Nondestructive prediction method, device, storage medium and electronic equipment for component porosity

Abstract

The present application discloses a non-destructive prediction method, device, storage medium, and electronic device for component porosity. The method comprises: placing a target component to be detected on a detection platform for frequency detection to obtain a frequency detection result, wherein a laser excitation probe and a three-dimensional laser vibrometer are placed on both sides of the detection platform; determining the highest frequency signal in the frequency detection result; determining the ultrasonic signal corresponding to the highest frequency signal by the laser excitation probe, and receiving the ultrasonic signal by the three-dimensional laser vibrometer; processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determining a target deep learning network model based on the nonlinear zero-frequency wave signal, and predicting the porosity of the target component by the target deep learning network model. Through this application, the problem of low detection accuracy of the porosity of additively manufactured components in related technologies is solved.

Description

Technical Field

[0001] The present application relates to the field of nondestructive testing technology, and in particular to a method, device, storage medium and electronic device for nondestructive prediction of component porosity. Background Art

[0002] Currently, several methods are commonly used to test the porosity of complex, irregularly shaped additively manufactured components. Metallography is the most commonly used method for testing the porosity of additively manufactured components. This method uses the metallographic structure of the material being tested to assess porosity. However, the metallographic process is destructive, making this method incapable of nondestructive testing. Furthermore, this method is subject to certain randomness, requiring extensive metallographic acquisition and resulting in high testing costs. Computed tomography (CT) is a radiographic nondestructive testing method. CT can produce a three-dimensional image of the internal structure of the sample being tested, and through post-processing, the porosity of the sample can be determined. However, CT testing is inefficient, costly, and time-consuming, making it incapable of real-time testing. Furthermore, radiation exposure is harmful to the human body and requires strict safety precautions. Linear ultrasound is an ultrasonic nondestructive testing method. Linear ultrasound uses acoustic parameters, such as the velocity and attenuation of the sample being tested, to establish a mapping relationship between the internal porosity of the component, thereby enabling porosity detection. However, linear ultrasound is insensitive to micropores within additively manufactured components, resulting in low detection accuracy. Furthermore, its detection based on acoustic parameters such as velocity and attenuation has certain requirements for the geometry of the sample being tested, making it incapable of testing samples with complex geometries. Therefore, existing methods for testing the porosity of additively manufactured components cannot meet the requirements for efficient and high-precision nondestructive testing of additively manufactured samples with complex geometries.

[0003] Currently, no effective solution has been proposed to the problem of low accuracy in detecting the porosity of additively manufactured components in related technologies. Summary of the Invention

[0004] The main purpose of this application is to provide a non-destructive prediction method, device, storage medium and electronic equipment for component porosity, so as to solve the problem of low accuracy in porosity detection of additively manufactured components in related technologies.

[0005] To achieve the above-mentioned purpose, according to the first aspect of the present application, a non-destructive prediction method for component porosity is provided. The method comprises: placing a target component to be detected on a detection platform for frequency detection to obtain a frequency detection result, wherein a laser excitation probe and a three-dimensional laser vibrometer are placed on both sides of the detection platform; determining the highest frequency signal in the frequency detection result; determining the ultrasonic signal corresponding to the highest frequency signal by the laser excitation probe, and receiving the ultrasonic signal by the three-dimensional laser vibrometer; processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determining a target deep learning network model based on the nonlinear zero-frequency wave signal, and predicting the porosity of the target component by the target deep learning network model, wherein the target deep learning network model is used to map the corresponding relationship between the nonlinear zero-frequency wave signal and the porosity of the target component.

[0006] Furthermore, the ultrasonic signal is processed to obtain a nonlinear zero-frequency wave signal, including: performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal; and performing low-pass filtering on the noise-reduced ultrasonic signal to obtain a filtered nonlinear zero-frequency wave signal.

[0007] Furthermore, a target deep learning network model is determined based on the ultrasonic signal, including: determining a data set between the nonlinear zero-frequency wave signal and the porosity sample; dividing the data set into a training set and a test set; and training the initial deep learning network model based on the training set and the test set to obtain a target deep learning network model.

[0008] Furthermore, the target deep learning network model includes: an input layer, a convolutional neural network module, a gated recurrent unit module, a fully connected layer and an output layer connected in sequence, wherein the convolutional neural network module includes a first convolutional layer, a first pooling layer, a second convolutional layer and a second pooling layer connected in sequence, and the gated recurrent unit module includes a first gated recurrent layer, a first activation layer, a second gated recurrent layer and a second activation layer.

[0009] Furthermore, the porosity of the target component is predicted through a target deep learning network model, including: extracting local features of the nonlinear zero-frequency wave signal based on a convolutional neural network module; training the dependency relationship between local features based on a gated recurrent unit module; mapping the dependency relationship through a fully connected layer to obtain the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component; and predicting the porosity of the target component based on the correspondence relationship.

