Method, device and equipment for rapidly analyzing machining defects of remanufactured parts

Abstract

The invention relates to the field of acoustic nondestructive testing, and provides a rapid analysis method, device and equipment for machining defects of remanufactured parts, and the method comprises the following steps: establishing a finite element simulation model of a part to be analyzed; meshing the finite element simulation model into a finite element mesh and an infinite element mesh; applying a preset ultrasonic excitation signal to the excitation point; calculating the propagation process of the ultrasonic excitation signal in the to-be-analyzed part through a finite element simulation model, and collecting an ultrasonic echo signal at a signal collection point; extracting a plurality of characteristic parameters related to defect states from the collected ultrasonic echo signals; and inputting the plurality of extracted feature parameters into a pre-trained deep learning prediction model to obtain a residual service duration prediction result of the to-be-analyzed part. According to the method, the problem of insufficient ultrasonic detection accuracy in the prior art is solved, and the residual service time of the remanufactured part is quickly and accurately predicted.

Description

Technical Field

[0001] This invention relates to the field of acoustic nondestructive testing, and in particular to a method, apparatus and equipment for rapid analysis of defects in remanufactured parts. Background Technology

[0002] Acoustic nondestructive testing, especially ultrasonic testing technology, is a key means of assessing internal defects and structural integrity of mechanical components (parts particularly in the remanufacturing field). By analyzing the reflection, diffraction, and nonlinear effects generated when ultrasonic waves propagate through materials and interact with defects such as cracks, it is possible to locate and quantify defects and even predict the remaining service life of components.

[0003] Currently, ultrasonic testing methods based on finite element simulation have become an important tool for supplementing and guiding experimental research. Existing techniques typically employ pure finite element models to simulate the ultrasonic wave propagation process, setting low-reflection or absorbing boundary conditions at the model's external boundaries to approximate the propagation of sound waves in an infinite medium, thus avoiding interference from boundary reflected waves on the defect echo signal. Furthermore, after acquiring the simulated or experimental ultrasonic signal, an empirical correlation model is usually established based on one or a few acoustic characteristic parameters and the defect size or fatigue life for evaluation and prediction.

[0004] However, in simulation modeling, existing technologies struggle to balance computational accuracy and efficiency with traditional low-reflection boundary processing methods. Improper settings can easily lead to acoustic wave reflections interfering with the integrity of signal acquisition, failing to accurately simulate the true dissipation behavior of ultrasonic waves at component boundaries and affecting the reliability of simulation results. In signal analysis and lifespan prediction, analysis methods relying on single or limited characteristic parameters fail to fully utilize the rich multidimensional information contained in ultrasonic echo signals, resulting in limited ability to characterize complex defect states and insufficient accuracy in predicting the remaining service life of components. Summary of the Invention

[0005] This invention provides a method, apparatus, and equipment for rapid analysis of processing defects in remanufactured parts, which solves the problem of insufficient accuracy of ultrasonic testing in the prior art and enables rapid and accurate prediction of the remaining service life of remanufactured parts.

[0006] This invention provides a rapid analysis method for processing defects in remanufactured parts, comprising the following steps: A finite element simulation model of the part to be analyzed is established. The simulation model includes excitation points, signal acquisition points, an intermediate mesh component for characterizing the intermediate solid region of the part, and a boundary mesh component surrounding the intermediate mesh component. The finite element simulation model is meshed, wherein the intermediate mesh component is divided into a finite element mesh, and the boundary mesh component is divided into an infinite element mesh; A preset ultrasonic excitation signal is applied to the excitation point; The propagation process of the ultrasonic excitation signal in the part to be analyzed is calculated using the finite element simulation model, and ultrasonic echo signals are collected at the signal acquisition point. Multiple feature parameters related to the defect state are extracted from the acquired ultrasonic echo signals; The extracted feature parameters are input into a pre-trained deep learning prediction model to obtain the prediction result of the remaining service time of the part to be analyzed; wherein, the deep learning prediction model is a neural network model that integrates the temporal convolutional network TCN module and the Transformer module.

[0007] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided. The step of establishing a finite element simulation model of the part to be analyzed includes: presetting a crack model in the central region of the geometric model of the part to be analyzed; dividing the geometric model into regions on both sides of the crack model to obtain a central region containing the crack model and boundary regions located on both sides of the central region; assigning material properties to the central region and the boundary regions respectively, and setting excitation points and signal acquisition points on the central region to obtain the constructed finite element simulation model.

[0008] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided. The step of meshing the finite element simulation model includes: meshing the intermediate mesh component with a global size no greater than one-quarter of the ultrasonic wavelength; seeding the boundary mesh component with a mesh size larger than the global size, and then performing a first meshing; performing a second meshing on the boundary mesh component based on a sweeping method to generate a quadrilateral dominant mesh with the sweeping direction pointing outward from the model; and configuring the mesh element type of the boundary mesh component obtained after the second meshing as a two-dimensional acoustic infinite element.

