Modal prediction method and device for vehicle power assembly suspension bracket and vehicle

Abstract

The invention relates to the technical field of vehicle performance development, in particular to a modal prediction method and device for a vehicle power assembly suspension bracket and a vehicle, and the method comprises the steps: constructing a suspension bracket assembly calculation model; constructing an assembly mathematical model based on the suspension bracket assembly calculation model and the corresponding experiment working condition; and performing modal calculation based on the assembly mathematical model to obtain a modal prediction result of the vehicle power assembly suspension bracket. Therefore, the problems that the NVH performance development period of the power assembly suspension bracket is prolonged, the cost is increased and the rapid optimization of the NVH performance of the whole vehicle is restricted due to the fact that a modal prediction method of the power assembly suspension bracket in the related technology is difficult to balance the relationship among the prediction precision, the calculation efficiency and the development cost are solved.

Description

Technical Field

[0001] This application relates to the field of vehicle performance development technology, and in particular to a modal prediction method, device and vehicle for a vehicle powertrain suspension bracket. Background Technology

[0002] With the rapid development of the automotive industry and the increasing demands of consumers for vehicle quality, NVH (Noise, Vibration, and Harshness) performance has become a core indicator for measuring vehicle comfort and premium quality. In vehicle NVH performance development, the powertrain, as one of the largest sources of vibration and noise in the vehicle (contributing 40%-60%), has its vibration transmission characteristics controlled, making NVH development a crucial aspect. The powertrain mounting system, as a core component connecting the powertrain and the vehicle body, performs three functions: supporting the powertrain weight, isolating vibration transmission, and limiting powertrain displacement. Its dynamic characteristics (especially modal characteristics) directly determine the efficiency of powertrain vibration transmission to the vehicle body. If the modal parameters of the mounting system (such as natural frequencies and mode shapes) are coupled with those of the powertrain or vehicle body, resonance can occur, significantly amplifying in-vehicle vibration and noise (e.g., steering wheel vibration at idle, body resonance at specific speeds). Therefore, accurately predicting the modal characteristics of the mounting system is an important prerequisite for powertrain NVH performance development.

[0003] In related technologies, modal prediction of powertrain mounting brackets mainly relies on two types of methods: One type is the experimental testing method: by making physical prototypes of the suspension bracket, the modal parameters are measured in the laboratory using vibration tests (such as hammer impact method, shaking table method).

[0004] One type is the finite element simulation method: the geometric model of the suspension bracket is constructed based on 3D modeling software (such as CATIA), and modal calculations are performed using finite element analysis software (such as ANSYS, Abaqus).

[0005] However, while modal prediction methods for powertrain mounts can predict modal characteristics during the design phase and reduce reliance on physical samples, they suffer from drawbacks such as low computational efficiency, high modeling complexity, difficulty in parameter correction, and high computational cost. This makes it difficult to balance the relationship between "prediction accuracy, computational efficiency, and development cost," resulting in a longer development cycle and increased cost for the NVH performance of powertrain mounts, which restricts the rapid optimization of the overall vehicle's NVH performance.

[0006] Therefore, there is an urgent need for a powertrain suspension bracket modal prediction method that can significantly improve computational efficiency and reduce computational costs while ensuring prediction accuracy, so as to meet the high-efficiency and low-cost R&D needs of the modern automotive industry. Summary of the Invention

[0007] This application provides a modal prediction method, device, and vehicle for powertrain mounting brackets, in order to solve the problems that related modal prediction methods for powertrain mounting brackets are difficult to balance the relationship between "prediction accuracy, computational efficiency, and development cost," which leads to extended development cycles and increased costs for the NVH performance of powertrain mounting brackets, and restricts the rapid optimization of the NVH performance of the whole vehicle.

[0008] The first aspect of this application provides a modal prediction method for a vehicle powertrain suspension bracket, comprising the following steps: constructing a calculation model of the suspension bracket assembly; constructing a mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions; and performing modal calculations based on the mathematical model of the assembly to obtain the modal prediction results of the vehicle powertrain suspension bracket.