[0010] Furthermore, the ultrasonic signal is subjected to wavelet denoising processing to obtain a denoised ultrasonic signal, including: determining a wavelet function, and performing a multi-layer wavelet transform on the ultrasonic signal based on the wavelet function to obtain high-frequency coefficients and low-frequency coefficients of the ultrasonic signal; and processing the high-frequency coefficients and low-frequency coefficients respectively to obtain a denoised ultrasonic signal.

[0011] Furthermore, the target component is placed on a detection platform and kept in a fixed position during the detection process, wherein the laser excitation probe and the three-dimensional laser vibrometer are placed on both sides of the detection platform and kept at a fixed interval.

[0012] In order to achieve the above-mentioned purpose, according to the second aspect of the present application, a non-destructive prediction device for component porosity is provided. The device includes: a detection unit, which is used to place the target component to be detected on a detection platform for frequency detection to obtain a frequency detection result, wherein: a laser excitation probe and a three-dimensional laser vibrometer are placed on both sides of the detection platform; a first determination unit, which is used to determine the highest frequency signal in the frequency detection result; a receiving unit, which is used to determine the ultrasonic signal corresponding to the highest frequency signal through the laser excitation probe, and receive the ultrasonic signal through the three-dimensional laser vibrometer; a second determination unit, which is used to process the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determine a target deep learning network model based on the nonlinear zero-frequency wave signal, and predict the porosity of the target component through the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component.

[0013] Furthermore, the second determination unit includes: a processing module for performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal; a filtering module for performing low-pass filtering on the noise-reduced ultrasonic signal to obtain a filtered nonlinear zero-frequency wave signal.

[0014] Furthermore, the second determination unit includes: a determination module for determining the data set between the nonlinear zero-frequency wave signal and the porosity sample; a division module for dividing the data set into a training set and a test set; and a first training module for training the initial deep learning network model based on the training set and the test set to obtain a target deep learning network model.

[0015] Furthermore, the target deep learning network model includes: an input layer, a convolutional neural network module, a gated recurrent unit module, a fully connected layer and an output layer connected in sequence, wherein the convolutional neural network module includes a first convolutional layer, a first pooling layer, a second convolutional layer and a second pooling layer connected in sequence, and the gated recurrent unit module includes a first gated recurrent layer, a first activation layer, a second gated recurrent layer and a second activation layer.

[0016] Furthermore, the second determination unit includes: an extraction module for extracting local features of the nonlinear zero-frequency wave signal based on a convolutional neural network module; a second training module for training the dependency relationship between local features based on a gated recurrent unit module; a mapping module for mapping the dependency relationship through a fully connected layer to obtain a correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component; and a prediction module for predicting the porosity of the target component based on the correspondence relationship.

[0017] Furthermore, the processing module includes: a determination submodule, used to determine the wavelet function, and perform multi-layer wavelet transform on the ultrasonic signal based on the wavelet function to obtain the high-frequency coefficients and low-frequency coefficients of the ultrasonic signal; a processing submodule, used to process the high-frequency coefficients and low-frequency coefficients respectively to obtain the denoised ultrasonic signal.

[0018] Furthermore, the target component is placed on a detection platform and kept in a fixed position during the detection process, wherein the laser excitation probe and the three-dimensional laser vibrometer are placed on both sides of the detection platform and kept at a fixed interval.

[0019] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising: a processor; and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any of the above-mentioned methods for non-destructive prediction of component porosity is implemented.

[0020] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for non-destructive prediction of component porosity according to any one of the above items is implemented.

[0021] Through this application, the following steps are adopted: placing the target component to be detected on a detection platform for frequency detection to obtain a frequency detection result, wherein the detection platform is placed on both sides with: a laser excitation probe and a three-dimensional laser vibrometer; determining the highest frequency signal in the frequency detection result; determining the ultrasonic signal corresponding to the highest frequency signal through the laser excitation probe, and receiving the ultrasonic signal through the three-dimensional laser vibrometer; processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determining a target deep learning network model based on the nonlinear zero-frequency wave signal, and predicting the porosity of the target component through the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component, thereby solving the problem of low detection accuracy of the porosity of additively manufactured components in related technologies. Through the constructed target deep learning network model, the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component is mapped, thereby achieving the effect of improving the detection accuracy of the porosity of additively manufactured components. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0023] Figure 1 This is a flow chart of a nondestructive prediction method for component porosity provided in accordance with an embodiment of the present application;

[0024] Figure 2 is a schematic diagram of a nonlinear zero-frequency wave ultrasonic detection signal provided according to an embodiment of the present application;

[0025] Figure 3 is a schematic structural diagram of a target deep learning network model provided according to an embodiment of the present application;

[0026] Figure 4 This is a schematic diagram of a non-contact, non-destructive evaluation architecture for the porosity of complex, special-shaped components manufactured by additive manufacturing, provided in this application;

[0027] Figure 5 Schematic diagram of a nondestructive prediction device for component porosity provided in accordance with an embodiment of the present application;

[0028] Figure 6 Schematic diagram of the network architecture of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0030] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] According to an embodiment of the present application, a non-destructive prediction method for component porosity is provided.