[0009] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided, wherein configuring the mesh element type of the boundary mesh component obtained after the second meshing is configured as a two-dimensional acoustic infinite element, comprising: in finite element analysis software, initially defining the boundary mesh component as a two-dimensional acoustic quadrilateral element and generating a corresponding input file; editing the input file to locate the mesh element definition part related to the boundary mesh component; modifying the mesh element type identifier in the mesh element definition part to a preset code for characterizing the two-dimensional acoustic infinite element; and performing calculations based on the preset code of the two-dimensional acoustic infinite element, so that the boundary mesh component functions as a two-dimensional acoustic infinite element in the simulation.

[0010] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided. The step of acquiring ultrasonic echo signals at the signal acquisition point includes: applying the ultrasonic excitation signal in the form of a concentrated force at the excitation point; setting an explicit dynamic analysis step and configuring a fixed, small time increment step that is smaller than the period of the excitation signal; defining the displacement data sequence of the signal acquisition point with the small time increment step as the interval in the history output; and after performing finite element calculation, extracting the time-domain waveform from the displacement data sequence as an ultrasonic echo signal containing crack reflection and diffraction information.

[0011] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided. The method involves extracting multiple characteristic parameters related to the defect state from the acquired ultrasonic echo signal, including: performing time-frequency transformation on the time-domain waveform of the ultrasonic echo signal to obtain the corresponding frequency-domain signal; calculating multiple statistical characteristics of the time-domain and frequency-domain signals as energy characteristic parameters reflecting signal energy; calculating the nonlinear coefficient generated by the interaction between ultrasonic waves and defects based on harmonic analysis as a nonlinear characteristic parameter reflecting material damage; calculating the waveform distortion parameter between the time-domain waveform and the defect-free reference waveform based on a crack propagation theory model as a damage correlation parameter reflecting the degree of crack propagation; and combining the energy characteristic parameter, nonlinear characteristic parameter, and damage correlation parameter to form multiple characteristic parameters related to the defect state.

[0012] The present invention also provides a rapid analysis device for processing defects in remanufactured parts, comprising the following modules: The defect integrated modeling module is used to establish a finite element simulation model of the part to be analyzed. The simulation model includes excitation points, signal acquisition points, an intermediate mesh component for characterizing the intermediate solid region of the part, and a boundary mesh component surrounding the intermediate mesh component. The mesh generation module is used to perform mesh generation on the finite element simulation model, wherein the intermediate mesh component is divided into a finite element mesh, and the boundary mesh component is divided into an infinite element mesh. An ultrasonic excitation loading module is used to apply a preset ultrasonic excitation signal to the excitation point; The echo acquisition module is used to calculate the propagation process of the ultrasonic excitation signal in the part to be analyzed through the finite element simulation model, and to acquire ultrasonic echo signals at the signal acquisition point. The feature extraction module is used to extract multiple feature parameters related to the defect state from the acquired ultrasonic echo signal; The life prediction module is used to input the extracted feature parameters into a pre-trained deep learning prediction model to obtain the prediction result of the remaining service life of the part to be analyzed; wherein, the deep learning prediction model is a neural network model that integrates a temporal convolutional network (TCN) module and a Transformer module.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the rapid analysis method for remanufactured part processing defects as described above.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the rapid analysis method for remanufactured part defects as described above.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the rapid analysis method for remanufactured part processing defects as described above.

[0016] This invention provides a rapid analysis method, apparatus, and equipment for remanufactured parts, offering the following advantages: By combining finite element meshes with infinite element meshes to divide the simulation model, the technical challenge of accurately simulating boundary acoustic wave reflections in traditional pure finite element simulations is effectively solved. Configuring infinite element meshes for boundary mesh components enables a more realistic simulation of ultrasonic energy dissipation and wavefront divergence at component boundaries, resulting in purer and more complete defect echo signals at signal acquisition points, significantly improving the accuracy and reliability of ultrasonic propagation simulation. Multiple-dimensional feature parameters are extracted from the acquired ultrasonic echo signals, constructing a more comprehensive and in-depth characterization of the defect state. By employing a deep learning prediction model that integrates a temporal convolutional network (TCN) module and a Transformer module, the local temporal evolution patterns and global correlations inherent in multiple feature parameters can be collaboratively captured, thereby achieving a more accurate prediction of the remaining service life of the parts. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1This is a flowchart illustrating the rapid analysis method for processing defects in remanufactured parts provided by the present invention.

[0019] Figure 2 This is a schematic diagram of the grid division provided by the present invention.

[0020] Figure 3 This is a waveform diagram of the ultrasonic excitation signal provided by the present invention.

[0021] Figure 4 This is a schematic diagram of the ultrasonic wave propagation process provided by the present invention.

[0022] Figure 5 This is a diagram of the Transformer-TCN model architecture for predicting remaining service time provided by the present invention.

