Vibration amplitude prediction method and device and vibration amplitude prediction system

Abstract

The invention provides a vibration amplitude prediction method, a vibration amplitude prediction device and a vibration amplitude prediction system. The method comprises the following steps: acquiring an actual measurement value of a vibration amplitude of an actual measurement point of measured equipment; according to the distance attenuation coefficient, the structural constraint coefficient, the resonance factor and the transfer constraint coefficient of the actual measurement point, calculating a predicted value of the vibration amplitude of the predicted point; and correcting the predicted value to obtain a corrected predicted value. The method solves the problem of difficulty in predicting the vibration hot area applied to the actual engineering field in the prior art.

Description

Technical Field

[0001] This application relates to the field of intelligent manufacturing technology, and more specifically, to a method for predicting vibration amplitude, a device for predicting vibration amplitude, a computer program product, and a system for predicting vibration amplitude. Background Technology

[0002] In modern engineering, accurate estimation and prediction of system vibration is crucial, especially in defense, security monitoring, autonomous vehicles, and a wide range of industrial applications. Vibration analysis not only reveals the health status of a system but also plays an irreplaceable role in preventing failures, optimizing design, and ensuring operational safety. Traditional vibration prediction methods mainly rely on modal analysis, which involves obtaining the system's natural frequencies, damping ratios, and modal shapes through experiments or simulations, and then predicting vibrations based on these modal parameters.

[0003] Modal analysis is divided into two categories: experimental modal analysis (EMA) and operating modal analysis (ODS). EMA is typically performed under laboratory conditions, determining modal parameters by exciting the structure and measuring its response; while ODS estimates modal parameters by using vibration records of the system in its natural operating environment under natural operating conditions. Both EMA and ODS require collecting vibration data of the entire structure, especially for large or complex structures. This means deploying a large number of sensors to cover all potentially important measurement points, followed by detailed data analysis.

[0004] However, current vibration prediction methods rely on a large amount of preliminary design data, which leads to low vibration prediction efficiency and limited prediction methods in actual engineering applications. Summary of the Invention

[0005] The main objective of this application is to provide a method, device, computer program, and system for predicting vibration amplitude, so as to at least solve the problems of insufficient vibration prediction and long cycle in the prior art.

[0006] To achieve the above objectives, according to one aspect of this application, a method for predicting vibration amplitude is provided, comprising: acquiring measured values ​​of vibration amplitude at measured points of a device under test; calculating predicted values ​​of the vibration amplitude at predicted points based on distance attenuation coefficients, structural constraint coefficients, resonance factors, and transmission constraint coefficients of the measured points, wherein the distance attenuation coefficient is a coefficient measuring the change of vibration energy with propagation distance, the structural constraint coefficient is a coefficient measuring the change of vibration energy with structural geometric characteristics, the resonance factor is a coefficient simulating the resonance phenomenon of a structure at a specific distance, and the transmission constraint coefficient is calculated based on the distance attenuation coefficient, the measured values ​​of vibration amplitude at measured points ... The transmission coefficient of vibration change is calculated from the structural constraint coefficient and the resonance factor, wherein the measured point and the predicted point are located at different positions; the predicted value is corrected to obtain a corrected predicted value, wherein the corrected predicted value is used to provide data for the prediction model to predict the vibration amplitude, so that the vibration characteristics of the device under test can be evaluated using the corrected predicted value; the historical measured values ​​and the corresponding prediction labels are combined into a training set, and the prediction model is trained using the training set to obtain a vibration prediction model, wherein the prediction label is the historical corrected predicted value corresponding to the historical measured value in the training set.

[0007] According to another aspect of this application, a vibration amplitude prediction device is provided, comprising: an acquisition unit for acquiring measured values ​​of vibration amplitude at a measured point of a device under test; and a first calculation unit for calculating predicted values ​​of the vibration amplitude at a prediction point based on a distance attenuation coefficient, a structural constraint coefficient, a resonance factor, and a transmission constraint coefficient at the measured point, wherein the distance attenuation coefficient is a coefficient measuring the change of vibration energy with propagation distance, the structural constraint coefficient is a coefficient measuring the change of vibration energy with structural geometric characteristics, the resonance factor is a coefficient simulating the resonance phenomenon of a structure at a specific distance, and the transmission constraint coefficient is calculated based on the distance attenuation coefficient, the structural constraint coefficient, a resonance factor, and a transmission constraint coefficient at the measured point. The structural constraint coefficient and the resonance factor are used to calculate the vibration change transmission coefficient, wherein the measured point and the predicted point are located at different positions; a correction unit is used to correct the predicted value to obtain a corrected predicted value, wherein the corrected predicted value is used to provide data for the prediction model to predict the vibration amplitude, so that the vibration characteristics of the tested equipment can be evaluated using the corrected predicted value; historical measured values ​​and corresponding prediction labels are combined into a training set, and the prediction model is trained using the training set to obtain a vibration prediction model, wherein the prediction label is the historical corrected predicted value corresponding to the historical measured value in the training set.

[0008] According to another aspect of this application, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps of any of the vibration amplitude prediction methods.

[0009] According to another aspect of this application, a vibration amplitude prediction system is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing any of the vibration amplitude prediction methods described above.

[0010] By applying the technical solution of this application, the concept of physical constraints is introduced, including distance attenuation coefficient, structural constraint coefficient, resonance factor and transmission constraint coefficient. Vibration is predicted through multiple influences, which can ensure that the predicted value follows multiple physical laws. At the same time, the predicted value is also corrected to identify prediction differences, further ensuring that the predicted value is closer to the actual vibration behavior, reducing prediction error, and thus improving the accuracy of vibration prediction. Attached Figure Description

[0011] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0012] Figure 1 A hardware structure block diagram of a mobile terminal for performing a method for predicting vibration amplitude according to an embodiment of this application is shown.

[0013] Figure 2 A flowchart illustrating a method for predicting vibration amplitude according to an embodiment of this application is shown.

[0014] Figure 3 The geometric model and orientation diagram of the surface of the object under test are shown.

[0015] Figure 4 A schematic diagram of the vibration adaptive analysis technology process based on multi-source data fusion is shown.

[0016] Figure 5 A schematic diagram showing the vibration prediction results and directions is provided.

[0017] Figure 6 A schematic diagram of the vibration hotspot distribution is shown;

[0018] Figure 7 This diagram illustrates the distribution of uncertainty in the global prediction.

[0019] Figure 8 A schematic diagram showing the predicted vibration amplitude and uncertainty results is provided.

[0020] Figure 9 A structural block diagram of a vibration amplitude prediction device provided according to an embodiment of this application is shown.

[0021] The above figures include the following reference numerals:

[0022] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. Detailed Implementation

[0023] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, 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 for the embodiments of this application 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.

[0026] Vibration estimation of a system is a technique used in engineering applications to determine the vibration source and vibration response, and it has broad application prospects in fields such as security, autonomous driving, and industrial applications. Currently, there are many algorithms and techniques for system vibration estimation on the market, but their main shortcomings and limitations include:

[0027] 1. Computational complexity: Vibration prediction often requires combining the system's modal data, which requires a lot of preliminary work such as modal testing or modal simulation. In actual engineering field analysis, it is difficult to implement on equipment with limited resources, and the results cannot be quickly evaluated in a short time.