[0033] Figure 1 FIG. 1 is a flow chart of a nondestructive prediction method for component porosity according to an embodiment of the present application. Figure 1 As shown, the method includes the following steps:

[0034] In step S101, a target component to be detected is placed on a detection platform for frequency detection to obtain a frequency detection result, wherein a laser excitation probe and a three-dimensional laser vibrometer are placed on both sides of the detection platform.

[0035] Among them, the target component can be a complex and special-shaped component manufactured by additive manufacturing. In this case, the complex and special-shaped component manufactured by additive manufacturing to be inspected is placed on the inspection platform, and the positions of the laser excitation probe and the three-dimensional laser vibrometer are fixed. The complex and special-shaped component manufactured by additive manufacturing to be inspected is located between the laser excitation probe and the three-dimensional laser vibrometer. This not only improves the accuracy and reliability of the inspection, reduces production costs, but also improves production efficiency.

[0036] Step S102: determining the highest frequency signal in the frequency detection result.

[0037] For example, ultrasonic sweep detection is performed on the complex and special-shaped additively manufactured components being tested, with a sweep frequency range of 1-10 MHz. Based on the ultrasonic sweep detection results, the frequency signal f1 with the strongest response in the detection signal spectrum is selected as the ultrasonic excitation frequency for subsequent porosity detection. In this case, by detecting the highest frequency signal, that is, the frequency signal f1 with the strongest response, the sensitivity of detection can be significantly improved, the detection efficiency can be optimized, and the reliability of the detection results can be improved.

[0038] It should be noted that, due to the manufacturing principle of additively manufactured components, the surface roughness of the components is higher than that of machined specimens. Conventional contact ultrasonic nondestructive testing cannot guarantee consistent coupling conditions between the probe and the specimen being tested during each test, meaning the robustness of the test results cannot be guaranteed. Furthermore, for complex and irregularly shaped components, it is even more difficult to ensure the coupling conditions between the contact probe and the component surface. Therefore, the use of laser ultrasound in this case can effectively reduce the impact of the surface roughness and complex geometric structure of the specimen being tested on the test results.

[0039] Step S103 : determining the ultrasonic signal corresponding to the highest frequency signal by using a laser excitation probe, and receiving the ultrasonic signal by using a three-dimensional laser vibrometer.

[0040] Among them, a long-period ultrasonic detection signal with a frequency of f1 and a period of 200 is excited by a laser excitation probe, and the corresponding ultrasonic signal is received by a three-dimensional laser vibrometer. The duration of receiving the ultrasonic signal is 5 times that of the exciting ultrasonic signal, ensuring that the ultrasonic wave fully interacts with the micropores in the inspected component in the inspected sample. That is, this case uses a laser excitation probe to excite ultrasonic waves at a specific frequency in the inspected sample, and uses a three-dimensional laser vibrometer to receive the ultrasonic signal, which not only realizes non-contact, high-precision and high-efficiency detection, but also has rich waveform and wide-band detection capabilities.

[0041] Step S104: Process the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determine the target deep learning network model based on the nonlinear zero-frequency wave signal, and predict the porosity of the target component through the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component.

[0042] Specifically, the ultrasonic signal is processed to obtain a nonlinear zero-frequency wave signal, which can be obtained by the following steps: performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal; and performing low-pass filtering on the noise-reduced ultrasonic signal to obtain a filtered nonlinear zero-frequency wave signal.

[0043] Exemplarily, the received ultrasonic signal is processed by wavelet transformation, and the signal after wavelet threshold noise reduction is further low-pass filtered, and the filtering range is f<1 / 20f1, and its cut-off frequency is 1 / 20 of the selected ultrasonic excitation frequency f1. After that, the filtered nonlinear zero-frequency wave signal is obtained, and the nonlinear zero-frequency wave signal in the detection signal obtained after filtering is recorded as S1. In this case, by processing the ultrasonic signal, a nonlinear zero-frequency wave signal is obtained, which can significantly improve the signal quality, maintain the important characteristics of the signal, remove high-frequency noise and protect the low-frequency effective signal.

