder completely melts during the printing process, the resulting printed components exhibit metallurgical bonding and excellent mechanical properties. Despite these advantages, SLM still has several drawbacks. During SLM printing, the printing parameters have a significant impact on the sample forming quality. Inappropriate printing parameters can introduce defects such as porosity, keyholes, and cracks into the sample. Selective laser melting (SLM) printing is a highly complex physicochemical and metallurgical process. Under the influence of a high-speed laser, the solid-liquid-solid phase transformation of the metal is very brief, involving the transfer of heat, mass, and momentum. Therefore, certain defects, especially porosity, are inevitably introduced during the printing process. These defects are small, typically 50-100 μm in size. These defects significantly reduce the mechanical properties of the printed specimen, particularly its fatigue performance. Due to the limited number of defects introduced during printing, conventional non-destructive testing methods are insufficient to guarantee the accuracy and robustness of the results. Therefore, high-precision density testing of SLM specimens is essential.

[0003] Currently, the commonly used methods for density testing of selected area laser melting printed samples include: The drainage method. The drainage method is simple to operate and requires little equipment. However, its results have poor robustness and are easily affected by external conditions, resulting in large errors. During the drainage method, the test sample needs to be completely immersed in the liquid and cannot touch the container holding the liquid. Therefore, due to the limitation of the precision balance's range, the drainage method cannot test large-volume samples. The metallographic method is based on metallographic images of any cross-section of the test sample, and the density of the test sample is calculated through post-processing. This method has a certain degree of randomness. When a small number of cross-sections are selected, the results obtained by the metallographic method are unreliable. The metallographic preparation process is complex and requires high precision; large-scale preparation is costly. Furthermore, metallographic testing is a destructive testing method. CT testing is a non-destructive radiographic testing method. CT testing can obtain three-dimensional images of the internal structure of the test sample, and the density of the test sample can be obtained through post-processing. However, the CT method is costly, greatly limited by equipment; the size of the equipment determines the size of the test sample, and in-situ testing is difficult. Furthermore, radiation is harmful to the human body and requires strict protection. Therefore, existing methods for density testing of selective laser melting components cannot meet the needs for high-volume, efficient, accurate, and non-destructive testing of printed samples of arbitrary geometries. Summary of the Invention

[0004] A brief overview of embodiments of the invention is provided below to provide a basic understanding of certain aspects of the invention. It should be understood that this overview is not an exhaustive summary of the invention. It is not intended to identify key or essential parts of the invention, nor is it intended to limit the scope of the invention. Its purpose is merely to present certain concepts in a simplified form as a prelude to the more detailed description that follows.

[0005] To address the shortcomings of existing technologies, this invention provides a non-destructive evaluation method and apparatus for the density of additively manufactured components produced by selective laser melting (SLM). This solves the problem that existing SLM component density testing methods cannot meet the requirements for non-contact, high-volume, efficient, high-accuracy, and non-destructive testing of additively manufactured components of arbitrary geometric shapes.

[0006] According to one aspect of this application, a non-destructive evaluation method for the density of a selected area laser melting additive manufacturing component is provided, comprising:

[0007] Step 1: Select an appropriate ultrasonic excitation frequency f0 and detection distance l based on the geometry of the test piece; obtain the phase value of the received ultrasonic signal at the ultrasonic excitation frequency f0 under defect-free conditions through theoretical calculation. ;

[0008] Step 2: Fix the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer at both ends of the test area of ​​the test specimen;

[0009] Step 3: Based on the ultrasonic excitation frequency f0 and the wave velocity of the ultrasonic wave at that frequency, calculate the incident angle and receiving angle of the air-coupled ultrasonic excitation and receiving transducers; excite ultrasonic guided waves at a specific frequency in the workpiece under test through the air-coupled ultrasonic excitation transducer, and receive the ultrasonic signal through the air-coupled ultrasonic receiving transducer; analyze the phase change of the ultrasonic guided waves at the specific frequency, and use a convolutional neural network model to obtain the mapping relationship between the phase change and the density of the test piece under test, thereby realizing the prediction of the density of the selected area laser melting sample.

