Real-time monitoring method and device for structural state of exhaust nozzle based on digital twinning

Abstract

The invention provides an exhaust nozzle structure state real-time monitoring method and device based on digital twinning, and the method comprises the steps: obtaining the finite element simulation data of an exhaust nozzle, training a radial basis function proxy model of the exhaust nozzle based on the finite element simulation data, and enabling the radial basis function proxy model to obtain a radial basis function proxy model of the exhaust nozzle; comprising a radial basis function model for realizing displacement frequency response prediction; experimental data are introduced, and shape parameters of a Gaussian kernel function in the radial basis function model for displacement frequency response prediction are dynamically corrected; and on the basis of the corrected radial basis function agent model, a front-end and rear-end interactive visual interface is built to monitor the structural state of the exhaust nozzle in real time. According to the method, the key problem that the digital twin model is difficult to consider both the prediction precision and the real-time performance is effectively solved, and the engineering requirement of real-time monitoring is met with high requirements while the prediction precision is ensured.

Description

A method and device for real-time monitoring of tail nozzle structural status based on digital twin Technical Field

[0001] This application belongs to the field of performance measurement technology for aero-engine auxiliary devices, specifically relating to a method and device for real-time monitoring of the tail nozzle structure status based on digital twins. Background Technology

[0002] As a crucial load-bearing component, the structural health of an aero-engine's exhaust nozzle directly impacts the engine's overall safety and reliability. During operation, this structure must withstand complex vibration loads and thermal cycling caused by high-temperature, high-pressure exhaust gas flow. Damage or degradation can lead to resonance, fracture, and other problems, severely affecting the engine's service life. Therefore, developing effective real-time structural condition monitoring technologies is of significant engineering value for preventing structural failures and improving engine operational safety.

[0003] Digital twin technology provides a new technical approach for structural health monitoring by constructing high-precision virtual models of physical entities and achieving data interaction and synchronous mapping between virtual and real spaces. Currently, its application in the aero-engine field mainly focuses on predicting overall engine performance parameters (such as exhaust temperature and rotor speed) and diagnosing faults in rotating mechanical components (such as blades and bearings). However, research on real-time monitoring of vibration responses in static structural components such as exhaust nozzles remains relatively lacking. In existing structural digital twin modeling methods, the difficulty in simultaneously achieving high precision and real-time performance is a common technical bottleneck. While high-fidelity simulation can provide accurate response predictions, its low computational efficiency cannot meet the demands of real-time monitoring.

[0004] Chinese patent CN202211631773.0, entitled "A Digital Twin Model for Aircraft Engines," achieves higher prediction accuracy by considering the performance impact between historical flight sorties. Although it proposes a digital twin framework that considers the impact of historical sorties, its application focuses on the prediction of overall engine performance. The data input layer and coupling network layer it uses are mainly for system-level parameters and do not involve the dynamic response monitoring of specific structural components such as the tail nozzle. In particular, it lacks the ability to model key indicators such as structural vibration characteristics and modal parameters. Chinese patent application CN202310540741.8, entitled "A Digital Twin Parameter Prediction Method for Aircraft Engines Based on CEEMDAN-Informer," improves the accuracy of air path parameter prediction through signal decomposition and feature optimization. This method focuses on air path parameter prediction. Although the complex signal processing method it uses improves the prediction accuracy to a certain extent, the multi-layer processing flow leads to high computational complexity, making it difficult to meet the timeliness requirements of real-time monitoring. More importantly, existing technologies all share the following common problems: First, they fail to effectively resolve the core contradiction of balancing high precision and real-time performance in structural digital twins, especially the need for rapid calculation of the dynamic response of complex structures such as tail nozzles; second, they lack an effective verification mechanism for physical realism and fail to establish a closed-loop correction process between simulation data and experimental measurement data; third, the proposed digital twin models remain at the prediction level and lack a complete technical chain from prediction results to visualization and monitoring. Summary of the Invention

[0005] To address the aforementioned technical issues, this application proposes a method and device for real-time monitoring of the tailpipe structure status based on digital twins.

[0006] Firstly, this application proposes a method for real-time monitoring of the tailpipe structure status based on digital twins, including:

[0007] The finite element simulation data of the tail nozzle is obtained, and a radial basis function surrogate model of the tail nozzle is trained based on the finite element simulation data; the radial basis function surrogate model includes a radial basis function model for realizing displacement frequency response prediction.