[0044] It should be noted that when ultrasonic waves propagate within the component being inspected, in addition to the excitation frequency, they also generate nonlinear responses at frequencies other than the excitation frequency, such as higher-order harmonics, subharmonics, and nonlinear zero-frequency waves. These acoustic nonlinear responses are highly sensitive to microstructural changes and microdamage within the material being inspected. Using acoustic nonlinear responses can effectively characterize material nonlinearities. The nonlinear zero-frequency wave has a very low frequency, resulting in minimal acoustic attenuation, significantly increasing the detection range and improving inspection efficiency. During the manufacturing process of additively manufactured components, microvoids are inevitably generated within them. These microvoids are very small, far less than half the length of conventional linear ultrasonic waves. Therefore, linear ultrasonic waves cannot detect these microvoids. The intensity of the acoustic nonlinear response is related to the nonlinearity of the material. As the number of microvoids within the component being inspected increases, the intensity of the nonlinear zero-frequency wave generated by the fundamental ultrasonic wave during propagation also increases. Therefore, the nonlinear zero-frequency wave can be used to characterize microvoids within the specimen being inspected. Due to its unique processing technology, additive manufacturing offers significant processing flexibility, enabling the fabrication of components with complex geometries that are difficult to fabricate using conventional subtractive methods. Conventional ultrasonic testing has certain requirements for the geometric structure of the component being tested, and changes in the geometric structure will change the propagation path of the sound. When a long-period resonant signal is excited (that is, the frequency with the strongest response in the frequency domain of the swept-frequency detection signal), the energy of the detection signal is improved, and the intensity of the corresponding nonlinear zero-frequency wave is also greatly increased. The longer detection time (the time for receiving the ultrasonic signal is 5 times that of the exciting ultrasonic signal) ensures that the nonlinear zero-frequency wave generated by the detection signal is fully propagated in the complex and special-shaped additively manufactured component being tested, and fully interacts with its internal micropores, so that the nonlinear zero-frequency wave signal in the detection signal can contain a large amount of information related to the micropores. The nonlinear zero-frequency wave ultrasonic detection signal is such as Figure 2 shown.

[0045] In some embodiments, a target deep learning network model is determined based on an ultrasonic signal, which can be obtained by the following steps: determining a data set between a nonlinear zero-frequency wave signal and a porosity sample; dividing the data set into a training set and a test set; and training an initial deep learning network model based on the training set and the test set to obtain a target deep learning network model.

[0046] For example, by changing the additive manufacturing preparation process, samples with different porosities xi are obtained, and the nonlinear zero-frequency wave signals Si corresponding to the samples with different porosities are obtained to form a porosity-nonlinear zero-frequency wave signal data set. The data set is divided into a training set and a test set to obtain the target deep learning network model of this application, and the trained deep learning network model is used to predict the porosity of complex special-shaped components manufactured by additive manufacturing.

[0047] In some embodiments, the target deep learning network includes: an input layer, a convolutional neural network module, a gated recurrent unit module, a fully connected layer and an output layer connected in sequence, wherein the convolutional neural network module includes a first convolutional layer, a first pooling layer, a second convolutional layer and a second pooling layer connected in sequence, and the gated recurrent unit module includes a first gated recurrent layer, a first activation layer, a second gated recurrent layer and a second activation layer.

[0048] Specifically, the convolutional neural network (CNN)-gated recurrent unit (GRU) network model in this case includes a sequentially connected input layer, a CNN module (convolutional neural network module), a GRU module (gated recurrent unit module), a fully connected layer, and an output layer. The input layer receives the porosity-nonlinear zero-frequency wave signal information set; the CNN module is used to extract local features from the detection signal; the GRU module is used to extract and learn long-term dependencies in the local features; the fully connected layer is used to connect neurons between layers, construct a mapping relationship between porosity and detection signals, and predict porosity; and the output layer is used to output the prediction results of the deep learning network model.

[0049] In some embodiments, the porosity of the target component is predicted by a target deep learning network model through the following steps: extracting local features of the nonlinear zero-frequency wave signal based on a convolutional neural network module; training the dependency relationship between the local features based on a gated recurrent unit module; mapping the dependency relationship through a fully connected layer to obtain the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component; and predicting the porosity of the target component based on the correspondence relationship.

[0050] For example, Figure 3As shown, the CNN module includes a first convolutional layer, a first pooling layer, a second convolutional layer, and a second pooling layer, which are connected in sequence. The first convolutional layer is used to process the porosity-detection signal information set and extract local features of the data; the first pooling layer is used to reduce the matrix size of the extracted local features while retaining important information; the second convolutional layer is used to further extract important information from the local features; and the second pooling layer is used to convert the important information into a one-dimensional matrix. The GRU module includes a first GRU layer (first gated recurrent layer), a first ReLU layer (first activation layer), a second GRU layer (second gated recurrent layer), and a second ReLU layer (second activation layer), which are connected in sequence. The first GRU layer is used to process the one-dimensional matrix of important information extracted by the CNN module, and to capture and learn the long-term dependencies in the above information; the first ReLU layer is used to introduce nonlinearity into the network and increase the nonlinear prediction function of the network; the second GRU layer is used to further capture and learn the long-term dependencies of the extracted features, and map the long-term dependencies through the fully connected layer to obtain the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component, thereby realizing the prediction of the porosity, avoiding the gradient vanishing phenomenon of the overall network. At the same time, the second ReLU layer can also reduce the interdependence between features and alleviate the occurrence of overfitting.

[0051] In some embodiments, wavelet noise reduction processing is performed on the ultrasonic signal to obtain the noise-reduced ultrasonic signal, which can be obtained by the following steps: determining the wavelet function, and performing multi-layer wavelet transform on the ultrasonic signal based on the wavelet function to obtain the high-frequency coefficients and low-frequency coefficients of the ultrasonic signal; processing the high-frequency coefficients and low-frequency coefficients respectively to obtain the noise-reduced ultrasonic signal.