[0010] Furthermore, in step 3, an ultrasonic guided wave at a specific frequency is excited in the workpiece under test using an air-coupled ultrasonic excitation transducer, and the ultrasonic signal is received using an air-coupled ultrasonic receiving transducer. The phase change of the ultrasonic guided wave at the specific frequency is analyzed, and the mapping relationship between the phase change and the density of the test piece is obtained using a convolutional neural network model, thereby achieving the prediction of the density of the selected area laser melting sample. Specifically, this includes:

[0011] An ultrasonic guided wave with an ultrasonic excitation frequency of f0 is excited by an air-coupled ultrasonic excitation transducer, and the corresponding ultrasonic signal is received by an air-coupled ultrasonic receiving transducer.

[0012] The received ultrasonic signal is subjected to FFT filtering and FFT processing to obtain the phase value at frequency f0 in the received ultrasonic signal. ;

[0013] By calculating the phase difference ( Qualitative analysis of the density of the tested sample;

[0014] By changing the selected area laser melting printing parameters, different densities can be prepared by printing. The samples were analyzed, and the phase differences corresponding to samples with different densities were obtained. ;

[0015] By detecting the phase values ​​of samples at different densities This constitutes a density-phase difference information set;

[0016] A dataset is constructed based on the density-phase difference information set, and the dataset is divided into a training set and a test set;

[0017] Convolutional neural network model is constructed, and the convolutional neural network is trained and tested using the training set and the test set respectively to obtain a trained convolutional neural network model;

[0018] The density of selected area laser melting components is predicted using a trained convolutional neural network model.

[0019] Furthermore, the ultrasonic guided wave excited by the air-coupled ultrasonic excitation transducer has a frequency of ultrasonic excitation frequency f0, and the initial phase of the ultrasonic guided wave is 0° in order to calculate the phase difference caused by the density difference.

[0020] Furthermore, the convolutional neural network model is used to obtain the mapping relationship between phase difference and the packing density of the tested component. The convolutional neural network model includes an input layer, a convolutional layer, a first fully connected layer, a first discard layer, a second fully connected layer, a second discard layer, and a regression layer connected in sequence. The input layer receives the obtained packing density-phase difference information; the convolutional layer processes the phase difference-packing density data and extracts data features; the first fully connected layer connects neurons between layers to construct the relationship between phase difference and packing density; the first discard layer randomly filters and removes some neurons to prevent overfitting of the training set; the second fully connected layer connects the remaining neurons to further construct the relationship between phase difference and packing density; the second discard layer also randomly filters and removes some neurons to prevent overfitting of the training set; and the regression layer predicts the packing density based on the phase difference information of the tested component.

[0021] Furthermore, the theoretical calculation formula in step 1 is as follows: Where l is the selected detection length, λ is the wavelength of the selected excitation frequency f0, and n is the number of phase rotations in the detection length.

[0022] An air-coupled ultrasonic excitation transducer is an air-coupled ultrasonic excitation probe, and an air-coupled ultrasonic receiving transducer is an air-coupled ultrasonic receiving probe. The air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer together constitute an air-coupled ultrasonic transducer (or simply an ultrasonic transducer). The distance between the air-coupled ultrasonic excitation probe and the air-coupled ultrasonic receiving probe placed above the sample being tested remains fixed.

[0023] Furthermore, to ensure a constant coupling state between the air-coupled ultrasonic transducer and the test specimen, the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer are placed at fixed positions above both ends of the test area of ​​the test specimen during the testing process. Specifically, during the testing of different sample densities, the distance between the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer remains constant, and the distance between the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer placed above the test specimen remains fixed.

[0024] More preferably, the air-coupled ultrasonic excitation transducer is an air-coupled piezoelectric ultrasonic excitation transducer, and the air-coupled ultrasonic receiving transducer is an air-coupled piezoelectric ultrasonic receiving transducer.