[0028] 2. High hardware and software requirements: To accurately obtain the vibration characteristics of a system under working conditions, it is usually necessary to traverse all measurement points to perform ODS analysis and modal analysis. This requires professional testing equipment and software, which places high demands on equipment and personnel, is time-consuming, and has a high entry threshold for analysis.

[0029] The aforementioned drawbacks mean that global vibration estimation methods have high equipment / method investment costs, setting a high software and hardware entry barrier for their application in industrial / civilian fields, and their cost-effectiveness in practical engineering applications is not high.

[0030] As described in the background section, the accuracy of vibration prediction in the prior art is low. To solve the above problems, embodiments of this application provide a vibration amplitude prediction method, a vibration amplitude prediction device, a computer program product, and a vibration amplitude prediction system.

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0032] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a vibration amplitude prediction method according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0033] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the vibration amplitude prediction method in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] This embodiment provides a method for predicting vibration amplitude that runs on a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0035] Figure 2 This is a schematic flowchart of a vibration amplitude prediction method according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0036] Step S201: Obtain the measured value of the vibration amplitude at the measured point of the device under test;

[0037] Specifically, in practice, vibration sensors need to be installed on the device under test to collect vibration data from multiple measurement points. These sensors can be vibration pens or triaxial accelerometers, capable of measuring the vibration amplitude of the tested structure in three directions. The data acquisition process ensures the quality of the foundational data for subsequent prediction models, providing an accurate source of information for follow-up steps.

[0038] In the above embodiments, by using direct measured data, it is possible to ensure that the initial parameter settings of the prediction model match the actual vibration conditions, thereby reducing the possible deviation between the model and the real world and enhancing the prediction accuracy of the model.

[0039] Step S202: Based on the distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient of the measured point, calculate the predicted value of the vibration amplitude of the predicted point. The distance attenuation coefficient is a coefficient that measures the change of vibration energy with the propagation distance. The structural constraint coefficient is a coefficient that measures the change of vibration energy with the geometric characteristics of the structure. The resonance factor is a coefficient that simulates the resonance phenomenon of the structure at a specific distance. The transmission constraint coefficient is a transmission coefficient of vibration change calculated based on the distance attenuation coefficient, the structural constraint coefficient, and the resonance factor. The measured point and the predicted point are located at different positions.

[0040] Specifically, by defining distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient, the model can estimate the vibration amplitude at non-measurement points based on limited measured point data. The setting of these coefficients fully considers the propagation characteristics of vibration signals in the structure, including natural attenuation with increasing distance, energy absorption or amplification effects of specific structural characteristics, and the influence of resonance on energy propagation.

[0041] In the above embodiments, by introducing physical constraints, the model prediction results are more closely aligned with physical laws, avoiding non-physical biases that may arise from purely data-driven predictions, and improving the reliability and accuracy of the predictions. The calculation of the transfer constraint coefficients ensures that the model has a reasonable mathematical expression for the transfer of vibration energy, further enhancing the physical meaning of the predictions.

[0042] Step S203: Correct the predicted value to obtain a corrected predicted value. The corrected predicted value is used to provide data for the prediction model to predict the vibration amplitude, so that the vibration characteristics of the tested equipment can be evaluated using the corrected predicted value. The historical measured values ​​and the corresponding prediction labels are combined into a training set. The prediction model is trained using the training set to obtain a vibration prediction model. The prediction label is the historical corrected predicted value corresponding to the historical measured value in the training set.

[0043] Specifically, after the initial prediction, the prediction results are corrected by comparing the measured values ​​(if any) at non-measurement points with the predicted values. The corrected prediction values ​​are used as feedback input to retrain and optimize the prediction model. This process ensures that the model can improve itself over time and gradually reduce prediction errors.

[0044] In the above embodiments, through continuous model calibration, the prediction model can adapt to vibration characteristics under different working conditions and load scenarios, improving the model's generalization ability. The existence of the correction mechanism is equivalent to introducing a learning process; the model can not only process static data but also improve itself based on dynamic calibration data, ultimately achieving higher prediction accuracy.

[0045] This embodiment introduces the concept of physical constraints, including distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient. By using multiple influences to predict vibration, it can be ensured that the predicted value follows various physical laws. At the same time, the predicted value is also corrected to identify prediction differences, further ensuring that the predicted value is closer to the actual vibration behavior, reducing prediction errors, and thus improving the accuracy of vibration prediction.

[0046] Obtain a prediction model, which can be one of the following: SVR model, random forest model, CNN model, Kalman filter model, Bayesian regression model, or FEA model; assemble historical measured values ​​and corresponding prediction labels into a training set, and train the prediction model using the training set to obtain the vibration prediction model, where the prediction labels are the historically corrected prediction values ​​corresponding to the historical measured values ​​in the training set; input the measured values ​​into the vibration prediction model to obtain the prediction values ​​corresponding to the measured values.

[0047] In the specific implementation process, based on the distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient of the above-mentioned measured points, the predicted value of the vibration amplitude of the predicted point is calculated, including: calculating the product of the negative of the attenuation coefficient and the Euclidean distance to obtain a first value, wherein the Euclidean distance is the difference between the coordinates of the two above-mentioned measured points in space; calculating the natural constant raised to the power of the first value to obtain the distance attenuation coefficient; calculating the sum of the curvature feature and the minimum positive number to obtain a second value, wherein the curvature feature characterizes the degree of curvature of the region where the above-mentioned measured point is located; calculating the quotient of the minimum curvature feature and the second value to obtain the structural constraint coefficient; calculating the quotient of the Euclidean distance and a first preset value to obtain a third value; calculating a first preset multiple of π to obtain a fourth value; calculating the product of the third value and the fourth value to obtain a fifth value; calculating the sine function value of the fifth value to obtain a sixth value; calculating the sixth value and the second... The sum of preset values ​​yields the seventh value; the second preset multiple of the structural amplification factor is calculated to obtain the eighth value; the product of the seventh and eighth values ​​is calculated to obtain the ninth value; the sum of the ninth and third preset values ​​is calculated to obtain the resonance factor; the product of the distance attenuation factor and the resonance factor is calculated to obtain the tenth value; the difference between the fourth preset value and the structural constraint factor is calculated to obtain the eleventh value; the product of the tenth and eleventh values ​​is calculated to obtain the transmission constraint factor; the product of the multi-head prediction fusion result and the transmission constraint factor is calculated to obtain the twelfth value, wherein the multi-head prediction fusion result is calculated based on the square of the vibration amplitude of multiple measured points; the product of the twelfth value and the direction vector is calculated to obtain the thirteenth value, wherein the direction vector is the unit vector of vibration in the vibration direction; the weighted average of multiple thirteenth values ​​is calculated to obtain the predicted value.

[0048] In this scheme, the distance attenuation coefficient is calculated to accurately reflect the natural attenuation law of vibration energy with distance, ensuring the physical rationality of non-measuring point predictions. The calculation of the structural constraint coefficient quantifies the influence of structural geometry on vibration propagation, especially in areas with large structural curvature, effectively reducing the predicted value and simulating the natural attenuation of vibration energy at structural bends in reality. The calculation of the resonance factor simulates the resonance characteristics of the structure, periodically amplifying the vibration amplitude at specific distances, so that the prediction results can reflect the resonance effect of the actual structure. By calculating the transfer constraint coefficient, the influence of distance attenuation, structural constraints, and resonance factor is integrated, providing a mathematical model that comprehensively considers physical constraints for non-measuring point predictions. By combining the multi-head prediction fusion results with the transfer constraint coefficient and further multiplying them with the direction vector, the predicted value is finally obtained through weighted averaging, achieving accurate estimation of the vibration amplitude at non-measuring points.