[0023] Figure 6 This is a schematic diagram of the structure of the rapid analysis device for processing defects in remanufactured parts provided by the present invention.

[0024] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0026] The following is combined Figures 1-7 The embodiments of the present invention are described in detail.

[0027] The rapid analysis method for remanufactured parts processing defects provided in this embodiment of the invention is executed by a rapid analysis device for remanufactured parts processing defects. This device can be configured in a computer, which can be a local computer or a cloud computer. The local computer can be a computer, tablet, etc., and no specific limitation is made here.

[0028] Figure 1 This is a flowchart illustrating the rapid analysis method for processing defects in remanufactured parts provided by the present invention, as shown below. Figure 1 As shown, the method includes the following steps: S110. Establish a finite element simulation model of the part to be analyzed. The simulation model includes excitation points, signal acquisition points, an intermediate mesh component used to characterize the intermediate solid region of the part, and a boundary mesh component surrounding the intermediate mesh component.

[0029] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided, which establishes a finite element simulation model of the part to be analyzed, including: pre-setting a crack model in the central region of the geometric model of the part to be analyzed; dividing the geometric model into regions on both sides of the crack model to obtain a central region containing the crack model and boundary regions located on both sides of the central region; assigning material properties to the central region and the boundary regions respectively, and setting excitation points and signal acquisition points on the central region to obtain the constructed finite element simulation model.

[0030] Specifically, such as Figure 2 The diagram shows a schematic of the mesh generation. Taking the simulation analysis of a remanufactured part in a two-dimensional plane as an example, firstly, a rectangular plate geometric model is created in the computer-aided engineering software to represent the matrix of the part to be analyzed. To simulate real defects inside the part, an elliptical crack model is preset in the central region of this rectangular geometric model. The dimensions of this crack model are set as follows: major radius 50µm, minor radius 10nm, to accurately simulate the micron / nanoscale crack characteristics that may occur in actual processing.

[0031] Next, the geometric model is structurally divided into regions. Using the elliptical crack model as the central reference, the entire model is divided into three continuous regions along the thickness direction of the part (i.e., the vertical direction of the rectangle). The central portion containing the elliptical crack is defined as the middle region, which represents the solid material portion of the part. On the upper and lower sides of the middle region, a narrow strip-shaped region parallel to the long side of the rectangle is defined as the upper boundary region and the lower boundary region, respectively. The width of these two boundary regions can be parameterized; in this embodiment, it is set to 0.4 mm. The principle is to set it to be less than half the width of the original rectangle to ensure that the middle region occupies the main part of the model. After completing the geometric division, material properties are assigned to the model. Taking 316L stainless steel as an example, key material parameters such as density (ρ), elastic modulus (E), and Poisson's ratio (ν) are assigned to the middle region, upper boundary region, and lower boundary region, respectively. Finally, in the central region, based on the actual arrangement of the ultrasonic transducers, two key locations were set on the model: one point was defined as the excitation point for applying ultrasonic loads subsequently; the other point was defined as the signal acquisition point for acquiring ultrasonic echo signals. At this point, a finite element simulation model integrating preset cracks, partitioned structures, and material properties was completed.

[0032] This embodiment constructs a simulation analysis object with a clear structure and well-defined physical meaning by pre-setting a crack in the geometric model and performing two-sided region division. The part is clearly divided into an intermediate solid region containing defects and upper and lower boundary regions for special boundary treatment, laying a precise geometric foundation for subsequent innovative hybrid mesh generation (finite element method for the intermediate region and infinite element method for the boundary regions). This method ensures that the crack is contained within the core region requiring detailed analysis, while reserving independent and controllable operational space for special treatment of the boundary regions (configuring infinite element methods to eliminate artificial boundary reflections), thus guaranteeing the reliability and accuracy of subsequent ultrasonic wave propagation simulation results from the model's source.

[0033] S120. Mesh the finite element simulation model, wherein the intermediate mesh components are divided into finite element meshes, and the boundary mesh components are divided into infinite element meshes.

[0034] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided, which involves meshing a finite element simulation model, including: meshing the intermediate mesh components with a global size no greater than one-quarter of the ultrasonic wavelength; seeding the boundary mesh components with a mesh size larger than the global size, and then performing a first meshing; performing a second meshing on the boundary mesh components based on a sweeping method to generate a quadrilateral dominant mesh with the sweeping direction pointing outward from the model; and configuring the mesh element type of the boundary mesh components obtained after the second meshing as a two-dimensional acoustic infinite element.

[0035] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided, which configures the mesh element type of the boundary mesh component obtained after the second mesh generation as a two-dimensional acoustic infinite element. The method includes: in finite element analysis software, initially defining the boundary mesh component as a two-dimensional acoustic quadrilateral element and generating a corresponding input file; editing the input file to locate the mesh element definition section related to the boundary mesh component; modifying the mesh element type identifier in the mesh element definition section to a preset code used to characterize the two-dimensional acoustic infinite element; and performing calculations based on the preset code of the two-dimensional acoustic infinite element, so that the boundary mesh component functions as a two-dimensional acoustic infinite element in the simulation.