[0009] Through the aforementioned technical means, the embodiments of this application can reproduce the physical characteristics of the suspension bracket by setting up a computational model (such as a finite element model) for the suspension bracket assembly. Combined with corresponding experimental conditions (static load, dynamic excitation, ultimate impact), the actual stress scenario is anchored, thereby providing a reliable data foundation for subsequent modeling. The assembly mathematical model can further integrate the learning capabilities of deep neural networks, which not only retains the mechanical accuracy of the computational model, but also achieves the capture of nonlinear characteristics under complex working conditions and the bidirectional improvement of computational efficiency and iterative optimization through "input-output" label mapping and data-driven training. Compared with modal calculations that rely solely on finite element models, the assembly mathematical model can shorten the modal prediction time to the millisecond level through the fast inference capability of neural networks, supporting high-frequency parameter iteration in the suspension bracket design stage. At the same time, the model can be continuously optimized through test sets, gradually adapting to uncovered working conditions. The generalization ability is continuously enhanced with the accumulation of data, reducing reliance on physical experiments and effectively shortening the R&D cycle.

[0010] Optionally, in one embodiment of this application, the step of constructing a mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions includes: determining the experimental conditions based on the calculation model of the suspension bracket assembly, and performing model preprocessing using a finite element preprocessor to obtain a processed finite element model.

[0011] Through the above-mentioned technical means, the embodiments of this application can construct the mathematical model of the assembly by "determining the appropriate experimental conditions based on the calculation model of the suspension bracket assembly, and then combining the finite element preprocessing component to complete the model preprocessing to obtain the processed finite element model", thereby realizing the deep coupling between the experimental conditions and the calculation model.

[0012] Optionally, in one embodiment of this application, the step of constructing a mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions further includes: defining input and output labels for the finite element model; generating a training set and a validation set for training a deep neural network based on the input and output labels; and constructing the mathematical model of the assembly using the training set and the validation set.

[0013] Through the above technical solution, in the process of constructing the mathematical model of the assembly based on the computational model of the suspension bracket assembly and experimental conditions, the embodiments of this application define input and output labels for the finite element model, which can clearly define the mechanical parameter input and performance result output of the model, providing accurate data mapping relationships for data-driven modeling. Then, training sets and validation sets can be generated based on these labels to train deep neural networks and construct the assembly mathematical model. This not only leverages the richness and accuracy of finite element simulation data to allow the neural network to fully learn the complex mechanical properties of the suspension bracket, but also ensures the model's generalization ability through the validation set. The resulting assembly mathematical model not only avoids the problems of low computational efficiency and reliance on hardware resources of traditional finite element models, but also can quickly output performance prediction results under experimental conditions, greatly improving the efficiency of suspension bracket mechanical property analysis. At the same time, it provides efficient model support for subsequent optimization of operating condition parameters and rapid performance iteration, further shortening the R&D cycle and reducing computational costs.

[0014] Optionally, in one embodiment of this application, the step of constructing an assembly mathematical model based on the suspension bracket assembly calculation model and the corresponding experimental conditions further includes: obtaining the output result of the assembly mathematical model using a test set; and optimizing the assembly mathematical model using the output result.

[0015] Through the above-mentioned technical means, the embodiments of this application can obtain the model output results through the test set and continuously optimize them during the process of constructing the mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the experimental conditions, thereby significantly improving the generalization ability and prediction accuracy of the model.

[0016] Optionally, in one embodiment of this application, before performing modal calculations based on the assembly mathematical model, the method further includes: accepting user adjustment parameters; and updating the assembly mathematical model based on the adjustment parameters.

[0017] Through the above-mentioned technical means, the embodiments of this application can enable the assembly mathematical model to reflect the user's design intent in real time through the dynamic update mechanism that accepts the user's adjustment parameters, thereby ensuring that subsequent modal calculations are carried out based on the latest parameters and improving the model's ability to support the design iteration of the suspension bracket.

[0018] A second aspect of this application provides a modal prediction device for a vehicle powertrain suspension bracket, comprising: a first construction module for constructing a calculation model of the suspension bracket assembly; a second construction module for constructing a mathematical model of the assembly based on the calculation model of the suspension bracket assembly and corresponding experimental conditions; and a prediction module for performing modal calculations based on the mathematical model of the assembly to obtain modal prediction results for the vehicle powertrain suspension bracket.