[0008] Experimental data were introduced to dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction;

[0009] Based on the corrected radial basis function surrogate model, a visual interface with front-end and back-end interaction is built to monitor the tail nozzle structure status in real time.

[0010] Optionally, the Gaussian kernel function is expressed as follows:

[0011] ;

[0012] in, Let r be the kernel function used to measure the correlation between the evaluation point and the sample point in the radial basis function surrogate model, and r be the Euclidean distance between the evaluation point and the sample point in the radial basis function surrogate model. Here, is the shape parameter, and e is the natural logarithm.

[0013] Optionally, obtaining the finite element simulation data of the tail nozzle includes:

[0014] Use SolidWorks software to create a 3D model of the tail nozzle structure;

[0015] The 3D model is imported into the finite element analysis software, preprocessing is completed, and finite element simulation is performed to obtain the finite element model. The preprocessing includes defining geometric and material parameters, setting contact, meshing, and applying boundary constraints.

[0016] In the harmonic response analysis module of the finite element analysis software, a horizontal acceleration excitation is applied to the structural base of the finite element model, and the finite element model is simulated to obtain finite element simulation data under multiple working conditions.

[0017] Optionally, the radial basis function surrogate model further includes:

[0018] The radial basis function model for realizing the modal response is a radial basis function model using a thin plate spline kernel function without shape parameters.

[0019] Optionally, the step of introducing experimental data to dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction includes:

[0020] Displacement frequency response curve data under multiple excitation conditions were collected from the experiment. After acceleration-to-displacement, frequency axis alignment and resampling preprocessing, an experimental verification set consistent with the finite element simulation data space was formed.

[0021] Regarding the shape parameters of the Gaussian kernel function The optimal shape parameter values ​​were obtained by using a grid search strategy based on data within the experimental validation set.

[0022] An interpolation method is used to establish a continuous mapping relationship between the excitation level and the optimal shape parameter value, thereby enabling the estimation of any excitation value.

[0023] Optionally, the shape parameters of the Gaussian kernel function are... Optimization is achieved using a grid search strategy based on data within the experimental validation set, including:

[0024] For different excitation conditions, a grid search strategy is used to traverse the training radial basis function model for displacement frequency response prediction within a preset range of shape parameter values ​​to obtain the predicted values ​​for different excitations. The peak displacement error between the predicted values ​​and the experimental verification data is calculated, and the shape parameter value that minimizes the peak displacement error is selected as the optimal shape parameter value for the corresponding condition.

[0025] Secondly, a real-time monitoring device for the structural status of a tailpipe based on digital twins is provided, comprising:

[0026] The model training module is used to acquire finite element simulation data of the tail nozzle and train a radial basis function surrogate model of the tail nozzle based on the finite element simulation data. The radial basis function surrogate model includes a radial basis function model for realizing displacement frequency response prediction.

[0027] The model calibration module is used to introduce experimental data and dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction.

[0028] The interface building module is used to build a visual interface for real-time monitoring of the tail nozzle structure status based on the corrected radial basis function surrogate model.

[0029] Thirdly, a computer storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0030] Fourthly, an electronic device is provided, 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 real-time monitoring method for tailpipe structure status based on digital twin as described in the first aspect.

[0031] Fifthly, a computer program product is provided, including a computer program or instructions that, when executed by a processor, implement the steps of the real-time monitoring method for the tailpipe structure status based on digital twins as described in the first aspect.

[0032] Beneficial effects

[0033] This application proposes a method and device for real-time monitoring of tailpipe structure status based on digital twins. By integrating simulation, radial basis function surrogate model and measured data correction, and combining it with a real-time interactive visualization interface, it effectively solves the key problem of the difficulty in balancing prediction accuracy and real-time performance of digital twin models. The constructed radial basis function (RBF) surrogate model maintains accuracy comparable to high-fidelity finite element simulation while reducing prediction time from hours to seconds. Furthermore, by combining real experimental data to dynamically correct model parameters, the amplitude error is controlled below 10%, significantly improving the mapping accuracy of physical entities. Attached Figure Description

[0034] Figure 1 is a flowchart of the real-time monitoring method for tail nozzle structure status based on digital twin according to an embodiment of this application;