[0049] Constraint equations defined by the physical model: Predicting the system based on the physical model mainly includes five aspects: distance attenuation (vibration attenuates with distance), structural constraints, resonance factor, transfer function constraints, and correction feedback.

[0050] The constraint equation based on vibration attenuation: In the transfer function, the distance attenuation constraint is the basic attenuation of vibration as it propagates from the measuring point to other points. The greater the distance, the greater the attenuation of the vibration amplitude, as expressed in equation (4):

[0051] ;in, This is the attenuation coefficient, which is set to 0.1 here;

[0052] Structural constraints: Structural complexity constraints adjust the propagation of vibrations through local geometric features (curvature). In regions with high curvature (such as corners), vibrations are weakened, as expressed in equation (5):

[0053] ;in, The eigenvalues ​​represent the eigenvalues ​​of the local point cloud covariance matrix. To represent the smallest eigenvalue, a small positive number is introduced. Avoid dividing by 0 in equation (5);

[0054] Resonance factor: Simulates the resonance phenomenon of a structure at a specific distance, so that the vibration propagation is not a simple exponential decay, but has a periodic amplification, expressed as equation (6):

[0055] ;in, This is the structural magnification factor, which is taken as 0.3 here.

[0056] Calculate the transfer function: Based on equations (4), (5), and (6), calculate the transfer function from measuring point j to non-measuring point i, as shown in equation (7): ;

[0057] For each non-measure point i, the multi-head attention mechanism generates H predicted values, as shown in equation (16): .

[0058] In some embodiments, before calculating the weighted average of multiple thirteenth values ​​to obtain the predicted value, the method further includes: calculating a first initial weight based on distance; calculating a second initial weight based on amplitude; calculating a weighted average of the first initial weight and the second initial weight to obtain a third initial weight; and performing softmax normalization on the third initial weight to obtain a target weight value, wherein the target weight value is the weight coefficient of the thirteenth value.

[0059] In this scheme, by introducing a first initial weight and a second initial weight, the model can more finely adjust the contribution of different measurement points to the prediction results, thereby improving the physical relevance and accuracy of the prediction results. This ensures the scientific nature of the weight allocation and the interpretability of the prediction results.

[0060] Vibration prediction based on multi-head attention consideration: For non-measurement point i, its weight estimation is first calculated by multi-head attention of vibration of all measurement points, where head 1 represents distance-based attention, as shown in equation (8): ;

[0061] in, Defined as a distance weight parameter.

[0062] The first 2 represents amplitude-based attention, as shown in equation (9): ;

[0063] in, a j Indicates the first j Vibration amplitude at each measuring point; M Indicates the number of measurement points.

[0064] The first 3 represents the combined attention of distance and vibration, as in equation (10):

[0065] ,in, =0.4, defined as the vibration weighting parameter.

[0066] The first 4 represents structural complexity attention, as shown in equation (5).

[0067] The weights of each attention head h are subjected to softmax normalization, as shown in Equation (11):

[0068] .

[0069] In the specific implementation process, before calculating the product of the twelfth value and the direction vector to obtain the thirteenth value, the method further includes: calculating the sum of the measured values ​​of the vibration amplitude of the measured point in all directions to obtain the fourteenth value; calculating the square of the measured values ​​of the vibration amplitude of the measured point in each direction to obtain multiple fifteenth values; calculating the sum of the multiple fifteenth values ​​to obtain the sixteenth value; calculating the square root of the sixteenth value to obtain the seventeenth value; and calculating the quotient of the fourteenth value and the seventeenth value to obtain the direction vector.

[0070] In this scheme, the calculation of the fourteenth numerical step provides a more comprehensive reflection of the overall vibration amplitude of the measured point, offering a numerical foundation that integrates vibration information from multiple directions for subsequent direction vector calculations. This ensures that the direction vector not only points to the main vibration direction but also considers the total vibration intensity. The calculation of the fifteenth numerical step accurately quantifies the vibration energy of the measured point in each direction. The square of the vibration amplitude reflects the magnitude of the vibration energy; this calculation provides detailed information on the energy distribution for constructing the direction vector, ensuring that the length of the direction vector (i.e., the norm of the direction vector) represents the vibration energy of the measured point in three-dimensional space, thus improving the physical relevance of the prediction results. The calculation of the sixteenth numerical step standardizes the length of the direction vector, ensuring that it accurately reflects the vibration energy of the measured point and serves as a guiding vector for subsequent prediction calculations, improving prediction accuracy. The calculation of the seventeenth numerical step guarantees the normalization of the direction vector, making comparisons between different measured points comparable and avoiding inconsistencies in direction vector length due to differences in vibration energy magnitude at different points. This improves the consistency and determinism of the global vibration prediction. By calculating the direction vector, the vibration direction and energy level of the measured point can be accurately described, providing dual guidance of direction and energy for vibration prediction at non-measured points, thus improving the accuracy and robustness of vibration prediction.

[0071] The contribution of each directional component is expressed as equation (12): ;

[0072] in, This represents the direction vector of the j-th measurement point.

[0073] The composite vibration prediction is expressed as equation (13): ;

[0074] The result of multi-head prediction fusion is expressed as equation (14): ;

[0075] The direction vector is expressed as equation (15): .

[0076] In some embodiments, before calculating the curvature feature and the sum of the smallest positive numbers to obtain the second value, the method further includes: calculating the difference between the coordinates of the j-th measured point and the i-th measured point in space to obtain a first initial curvature; calculating the difference between the coordinates of the j+1-th measured point and the i-th measured point in space to obtain a second initial curvature; calculating the product of the first initial curvature and the second initial curvature to obtain an eighteenth value; calculating the norm of the eighteenth value to obtain a nineteenth value; calculating the quotient of the eighteenth value and the nineteenth value to obtain a normal vector; and calculating the norm of the difference between the normal vectors of the coordinates of the j-th and i-th measured points in space to obtain the curvature feature.

[0077] In this scheme, the calculation of the second initial curvature allows for a deeper analysis of the continuous changes in local geometric features, ensuring that the model can more accurately capture subtle differences in structural characteristics between measured points. This is crucial for assessing the propagation characteristics of vibration at structural inflection points. The calculation of the eighteenth numerical value enhances the model's understanding of local spatial curvature relationships, enabling it to fully consider the complex interactions of spatial geometric features when predicting vibration propagation, thus improving the physical relevance and accuracy of the prediction. The calculation of the nineteenth numerical value standardizes the representation of curvature relationship strength, ensuring that the model can compare and analyze curvature features on a uniform scale regardless of changes in the relative positions of the measured points, avoiding calculation errors caused by differences in measured point positions. The calculation of the normal vector accurately describes the directionality of local curvature, allowing the model to fully consider the geometric direction information of the structure when assessing vibration propagation, enhancing the physical interpretability of the prediction. The calculation of curvature features comprehensively reflects the impact of structural complexity on vibration propagation, ensuring that the model considers not only the intensity and direction of vibration but also the geometric features of the structure when predicting vibration, improving the accuracy and robustness of global vibration estimation.