[0019] Through the aforementioned technical means, the embodiments of this application can reproduce the physical characteristics of the suspension bracket by setting up a computational model (such as a finite element model) for the suspension bracket assembly. Combined with corresponding experimental conditions (static load, dynamic excitation, ultimate impact), the actual stress scenario is anchored, thereby providing a reliable data foundation for subsequent modeling. The assembly mathematical model can further integrate the learning capabilities of deep neural networks, which not only retains the mechanical accuracy of the computational model, but also achieves the capture of nonlinear characteristics under complex working conditions and the bidirectional improvement of computational efficiency and iterative optimization through "input-output" label mapping and data-driven training. Compared with modal calculations that rely solely on finite element models, the assembly mathematical model can shorten the modal prediction time to the millisecond level through the fast inference capability of neural networks, supporting high-frequency parameter iteration in the suspension bracket design stage. At the same time, the model can be continuously optimized through test sets, gradually adapting to uncovered working conditions. The generalization ability is continuously enhanced with the accumulation of data, reducing reliance on physical experiments and effectively shortening the R&D cycle.

[0020] Optionally, in one embodiment of this application, the second construction module includes: a processing unit, used to determine the experimental conditions based on the calculation model of the suspension bracket assembly, and to perform model preprocessing using a finite element preprocessor to obtain a processed finite element model.

[0021] Through the above-mentioned technical means, the embodiments of this application can construct the mathematical model of the assembly by "determining the appropriate experimental conditions based on the calculation model of the suspension bracket assembly, and then combining the finite element preprocessing component to complete the model preprocessing to obtain the processed finite element model", thereby realizing the deep coupling between the experimental conditions and the calculation model.

[0022] Optionally, in one embodiment of this application, the second construction module further includes: a definition unit for defining input and output labels for the finite element model; and a construction unit for generating a training set and a validation set for training a deep neural network based on the input and output labels, and constructing the assembly mathematical model using the training set and the validation set.

[0023] Through the aforementioned technical means, in the process of constructing the mathematical model of the assembly based on the computational model of the suspension bracket assembly and experimental conditions, this application embodiment defines input and output labels for the finite element model, which clearly defines the mechanical parameter input and performance result output of the model, providing a precise data mapping relationship for data-driven modeling. Furthermore, training and validation sets can be generated based on these labels to train a deep neural network and construct the assembly mathematical model. This leverages the richness and accuracy of finite element simulation data, allowing the neural network to fully learn the complex mechanical properties of the suspension bracket, while the validation set ensures the model's generalization ability. The resulting assembly mathematical model not only avoids the problems of low computational efficiency and reliance on hardware resources inherent in traditional finite element models, but also quickly outputs performance prediction results under experimental conditions, significantly improving the efficiency of suspension bracket mechanical property analysis. Simultaneously, it provides efficient model support for subsequent optimization of operating parameters and rapid performance iteration, further shortening the R&D cycle and reducing computational costs.

[0024] Optionally, in one embodiment of this application, the second building module is further configured to: obtain the output result of the assembly mathematical model using a test set; and optimize the assembly mathematical model using the output result.

[0025] Through the above-mentioned technical means, the embodiments of this application can obtain the model output results through the test set and continuously optimize them during the process of constructing the mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the experimental conditions, thereby significantly improving the generalization ability and prediction accuracy of the model.

[0026] Optionally, in one embodiment of this application, it further includes: a receiving module, configured to receive user adjustment parameters before performing modal calculations based on the assembly mathematical model; and an updating module, configured to update the assembly mathematical model based on the adjustment parameters.

[0027] Through the above-mentioned technical means, the embodiments of this application can enable the assembly mathematical model to reflect the user's design intent in real time through the dynamic update mechanism that accepts the user's adjustment parameters, thereby ensuring that subsequent modal calculations are carried out based on the latest parameters and improving the model's ability to support the design iteration of the suspension bracket.

[0028] A third aspect of this application provides a vehicle, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the modal prediction method for vehicle powertrain suspension brackets as described in the above embodiments.

[0029] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described modal prediction method for vehicle powertrain suspension brackets.

[0030] A fifth aspect of this application provides a computer program product that stores a computer program that, when executed by a processor, implements the above-described modal prediction method for vehicle powertrain suspension brackets.

[0031] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0032] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a modal prediction method for a vehicle powertrain suspension bracket according to an embodiment of this application; Figure 2 This is a flowchart of a modal prediction method for a vehicle powertrain mounting bracket according to a specific embodiment of this application; Figure 3 This is a schematic diagram of the modal prediction device for a vehicle powertrain suspension bracket according to an embodiment of this application; Figure 4 This is a structural schematic diagram of a vehicle provided according to an embodiment of this application.