[0052] Exemplarily, the specific process of wavelet change noise reduction processing is as follows: the wavelet function selected is the "db4" wavelet, the decomposition layer number is 4, and the high-frequency coefficients and low-frequency coefficients of the signal are obtained; the high-frequency coefficients of each layer are further subjected to wavelet threshold processing, and the wavelet hard threshold noise reduction method is adopted; finally, the low-frequency coefficients and the high-frequency coefficients after threshold processing are reconstructed to obtain the noise-reduced signal. Wavelet noise reduction processing of ultrasonic signals can significantly improve signal quality, improve signal feature recognition, and improve detection accuracy and reliability.

[0053] In some embodiments, the target component is placed on a detection platform and kept in a fixed position during the detection process, wherein the laser excitation probe and the three-dimensional laser vibrometer are placed on both sides of the detection platform and kept at a fixed interval.

[0054] That is, to ensure a constant coupling state between the laser excitation probe and the three-dimensional laser vibrometer and the test piece, the laser excitation probe and the three-dimensional laser vibrometer are placed at fixed positions at both ends of the test area of ​​the test sample during the detection process. Specifically, during the process of detecting the porosity of different samples, the distance between the laser excitation probe and the three-dimensional laser vibrometer remains unchanged, and the distance between the laser excitation probe and the three-dimensional laser vibrometer at both ends of the test sample remains fixed.

[0055] Optionally, Figure 4 This application provides a schematic diagram of a non-contact non-destructive evaluation framework for the porosity of complex special-shaped components manufactured by additive manufacturing. Figure 4 As shown, 1—computer, 2—pulsed solid-state laser, 3—laser excitation probe, 4—AM complex and irregularly shaped component, 5—testing platform, 6—3D laser vibrometer, 7—low-pass filter, 8—power amplifier, and 9—oscilloscope. AM complex and irregularly shaped component 4 is placed on testing platform 5 and remains fixed during testing. Laser excitation probe 3 and 3D laser vibrometer 6 are placed on either side of testing platform 5, spaced a fixed distance apart. Computer 1 controls pulsed solid-state laser 2 to generate an ultrasonic signal of appropriate frequency, which is then directed into AM complex and irregularly shaped component 4 via laser excitation probe 3. 3D laser vibrometer 6 is placed on the other side of AM complex and irregularly shaped component 4. The test signal is received by 3D laser vibrometer 6 and then passed through low-pass filter 7 and power amplifier 8 to remove high-frequency noise and the fundamental frequency signal. The processed signal is then fed back to 3D laser vibrometer 6 and further to computer 1 and oscilloscope 9 for subsequent signal analysis. In addition, the cutoff frequency of the low-pass filter used is 1 / 20 of the selected ultrasonic excitation frequency f1.

[0056] The embodiment of the present application provides a non-destructive prediction method for component porosity, which obtains a frequency detection result by placing the target component to be detected on a detection platform for frequency detection, wherein the detection platform is provided with: a laser excitation probe and a three-dimensional laser vibrometer; determining the highest frequency signal in the frequency detection result; determining the ultrasonic signal corresponding to the highest frequency signal by the laser excitation probe, and receiving the ultrasonic signal by the three-dimensional laser vibrometer; processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determining a target deep learning network model based on the nonlinear zero-frequency wave signal, and predicting the porosity of the target component by the target deep learning network model, wherein the target deep learning network model is used to map the corresponding relationship between the nonlinear zero-frequency wave signal and the porosity of the target component, thereby solving the problem of low detection accuracy of the porosity of additively manufactured components in related technologies. By constructing the target deep learning network model, the corresponding relationship between the nonlinear zero-frequency wave signal and the porosity of the target component is mapped, thereby achieving the effect of improving the detection accuracy of the porosity of additively manufactured components.

[0057] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0058] The present application also provides a nondestructive prediction device for component porosity. It should be noted that the nondestructive prediction device for component porosity in the present application can be used to execute the nondestructive prediction method for component porosity provided in the present application. The following describes the nondestructive prediction device for component porosity provided in the present application.

[0059] Figure 5 Schematic diagram of a nondestructive prediction device for component porosity according to an embodiment of the present application. Figure 5 As shown, the device 500 includes: a detection unit 501, a first determination unit 502, a receiving unit 503, and a second determination unit 504.

[0060] Specifically, the detection unit 501 is used to place the target component to be detected on the detection platform for frequency detection to obtain the frequency detection result, wherein the detection platform is placed on both sides with: a laser excitation probe and a three-dimensional laser vibrometer;

[0061] A first determining unit 502 is configured to determine the highest frequency signal in the frequency detection result;

[0062] A receiving unit 503 is configured to determine the ultrasonic signal corresponding to the highest frequency signal through a laser excitation probe, and receive the ultrasonic signal through a three-dimensional laser vibrometer;

[0063] The second determination unit 504 is used to process the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determine the target deep learning network model based on the nonlinear zero-frequency wave signal, and predict the porosity of the target component through the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component.