[0025] According to another aspect of this application, a non-destructive evaluation device for the density of selected area laser melting additive manufacturing components is provided, comprising a signal generator / receiver, an air-coupled ultrasonic excitation transducer, an air-coupled ultrasonic receiving transducer, a power amplifier, a bandpass filter, a scanning platform, an oscilloscope, and a computer; the signal generator / receiver is connected to the bandpass filter, the computer, and the oscilloscope respectively; the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer are respectively disposed at both ends of the sample to be tested on the scanning platform; the bandpass filter is connected to the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer through the power amplifier. The transducers are connected; wherein, the signal generator / receiver excites an ultrasonic guided wave signal at a suitable frequency, which, after passing through the bandpass filter to ensure the purity of the excitation signal frequency band and being amplified by the power amplifier, is connected to the air-coupled ultrasonic excitation transducer, so that the ultrasonic signal is introduced to one end of the sample being tested; the air-coupled ultrasonic receiving transducer located at the other end of the sample being tested receives the propagated ultrasonic signal, which, after being amplified by the power amplifier and filtered by the bandpass filter to remove noise and clutter signals, is sent to the signal generator / receiver, as well as to an oscilloscope and a computer for signal analysis. The bandpass filter used has a bandpass range that includes the selected ultrasonic excitation frequency f0, and the center frequency of the bandpass filter is close to the selected ultrasonic excitation frequency f0.

[0026] This invention discloses a non-destructive evaluation method and apparatus for the density of components manufactured by selective laser melting additive manufacturing. Compared with the prior art, this invention has the following advantages:

[0027] By selecting an appropriate ultrasonic frequency and testing area based on the geometry of the sample being tested, the ultrasonic phase angle at a specific frequency under defect-free conditions is obtained through theoretical calculations and testing. And in defective conditions, ultrasonic phase angle at a specific frequency By calculating the phase difference The density of the tested samples was qualitatively analyzed. A density-phase difference dataset was established by preparing and detecting the phase difference at specific frequencies of selected area laser melting samples with different densities. Based on this dataset, a convolutional neural network model was constructed, trained, and tested. The trained convolutional neural network model can obtain the mapping relationship between phase difference and density, enabling the prediction of the density of selected area laser melting components.

[0028] This invention, based on the principle of ultrasonic diffraction, overcomes the problem of low detection efficiency in traditional methods for detecting the density of selective laser melting (SLM) samples, enabling large-scale non-destructive testing of SLM samples. It employs a convolutional neural network to predict the density of the tested samples, improving the accuracy and precision of the detection results. The use of an air-coupled ultrasonic transducer enables non-contact testing. Furthermore, the use of ultrasonic guided waves eliminates limitations imposed by the geometry of the testing equipment and workpiece, increasing the flexibility of density detection. This invention can be used for on-site, rapid, and real-time detection and evaluation of the density of SLM samples. Attached Figure Description

[0029] The present invention can be better understood by referring to the description given below in conjunction with the accompanying drawings, in which the same or similar reference numerals are used throughout the drawings to denote the same or similar parts. These drawings, together with the following detailed description, are incorporated in and form part of this specification, and are used to further illustrate preferred embodiments of the invention and explain the principles and advantages of the invention. In the drawings:

[0030] Figure 1 This is a schematic diagram of the non-destructive evaluation device for the density of selected area laser melting additive manufacturing components according to the present invention.

[0031] Among them, 1—signal generator / receiver, 2—bandpass filter, 3—signal amplifier, 4—ultrasonic excitation transducer, 5—ultrasonic receiving transducer, 6—selective laser melting additive manufacturing sample, 7—scanning platform, 8—oscilloscope, 9—computer;

[0032] Figure 2 This is a flowchart illustrating the non-destructive evaluation method for the density of selected area laser melting additive manufacturing components according to the present invention.

[0033] Figure 3 This is a flowchart of the convolutional neural network algorithm in the non-destructive evaluation method for the density of selected area laser melting additive manufacturing components of the present invention;

[0034] Figure 4 This is a schematic diagram illustrating the change in ultrasonic phase caused by defects in the non-destructive evaluation method for the density of selected area laser melting additive manufacturing components according to the present invention. Detailed Implementation

[0035] Embodiments of the present invention will now be described with reference to the accompanying drawings. Elements and features described in one drawing or embodiment of the invention may be combined with elements and features shown in one or more other drawings or embodiments. It should be noted that, for clarity, representations and descriptions of components and processes unrelated to the present invention and known to those skilled in the art have been omitted from the drawings and description.

[0036] In the description of this invention, it should be understood that the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, unless otherwise explicitly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0037] This invention provides a non-destructive testing method and apparatus for the density of selective laser melting (SLM) additive manufacturing components. The method uses an air-coupled ultrasonic transducer to excite ultrasonic guided waves at a specific frequency within the workpiece. The transducer receives the ultrasonic signals, analyzes the phase change of the ultrasonic guided waves at the specific frequency, and uses a convolutional neural network model to obtain the mapping relationship between the phase change and the density of the test piece, thus predicting the density of the SLM sample. This invention overcomes the problems of low detection efficiency, high cost, and the need for destructive testing in conventional SLM component density detection methods, achieving non-contact, high-precision non-destructive testing of SLM component density, improving detection efficiency, accuracy, and flexibility.