[0035] Figure 2 is a shape parameter calibration diagram of the proxy model in the embodiment of this application; where (a) is a simulation diagram of the shape parameter-percentage error relationship of 1g excitation, (b) is a simulation diagram of the amplitude-frequency characteristic curve of 1g excitation, (c) is a simulation diagram of the shape parameter-percentage error relationship of 2g excitation, (d) is a simulation diagram of the amplitude-frequency characteristic curve of 2g excitation, (e) is a simulation diagram of the shape parameter-percentage error relationship of 3g excitation, and (f) is a simulation diagram of the amplitude-frequency characteristic curve of 3g excitation;

[0036] Figure 3 is a schematic diagram of the three-dimensional visualization interface of the digital twin system according to an embodiment of this application; wherein (a) is the initial running diagram of the twin interface, and (b) is the real-time monitoring diagram of the twin interface;

[0037] Figure 4 is a schematic diagram of mode prediction for each order according to an embodiment of this application; wherein (a) is a schematic diagram of first-order mode prediction, (b) is a schematic diagram of second-order mode prediction, (c) is a schematic diagram of third-order mode prediction, and (d) is a schematic diagram of fourth-order mode prediction.

[0038] Figure 5 is a schematic diagram of frequency response curve prediction according to an embodiment of this application; wherein (a) is a simulation diagram of amplitude-frequency characteristic curve of 1.5g excitation, and (b) is a simulation diagram of amplitude-frequency characteristic curve of 2.5g excitation;

[0039] Figure 6 is a flowchart of a method for real-time monitoring of the tailpipe structure status based on digital twins, according to a specific example of this application. Detailed Implementation

[0040] The terms "first," "second," "third," "fourth," etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The technical solutions of the embodiments of this application will now be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0041] Example 1

[0042] This application discloses a method for real-time monitoring of the structural status of a tailpipe based on digital twins, as shown in Figure 1, including the following steps:

[0043] Step S1: Obtain finite element simulation data of the tail nozzle, and train a radial basis function surrogate model of the tail nozzle based on the finite element simulation data; the radial basis function surrogate model includes a radial basis function model for realizing displacement frequency response prediction.

[0044] In the specific implementation process, the following steps shall be performed:

[0045] Step S1.1: Simplify the tail nozzle model and prepare pre-training simulation data;

[0046] First, a three-dimensional model of the bolted flange boundary structure of the tailpipe was created using SolidWorks software and imported into the finite element analysis software. Preprocessing was completed, including defining geometric and material parameters, setting contact, mesh generation, and applying boundary constraints. Then, in the harmonic response analysis module of the finite element analysis software, a horizontal acceleration excitation was applied to the structural base of the finite element model to obtain displacement response data at specified measurement points.

[0047] Specifically, in this embodiment, in order to meet the real-time requirements of the digital twin model, the mesh of the tail nozzle finite element model established in the finite element analysis software is simplified, the element size is set to 50 mm, and finally a simplified model containing 5376 elements and 12880 nodes is obtained.

[0048] The finite element model is the tailpipe model that includes the bolt flange boundary;

[0049] By performing multi-condition finite element simulations on the finite element model, the mode response and displacement frequency response curve data of the tail nozzle are extracted, providing a training basis for the construction of the surrogate model.

[0050] Step S1.2: Construct the proxy model;

[0051] Based on the multi-condition finite element simulation data samples obtained in step S1.1, a unified preprocessing is first performed, including removing duplicate nodes to reduce computational resource consumption, and using the K-nearest neighbor algorithm to align node indices to ensure that the data under different conditions strictly correspond on the same node.

[0052] Removing duplicate nodes is a common existing technique for filtering out duplicate samples in the data preprocessing process, so it will not be described in detail here in this embodiment;

[0053] The K-nearest neighbors algorithm is an instance-based supervised learning method that is widely used in classification and regression problems. Its basic assumption is that similar samples are close to each other in the feature space, that is, the category or numerical attribute of a sample can be inferred from its K nearest neighbors.

[0054] The distance between samples represents their similarity, and the commonly used metric is the Euclidean distance for two n-dimensional vectors. and The basic form of its Euclidean distance can be expressed as:

[0055] ;

[0056] in, This represents the component of vector x in the i-th dimension; This represents the component of vector y in the i-th dimension; i is a positive integer from 1 to n, representing the index of the dimension; n is the number of dimensions of the vector, that is, the number of components contained in each vector.