[0064] The nondestructive prediction device for component porosity provided in the embodiment of the present application includes a detection unit 501 for placing the target component to be detected on a detection platform for frequency detection to obtain a frequency detection result, wherein a laser excitation probe and a three-dimensional laser vibrometer are placed on both sides of the detection platform; a first determination unit 502 for determining the highest frequency signal in the frequency detection result; a receiving unit 503 for determining the ultrasonic signal corresponding to the highest frequency signal through the laser excitation probe and receiving the ultrasonic signal through the three-dimensional laser vibrometer; a second determination unit 504 for processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determining a target deep learning network model based on the nonlinear zero-frequency wave signal, and predicting the porosity of the target component through the target deep learning network model, wherein the target deep learning network model is used to map the corresponding relationship between the nonlinear zero-frequency wave signal and the porosity of the target component, thereby solving the problem of low detection accuracy of the porosity of additively manufactured components in the related art. By constructing a target deep learning network model, the corresponding relationship between the nonlinear zero-frequency wave signal and the porosity of the target component is mapped, thereby achieving the effect of improving the detection accuracy of the porosity of the additively manufactured component.

[0065] Optionally, in the non-destructive prediction device for component porosity provided in an embodiment of the present application, the second determination unit includes: a processing module for performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal; and a filtering module for performing low-pass filtering on the noise-reduced ultrasonic signal to obtain a filtered nonlinear zero-frequency wave signal.

[0066] Optionally, in the non-destructive prediction device for component porosity provided in an embodiment of the present application, the second determination unit includes: a determination module for determining a data set between a nonlinear zero-frequency wave signal and a porosity sample; a division module for dividing the data set into a training set and a test set; and a first training module for training an initial deep learning network model based on the training set and the test set to obtain a target deep learning network model.

[0067] Optionally, in the non-destructive prediction device for component porosity provided in an embodiment of the present application, the target deep learning network model includes: an input layer, a convolutional neural network module, a gated recurrent unit module, a fully connected layer and an output layer connected in sequence, wherein the convolutional neural network module includes a first convolutional layer, a first pooling layer, a second convolutional layer and a second pooling layer connected in sequence, and the gated recurrent unit module includes a first gated recurrent layer, a first activation layer, a second gated recurrent layer and a second activation layer.

[0068] Optionally, in the non-destructive prediction device for component porosity provided in an embodiment of the present application, the second determination unit includes: an extraction module for extracting local features of the nonlinear zero-frequency wave signal based on a convolutional neural network module; a second training module for training the dependency relationship between local features based on a gated recurrent unit module; a mapping module for mapping the dependency relationship through a fully connected layer to obtain a correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component; and a prediction module for determining the porosity of the target component based on the correspondence relationship.

[0069] Optionally, in the non-destructive prediction device for component porosity provided in an embodiment of the present application, the processing module includes: a determination submodule, used to determine the wavelet function, and perform a multi-layer wavelet transform on the ultrasonic signal based on the wavelet function to obtain the high-frequency coefficients and low-frequency coefficients of the ultrasonic signal; a processing submodule, used to process the high-frequency coefficients and the low-frequency coefficients respectively to obtain the denoised ultrasonic signal.

[0070] Optionally, in the non-destructive prediction device for component porosity provided in an embodiment of the present application, the target component is placed on a detection platform and remains in a fixed position during the detection process, wherein the laser excitation probe and the three-dimensional laser vibrometer are placed on both sides of the detection platform and are kept at a fixed interval.

[0071] The nondestructive prediction device for component porosity includes a processor and a memory. The detection unit 501, the first determination unit 502, the receiving unit 503, and the second determination unit 504 are all stored in the memory as program units. The processor executes the program units stored in the memory to implement corresponding functions.

[0072] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the porosity of the component can be non-destructively predicted by adjusting the kernel parameters.

[0073] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0074] In an exemplary embodiment of the present application, a computer storage medium capable of implementing the above method is also provided. A program product capable of implementing the above method of this specification is stored thereon. In some possible embodiments, various aspects of the present application can also be implemented in the form of a program product, which includes program code. When the program product is run on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments of the present application described in the "Exemplary Method" section above. For example, the following steps can be performed: placing the target component to be detected on a detection platform for frequency detection to obtain a frequency detection result, wherein the detection platform is placed on both sides with: a laser excitation probe and a three-dimensional laser vibrometer; determining the highest frequency signal in the frequency detection result; determining the ultrasonic signal corresponding to the highest frequency signal by the laser excitation probe, and receiving the ultrasonic signal by the three-dimensional laser vibrometer; processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determining a target deep learning network model based on the nonlinear zero-frequency wave signal, and predicting the porosity of the target component by the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component.

[0075] In an optional implementation: performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal; and performing low-pass filtering on the noise-reduced ultrasonic signal to obtain a filtered nonlinear zero-frequency wave signal.