[0038] Example 1

[0039] This invention provides a non-destructive evaluation device for the density of selected area laser melting additive manufacturing components, the structure of which is shown in the figure below. Figure 1 As shown, the system includes a signal generator / receiver 1, a signal amplifier 3 (power amplifier), a bandpass filter 2, an air-coupled piezoelectric ultrasonic excitation transducer 4, an air-coupled piezoelectric ultrasonic receiving transducer 5, a selective laser melting additive manufacturing sample 6, a scanning platform 7, an oscilloscope 8, and a computer 9. The air-coupled piezoelectric ultrasonic excitation transducer 4 and the air-coupled piezoelectric ultrasonic receiving transducer 5 are placed on the scanning platform 7 at a fixed distance above the area to be tested on the sample. In this embodiment, two signal amplifiers 3 and two bandpass filters 2 are provided.

[0040] Signal generator / receiver 1 excites an ultrasonic signal of a suitable frequency. After the excitation signal is purified by bandpass filter 2 and amplified by signal amplifier 3, it is connected to an air-coupled piezoelectric ultrasonic excitation transducer 4 at one end of the area to be tested, and the ultrasonic signal is introduced into the test specimen 6 at a suitable incident angle. An air-coupled piezoelectric ultrasonic receiving transducer 5 is placed above the other end of the area to be tested. Both air-coupled piezoelectric ultrasonic excitation transducer 4 and air-coupled piezoelectric ultrasonic receiving transducer 5 are fixed on the scanning platform 7. The detected propagation signal is amplified by signal amplifier 3 and filtered by bandpass filter 2 to remove noise and clutter signals before being sent to signal generator / receiver 1. The received signal is simultaneously sent to oscilloscope 8. The signal received by signal generator / receiver 1 is further input to computer 9 for signal analysis. In addition, the bandpass filter used has a bandpass range that includes the selected ultrasonic excitation frequency f0, and the center frequency of the bandpass filter is close to the selected ultrasonic excitation frequency f0.

[0041] Example 2

[0042] This invention provides a non-destructive evaluation method for the density of selective laser melting additive manufacturing components. The method obtains the ultrasonic phase angle at a specific frequency under defect-free conditions through theoretical calculations and testing. And in defective conditions, ultrasonic phase angle at a specific frequency Preparation of materials with different densities The sample is melted by laser in the lower selected area, and the above operation is repeated to obtain the phase difference. A density-phase difference dataset was established. Based on a well-trained convolutional neural network model, the density of selected area laser melting components can be predicted. The implementation of this method is based on: (1) When ultrasonic waves encounter a defect smaller than 1 / 2 of their wavelength, they will diffract, that is, bypass the obstacle and continue to propagate in the direction of incidence. During this process, the distance of ultrasonic wave propagation increases, which leads to an increase in ultrasonic phase angle. The more defect volume and number contained in the ultrasonic wave propagation path, the more obvious the increase in phase angle; (2) Using ultrasonic guided waves for detection does not require damaging the test sample, which can achieve non-destructive testing, and there are no special requirements for the geometry of the test sample; (3) A well-trained convolutional neural network model can improve the reliability of the prediction results.

[0043] For details, see Figure 2 The flowchart of the non-destructive evaluation method for the density of selected area laser melting additive manufacturing components in this embodiment includes the following steps:

[0044] Process 1) Select a suitable ultrasonic excitation frequency f0 and detection area based on the geometry of the test piece 6, and further plan the ultrasonic detection path and distance l;

[0045] Process 2) Through theoretical calculations, the phase value at frequency f0 in the received ultrasonic signal under defect-free conditions is obtained. ;

[0046] Process 3) Based on the selected ultrasonic excitation frequency f0 and the ultrasonic wave velocity at that frequency, the incident angle and receiving angle of the air-coupled ultrasonic excitation transducer and the receiving transducer are calculated.

[0047] Step 4) Place the air-coupled ultrasonic excitation transducer 4 and the air-coupled ultrasonic receiving transducer 5 above both ends of the selected area to be tested, and ensure that the height of the excitation transducer and the receiving transducer from the sample to be tested is consistent.