[0078] Parameterization of measurement points: M vibration sensors are arranged on the test structure (each sensor measures the amplitudes x(t), y(t), and z(t) in the x, y, and z directions of the test structure) to obtain the characteristic parameters as shown in equation (1): Where u(t) represents the normalized composite direction vector, The formulas for representing the composite amplitude are as follows:

[0079] , ;

[0080] Geometric feature extraction of the test sample wireframe model: Based on the actual test subject structure, a wireframe model of the test sample is established, which directly defines the spatial layout and geometric features of the structure. The main parameters are:

[0081] (1) Euclidean distance d ij : ,in, and These are the coordinates of points i and j in the wireframe model, respectively.

[0082] (2) Curvature characteristics Estimation formula for 3D wireframe models (3): ,in, Represents the normal vector of adjacent edges: ;

[0083] (3) Average normal vector ; where weight Take the area of ​​adjacent triangles; n represents the number of adjacent points.

[0084] In the specific implementation process, the above predicted value is corrected to obtain the corrected predicted value, including: calculating the difference between the kth measured value and the kth predicted value to obtain the kth prediction error; calculating the product of the above prediction error and the above distance attenuation coefficient to obtain the twentieth value; calculating the weighted average of multiple above twentieth values ​​to obtain the twenty-first value; and calculating the sum of the above twenty-first value and the above multi-head prediction fusion result to obtain the above corrected predicted value.

[0085] In this scheme, the deviation between the predicted value and the actual vibration state can be quantified by calculating the prediction error. This deviation forms the basis of the subsequent correction mechanism, ensuring that the correction process can make specific adjustments for under- or over-prediction. By introducing a distance attenuation coefficient, the influence of the prediction error on the correction of predicted values ​​in non-measurement areas can be adjusted, ensuring that the correction mechanism considers both the prediction error and the physical characteristics of vibration propagation across the entire domain. Through weighted averaging, measurement information from multiple correction points can be integrated to obtain a guiding value for global correction, improving the robustness and global accuracy of the model correction. The corrected predicted value can further improve the accuracy and reliability of global vibration prediction. This final correction value integrates the original prediction information and error correction based on measured data, ensuring that the model prediction results are closer to the actual vibration characteristics.

[0086] In some embodiments, after correcting the predicted values ​​to obtain corrected predicted values, the method further includes: calculating the mean of all the predicted values ​​to obtain a twenty-second value; calculating the standard deviation of all the predicted values ​​to obtain a twenty-third value; calculating the product of the twenty-third value and a fifth preset value to obtain a twenty-fourth value; calculating the sum of the twenty-second value, a sixth preset value, and the twenty-fourth value to obtain an initial uncertainty score; calculating the absolute value of the difference between the corrected predicted value and the multi-head prediction fusion result to obtain a twenty-fifth value; calculating the product of the twenty-fifth value and a seventh preset value to obtain a corrected uncertainty score; and calculating the sum of the initial uncertainty score and the corrected uncertainty score to obtain a target uncertainty score. The parameters of the vibration prediction model are adjusted according to the target uncertainty score to obtain an optimized vibration prediction model. The parameters of the vibration prediction model include one or more of a loss function, learning rate, activation function, and regularization coefficient. The optimized vibration prediction model is trained using the training set to obtain a target vibration prediction model.

[0087] In this scheme, the calculation of the twenty-second numerical value comprehensively assesses the baseline level of the global vibration prediction, providing a central trend reference point for subsequent uncertainty analysis. The calculation of the twenty-third numerical value quantifies the volatility of the global prediction values, thereby assessing the stability of the prediction results and the uncertainty of the prediction model. The calculation of the twenty-fourth numerical value adjusts the weight of the initial uncertainty score, ensuring the model can flexibly adjust its sensitivity to uncertainty analysis based on the volatility of actual prediction results. The calculation of the initial uncertainty score quantifies the overall uncertainty of the prediction results, providing a quantitative basis for subsequent model parameter optimization. The calculation of the twenty-fifth numerical value evaluates the effectiveness of the model correction mechanism, ensuring the corrected prediction values ​​more closely reflect actual vibration characteristics. The calculation of the correction uncertainty score quantifies the impact of model correction on prediction uncertainty, ensuring improved reliability of the prediction results. The calculation of the target uncertainty score comprehensively assesses the uncertainty level of the global prediction results, providing engineers with a key indicator for measuring the reliability of global predictions.

[0088] Obtain a prediction model, which can be one of the following: SVR model, Random Forest model, CNN model, Kalman Filter model, Bayesian Regression model, or FEA model. Use historical measured values ​​and corresponding predicted labels to form a training set. Train the prediction model using the training set to obtain a vibration prediction model, where the predicted labels are the historically corrected predicted values ​​corresponding to the historical measured values ​​in the training set. Adjust the parameters of the vibration prediction model according to the target uncertainty score to obtain an optimized vibration prediction model. The parameters of the vibration prediction model include one or more of the loss function, learning rate, activation function, and regularization term coefficient. Train the optimized vibration prediction model using the training set to obtain the target vibration model. Input the measured values ​​into the target vibration model to obtain the predicted values ​​corresponding to the measured values.

[0089] The initial uncertainty is expressed as equation (17): ;

[0090] in: This represents the predicted mean; This represents the standard deviation of the predicted values.

[0091] Introduce k calibration measurement points (k < M) in addition to the M main measurement points, and use the measured results to correct the predicted value, as shown in equation (18): ;

[0092] in, a i The result of the calculation is given by equation (14); w k Define equation (11); Indicates the prediction error of the calibration measuring point: ,in, This represents the k-th measured value; This represents the k-th predicted value.

[0093] Based on the theory of elastic wave transmission, the vibration energy decreases exponentially with distance, which is consistent with the definition of equation (4).

[0094] Corrected prediction uncertainty: Based on the uncertainty results, the reliability of the predicted values ​​is judged. If the uncertainty is generally high in some areas, it indicates that the model's predictive ability in these areas is insufficient, and more test data is needed for parameter callback.

[0095] The total uncertainty after correction is expressed as equation (19):

[0096] Among them, correcting uncertainty .

[0097] Process Implementation: Taking a permanent magnet motor as an example, equivalent to a cuboid, a geometric model of the surface of the object being measured is established as follows: Figure 3 As shown.