[0036] Specifically, based on the completed finite element simulation model, the wavelength λ of the ultrasonic wave in the part material is first calculated according to the frequency (e.g., 5MHz) of the selected ultrasonic excitation signal and its propagation speed in the material. Subsequently, the intermediate mesh component representing the intermediate solid region of the part is meshed, and the global size is less than λ / 4, for example, set to 0.03mm in this embodiment, to ensure sufficient spatial resolution for high-frequency ultrasonic waves.

[0037] For the upper and lower boundary mesh components surrounding the middle mesh component, since they will ultimately be configured as infinite elements and have lower requirements for computational accuracy, a mesh size larger than the aforementioned global size is used for meshing. Specifically, on the wide edges (i.e., edges along the boundary length) of each of these two boundary mesh components, local seeds are set by number, for example, setting the seed count to 1, and then the first meshing is performed to generate a preliminary mesh. Next, a second meshing is performed on these two boundary mesh components using a sweeping method. The key is to set the sweep direction to point outwards from the model (i.e., away from the middle mesh component) to generate a mesh dominated by quadrilateral elements, with the sweep direction conforming to the requirements of infinite element calculation. In the graphical interface of finite element analysis software (such as ABAQUS), the element type of the boundary mesh components obtained after the second meshing is initially set to a conventional two-dimensional acoustic quadrilateral element (e.g., AC2D4R element).

[0038] After completing the above graphical interface operations, the corresponding input file (INP file) is generated. Then, open the INP file with a text editor, locate the mesh element definition section related to the boundary mesh component (usually identified by keywords such as "*ELEMENT,TYPE=AC2D4R"). Modify the element type identifier code in this section to the specific preset code used by the software to represent two-dimensional acoustic infinite element (for example, modify it to "ACIN2D4" or the infinite element code corresponding to the software version). Save the modified INP file and submit it to the finite element analysis solver for calculation. At this point, the boundary mesh component is effective as a two-dimensional acoustic infinite element in the simulation.

[0039] This embodiment achieves a precise combination of finite and infinite element methods by employing a hybrid mesh generation method that uses different regions and strategies, and cleverly combines the initial setup via a graphical interface with post-processing modifications to the input file. This ensures both the high-precision simulation of the interaction between ultrasound and defects in the central region and the efficient resolution of the boundary acoustic wave reflection problem. By configuring the boundary region as an infinite element, the acoustic wave energy propagating there is reasonably dissipated and diffused during calculation, thus eliminating non-physical reflections caused by artificial boundaries. This significantly improves the realism and accuracy of the ultrasound propagation simulation results, while avoiding excessive computational burden caused by using overly dense meshes or complex boundary conditions in the entire model.

[0040] S130. Apply a preset ultrasonic excitation signal to the excitation point.

[0041] Specifically, after the finite element simulation model has been meshed and the excitation points defined, this embodiment uses a sinusoidal Gaussian signal with an eight-cycle carrier wave nested within a Hanning amplitude window as the preset ultrasonic excitation signal, the waveform of which is shown in the figure below. Figure 3As shown. The ultrasonic excitation signal is composed of a Gaussian envelope sine wave with a center frequency of f0 (e.g., f0 = 5MHz) multiplied by a Hanning window function. It has the characteristics of high frequency, narrow bandwidth, and rapid attenuation, good directionality, and strong anti-interference ability. In the finite element analysis software, the calculated time-domain data sequence of the ultrasonic excitation signal is applied to the defined excitation point in the form of a concentrated force. When applying the load, the direction of the force is set to be perpendicular to the normal direction of the surface where the excitation point is located, to simulate the excitation mode of an actual ultrasonic longitudinal wave probe, such as... Figure 4 This is a schematic diagram of the ultrasonic wave propagation process. Meanwhile, based on the stability requirements of explicit dynamic analysis, the fixed time increment step size of the analysis step is set to be much smaller than the excitation signal period (for example, for a 5MHz signal, the period is 200ns, and the time increment step size is set to 0.1ps) to ensure that the propagation process of the high-frequency signal can be calculated stably and accurately.

[0042] This embodiment provides a high-quality, high-fidelity acoustic excitation source for subsequent defect detection simulation by applying a carefully designed ultrasonic excitation signal with a specific waveform. The signal, modulated with a Gaussian envelope and Hanning window, exhibits concentrated energy and steep start and end points in the time domain, and a prominent main lobe and good side lobe suppression in the frequency domain. This effectively enhances the penetration and directionality of the ultrasonic wave while reducing out-of-band noise interference. Applying this signal precisely as a concentrated force to the excitation point simulates the point source excitation of an actual probe, laying a crucial foundation for obtaining clear ultrasonic echo signals containing rich defect information at the signal acquisition point. This improves the reliability of the entire simulation process and the sensitivity of defect analysis.