[0033] Explanation of reference numerals in the attached figures: 10 - Modal prediction device for vehicle powertrain suspension bracket; 100 - First building module; 200 - Second building module; 300 - Prediction module. Detailed Implementation

[0034] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0035] The modal prediction method, apparatus, and vehicle for a vehicle powertrain suspension bracket according to embodiments of this application are described below with reference to the accompanying drawings. Addressing the problem mentioned in the background section that modal prediction methods for powertrain suspension brackets struggle to balance prediction accuracy, computational efficiency, and development cost, leading to prolonged NVH performance development cycles and increased costs, thus hindering rapid optimization of overall vehicle NVH performance, this application provides a modal prediction method for a vehicle powertrain suspension bracket. In this method, a mathematical model of the assembly is set to characterize the structural features of components and the assembly, and the modal calculation results of the suspension bracket are predicted by combining input material properties, loads, and other parameters, thereby providing a reference for modal optimization of the suspension bracket. This method is based on deep learning for modal prediction, avoiding extensive preliminary calculations and providing a reference for subsequent modal calculations of the suspension bracket, thus improving computational efficiency. This solves the problem that the modal prediction method for powertrain mounts in related technologies is difficult to balance the relationship between "prediction accuracy, computational efficiency and development cost", which leads to a longer development cycle and increased cost of powertrain mount NVH performance, and restricts the rapid optimization of vehicle NVH performance.

[0036] Specifically, Figure 1 This is a flowchart illustrating a modal prediction method for a vehicle powertrain suspension bracket provided in an embodiment of this application.

[0037] like Figure 1 As shown, the modal prediction method for the vehicle powertrain suspension bracket includes the following steps: In step S101, a calculation model of the suspension bracket assembly is constructed.

[0038] The calculation model of the suspension bracket assembly can be understood as a digital simulation model used to analyze the mechanical characteristics (such as modal, stiffness, and strength) of the powertrain suspension bracket system.

[0039] In actual implementation, the embodiments of this application can convert CAD drawings into mathematical models that can be used for calculation to establish a calculation model of the suspension bracket assembly.

[0040] The embodiments of this application can determine the test conditions by constructing a calculation model of the suspension bracket assembly, thereby further constructing a simplified mathematical model of the assembly.

[0041] In step S102, a mathematical model of the assembly is constructed based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions.

[0042] In the construction of the mathematical model for the suspension bracket assembly, the "test conditions" refer to the loads, constraints, and environmental conditions that the suspension bracket experiences during actual vehicle installation and use. This is the core basis for transforming the physical world's usage scenarios into "input parameters" that the mathematical model can recognize. Essentially, by defining standardized and reproducible test scenarios, it ensures that the computational boundaries of the mathematical model are highly consistent with actual working conditions. This allows the simulation results (such as modal frequencies, stress distribution, and vibration transmissibility) to accurately reflect the true mechanical behavior of the suspension bracket, providing effective support for subsequent performance optimization and reliability verification.

[0043] Optionally, in one embodiment of this application, a mathematical model of the assembly is constructed based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions, including: determining the experimental conditions based on the calculation model of the suspension bracket assembly, and performing model preprocessing using a finite element preprocessor to obtain the processed finite element model.

[0044] Specifically, in actual implementation, the embodiments of this application determine the test conditions based on the calculation model of the suspension bracket assembly. First, the core characteristics of the model (such as the simulation capability determined by the unit type, the accuracy of connection modeling, and the applicable scope of the material model) can be analyzed to clarify the types of loads and constraint boundaries that can be accurately calculated. Then, the performance requirements can be further transformed into quantitative parameters (such as setting the static load at 1.2 times the powertrain weight, matching the dynamic excitation frequency to the powertrain vibration order, and taking the ultimate impact load 3-5 times the static load) by referring to powertrain parameters (such as setting the static load at 1.2 times the powertrain weight, matching the dynamic excitation frequency to the powertrain vibration order, and taking the ultimate impact load 3-5 times the static load) through model pre-simulation to verify the rationality of the parameters (such as stress not exceeding material limits and vibration transmissibility convergence) to ensure that the test conditions are both consistent with actual usage and match the capabilities of the calculation model, supporting the consistency verification between simulation and physical testing.

[0045] Furthermore, in this embodiment, the determined test conditions can be imported into the model, and the model preprocessing can be performed using the finite element processing component Hypermesh, thereby outputting a simplified mathematical model of the assembly. Model preprocessing can include model simplification, such as chamfer removal, bolt hole simplification, etc., and material properties can be imported after simplification to establish connection relationships, etc.