[0098] Measurement point location and amplitude information: This example includes the location information of 5 measurement points. Table 1 shows the amplitude and location information of the registered points, and their amplitudes and the positions of the measurement points in the geometric model are shown in Table 1:

[0099] Table 1: Amplitude and Location Information of Main Measurement Points

[0100]

[0101] Correction point location and amplitude information: This example includes the location information of two correction points. Table 2 shows the amplitude and location information of the correction measurement points, and their amplitude and the location of the measurement points in the geometric model are shown in Table 2:

[0102] Table 2: Amplitude and Location Information of Calibration Measurement Points

[0103]

[0104] The vibration prediction process based on multivariate data fusion calculation is as follows: Figure 4As shown, first construct the geometric model of the object to be measured, which contains the coordinate space of n points; measure the amplitude of 5 points, and measure the x, y, z directions of each point; calculate the transfer function from the five points to the remaining n-5 points according to formula (7); according to the definition of multi-head attention mechanism, use four weight functions (distance, amplitude, structure, distance + amplitude) and normalize with softmax (formulas 5, 8, 9, 10, 11); calculate the sum of the vibration prediction (formula 12) (formula 13) for each non-measurement point according to the coordinates and amplitude of the 5 points, and assign weights through multi-head attention mechanism (formula 14); select two points as correction points in the remaining n-5 non-measurement points to correct the prediction model (formula 18); determine whether the sampling of the remaining n-7 measurement points is consistent with the actual measurement results (whether the hot zone distribution is accurate). If yes, end; if no, adjust the weight parameters and return to the normalization calculation stage.

[0105] Vibration adaptive prediction results based on multivariate data fusion calculation: After obtaining the wireframe model of the surface of the object under test, the main measuring points, and the calibration measuring points, the solution can be performed according to the process. Table 3 shows the vibration prediction results (partial measuring points). The prediction results are shown in Table 3.

[0106] Table 3: Vibration Prediction Results (Partial Measurement Points)

[0107]

[0108] The vibration prediction results and direction distribution in the global visualization results are as follows: Figure 5 As shown, the spatial distribution of vibration-related points is displayed using three-dimensional X, Y, and Z coordinates, including the main measuring point, the calibration measuring point, and the points corresponding to the vibration amplitude. Some points are marked with the specific values ​​of the vibration amplitude (such as 2.116, 1.90, etc.), and the maximum value of the vibration amplitude (2.409) is also marked; the amplitude and direction information of the predicted vibration are also shown.

[0109] Vibration hotspot distribution as follows Figure 6 As shown, the region distribution is displayed using three-dimensional X, Y, and Z coordinates. Figure 6 It includes the locations of 5 types of hotspot areas, main measuring points, and calibration measuring points. Each location is marked with its corresponding hotspot intensity value, and the maximum hotspot intensity value for different areas is also marked. It predicts the range of possible vibration hotspot areas of the system.

[0110] Uncertainty distribution of measurement point prediction, as follows Figure 7 As shown: The spatial distribution of relevant points is displayed using three-dimensional X, Y, and Z coordinates. Figure 7 It includes the locations corresponding to uncertainties, high uncertainty points, main measurement points, and correction measurement points.

[0111] High uncertainty points are identified, the confidence level of the prediction results is quantified, and a reliable uncertainty estimate is provided for each prediction node. The relationship between the vibration amplitude and uncertainty at the predicted measurement point is as follows: Figure 8 As shown: Using "Measurement Point Number (1-48)" as the horizontal axis, the trends of both "Vibration Prediction Value" and "Prediction Uncertainty" are displayed. Figure 8 It includes the main measuring point (measured value), the calibration measuring point (measured value), the predicted value, the prediction uncertainty, and the data points corresponding to the hot spot area; the statistics box in the upper left corner lists the relevant statistical information: average vibration value (0.5634), maximum vibration value (2.2138), average uncertainty (0.1087), and maximum uncertainty (0.1815).

[0112] At the same time, high uncertainty measurement points are plotted to remind application personnel to pay special attention to them.

[0113] This method is not only for predicting the results of a single vibration response value, but can also be extended to analyze each vibration peak obtained by spectral analysis, and calculate the vibration prediction results of the whole domain under the vibration response at different frequency points.

[0114] This invention, from the perspective of practical engineering applications, proposes a vibration intelligent prediction method based on multivariate data fusion and physical constraints, using commonly used vibration pens or triaxial accelerometers in common application scenarios, to achieve full-domain vibration estimation of complex models. This method has low hardware requirements and fast computation speed, requiring only a limited number of measurement points to quickly estimate the full-domain vibration of systems under different load scenarios. It has extremely high practical application value in engineering applications where rapid assessment of vibration hotspots is required.

[0115] This invention proposes a vibration intelligent prediction method based on multi-data fusion and physical constraints. The core technology lies in utilizing intelligent multi-data fusion to achieve rapid prediction of the vibration response of a test sample without a physical model. The technical effects are as follows:

[0116] 1. Multi-source information fusion: Simultaneously utilize measurement point data and geometric information to improve prediction reliability;

[0117] 2. Physical constraints: The physical laws of vibration propagation are incorporated into the calculation model to avoid non-physical errors introduced by purely data-driven approaches;

[0118] 3. Attention mechanism: Adaptively learns the weight influence of different measurement points on different regions. The attention mechanism combined with physical constraints is more in line with the actual vibration characteristics.

[0119] Interpretability: Through uncertainty analysis, it provides a confidence estimate for each prediction, helping engineers determine the reliability of the results.

[0120] This application also provides a vibration amplitude prediction device. It should be noted that the vibration amplitude prediction device of this application can be used to execute the vibration amplitude prediction method provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0121] The vibration amplitude prediction device provided in the embodiments of this application will be described below.

[0122] Figure 9 This is a structural block diagram of a vibration amplitude prediction device according to an embodiment of this application. Figure 9 As shown, the device includes:

[0123] Acquisition unit 10 is used to acquire the measured value of the vibration amplitude at the actual measurement point of the device under test;

[0124] The first calculation unit 20 is used to calculate the predicted value of the vibration amplitude of the prediction point based on the distance attenuation coefficient, structural constraint coefficient, resonance factor and transmission constraint coefficient of the measured point. The distance attenuation coefficient is a coefficient that measures the change of vibration energy with the propagation distance. The structural constraint coefficient is a coefficient that measures the change of vibration energy with the geometric characteristics of the structure. The resonance factor is a coefficient that simulates the resonance phenomenon of the structure at a specific distance. The transmission constraint coefficient is a transmission coefficient of vibration change calculated based on the distance attenuation coefficient, the structural constraint coefficient and the resonance factor. The measured point and the prediction point are located at different positions.

[0125] The correction unit 30 is used to correct the predicted value to obtain a corrected predicted value. The corrected predicted value is used to provide data for the prediction model to predict the vibration amplitude, so that the vibration characteristics of the device under test can be evaluated using the corrected predicted value. The historical measured values ​​and the corresponding prediction labels are combined into a training set. The prediction model is trained using the training set to obtain a vibration prediction model. The prediction label is the historical corrected predicted value corresponding to the historical measured value in the training set.

[0126] This embodiment introduces the concept of physical constraints, including distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient. By using multiple influences to predict vibration, it can be ensured that the predicted value follows various physical laws. At the same time, the predicted value is also corrected to identify prediction differences, further ensuring that the predicted value is closer to the actual vibration behavior, reducing prediction errors, and thus improving the accuracy of vibration prediction.