[0057] The algorithm's execution flow includes: calculating the distance between the test sample and each sample in the training set; selecting the K nearest samples as its nearest neighbor set based on their distance; and using a majority voting mechanism to determine the class label in the classification task. Unlike models that require explicit training (such as linear regression and support vector machines), KNN is a lazy learning algorithm. It only stores the labeled training dataset during the training phase, deferring all computation to the next stage, making it highly intuitive and easy to implement.

[0058] Secondly, the processed training data is trained using the RBF model. The RBF model is a feedforward neural network that uses radial basis functions as activation functions. Thanks to its unique network structure and high learning efficiency, the RBF model has been widely used in various fields such as function approximation, time series prediction, and pattern recognition. Its basic form can be expressed as:

[0059] ;

[0060] in: To predict the output value; It is used to measure evaluation points With sample points Kernel function for correlation; These are the weight coefficients corresponding to the k-th kernel function; It is the number of kernel functions; k takes the value of a positive integer from 1 to n.

[0061] In the RBF model, the choice of kernel function is crucial because it determines how the similarity between two points is measured and how the influence is weighted according to distance. Different kernel functions result in models with different properties and applicable scenarios. For different target physical quantities, appropriate kernel functions are selected to construct multiple surrogate models.

[0062] For modal response prediction, this embodiment uses a thin-plate spline kernel function with global smoothness and no additional shape parameters to construct a radial basis function model. The mathematical expression of this kernel function is as follows:

[0063] ;

[0064] For displacement frequency response prediction, a Gaussian kernel function with good local support characteristics and the ability to flexibly capture local details of the frequency response curve is used. Its mathematical expression is:

[0065] ;

[0066] Where r is a point With point The Euclidean distance; It is a shape parameter that controls the width of the kernel function, i.e., the rate at which the influence decays with distance.

[0067] Finally, the construction of a surrogate model capable of rapidly predicting the dynamic response of a structure was completed through the radial basis function model for predicting modal response and the radial basis function model for predicting displacement frequency response constructed above.

[0068] Step S2: Introduce experimental data and dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction;

[0069] In this specific implementation, addressing the issue that the shape factor in the radial basis function significantly impacts prediction accuracy, experimental data is introduced, and a grid search strategy is employed to optimize this parameter, thereby significantly improving the prediction accuracy of the surrogate model. Therefore, the following model parameter optimization process is performed:

[0070] Step S2.1: Collect displacement frequency response curve data under multiple excitation conditions from the experiment. After acceleration to displacement conversion, frequency axis alignment and resampling preprocessing, an experimental verification set consistent with the finite element simulation data space is formed. Frequency alignment and resampling are common existing techniques for unifying the time scale of time series data, which will not be described in detail here in this embodiment.

[0071] Step S2.2: Regarding the shape parameters of the Gaussian kernel function The optimal shape parameter values ​​were obtained by using a grid search strategy based on data within the experimental validation set.

[0072] The grid search strategy is a commonly used existing technique for model hyperparameter optimization, which will not be described in detail here in this embodiment;

[0073] Specifically, for different excitation conditions, a grid search strategy is used to traverse the radial basis function model for displacement frequency response prediction within a preset range of shape parameter values ​​to obtain predicted values ​​for different excitations. The peak displacement error between the predicted values ​​and the experimental verification data is calculated, and the shape parameter value that minimizes the peak displacement error is selected as the optimal shape parameter value for the corresponding condition.

[0074] Step S2.3: Use interpolation to establish a continuous mapping relationship between the excitation level and the optimal shape parameter value, thereby achieving the estimation of any excitation value.

[0075] Figure 2 shows the calibration diagram of the surrogate model's shape parameters. This process is used to improve the prediction accuracy of the surrogate model for real physical systems using experimental data. First, experimental data is collected and preprocessed to construct a validation set. Then, the shape parameters of the Gaussian kernel function are optimized based on a grid search strategy. As shown in Figure 2, (a) is the simulation diagram of the shape parameter-percentage error relationship of the 1g excitation, (b) is the simulation diagram of the amplitude-frequency response curve of the 1g excitation, (c) is the simulation diagram of the shape parameter-percentage error relationship of the 2g excitation, (d) is the simulation diagram of the amplitude-frequency response curve of the 2g excitation, (e) is the simulation diagram of the shape parameter-percentage error relationship of the 3g excitation, and (f) is the simulation diagram of the amplitude-frequency response curve of the 3g excitation. Finally, a continuous mapping relationship is established through interpolation. The specific steps are as follows:

[0076] First, displacement frequency response curve data under multiple excitation conditions are collected from the experiment. After acceleration to displacement, frequency axis alignment and resampling preprocessing, an experimental verification set consistent with the finite element simulation data space is formed.