[0048] Step 5) An ultrasonic wave with an excitation frequency of f0 and an initial phase of 0° is applied before the excitation transducer to ensure a pure excitation signal, unaffected by equipment interference, and is amplified by a signal amplifier 3. The signal is received by the receiving transducer 5, after which a bandpass filter 2 is applied to filter out noise, and the signal is amplified by a signal amplifier 3. Finally, the ultrasonic wave signal is transmitted to the signal generator / receiver 1, oscilloscope 8, and computer 9, respectively. The receiving transducer takes the average of multiple signal receptions.

[0049] Step 6) Perform FFT filtering on the received signal, and then perform FFT processing on the signal again to obtain the phase angle at frequency f0. According to the formula Calculate the phase difference ;

[0050] Process 7) Printing to prepare materials with different densities Selective laser melting printing of the sample was performed, and the above operations 1)-6) were repeated to obtain the phase difference corresponding to samples with different densities. Different densities of printed samples can be obtained by changing different selective laser melting printing parameters, including laser power, laser scanning speed, laser scanning spacing, and printing layer thickness.

[0051] Step 8) Construct a dataset based on the obtained density-phase difference data, and divide the dataset into a training set and a test set. Construct a convolutional neural network model. Train and test the convolutional neural network using the training set and the test set respectively to obtain a trained convolutional neural network model;

[0052] Process 9) Use the trained neural network model to predict the density of selected area laser melting printed components.

[0053] The process of predicting the density of the detected component using the convolutional neural network model in process 8) of this invention is as follows: Figure 3As shown, a convolutional neural network (CNN) is used to obtain the mapping relationship between phase difference and packing density information of the test sample. This CNN model includes an input layer, a convolutional layer, a first fully connected layer, a first discard layer, a second fully connected layer, a second discard layer, and a regression layer. The input layer receives the obtained packing density-phase difference information; the convolutional layer processes the phase difference-packing density data and extracts data features; the first fully connected layer connects neurons between layers to construct the relationship between phase difference and packing density; the first discard layer randomly selects and removes some neurons to prevent overfitting of the training set; the second fully connected layer connects the remaining neurons to further construct the relationship between phase difference and packing density; the second discard layer also randomly selects and removes some neurons to prevent overfitting of the training set; and the regression layer predicts packing density based on the phase difference information of the test sample.

[0054] The testing principle of this invention is as follows:

[0055] The signal generator controls the ultrasonic excitation transducer to emit an ultrasonic guided wave of a pre-set specific frequency f0. The detection distance is a pre-set detection distance l. If there are no defects inside the sample being tested, according to the formula... (l is the selected detection length, Where f is the wavelength of the selected excitation frequency f0, and n is the number of complete phase rotations within the detection length. (If the phase angle is less than one revolution in the detection length), then the phase angle of the received ultrasonic signal at a specific frequency is: If the sample being tested contains minute printing defects, such as pores and keyholes, these defects are too small (much smaller than half the length of the ultrasonic wave) and numerous to be distinguishable from the ultrasonic time-domain signal. When an ultrasonic wave encounters an obstacle smaller than half its wavelength, it diffracts, meaning the wave bypasses the obstacle and continues propagating along its original incident direction. During this process, the propagation distance of the ultrasonic wave increases by an amount that is... Because the defect volume is small and the wave velocity is fast, the change over time is... t( The change in the phase velocity (where c is the wave velocity of the ultrasonic guided wave used) is also relatively small, so this difference cannot be observed in the time domain signal. However, the change is significant in the phase domain, with a phase difference of [missing value]. ( Therefore, the phase angle of the ultrasonic signal received by the test piece with internal defects at a specific frequency is... The diagram is as follows Figure 4As shown, the phase difference is directly related to the number of defects within the detection range. The more internal defects the tested sample has, the more significant the phase angle difference becomes. Therefore, by detecting the phase angle at a specific frequency of the received ultrasound, the density of the defects inside the tested sample can be accurately reflected, avoiding excessive internal defects in the printed component and preventing the degradation of the mechanical properties of the tested sample.