[0043] S140. Calculate the propagation process of the ultrasonic excitation signal in the part to be analyzed using a finite element simulation model, and collect ultrasonic echo signals at the signal acquisition point.

[0044] According to the present invention, a rapid analysis method for processing defects in remanufactured parts is provided, which involves acquiring ultrasonic echo signals at signal acquisition points, including: applying an ultrasonic excitation signal in the form of a concentrated force at the excitation point; setting an explicit dynamic analysis step and configuring a fixed, small time increment step that is smaller than the period of the excitation signal; defining a displacement data sequence of the signal acquisition point with the small time increment step as the interval in the history output; and after performing finite element calculation, extracting the time-domain waveform from the displacement data sequence as an ultrasonic echo signal containing information on crack reflection and diffraction.

[0045] Specifically, after the ultrasonic excitation signal has been applied to the finite element simulation model, first, an explicit dynamic analysis step is set up, for example, using the ABAQUS / Explicit solver. To ensure the stability and accuracy of the high-frequency ultrasonic propagation calculation, according to the Courant-Friedrichs-Lewy (CFL) stability condition, a fixed and very small time increment step much smaller than the excitation signal period must be configured. For an excitation signal with a central frequency of f0 (for example, f0 = 5 MHz and a period T = 1 / f0 = 200 ns), in this embodiment, the fixed time increment step size Δt is set to 0.1 ps, satisfying the condition of Δt << T. At the same time, the total time length of the analysis step is set to a duration sufficient for the ultrasonic excitation signal to propagate completely and be collected, for example, 5 ns. When defining the output requests of the analysis step, variables such as stress and strain in the entire model area are recorded in the field output for post-processing cloud map display; in the history output, the key setting is to continuously output the displacement data of the predefined signal acquisition points at intervals of the small time increment step Δt. After performing the finite element calculation, the solver will generate the displacement values of the signal acquisition points at each time increment step, forming a displacement data sequence with high time resolution. Subsequently, this sequence is extracted from the calculation results and used as the original time-domain waveform data. This waveform data directly reflects the complete acoustic response of the excitation signal after being emitted from the excitation point, interacting with the elliptical crack (generating reflection, diffraction, scattering) during propagation, and finally reaching the signal acquisition point, thus constituting an ultrasonic echo signal containing rich defect information.

[0046] In this embodiment, by setting up an explicit dynamic analysis step and using a very small fixed time increment step much smaller than the signal period for high-resolution calculation, the numerical stability and time accuracy of the high-frequency ultrasonic propagation simulation are ensured. By accurately capturing the displacement sequence of the signal acquisition points changing at extremely small time steps in the history output, the original time-domain response data with high fidelity is obtained. This method directly extracts the complete waveform information containing all the acoustic effects caused by the crack, providing an original signal basis with no distortion and high signal-to-noise ratio for subsequent high-precision feature extraction and life prediction, thus ensuring the accuracy and reliability of defect analysis and remaining life assessment.

[0047] S150. Extract multiple characteristic parameters related to the defect state from the collected ultrasonic echo signal.

[0048] According to the present invention, a rapid analysis method for processing defects in remanufactured parts extracts multiple characteristic parameters related to the defect state from the acquired ultrasonic echo signal, including: performing time-frequency transformation on the time-domain waveform of the ultrasonic echo signal to obtain the corresponding frequency-domain signal; calculating multiple statistical characteristics of the time-domain signal and the frequency-domain signal as energy characteristic parameters reflecting signal energy; calculating the nonlinear coefficient generated by the interaction between ultrasonic waves and defects based on harmonic analysis as a nonlinear characteristic parameter reflecting material damage; calculating the waveform distortion parameter between the time-domain waveform and the defect-free reference waveform based on the crack propagation theory model as a damage correlation parameter reflecting the degree of crack propagation; and combining the energy characteristic parameter, the nonlinear characteristic parameter, and the damage correlation parameter to form multiple characteristic parameters related to the defect state.

[0049] Specifically, the time-domain waveform s(t) of the ultrasonic echo signal is transformed using a time-frequency method, such as Fast Fourier Transform (FFT), to obtain its corresponding frequency-domain signal. Then, based on the time-domain waveform s(t), its root mean square (RMS) value, peak amplitude, and waveform factor (the ratio of peak value to RMS value) are calculated; based on the frequency-domain signal S(f), its spectral center frequency, spectral bandwidth, and spectral centroid are calculated. These statistics are used together as energy characteristic parameters reflecting the overall energy distribution and concentration of the signal. Next, based on the principle of harmonic analysis, the nonlinear effects generated by the interaction between the ultrasonic wave and the crack defect are calculated. Specifically, the amplitude ratio of the second harmonic component to the fundamental component, i.e., the relative nonlinear coefficient, is calculated. This coefficient reflects the enhancement of the nonlinear acoustic behavior of the material due to damage and serves as a nonlinear characteristic parameter. Finally, based on the physical model of crack propagation, waveform distortion parameters are calculated. The current defective time-domain waveform is compared with a defect-free baseline waveform obtained through the same simulation. The cross-correlation coefficient between the two is calculated, and the normalized root mean square error is also calculated. These parameters reflect the changes in signal propagation paths and the degree of energy scattering caused by the presence of cracks, serving as damage-related parameters. Finally, the aforementioned energy characteristic parameters, nonlinear characteristic parameters, and damage-related parameters are combined to form a multidimensional feature vector, which serves as multiple defect-state-related feature parameters for lifetime prediction.