[0127] In the specific implementation process, the first calculation unit includes a first calculation module, a second calculation module, a third calculation module, a fourth calculation module, a fifth calculation module, a sixth calculation module, a seventh calculation module, an eighth calculation module, a ninth calculation module, a tenth calculation module, an eleventh calculation module, a twelfth calculation module, a thirteenth calculation module, a fourteenth calculation module, a fifteenth calculation module, a sixteenth calculation module, a seventeenth calculation module, and an eighteenth calculation module. The first calculation module is used to calculate the product of the negative attenuation coefficient and the Euclidean distance to obtain a first value, wherein the Euclidean distance is the difference in the coordinates of the two measured points in space; the second calculation module is used to calculate... The distance attenuation coefficient is obtained by raising the first value of the natural constant to a power; the third calculation module is used to calculate the sum of the curvature feature and the minimum positive number to obtain the second value, wherein the curvature feature characterizes the degree of curvature of the region where the measured point is located; the fourth calculation module is used to calculate the quotient of the minimum curvature feature and the second value to obtain the structural constraint coefficient; the fifth calculation module is used to calculate the quotient of the Euclidean distance and the first preset value to obtain the third value; the sixth calculation module is used to calculate the first preset multiple of π to obtain the fourth value; the seventh calculation module is used to calculate the product of the third value and the fourth value to obtain the fifth value; the eighth calculation module is used to calculate... The sine function value of the fifth value above is used to obtain the sixth value; the ninth calculation module is used to calculate the sum of the sixth value and the second preset value to obtain the seventh value; the tenth calculation module is used to calculate the second preset multiple of the structural amplification factor to obtain the eighth value; the eleventh calculation module is used to calculate the product of the seventh value and the eighth value above to obtain the ninth value; the twelfth calculation module is used to calculate the sum of the ninth value and the third preset value to obtain the resonance factor; the thirteenth calculation module is used to calculate the product of the distance attenuation coefficient and the resonance factor above to obtain the tenth value; the fourteenth calculation module is used to calculate the difference between the fourth preset value and the structural constraint coefficient above. The eleventh value is obtained by calculating the product of the tenth and eleventh values. The fifteenth calculation module is used to calculate the product of the tenth and eleventh values ​​to obtain the transfer constraint coefficient. The sixteenth calculation module is used to calculate the product of the multi-head prediction fusion result and the transfer constraint coefficient to obtain the twelfth value, wherein the multi-head prediction fusion result is calculated based on the square of the vibration amplitude of multiple measured points. The seventeenth calculation module is used to calculate the product of the twelfth value and the direction vector to obtain the thirteenth value, wherein the direction vector is the unit vector of the vibration direction. The eighteenth calculation module is used to calculate the weighted average of multiple thirteenth values ​​to obtain the predicted value.

[0128] In this scheme, the distance attenuation coefficient is calculated to accurately reflect the natural attenuation law of vibration energy with distance, ensuring the physical rationality of non-measuring point predictions. The calculation of the structural constraint coefficient quantifies the influence of structural geometry on vibration propagation, especially in areas with large structural curvature, effectively reducing the predicted value and simulating the natural attenuation of vibration energy at structural bends in reality. The calculation of the resonance factor simulates the resonance characteristics of the structure, periodically amplifying the vibration amplitude at specific distances, so that the prediction results can reflect the resonance effect of the actual structure. By calculating the transfer constraint coefficient, the influence of distance attenuation, structural constraints, and resonance factor is integrated, providing a mathematical model that comprehensively considers physical constraints for non-measuring point predictions. By combining the multi-head prediction fusion results with the transfer constraint coefficient and further multiplying them with the direction vector, the predicted value is finally obtained through weighted averaging, achieving accurate estimation of the vibration amplitude at non-measuring points.

[0129] In some embodiments, the above-described apparatus further includes a second calculation unit, a third calculation unit, a fourth calculation unit, and a processing unit. The second calculation unit is used to calculate a first initial weight based on distance before calculating a weighted average of multiple of the thirteenth values ​​to obtain the predicted value. The third calculation unit is used to calculate a second initial weight based on amplitude. The fourth calculation unit is used to calculate a weighted average of the first initial weight and the second initial weight to obtain a third initial weight. The processing unit is used to perform softmax normalization on the third initial weight to obtain a target weight value, wherein the target weight value is the weight coefficient of the thirteenth value.

[0130] In this scheme, by introducing a first initial weight and a second initial weight, the model can more finely adjust the contribution of different measurement points to the prediction results, thereby improving the physical relevance and accuracy of the prediction results. This ensures the scientific nature of the weight allocation and the interpretability of the prediction results.

[0131] In the specific implementation process, the above-mentioned device further includes a fifth calculation unit, a sixth calculation unit, a seventh calculation unit, an eighth calculation unit, and a ninth calculation unit. The fifth calculation unit is used to calculate the sum of the measured values ​​of the vibration amplitude of the above-mentioned measurement point in all directions before calculating the product of the twelfth value and the direction vector to obtain the thirteenth value, thereby obtaining the fourteenth value. The sixth calculation unit is used to calculate the square of the measured value of the vibration amplitude of the above-mentioned measurement point in each direction, thereby obtaining multiple fifteenth values. The seventh calculation unit is used to calculate the sum of the multiple fifteenth values, thereby obtaining the sixteenth value. The eighth calculation unit is used to calculate the square root of the sixteenth value, thereby obtaining the seventeenth value. The ninth calculation unit is used to calculate the quotient of the fourteenth value and the seventeenth value, thereby obtaining the direction vector.

[0132] In this scheme, the calculation of the fourteenth numerical step provides a more comprehensive reflection of the overall vibration amplitude of the measured point, offering a numerical foundation that integrates vibration information from multiple directions for subsequent direction vector calculations. This ensures that the direction vector not only points to the main vibration direction but also considers the total vibration intensity. The calculation of the fifteenth numerical step accurately quantifies the vibration energy of the measured point in each direction. The square of the vibration amplitude reflects the magnitude of the vibration energy; this calculation provides detailed information on the energy distribution for constructing the direction vector, ensuring that the length of the direction vector (i.e., the norm of the direction vector) represents the vibration energy of the measured point in three-dimensional space, thus improving the physical relevance of the prediction results. The calculation of the sixteenth numerical step standardizes the length of the direction vector, ensuring that it accurately reflects the vibration energy of the measured point and serves as a guiding vector for subsequent prediction calculations, improving prediction accuracy. The calculation of the seventeenth numerical step guarantees the normalization of the direction vector, making comparisons between different measured points comparable and avoiding inconsistencies in direction vector length due to differences in vibration energy magnitude at different points. This improves the consistency and determinism of the global vibration prediction. By calculating the direction vector, the vibration direction and energy level of the measured point can be accurately described, providing dual guidance of direction and energy for vibration prediction at non-measured points, thus improving the accuracy and robustness of vibration prediction.

[0133] In some embodiments, the above-described apparatus further includes a tenth calculation unit, an eleventh calculation unit, a twelfth calculation unit, a thirteenth calculation unit, a fourteenth calculation unit, and a fifteenth calculation unit. The tenth calculation unit is used to calculate the difference between the coordinates of the j-th measured point and the i-th measured point in space before calculating the curvature feature and the sum of the smallest positive number to obtain the second value, thereby obtaining a first initial curvature. The eleventh calculation unit is used to calculate the difference between the coordinates of the j+1-th measured point and the i-th measured point in space, thereby obtaining a second initial curvature. The twelfth calculation unit is used to calculate the product of the first initial curvature and the second initial curvature, thereby obtaining an eighteenth value. The thirteenth calculation unit is used to calculate the norm of the eighteenth value, thereby obtaining a nineteenth value. The fourteenth calculation unit is used to calculate the quotient of the eighteenth value and the nineteenth value, thereby obtaining a normal vector. The fifteenth calculation unit is used to calculate the norm of the difference between the normal vectors of the coordinates of the j-th measured point and the i-th measured point in space, thereby obtaining the curvature feature.