[0076] In an optional embodiment: determining a data set between a nonlinear zero-frequency wave signal and a porosity sample; dividing the data set into a training set and a test set; training an initial deep learning network model based on the training set and the test set to obtain a target deep learning network model.

[0077] In an optional embodiment: it includes an input layer, a convolutional neural network module, a gated recurrent unit module, a fully connected layer and an output layer connected in sequence, wherein the convolutional neural network module includes a first convolutional layer, a first pooling layer, a second convolutional layer and a second pooling layer connected in sequence, and the gated recurrent unit module includes a first gated recurrent layer, a first activation layer, a second gated recurrent layer and a second activation layer.

[0078] In an optional embodiment: local features of the nonlinear zero-frequency wave signal are extracted based on a convolutional neural network module; dependencies between local features are trained based on a gated recurrent unit module; the dependencies are mapped through a fully connected layer to obtain a correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component; and the porosity of the target component is predicted based on the correspondence.

[0079] In an optional embodiment: a wavelet function is determined, and a multi-layer wavelet transform is performed on the ultrasonic signal based on the wavelet function to obtain high-frequency coefficients and low-frequency coefficients of the ultrasonic signal; the high-frequency coefficients and low-frequency coefficients are processed separately to obtain a noise-reduced ultrasonic signal.

[0080] In an optional embodiment, the target component is placed on a detection platform and kept in a fixed position during the detection process, wherein the laser excitation probe and the three-dimensional laser vibrometer are placed on both sides of the detection platform and kept at a fixed interval.

[0081] In an optional embodiment, the embodiments of the present application may further include a program product for implementing the above method, which may be a portable compact disc read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present application is not limited thereto. In this document, a readable storage medium may be any tangible medium containing or storing a program, which may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0082] The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0083] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0084] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0085] The program code for performing the operations of the present application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, and the like, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, as a stand-alone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0086] In addition, in an exemplary embodiment of the present application, an electronic device capable of implementing the above method is also provided.

[0087] Those skilled in the art will appreciate that various aspects of the present application can be implemented as systems, methods, or program products. Therefore, various aspects of the present application can be specifically implemented in the following forms: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation that combines hardware and software aspects, which may be collectively referred to herein as a "circuit," "module," or "system."

[0088] Refer to the following Figure 6 hereinafter, an electronic device 600 according to this embodiment of the present application is described. Figure 6 The electronic device 600 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0089] like Figure 6 As shown, electronic device 600 is implemented as a general-purpose computing device. Components of electronic device 600 may include, but are not limited to, the aforementioned at least one processing unit 610, the aforementioned at least one storage unit 620, a bus 630 connecting various system components (including storage unit 620 and processing unit 610), and a display unit 640.

[0090] The storage unit stores a program code, and the program code can be executed by the processing unit 610, so that the processing unit 610 executes the steps according to various exemplary embodiments of the present application described in the above "Exemplary Method" section of this specification. For example, the processing unit 610 can execute the following steps: placing the target component to be detected on the detection platform for frequency detection to obtain a frequency detection result, wherein the detection platform is placed on both sides with: a laser excitation probe and a three-dimensional laser vibrometer; determining the highest frequency signal in the frequency detection result; determining the ultrasonic signal corresponding to the highest frequency signal through the laser excitation probe, and receiving the ultrasonic signal through the three-dimensional laser vibrometer; processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determining a target deep learning network model based on the nonlinear zero-frequency wave signal, and predicting the porosity of the target component through the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component.

[0091] In an optional implementation: performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal; and performing low-pass filtering on the noise-reduced ultrasonic signal to obtain a filtered nonlinear zero-frequency wave signal.

[0092] In an optional embodiment: determining a data set between a nonlinear zero-frequency wave signal and a porosity sample; dividing the data set into a training set and a test set; training an initial deep learning network model based on the training set and the test set to obtain a target deep learning network model.

[0093] In an optional embodiment: it includes an input layer, a convolutional neural network module, a gated recurrent unit module, a fully connected layer and an output layer connected in sequence, wherein the convolutional neural network module includes a first convolutional layer, a first pooling layer, a second convolutional layer and a second pooling layer connected in sequence, and the gated recurrent unit module includes a first gated recurrent layer, a first activation layer, a second gated recurrent layer and a second activation layer.

[0094] In an optional embodiment: local features of the nonlinear zero-frequency wave signal are extracted based on a convolutional neural network module; dependencies between local features are trained based on a gated recurrent unit module; the dependencies are mapped through a fully connected layer to obtain a correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component; and the porosity of the target component is predicted based on the correspondence.

[0095] In an optional embodiment: a wavelet function is determined, and a multi-layer wavelet transform is performed on the ultrasonic signal based on the wavelet function to obtain high-frequency coefficients and low-frequency coefficients of the ultrasonic signal; the high-frequency coefficients and low-frequency coefficients are processed separately to obtain a noise-reduced ultrasonic signal.

[0096] In an optional embodiment, the target component is placed on a detection platform and kept in a fixed position during the detection process, wherein the laser excitation probe and the three-dimensional laser vibrometer are placed on both sides of the detection platform and kept at a fixed interval.