[0077] Subsequently, the shape parameters of the Gaussian kernel surrogate model were addressed. Optimization was performed using a grid search strategy based on experimental validation set data: for 1g, 2g, and 3g excitation conditions, under preset conditions... The training model is traversed within the candidate range, and the peak displacement error is calculated based on experimental data. The model that minimizes the prediction deviation, i.e., the peak displacement error, is selected. The value is taken as the optimal shape parameter value for this working condition.

[0078] Through this optimization, the peak displacement error was successfully reduced to below 10%. Finally, given that experimental data typically only cover discrete operating points, an interpolation method was used to establish the excitation level and optimal... A continuous mapping relationship between values ​​is established, thereby enabling a reasonable estimate of any excitation value.

[0079] Step S3: Based on the corrected radial basis function surrogate model, build a front-end and back-end interactive visualization interface to monitor the tail nozzle structure status in real time.

[0080] Based on the established digital twin, namely the corrected tail nozzle, the above-mentioned radial basis function surrogate model is constructed to build a front-end and back-end interactive visualization interface, and finally realize the real-time monitoring and display of the tail nozzle structural status; specifically, Figure 3 shows a schematic diagram of the three-dimensional visualization interface of the digital twin system, wherein the initial running diagram of the twin interface is shown in Figure 3(a), and the real-time monitoring diagram of the twin interface is shown in Figure 3(b).

[0081] This interface is used to achieve real-time visual monitoring of the tailpipe structure's vibration modes and frequency response. Specifically, the front-end is built using the Unity3D engine, and a data interaction channel is established with the Python back-end via the Socket communication protocol to achieve the visual rendering of the structural response prediction results. The specific implementation process is as follows:

[0082] First, build a visualization system for the client-server architecture;

[0083] Specifically, the Unity3D engine is used to develop a 3D visualization front-end, and a data connection is established with the Python back-end through the Socket communication protocol;

[0084] The system uses a Socket connection manager to uniformly manage TCP connections, avoiding conflicts caused by concurrent access from multiple components, and configures a receive buffer to accommodate the transmission requirements of high-dimensional response data.

[0085] By developing an integrated visualization system and adopting Socket communication and asynchronous collaboration mechanisms, a closed-loop monitoring system from automatic prediction to real-time visualization was achieved. This system demonstrated good real-time performance, interactivity, and robustness in tailpipe structure vibration monitoring, meeting the needs of high-fidelity real-time monitoring in practical engineering.

[0086] Figure 6 is a flowchart of a real-time monitoring method for the tailpipe structure based on digital twins, as shown in a specific example of the method in this embodiment. Specifically, it includes: a surrogate model training data extraction stage, a surrogate model construction and dynamic parameter optimization stage, and a visualization interface construction stage. In the surrogate model training data extraction stage, simplified mesh data (i.e., a simplified tailpipe model), multi-condition vibration mode data, and multi-condition displacement frequency response data are acquired. In the surrogate model construction and dynamic parameter optimization stage, an RBF model is trained based on finite metadata, followed by vibration mode prediction and frequency response curve prediction. Simultaneously, a mesh search algorithm is used to correct the shape parameters of the predicted frequency response curve based on experimental data. In the visualization interface construction stage, a Python backend and a Unity frontend are built, and communication between the Python backend and the Unity frontend is achieved through a Socket communication protocol to realize real-time visualization monitoring of the tailpipe structure's vibration mode and frequency response.

[0087] In constructing a digital twin of an aero-engine exhaust nozzle, the method of this application first trains a radial basis function surrogate model based on simplified finite element simulation data to achieve rapid prediction. Then, real experimental data is introduced to dynamically correct the model shape parameters to improve prediction accuracy. Finally, a visualization platform integrating the front and back ends is used to realize real-time rendering and interaction of mode shape and displacement frequency response, thereby meeting the engineering requirements for real-time monitoring while ensuring prediction accuracy.

[0088] Example 2

[0089] This embodiment proposes a real-time monitoring device for the structural status of a tailpipe based on digital twins, including:

[0090] The model training module is used to acquire finite element simulation data of the tail nozzle and train a radial basis function surrogate model of the tail nozzle based on the finite element simulation data. The radial basis function surrogate model includes a radial basis function model for realizing displacement frequency response prediction.