[0056] Due to the manufacturing principle of additively manufactured specimens, the surface roughness of the components is higher than that of machined specimens. Using contact ultrasonic non-destructive testing (NDT) cannot guarantee consistent coupling conditions between the probe and the tested specimen during each test, thus compromising the robustness of the test results. Therefore, this invention employs an air-coupled piezoelectric ultrasonic receiving transducer. Using an air-coupled ultrasonic transducer effectively reduces the impact of the surface roughness of the tested specimen on the test results. This invention performs non-destructive evaluation of the density of additively manufactured specimens based on phase difference. To ensure the accuracy of the test results, a narrow-band ultrasonic signal is required for both excitation and reception. Therefore, a bandpass filter needs to be connected before the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer to ensure the reliability of the test results. The bandpass filter used in this invention has a bandpass range that includes the selected ultrasonic excitation frequency f0, and the center frequency of the bandpass filter is close to the selected ultrasonic excitation frequency f0. During the preparation of additively manufactured specimens, due to the large temperature gradient, defects inevitably occur inside the specimen. The three-dimensional geometry of these defects is not fixed. Therefore, when using phase difference to non-destructively evaluate the density of additively manufactured samples, linear fitting cannot establish an accurate mapping relationship between the phase difference and the density of the sample being tested. To improve the accuracy of the detection results, this invention employs a convolutional neural network model. The experimentally obtained phase difference and the density of the sample being tested are used to construct a corresponding dataset. This dataset is divided into a training set and a test set. A convolutional neural network model is then constructed. The convolutional neural network is trained and tested using the training and test sets, respectively. The convolutional neural network, after training and testing, can effectively improve the detection results and reduce detection errors.

[0057] This invention is based on the principle of ultrasonic diffraction, where minute defects lead to an increase in the ultrasonic phase angle at a selected frequency. It develops a non-destructive testing method and apparatus for selective laser melting (SLM) additive manufacturing components. This technology is highly sensitive to internal defects in SLM components and employs air-coupled ultrasonic guided wave technology, enabling non-contact non-destructive testing of a specific area of ​​a sample with arbitrary geometry. It allows for high-volume, efficient evaluation of the density of SLM samples. Furthermore, the use of a trained convolutional neural network model effectively improves the accuracy of the test results.

[0058] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0059] Furthermore, the method of the present invention is not limited to being executed in the chronological order described in the specification, but may also be executed in other chronological orders, in parallel, or independently. Therefore, the execution order of the method described in this specification does not constitute a limitation on the technical scope of the present invention.

[0060] Although the invention has been disclosed above through the description of specific embodiments, it should be understood that all the embodiments and examples described above are exemplary and not restrictive. Those skilled in the art can design various modifications, improvements, or equivalents to the invention within the spirit and scope of the appended claims. These modifications, improvements, or equivalents should also be considered to be included within the protection scope of the invention.

Claims

1. A method of non-destructive evaluation of the density of a selective laser melting additively manufactured component, characterized in that: The method comprises the following steps: Step 1: selecting a suitable ultrasonic excitation frequency f0 and a detection distance l according to the geometric shape of the test piece; By theoretical calculation, the phase value under the ultrasonic excitation frequency f0 in the received ultrasonic signal under the condition of no defects is obtained ; Step 2: fixing the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer at both ends of the detection area of the test piece; Step 3: based on the ultrasonic excitation frequency f0 and the wave speed of the ultrasonic wave at this frequency, the incident angle and the receiving angle of the air-coupled ultrasonic excitation and receiving transducers are calculated; the air-coupled ultrasonic excitation transducer excites ultrasonic guided waves at a specific frequency in the test piece, the air-coupled ultrasonic receiving transducer receives the ultrasonic signal, analyzes the phase change of the ultrasonic guided wave at the specific frequency, and uses a convolutional neural network model to obtain the mapping relationship between the phase change and the density of the test piece, thereby realizing the prediction of the density of the selective laser melting sample. In step 3, the air-coupled ultrasonic excitation transducer excites ultrasonic guided waves at a specific frequency in the test piece, the air-coupled ultrasonic receiving transducer receives the ultrasonic signal, analyzes the phase change of the ultrasonic guided wave at the specific frequency, and uses a convolutional neural network model to obtain the mapping relationship between the phase change and the density of the test piece, thereby realizing the prediction of the density of the selective laser melting sample. Specifically, it comprises: The air-coupled ultrasonic excitation transducer excites ultrasonic guided waves at an ultrasonic excitation frequency f0, and the air-coupled ultrasonic receiving transducer receives the corresponding ultrasonic signal; FFT filtering and FFT processing are performed on the received ultrasonic signal to obtain a phase value at the f0 frequency in the received ultrasonic signal ; By calculating the phase difference , , qualitatively analyzing the density of the test sample; Different density samples are printed by changing selective laser melting printing parameters Different density samples are printed by changing selective laser melting printing parameters Different density samples are printed by changing selective laser melting printing parameters, including laser power, laser scanning speed, laser scanning interval and printing layer thickness By detecting the phase difference of the sample under different densities , the density-phase difference information set is constituted; Based on the density-phase difference information set, a data set is constructed, and the data set is divided into a training set and a test set; A convolutional neural network model is constructed, and the training set and the test set are used to train and test the convolutional neural network respectively to obtain a trained convolutional neural network model; The trained convolutional neural network model is used to predict the density of the selective laser melting component.