[0134] In this scheme, the calculation of the second initial curvature allows for a deeper analysis of the continuous changes in local geometric features, ensuring that the model can more accurately capture subtle differences in structural characteristics between measured points. This is crucial for assessing the propagation characteristics of vibration at structural inflection points. The calculation of the eighteenth numerical value enhances the model's understanding of local spatial curvature relationships, enabling it to fully consider the complex interactions of spatial geometric features when predicting vibration propagation, thus improving the physical relevance and accuracy of the prediction. The calculation of the nineteenth numerical value standardizes the representation of curvature relationship strength, ensuring that the model can compare and analyze curvature features on a uniform scale regardless of changes in the relative positions of the measured points, avoiding calculation errors caused by differences in measured point positions. The calculation of the normal vector accurately describes the directionality of local curvature, allowing the model to fully consider the geometric direction information of the structure when assessing vibration propagation, enhancing the physical interpretability of the prediction. The calculation of curvature features comprehensively reflects the impact of structural complexity on vibration propagation, ensuring that the model considers not only the intensity and direction of vibration but also the geometric features of the structure when predicting vibration, improving the accuracy and robustness of global vibration estimation.

[0135] In the specific implementation process, the correction unit includes a nineteenth calculation module, a twentieth calculation module, a twenty-first calculation module, and a twenty-second calculation module. The nineteenth calculation module is used to calculate the difference between the kth measured value and the kth predicted value to obtain the kth prediction error. The twentieth calculation module is used to calculate the product of the above prediction error and the above distance attenuation coefficient to obtain the twentieth value. The twenty-first calculation module is used to calculate the weighted average of multiple above twentieth values ​​to obtain the twentieth value. The twenty-second calculation module is used to calculate the sum of the above twentieth value and the above multi-head prediction fusion result to obtain the above corrected prediction value.

[0136] In this scheme, the deviation between the predicted value and the actual vibration state can be quantified by calculating the prediction error. This deviation forms the basis of the subsequent correction mechanism, ensuring that the correction process can make specific adjustments for under- or over-prediction. By introducing a distance attenuation coefficient, the influence of the prediction error on the correction of predicted values ​​in non-measurement areas can be adjusted, ensuring that the correction mechanism considers both the prediction error and the physical characteristics of vibration propagation across the entire domain. Through weighted averaging, measurement information from multiple correction points can be integrated to obtain a guiding value for global correction, improving the robustness and global accuracy of the model correction. The corrected predicted value can further improve the accuracy and reliability of global vibration prediction. This final correction value integrates the original prediction information and error correction based on measured data, ensuring that the model prediction results are closer to the actual vibration characteristics.

[0137] In some embodiments, the above-described apparatus further includes a sixteenth calculation unit, a seventeenth calculation unit, an eighteenth calculation unit, a nineteenth calculation unit, a twentieth calculation unit, a twenty-first calculation unit, and a twenty-second calculation unit. The sixteenth calculation unit is used to correct the predicted values, and after obtaining the corrected predicted values, calculate the mean of all the predicted values ​​to obtain the twenty-second value. The seventeenth calculation unit is used to calculate the standard deviation of all the predicted values ​​to obtain the twenty-third value. The eighteenth calculation unit is used to calculate the product of the twenty-third value and a fifth preset value to obtain the twenty-fourth value. The nineteenth calculation unit is used to calculate the sum of the twenty-second value, the sixth preset value, and the twenty-fourth value to obtain an initial uncertainty score. The twentyth calculation unit... The calculation unit is used to calculate the absolute value of the difference between the above-corrected predicted value and the above-mentioned multi-head prediction fusion result, to obtain the twenty-fifth value; the twenty-first calculation unit is used to calculate the product of the above-mentioned twenty-fifth value and the seventh preset value, to obtain the corrected uncertainty score; the twenty-second calculation unit is used to calculate the sum of the above-mentioned initial uncertainty score and the above-mentioned corrected uncertainty score, to obtain the target uncertainty score, wherein the parameters of the above-mentioned vibration prediction model are adjusted according to the above-mentioned target uncertainty score to obtain the optimized vibration prediction model, wherein the above-mentioned parameters of the above-mentioned vibration prediction model include one or more of the loss function, learning rate, activation function, and regularization term coefficient, and the above-mentioned training set is used to train the above-mentioned optimized vibration prediction model to obtain the target vibration prediction model.

[0138] In this scheme, the calculation of the twenty-second numerical value comprehensively assesses the baseline level of the global vibration prediction, providing a central trend reference point for subsequent uncertainty analysis. The calculation of the twenty-third numerical value quantifies the volatility of the global prediction values, thereby assessing the stability of the prediction results and the uncertainty of the prediction model. The calculation of the twenty-fourth numerical value adjusts the weight of the initial uncertainty score, ensuring the model can flexibly adjust its sensitivity to uncertainty analysis based on the volatility of actual prediction results. The calculation of the initial uncertainty score quantifies the overall uncertainty of the prediction results, providing a quantitative basis for subsequent model parameter optimization. The calculation of the twenty-fifth numerical value evaluates the effectiveness of the model correction mechanism, ensuring the corrected prediction values ​​more closely reflect actual vibration characteristics. The calculation of the correction uncertainty score quantifies the impact of model correction on prediction uncertainty, ensuring improved reliability of the prediction results. The calculation of the target uncertainty score comprehensively assesses the uncertainty level of the global prediction results, providing engineers with a key indicator for measuring the reliability of global predictions.

[0139] The aforementioned vibration amplitude prediction device includes a processor and a memory. The acquisition unit, the first calculation unit, the correction unit, etc., are all stored as program units in the memory, and the processor executes the program units stored in the memory to achieve the corresponding functions. All of the above modules are located in the same processor; alternatively, the various modules may be located in different processors in any combination.

[0140] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and adjusting kernel parameters can address the challenges of predicting vibration thermal zones in practical engineering applications using existing technologies.

[0141] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0142] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the vibration amplitude prediction method.

[0143] This invention provides a processor for running a program, wherein the program executes the vibration amplitude prediction method.

[0144] This invention provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements at least the steps of a method for predicting vibration amplitude. The device described herein can be a server, PC, PAD, mobile phone, etc.

[0145] This application also provides a computer program product that, when executed on a data processing device, is adapted to perform the initialization steps of a prediction method with at least a dynamic amplitude value.

[0146] This application also provides a vibration amplitude prediction system, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include methods for performing any of the vibration amplitude prediction methods described above.

[0147] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0148] 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.