[0091] The model calibration module is used to introduce experimental data and dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction.

[0092] The interface building module is used to build a visual interface for real-time monitoring of the tail nozzle structure status based on the corrected radial basis function surrogate model.

[0093] In a specific example of this embodiment, the visual interface built using the interface building module includes three core functional sub-modules:

[0094] The mode shape cloud map real-time rendering submodule receives the operating parameters and mode shape order input by the user through a slider control and drop-down menu. The front end sends a request to the Python back end, which calls the proxy model to predict and return the nodal displacement array. After parsing the JSON data, the front end updates the grid vertex color mapping and geometric deformation in real time, and uses a heat map color scheme to intuitively display the displacement field distribution.

[0095] Frequency response curve dynamic drawing submodule: By continuously adjusting the excitation value through the slider, the front end dynamically draws a smooth frequency response curve in the canvas coordinate system and annotates the peak frequency and peak displacement in real time;

[0096] The experimental data automatic monitoring submodule: The Python backend continuously listens to the specified path through an independent thread. When a new experimental data file that conforms to the naming rules is detected, it automatically parses the operating conditions, calls the proxy model to predict the frequency response curve, and caches the results. The Unity frontend periodically checks whether there is new prediction data. If so, it automatically switches the display and updates the interface.

[0097] In this example, the system with the visual interface adopts asynchronous communication and coroutine mechanism to avoid blocking the main thread, and sets timeout and reconnection mechanism to ensure system robustness.

[0098] Figure 4 shows the predicted mode shapes for each order, where (a) is a schematic diagram of the first-order mode shape prediction, (b) is a schematic diagram of the second-order mode shape prediction, (c) is a schematic diagram of the third-order mode shape prediction, and (d) is a schematic diagram of the fourth-order mode shape prediction. This demonstrates the comparison results between the mode shape predictions of the surrogate model and the finite element model. Overall, the main mode shape characteristics predicted by the surrogate model are highly consistent with the ANSYS simulation results. Although there are still some deviations in local deformation amplitude and details, its prediction accuracy meets expectations, further verifying the effectiveness of the proposed method.

[0099] Figure 5 shows the predicted frequency response curves, where (a) is the simulated amplitude-frequency characteristic curve under 1.5g excitation and (b) is the simulated amplitude-frequency characteristic curve under 2.5g excitation. Analyzing the overall trend of the displacement frequency response curves reveals that under both 1.5g and 2.5g excitation conditions, the three curves from the ANSYS simulation, experimental tests, and the surrogate model exhibit similar regularities: the amplitude increases with frequency, rapidly increases and reaches a peak near the resonance frequency, and then gradually decreases. Furthermore, comparing the curve results shows that the predictions from the ANSYS simulation and the surrogate model are quite close to the experimental test results in terms of trend and resonance frequency.

[0100] The real-time monitoring device for tail nozzle structure status based on digital twin provided in this embodiment has the same technical features as the real-time monitoring method for tail nozzle structure status based on digital twin provided in Embodiment 1. Therefore, it can also solve the same technical problems and achieve the same technical effects.

[0101] Example 3

[0102] This application provides a computer storage medium storing a computer program that, when executed by a processor, implements the real-time monitoring method for tail nozzle structure status based on digital twin as described in Embodiment 1.

[0103] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0104] Therefore, this application also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the methods described in any embodiment of this application. The computer-readable storage medium can be configured in any device of this application.

[0105] Example 4

[0106] This application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the real-time monitoring method for tail nozzle structure status based on digital twin as described in Embodiment 1.

[0107] For example, it includes: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the method provided in the embodiments of this application. The methods described are included in the functional descriptions above and will not be repeated here.

[0108] The electronic device also includes input and output devices; the processor, storage device, input and output devices in the electronic device can be connected by a bus or other means.

[0109] A storage device, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and module units, such as program instructions corresponding to the methods in the embodiments of this application. The storage device may mainly include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the terminal, etc. Furthermore, the storage device may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the storage device may further include memory remotely located relative to the processor, and these remote memories can be connected via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The various embodiments in this application are described in a progressive manner; similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments.

[0110] Example 5

[0111] This embodiment provides a computer program product, including a computer program or instructions, which, when executed by a processor, implements the real-time monitoring method for tail nozzle structure status based on digital twin as described in Embodiment 1.