[0050] This embodiment employs multi-level and multi-dimensional quantitative processing of ultrasonic echo signals, extracting feature parameters from multiple physical perspectives, including energy statistics, nonlinear acoustics, and waveform distortion. This method not only utilizes traditional energy information but also reveals changes in microscopic material damage through nonlinear coefficients and directly quantifies the impact of macroscopic cracks on the sound wave propagation path through waveform distortion parameters. The resulting multi-dimensional feature vector comprehensively and complementaryly characterizes the size, type, and impact on material properties of the crack. Compared to single-parameter methods, it more accurately depicts the defect state, providing information-rich and physically meaningful input features for subsequent deep learning-based remaining lifetime prediction models, thereby significantly improving the accuracy and reliability of lifetime prediction.

[0051] S160. Input the extracted feature parameters into a pre-trained deep learning prediction model to obtain the prediction result of the remaining service life of the part to be analyzed. The deep learning prediction model is a neural network model that integrates the temporal convolutional network (TCN) module and the Transformer module.

[0052] Specifically, after obtaining multiple feature parameters related to the defect state, they are first standardized to ensure that all features are within a similar numerical range. Then, the processed feature vectors are input into a pre-trained deep learning prediction model. The core architecture of this model consists of a deeply coupled temporal convolutional network module and a Transformer module, as shown in the diagram below. Figure 5 As shown.

[0053] The Transformer module employs a two-layer encoder structure, with eight attention heads in each layer to capture the global correlations and dependencies between different feature parameters in the input feature vector. Its feedforward network uses the Sigmoid activation function and sets the dropout rate to 0.1 to prevent overfitting. After processing by the Transformer module, the features are reorganized and input into the temporal convolutional network module.

[0054] This TCN module contains two one-dimensional dilated causal convolutional layers. The first layer has a kernel size of 3, a dilation coefficient of 1, and 64 output channels; the second layer has a kernel size of 3, a dilation coefficient of 2, and 32 output channels. Each convolutional operation is followed by batch normalization, a sigmoid activation function, and a dropout rate of 0.1. Through dilated causal convolution, the TCN module effectively captures the local evolution patterns of features in a dimensional sequence (which can be viewed as a kind of implicit temporal relationship between features).

[0055] The output of the TCN module is compressed using an adaptive max-pooling layer, then connected to a fully connected layer with 128 neurons and activated by ReLU. Finally, a continuous value is generated through an output neuron (using a linear activation function), which represents the predicted remaining service life of the analyzed part. This model has been trained using a dataset containing numerous simulation samples with different crack states and their corresponding remaining service lives, enabling accurate mapping from multi-dimensional features to remaining service life.

[0056] This embodiment employs a deep learning prediction model that integrates a Temporal Convolutional Network (TCN) and a Transformer module, fully leveraging the advantages of TCN in capturing local sequence dependencies and Transformer in modeling global contextual relationships. This model can collaboratively extract deep, complex patterns reflecting crack evolution from multi-dimensional input defect features, synthesizing multiple physically meaningful features into an accurate lifetime prediction. Compared to single or shallow models, this deeply integrated model architecture significantly improves the ability to learn and represent complex nonlinear relationships (i.e., the mapping from multimodal defect features to remaining lifetime), thereby achieving more accurate and reliable intelligent prediction of the remaining service life of remanufactured parts.

[0057] The rapid analysis device for remanufactured parts processing defects provided by the present invention is described below. The rapid analysis device for remanufactured parts processing defects described below and the rapid analysis method for remanufactured parts processing defects described above can be referred to in correspondence.

[0058] like Figure 6 The image shows a rapid analysis device for processing defects in remanufactured parts provided by the present invention, comprising: The defect integrated modeling module 610 is used to establish a finite element simulation model of the part to be analyzed. The simulation model includes excitation points, signal acquisition points, intermediate mesh components used to characterize the intermediate solid region of the part, and boundary mesh components surrounding the intermediate mesh components. The mesh generation module 620 is used to generate meshes for the finite element simulation model, wherein the intermediate mesh components are divided into finite element meshes and the boundary mesh components are divided into infinite element meshes. The ultrasonic excitation loading module 630 is used to apply a preset ultrasonic excitation signal to the excitation point; The echo acquisition module 640 is used to calculate the propagation process of the ultrasonic excitation signal in the part to be analyzed through a finite element simulation model, and to acquire ultrasonic echo signals at the signal acquisition point. The feature extraction module 650 is used to extract multiple feature parameters related to the defect state from the acquired ultrasonic echo signal; The life prediction module 660 is used to input multiple extracted feature parameters into a pre-trained deep learning prediction model to obtain the prediction result of the remaining service life of the part to be analyzed; wherein, the deep learning prediction model is a neural network model that integrates the temporal convolutional network TCN module and the Transformer module.