[0149] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0150] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0151] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0152] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0153] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0154] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0155] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0156] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0157] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for predicting vibration amplitude, characterized in that, include: Obtain the measured values ​​of the vibration amplitude at the measured points of the device under test; Based on the distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient of the measured point, the predicted value of the vibration amplitude at the prediction point is calculated. The distance attenuation coefficient measures the change in vibration energy with propagation distance; the structural constraint coefficient measures the change in vibration energy with structural geometric characteristics; the resonance factor simulates the resonance phenomenon of the structure at a specific distance; and the transmission constraint coefficient is the transmission coefficient of vibration change calculated based on the distance attenuation coefficient, the structural constraint coefficient, and the resonance factor. The measured point and the prediction point are located at different positions. The predicted value is corrected to obtain a corrected predicted value, wherein the corrected predicted value is used to provide data for the prediction model to predict the vibration amplitude, so that the vibration characteristics of the device under test can be evaluated using the corrected predicted value. The historical measured values ​​and the corresponding prediction labels are combined to form a training set, and the prediction model is trained using the training set to obtain a vibration prediction model, wherein the prediction label is the historical corrected predicted value corresponding to the historical measured value in the training set.

2. The method according to claim 1, characterized in that, Based on the distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient of the measured point, the predicted value of the vibration amplitude at the predicted point is calculated, including: The first value is obtained by multiplying the negative of the attenuation coefficient and the Euclidean distance, where the Euclidean distance is the difference between the coordinates of the two measured points in space. The distance attenuation coefficient is obtained by raising the first value of the natural constant to a power of 1. The second value is obtained by calculating the sum of the curvature feature and the smallest positive number, wherein the curvature feature characterizes the degree of curvature of the region where the measured point is located; The structural constraint coefficient is obtained by calculating the quotient of the minimum curvature feature and the second value. Calculate the quotient of the Euclidean distance and the first preset value to obtain the third value; Calculate the first preset multiple of π to obtain the fourth value; Calculate the product of the third value and the fourth value to obtain the fifth value; Calculate the sine function value of the fifth value to obtain the sixth value; Calculate the sum of the sixth value and the second preset value to obtain the seventh value; Calculate the second preset factor of the structural magnification factor to obtain the eighth value; Calculate the product of the seventh value and the eighth value to obtain the ninth value; The resonance factor is obtained by calculating the sum of the ninth value and the third preset value; Calculate the product of the distance attenuation coefficient and the resonance factor to obtain the tenth value; Calculate the difference between the fourth preset value and the structural constraint coefficient to obtain the eleventh value; Calculate the product of the tenth value and the eleventh value to obtain the transitive constraint coefficient; The product of the multi-head prediction fusion result and the transfer constraint coefficient is calculated to obtain the twelfth value, wherein the multi-head prediction fusion result is calculated based on the square of the vibration amplitude of multiple measured points; Calculate the product of the twelfth value and the direction vector to obtain the thirteenth value, where the direction vector is the unit vector of vibration in the vibration direction; The predicted value is obtained by calculating a weighted average of the multiple thirteenth values.

3. The method according to claim 2, characterized in that, Before calculating a weighted average of the multiple thirteenth values ​​to obtain the predicted value, the method further includes: Calculate the first initial weight based on the distance; The second initial weight is calculated based on the amplitude. Calculate the weighted average of the first initial weight and the second initial weight to obtain the third initial weight; The third initial weight is subjected to softmax normalization to obtain the target weight value, wherein the target weight value is the weight coefficient of the thirteenth value.

4. The method according to claim 2, characterized in that, Before calculating the product of the twelfth value and the direction vector to obtain the thirteenth value, the method further includes: The sum of the measured values ​​of the vibration amplitude at the measured point in all directions is calculated to obtain the fourteenth value; Calculate the square of the measured value of the vibration amplitude at the measured point in each direction to obtain multiple fifteenth values; Calculate the sum of the multiple fifteenth values ​​to obtain the sixteenth value; Calculate the square root of the sixteenth value to obtain the seventeenth value; The direction vector is obtained by calculating the quotient of the fourteenth value and the seventeenth value.

5. The method according to claim 2, characterized in that, Before calculating the sum of the curvature feature and the smallest positive number to obtain the second value, the method further includes: Calculate the difference between the coordinates of measured point j in space and the coordinates of measured point i in space to obtain the first initial curvature; Calculate the difference between the coordinates of the measured point j+1 in space and the coordinates of the measured point i in space to obtain the second initial curvature; Calculate the product of the first initial curvature and the second initial curvature to obtain the eighteenth value; Calculate the norm of the eighteenth value to obtain the nineteenth value; Calculate the quotient of the eighteenth and nineteenth values ​​to obtain the normal vector; The curvature feature is obtained by calculating the norm of the difference between the normal vectors of the coordinates of measured point j and measured point i in space.

6. The method according to any one of claims 2 to 5, characterized in that, The predicted value is corrected to obtain the corrected predicted value, including: Calculate the difference between the kth measured value and the kth predicted value to obtain the kth prediction error; Calculate the product of the prediction error and the distance attenuation coefficient to obtain the twentieth value; Calculate the weighted average of multiple twentieth values ​​to obtain the twenty-first value; The corrected prediction value is obtained by summing the 21st value and the multi-head prediction fusion result.

7. The method according to any one of claims 2 to 5, characterized in that, After correcting the predicted value to obtain the corrected predicted value, the method further includes: Calculate the mean of all the predicted values ​​to obtain the twenty-second value; Calculate the standard deviation of all the predicted values ​​to obtain the twenty-third value; Calculate the product of the twenty-third value and the fifth preset value to obtain the twenty-fourth value; Calculate the sum of the 22nd value, the 6th preset value, and the 24th value to obtain the initial uncertainty score; Calculate the absolute value of the difference between the corrected prediction value and the multi-head prediction fusion result to obtain the twenty-fifth value; Calculate the product of the 25th value and the 7th preset value to obtain the corrected uncertainty score; The initial uncertainty score and the corrected uncertainty score are summed to obtain the target uncertainty score. The parameters of the vibration prediction model are adjusted according to the target uncertainty score to obtain an optimized vibration prediction model. The parameters of the vibration prediction model include one or more of a loss function, learning rate, activation function, and regularization term coefficient. The optimized vibration prediction model is trained using the training set to obtain the target vibration prediction model.

8. A device for predicting vibration amplitude, characterized in that, include: The acquisition unit is used to acquire the measured value of the vibration amplitude at the actual measurement point of the device under test. The first calculation unit is used to calculate the predicted value of the vibration amplitude at the prediction point based on the distance attenuation coefficient, structural constraint coefficient, resonance factor, and transmission constraint coefficient of the measured point. The distance attenuation coefficient is a coefficient that measures the change of vibration energy with propagation distance, the structural constraint coefficient is a coefficient that measures the change of vibration energy with structural geometric characteristics, the resonance factor is a coefficient that simulates the resonance phenomenon of the structure at a specific distance, and the transmission constraint coefficient is a transmission coefficient of vibration change calculated based on the distance attenuation coefficient, the structural constraint coefficient, and the resonance factor. The measured point and the prediction point are located at different positions. A correction unit is used to correct the predicted value to obtain a corrected predicted value. The corrected predicted value is used to provide data for the prediction model to predict the vibration amplitude, so that the vibration characteristics of the device under test can be evaluated using the corrected predicted value. The historical measured values ​​and corresponding prediction labels are combined into a training set, and the prediction model is trained using the training set to obtain a vibration prediction model. The prediction label is the historical corrected predicted value corresponding to the historical measured value in the training set.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the vibration amplitude prediction method according to any one of claims 1 to 7.

10. A vibration amplitude prediction system, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including a method for performing a vibration amplitude prediction method according to any one of claims 1 to 7.

Images