[0046] Through the aforementioned technical means, the embodiments of this application can construct a mathematical model of the assembly by "determining suitable experimental conditions based on the computational model of the suspension bracket assembly, and then completing model preprocessing with finite element preprocessing components to obtain a processed finite element model". This enables deep coupling between the experimental conditions and the computational model. It can clarify the boundaries of the operating condition parameters by leveraging the characteristics of the computational model, avoiding distortion caused by the operating conditions deviating from the model's simulation capabilities. Furthermore, it can optimize the model's geometry, mesh, connections, and physical properties through professional preprocessing, ensuring that the model accurately reflects the true mechanical properties of the assembly. Ultimately, the obtained finite element model can accurately meet the simulation analysis requirements under experimental conditions, significantly improving the consistency between simulation results and physical experiments, reducing research and development iterations caused by model-operating condition mismatch, while also reducing the cost of physical prototype manufacturing and testing, and shortening the development cycle of the suspension bracket's NVH performance and structural reliability.

[0047] Optionally, in one embodiment of this application, constructing a mathematical model of the assembly based on the computational model of the suspension bracket assembly and the corresponding experimental conditions further includes: defining input and output labels for the finite element model; generating a training set and a validation set for training a deep neural network based on the input and output labels; and constructing the mathematical model of the assembly using the training set and the validation set.

[0048] Before training the model, the embodiments of this application can first build a dataset. For example, the embodiments of this application can use the input and output labels of the model definition after HyperMesh preprocessing to divide it into a training set and a validation set, which are used as dataset 1 for training the deep neural network, thereby further constructing the assembly mathematical model.

[0049] As an feasible approach, input labels can correspond to the working condition driving parameters of the finite element model, that is, the external conditions that affect the mechanical state of the suspension bracket. The core is the load, constraint and environmental parameters in the experimental working condition, such as "powertrain static load (2450N)" and "bolt preload (8000N)" under static support working condition, "vibration excitation frequency (26.7Hz)" and "excitation force amplitude (125N)" under idle mode working condition, and "body side bracket fixed position" in the constraint conditions. These labels are directly related to the adjustable parameters of the "load application" and "constraint setting" modules in the finite element model.

[0050] The output labels correspond to the performance response results of the finite element model, that is, the mechanical performance of the suspension bracket under the input working conditions. The core is the NVH and structural performance indicators that need to be verified, such as: "maximum deformation of bracket (1.2mm)" and "maximum stress around bolt hole (320MPa)" under static working conditions, and "vibration transmissibility (28%)" and "first natural frequency (25Hz)" under dynamic working conditions. These labels are directly extracted from the core calculation results of the finite element model post-processing module (such as stress cloud diagram, displacement report, modal analysis table).

[0051] In short, the input label can be the "force and constraint conditions of the model" and the output label can be the "performance feedback results of the model". Together, they form a standardized "input-output" data pair, which provides clear data boundaries and physical meaning for the subsequent generation of training and validation sets for deep neural networks, ensuring that the data-driven model can accurately learn the complex mechanical properties of the suspension bracket.

[0052] Through the above technical solution, in the process of constructing the mathematical model of the assembly based on the computational model of the suspension bracket assembly and experimental conditions, the embodiments of this application define input and output labels for the finite element model, which can clearly define the mechanical parameter input (such as load, constraint) and performance result output (such as stress, modal frequency) of the model, providing accurate data mapping relationship for data-driven modeling. Then, training sets and validation sets can be generated based on the labels to train deep neural networks and construct assembly models. This not only leverages the richness and accuracy of finite element simulation data to allow the neural network to fully learn the complex mechanical properties of the suspension bracket, but also ensures the generalization ability of the model through the validation set. The resulting assembly mathematical model not only avoids the problems of low computational efficiency and reliance on hardware resources of traditional finite element models, but also can quickly output performance prediction results under experimental conditions, greatly improving the efficiency of suspension bracket mechanical property analysis. At the same time, it provides efficient model support for subsequent optimization of working condition parameters and rapid performance iteration, further shortening the R&D cycle and reducing computational costs.

[0053] In step S103, modal calculations are performed based on the mathematical model of the assembly to obtain the modal prediction results of the vehicle powertrain suspension bracket.