[0112] Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a computer program product.

[0113] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0114] The scope of protection of this application is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from the scope and spirit of this disclosure. If such modifications and variations fall within the scope of the technical content of this disclosure and its equivalents, then the intent of this disclosure also includes these modifications and variations.

Claims

1. A method for real-time monitoring of the structural status of a tailpipe based on digital twins, characterized in that, include: Finite element simulation data of the tail nozzle is acquired, and a radial basis function surrogate model of the tail nozzle is trained based on the finite element simulation data. The radial basis function surrogate model includes a radial basis function model for displacement frequency response prediction. Experimental data is introduced to dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction. Based on the corrected radial basis function surrogate model, a front-end and back-end interactive visualization interface is built to monitor the tail nozzle structural state in real time.

2. The method for real-time monitoring of the tail nozzle structure status based on digital twin according to claim 1, characterized in that, The Gaussian kernel function is expressed as follows: ;in, Let r be the kernel function used to measure the correlation between the evaluation point and the sample point in the radial basis function surrogate model, and r be the Euclidean distance between the evaluation point and the sample point in the radial basis function surrogate model. Here, is the shape parameter, and e is the natural logarithm.

3. The method for real-time monitoring of the tailpipe structure status based on digital twin according to claim 1, characterized in that, The process of obtaining finite element simulation data for the tail nozzle includes: establishing a three-dimensional model of the tail nozzle structure using SolidWorks software; importing the three-dimensional model into finite element analysis software, completing preprocessing, and performing finite element simulation to obtain the finite element model; the preprocessing includes defining geometric and material parameters, setting contact, mesh generation, and applying boundary constraints; applying horizontal acceleration excitation to the structural base of the finite element model in the harmonic response analysis module of the finite element analysis software, simulating the finite element model, and obtaining finite element simulation data under multiple working conditions.

4. The method for real-time monitoring of the tailpipe structure status based on digital twin according to claim 1, characterized in that, The radial basis function surrogate model further includes: a radial basis function model for predicting modal response, which is a radial basis function model using a thin plate spline kernel function without shape parameters.

5. The method for real-time monitoring of the tail nozzle structure status based on digital twin according to claim 1, characterized in that, The introduction of experimental data to dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction includes: collecting displacement frequency response curve data under multiple excitation conditions from experiments, and forming an experimental verification set consistent with the finite element simulation data space after acceleration-to-displacement, frequency axis alignment, and resampling preprocessing; and adjusting the shape parameters of the Gaussian kernel function. The optimal shape parameter values ​​are obtained by using a grid search strategy based on data within the experimental validation set. An interpolation method is then used to establish a continuous mapping relationship between the excitation level and the optimal shape parameter values, thereby enabling the estimation of any excitation value.

6. The method for real-time monitoring of the tail nozzle structure status based on digital twin according to claim 5, characterized in that, Regarding the shape parameters of the Gaussian kernel function The optimization is carried out using a grid search strategy based on data in the experimental verification set. This includes: for different excitation conditions, using the grid search strategy to traverse the radial basis function model for displacement frequency response prediction within a preset range of shape parameter values ​​to obtain the predicted values ​​for different excitations, calculating the peak displacement error between the predicted values ​​and the data in the experimental verification set, and selecting the shape parameter value that minimizes the peak displacement error as the optimal shape parameter value for the corresponding condition.

7. A real-time monitoring device for the structural status of a tailpipe based on digital twins, characterized in that, include: The model training module is used to acquire finite element simulation data of the tail nozzle and train a radial basis function surrogate model of the tail nozzle based on the finite element simulation data; the radial basis function surrogate model includes a radial basis function model for realizing displacement frequency response prediction; the model calibration module is used to introduce experimental data and dynamically correct the shape parameters of the Gaussian kernel function in the radial basis function model for displacement frequency response prediction; the interface building module is used to build a front-end and back-end interactive visualization interface based on the calibrated radial basis function surrogate model to monitor the tail nozzle structural state in real time.

8. A computer storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the real-time monitoring method for tailpipe structure status based on digital twin as described in any one of claims 1 to 6.

9. An electronic device, characterized in that, The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the real-time monitoring method for the tailpipe structure status based on digital twins as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the processor, they implement the steps of the real-time monitoring method for tail nozzle structure status based on digital twin as described in any one of claims 1 to 6.

Images