[0097] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache memory unit 6202 , and may further include a read-only memory unit (ROM) 6203 .

[0098] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0099] Bus 630 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0100] The electronic device 600 can also communicate with one or more external devices 700 (e.g., a keyboard, a pointing device, a Bluetooth device, etc.), one or more devices that enable a user to interact with the electronic device 600, and / or any device that enables the electronic device 600 to communicate with one or more other computing devices (e.g., a router, a modem, etc.). Such communication can occur via an input / output (I / O) interface 650. Furthermore, the electronic device 600 can communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network such as the Internet) via a network adapter 660. As shown, the network adapter 660 communicates with other modules of the electronic device 600 via a bus 630. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 600, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0101] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present application.

[0102] Furthermore, the above-mentioned figures are merely illustrative of the processes included in the methods according to exemplary embodiments of the present application and are not intended to be limiting. It is readily understood that the processes illustrated in the above-mentioned figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0103] Other embodiments of the present invention will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the claims.

Claims

1. A non-destructive prediction method for component porosity, characterized in that: include: The target component to be detected is placed on a detection platform for frequency detection to obtain a frequency detection result, wherein a laser excitation probe and a three-dimensional laser vibrometer are placed on both sides of the detection platform; Determining the highest frequency signal in the frequency detection results; Determining an ultrasonic signal corresponding to the highest frequency signal by the laser excitation probe, and receiving the ultrasonic signal by the three-dimensional laser vibrometer; The ultrasonic signal is processed to obtain a nonlinear zero-frequency wave signal, a target deep learning network model is determined based on the nonlinear zero-frequency wave signal, and the porosity of the target component is predicted by the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component.

2. The method according to claim 1, characterized in that Processing the ultrasonic signal to obtain a nonlinear zero-frequency wave signal includes: Performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal; The noise-reduced ultrasonic signal is low-pass filtered to obtain the filtered nonlinear zero-frequency wave signal.

3. The method according to claim 1, characterized in that Determining a target deep learning network model based on the ultrasonic signal includes: determining a data set between the nonlinear zero-frequency wave signal and a porosity sample; Dividing the data set into a training set and a test set; The initial deep learning network model is trained based on the training set and the test set to obtain the target deep learning network model.

4. The method according to claim 3, characterized in that The target deep learning network model includes: an input layer, a convolutional neural network module, a gated recurrent unit module, a fully connected layer and an output layer connected in sequence, wherein the convolutional neural network module includes a first convolutional layer, a first pooling layer, a second convolutional layer and a second pooling layer connected in sequence, and the gated recurrent unit module includes a first gated recurrent layer, a first activation layer, a second gated recurrent layer and a second activation layer.

5. The method according to claim 4, characterized in that Predicting the porosity of the target component by using the target deep learning network model includes: Extracting local features of the nonlinear zero-frequency wave signal based on the convolutional neural network module; Training the dependency relationship between the local features based on the gated recurrent unit module; Mapping the dependency relationship through the fully connected layer to obtain a corresponding relationship between the nonlinear zero-frequency wave signal and the porosity of the target component; The porosity of the target component is predicted according to the corresponding relationship.

6. The method according to claim 2, characterized in that Performing wavelet noise reduction processing on the ultrasonic signal to obtain a noise-reduced ultrasonic signal, including: Determining a wavelet function, and performing a multi-layer wavelet transform on the ultrasonic signal based on the wavelet function to obtain high-frequency coefficients and low-frequency coefficients of the ultrasonic signal; The high-frequency coefficients and the low-frequency coefficients are processed respectively to obtain the denoised ultrasonic signal.

7. The method according to claim 1, characterized in that The target component is placed on the detection platform and keeps a fixed position during the detection process, wherein the laser excitation probe and the three-dimensional laser vibrometer are placed on both sides of the detection platform and keep a fixed interval.

8. A nondestructive prediction device for component porosity, characterized in that: include: A detection unit is used to place the target component to be detected on a detection platform for frequency detection to obtain a frequency detection result, wherein a laser excitation probe and a three-dimensional laser vibrometer are placed on both sides of the detection platform; A first determining unit, configured to determine the highest frequency signal in the frequency detection result; a receiving unit, configured to determine an ultrasonic signal corresponding to the highest frequency signal through the laser excitation probe, and receive the ultrasonic signal through the three-dimensional laser vibrometer; The second determination unit is used to process the ultrasonic signal to obtain a nonlinear zero-frequency wave signal, determine a target deep learning network model based on the nonlinear zero-frequency wave signal, and predict the porosity of the target component through the target deep learning network model, wherein the target deep learning network model is used to map the correspondence between the nonlinear zero-frequency wave signal and the porosity of the target component.

9. A computer-readable storage medium, characterized in that The storage medium includes a stored program, wherein the program executes the non-destructive prediction method for component porosity according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: One or more processors, a memory, a display device, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the non-destructive prediction method of the component porosity according to any one of claims 1 to 7.

Images