[0059] Specifically, the functions of each module in the user account management system provided in this embodiment of the invention correspond one-to-one with the operation flow of each step in the above method-like embodiments, and the achieved effects are also the same. For details, please refer to the above embodiments, and this will not be repeated in this embodiment of the invention.

[0060] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7 As shown, the electronic device may include: a processor 710, a communications interface 720, a memory 730, and a communications bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other through the communications bus 740. The processor 710 can call logic instructions in the memory 730 to execute a rapid analysis method for processing defects in remanufactured parts. This method includes: establishing a finite element simulation model of the part to be analyzed, the simulation model including excitation points, signal acquisition points, an intermediate mesh component representing the intermediate solid region of the part, and boundary mesh components surrounding the intermediate mesh component; meshing the finite element simulation model, wherein the intermediate mesh component is divided into a finite element mesh, and the boundary mesh component is divided into an infinite element mesh; applying a preset ultrasonic excitation signal to the excitation points; calculating the propagation process of the ultrasonic excitation signal in the part to be analyzed using the finite element simulation model, and acquiring ultrasonic echo signals at the signal acquisition points; extracting multiple feature parameters related to the defect state from the acquired ultrasonic echo signals; inputting the extracted feature parameters into a pre-trained deep learning prediction model to obtain a prediction result of the remaining service life of the part to be analyzed; wherein the deep learning prediction model is a neural network model integrating a temporal convolutional network (TCN) module and a Transformer module.

[0061] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0062] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the rapid analysis method for processing defects of remanufactured parts provided by the above methods. The method includes: establishing a finite element simulation model of the part to be analyzed, the simulation model including excitation points, signal acquisition points, an intermediate mesh component for characterizing the intermediate solid region of the part, and a boundary mesh component surrounding the intermediate mesh component; meshing the finite element simulation model, wherein the intermediate mesh component is divided into a finite element mesh and the boundary mesh component is divided into an infinite element mesh; applying a preset ultrasonic excitation signal to the excitation points; calculating the propagation process of the ultrasonic excitation signal in the part to be analyzed through the finite element simulation model, and acquiring ultrasonic echo signals at the signal acquisition points; extracting multiple feature parameters related to the defect state from the acquired ultrasonic echo signals; inputting the extracted multiple feature parameters into a pre-trained deep learning prediction model to obtain the prediction result of the remaining service life of the part to be analyzed; wherein the deep learning prediction model is a neural network model that integrates a temporal convolutional network (TCN) module and a Transformer module.

[0063] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for rapid analysis of machining defects in remanufactured parts provided by the methods described above. This method includes: establishing a finite element simulation model of the part to be analyzed, the simulation model including excitation points, signal acquisition points, an intermediate mesh component for characterizing the intermediate solid region of the part, and boundary mesh components surrounding the intermediate mesh component; meshing the finite element simulation model, wherein the intermediate mesh component is divided into a finite element mesh, and the boundary mesh component is divided into an infinite element mesh; applying a preset ultrasonic excitation signal to the excitation points; calculating the propagation process of the ultrasonic excitation signal in the part to be analyzed using the finite element simulation model, and acquiring ultrasonic echo signals at the signal acquisition points; extracting multiple feature parameters related to the defect state from the acquired ultrasonic echo signals; inputting the extracted feature parameters into a pre-trained deep learning prediction model to obtain a prediction result of the remaining service life of the part to be analyzed; wherein the deep learning prediction model is a neural network model integrating a temporal convolutional network (TCN) module and a Transformer module.

[0064] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0065] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A rapid analysis method for processing defects in remanufactured parts, characterized in that, include: A finite element simulation model of the part to be analyzed is established. The simulation model includes excitation points, signal acquisition points, an intermediate mesh component for characterizing the intermediate solid region of the part, and a boundary mesh component surrounding the intermediate mesh component. The finite element simulation model is meshed, wherein the intermediate mesh component is divided into a finite element mesh, and the boundary mesh component is divided into an infinite element mesh; A preset ultrasonic excitation signal is applied to the excitation point; The propagation process of the ultrasonic excitation signal in the part to be analyzed is calculated using the finite element simulation model, and ultrasonic echo signals are collected at the signal acquisition point. Multiple feature parameters related to the defect state are extracted from the acquired ultrasonic echo signals; The extracted feature parameters are input into a pre-trained deep learning prediction model to obtain the prediction result of the remaining service time of the part to be analyzed; wherein, the deep learning prediction model is a neural network model that integrates the temporal convolutional network TCN module and the Transformer module.