2. The non-destructive evaluation method of claim 1, wherein: The ultrasonic guided wave excited by the air-coupled ultrasonic excitation transducer has a frequency of ultrasonic excitation frequency f0, and the initial phase of the ultrasonic guided wave caused by the density difference is 0°.

3. The non-destructive evaluation method of claim 1, wherein: The convolutional neural network model is used to obtain the mapping relationship between the phase difference and the density of the test component; the convolutional neural network model comprises an input layer, a convolutional layer, a first full connection layer, a first discard layer, a second full connection layer, a second discard layer and a regression layer connected in sequence; wherein the input layer inputs the obtained density-phase difference information; the convolutional layer is used to process the phase difference-density data information and extract data features; the first full connection layer is used to connect the neurons between layers and construct the relationship between the phase difference and the density; the first discard layer is used to randomly select and remove part of the neurons to prevent overfitting of the training set; the second full connection layer is used to connect the remaining neurons that have not been removed to further construct the relationship between the phase difference and the density; the second discard layer is also used to randomly select and remove part of the neurons to prevent overfitting of the training set; and the regression layer is used to predict the density based on the phase difference information of the test piece.

4. The non-destructive evaluation method of claim 1, wherein: In order to ensure the coupling state between the air-coupled ultrasonic transducer and the test sample, the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer are arranged at fixed positions above the test area of the test sample during the detection process. Specifically, during the detection of different sample densities, the distance between the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer remains unchanged, and the distance between the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer placed above the test sample remains fixed.

5. The non-destructive evaluation method of claim 1, wherein: The air-coupled ultrasonic excitation transducer is an air-coupled piezoelectric ultrasonic excitation transducer, and the air-coupled ultrasonic receiving transducer is an air-coupled piezoelectric ultrasonic receiving transducer.

6. An apparatus for non-destructive evaluation of the density of a selective laser melting additively manufactured component, characterized in that: The device comprises a signal generator / receiver, an air-coupled ultrasonic excitation transducer, an air-coupled ultrasonic receiving transducer, a power amplifier, a band-pass filter, a scanning platform, an oscilloscope and a computer. The signal generator / receiver is connected to the band-pass filter, the computer and the oscilloscope, respectively. The air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer are arranged at the two ends of the test sample on the scanning platform. The band-pass filter is connected to the air-coupled ultrasonic excitation transducer and the air-coupled ultrasonic receiving transducer through the power amplifier. The signal generator / receiver excites ultrasonic guided wave signals at a suitable frequency. After the band-pass filter ensures the purity of the excitation signal frequency band and the power amplifier amplifies the signal, the air-coupled ultrasonic excitation transducer is connected to the air-coupled ultrasonic excitation transducer, so that the ultrasonic wave signal is introduced into one end of the test sample. The air-coupled ultrasonic receiving transducer at the other end of the test sample receives the transmitted ultrasonic signal, which is amplified by the power amplifier and the band-pass filter to remove noise and clutter signals, and is sent to the signal generator / receiver, the oscilloscope and the computer for signal analysis. The non-destructive evaluation device performs the non-destructive evaluation method according to any one of claims 1-5.

7. The non-destructive evaluation apparatus of claim 6, wherein: The band-pass filter has a band-pass range containing the selected ultrasonic excitation frequency f0, and the center frequency of the band-pass filter is close to the selected ultrasonic excitation frequency f0.

Images