2. The rapid analysis method for processing defects in remanufactured parts according to claim 1, characterized in that, The establishment of the finite element simulation model of the part to be analyzed includes: A crack model is preset in the central region of the geometric model of the part to be analyzed; The geometric model is divided into regions on both sides of the crack model to obtain a central region containing the crack model and boundary regions located on both sides of the central region. Material properties are assigned to the intermediate region and the boundary region respectively, and excitation points and signal acquisition points are set on the intermediate region to obtain the constructed finite element simulation model.

3. The rapid analysis method for processing defects in remanufactured parts according to claim 1, characterized in that, The mesh generation of the finite element simulation model includes: The intermediate mesh component is meshed using a global size no larger than one-quarter of the ultrasonic wavelength; The boundary mesh component is seeded locally using a mesh size larger than the global size, and then the first mesh generation is performed; Based on the sweeping method, the boundary mesh component is subjected to a second meshing to generate a quadrilateral dominant mesh with the sweeping direction pointing outward from the model; Configure the mesh element type of the boundary mesh component obtained after the second meshing as a two-dimensional acoustic infinite element.

4. The rapid analysis method for processing defects in remanufactured parts according to claim 3, characterized in that, The configuration of the boundary mesh component obtained after the second mesh division as a two-dimensional acoustic infinite element includes: In the finite element analysis software, the boundary mesh component is initially defined as a two-dimensional acoustic quadrilateral element and the corresponding input file is generated; Edit the input file to locate the mesh cell definition section related to the boundary mesh component; Modify the grid cell type identifier in the grid cell definition section to a preset code used to characterize two-dimensional acoustic infinite elements; Calculations are performed based on the preset code of the two-dimensional acoustic infinite element, so that the boundary mesh component functions as a two-dimensional acoustic infinite element unit in the simulation.

5. The rapid analysis method for processing defects in remanufactured parts according to claim 1, characterized in that, The acquisition of ultrasonic echo signals at the signal acquisition point includes: The ultrasonic excitation signal is applied at the excitation point in the form of a concentrated force. Set an explicit dynamic analysis step, and configure a fixed, small time increment step that is smaller than the period of the excitation signal; In the process output, the displacement data sequence of the signal acquisition point is defined with the tiny time increment step as the interval; After performing finite element calculations, time-domain waveforms are extracted from the displacement data sequence as ultrasonic echo signals containing crack reflection and diffraction information.

6. The rapid analysis method for processing defects in remanufactured parts according to claim 1, characterized in that, The extraction of multiple feature parameters related to the defect state from the acquired ultrasonic echo signal includes: The time-domain waveform of the ultrasonic echo signal is transformed by time-frequency conversion to obtain the corresponding frequency-domain signal; Calculate multiple statistical characteristics of the time-domain and frequency-domain signals as energy characteristic parameters reflecting the signal energy; The nonlinear coefficient generated by the interaction between ultrasonic waves and defects is calculated based on harmonic analysis and used as a nonlinear characteristic parameter reflecting material damage. Based on the crack propagation theory model, the waveform distortion parameter between the time-domain waveform and the defect-free reference waveform is calculated as a damage correlation parameter reflecting the degree of crack propagation. The energy characteristic parameters, nonlinear characteristic parameters, and damage correlation parameters are combined to form multiple characteristic parameters related to the defect state.

7. A rapid analysis device for processing defects in remanufactured parts, characterized in that, include: The defect integrated modeling module is used to establish a finite element simulation model of the part to be analyzed. The simulation model includes excitation points, signal acquisition points, an intermediate mesh component for characterizing the intermediate solid region of the part, and a boundary mesh component surrounding the intermediate mesh component. The mesh generation module is used to perform mesh generation on the finite element simulation model, wherein the intermediate mesh component is divided into a finite element mesh, and the boundary mesh component is divided into an infinite element mesh. An ultrasonic excitation loading module is used to apply a preset ultrasonic excitation signal to the excitation point; The echo acquisition module is used to calculate the propagation process of the ultrasonic excitation signal in the part to be analyzed through the finite element simulation model, and to acquire ultrasonic echo signals at the signal acquisition point. The feature extraction module is used to extract multiple feature parameters related to the defect state from the acquired ultrasonic echo signal; The life prediction module is used to input the extracted feature parameters into a pre-trained deep learning prediction model to obtain the prediction result of the remaining service life of the part to be analyzed; wherein, the deep learning prediction model is a neural network model that integrates a temporal convolutional network (TCN) module and a Transformer module.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the rapid analysis method for remanufactured part processing defects as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the rapid analysis method for remanufactured part processing defects as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the rapid analysis method for remanufactured part processing defects as described in any one of claims 1 to 6.

Images