[0054] In practical implementation, the embodiments of this application can be used for modal calculations or performance analysis based on the mathematical model of the suspension bracket assembly (including finite element model and deep neural network model). The model input needs to be designed around the "core parameters affecting the mechanical state and modal characteristics of the suspension bracket", covering three dimensions: boundary constraints, load conditions, and structural and material properties. This ensures that the input parameters can both reproduce the actual working conditions and accurately drive the model to output the target performance results. Examples include constraint position and type, assembly gap parameters, static load, dynamic excitation load, structural geometric parameters, and material mechanical parameters. These parameters are used as input data, or as a test set. Furthermore, modal calculations can be performed on the high-performance computing platform Nastran based on the assembly mathematical model to output modal prediction results. It is important to note that as much sample data as possible should be prepared during the initial finite element model preparation process to ensure the accuracy of the results.

[0055] Optionally, in one embodiment of this application, constructing a mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions further includes: obtaining the output result of the mathematical model of the assembly using a test set; and optimizing the mathematical model of the assembly using the output result.

[0056] Those skilled in the art should understand that after inputting the test set data as input data into the constructed assembly mathematical model, the calculated output data will inevitably be obtained. In the embodiments of this application, the predicted calculated data can be further processed, input and output labels can be defined, each group of data can be arranged in vector form, and the data can be divided into training set and validation set to build dataset 2 for further training of deep neural network.

[0057] In this embodiment, the Adam algorithm can be used as the model training algorithm. A deep neural network model is built using Matlab's deep network designer. The calculation results are saved as a CSV file, read by Matlab, and converted into a numerical matrix. The data is then divided into training and validation sets. The validation set is used to predict the model, and test indicators are observed. The model is further optimized based on the prediction results. The assembly mathematical model can be understood as a composite model that integrates physical characteristics and data-driven capabilities. Its construction process includes steps such as "finite element model preprocessing → input and output label definition → neural network training". Among them, the deep neural network built by Matlab's deep network designer can be understood as the key module in the assembly mathematical model responsible for "learning the mapping relationship between working condition parameters and performance response" - it uses the structured data generated by the finite element model (converted into a numerical matrix after reading from a CSV file) as the training basis, optimizes the network parameters through the Adam algorithm, and finally has the ability to quickly predict the mechanical properties of the suspension bracket (such as modal frequencies and stress distribution).

[0058] Furthermore, in embodiments of this application, when employing a deep fully connected network architecture, the ReLU activation function is used, the number of fully connected layers is no less than 20, L2 regularization is used to prevent overfitting, and the cost function of the neural network is:

[0059] in, y i This represents the true value in the dataset. The predicted value output by the neural network. For network weights, is the regularization coefficient, and m and n represent the number of training samples and the number of weights, respectively. Through the aforementioned technical means, the embodiments of this application can, during the process of constructing a mathematical model of the assembly based on the computational model of the suspension bracket assembly and experimental conditions, obtain the model output results through a test set and continuously optimize it accordingly, thereby significantly improving the model's generalization ability and prediction accuracy. The test set can simulate new working conditions in actual engineering that were not involved in the training (such as uncovered powertrain vibration frequencies, changes in material properties under extreme temperatures), and by comparing the deviations between the model output results (such as modal frequencies, stress values) and real physical experimental data, accurately locate the model's shortcomings (such as insufficient prediction of material stiffness decay under high-temperature environments, etc.). (Calculation deviations in response to complex impact loads); Based on these deviations, the model can be optimized in a targeted manner—for example, adjusting the number of network layers and regularization coefficients of the deep neural network, correcting the temperature correction coefficients of material parameters in the finite element model, and supplementing training samples for key working conditions. This enables the assembly mathematical model to not only adapt to the trained working conditions but also stably cope with unseen new scenarios, ultimately achieving a continuous reduction in model prediction errors. This ensures that the model can provide reliable modal calculation and performance analysis results at different design stages and under different operating conditions of the suspension bracket, providing more accurate support for bracket structure optimization and NVH performance improvement.

[0060] Optionally, in one embodiment of this application, before performing modal calculations based on the assembly mathematical model, the method further includes: accepting user adjustment parameters; and updating the assembly mathematical model based on the adjustment parameters.

[0061] Specifically, before performing modal calculations based on the assembly mathematical model, a "receive user-adjusted parameters → update model" step is introduced, enabling the model to dynamically adapt to design changes. First, the model receives user-inputted adjustment parameters (e.g., adjustments to the structural parameters of a suspension bracket, such as increasing the bracket thickness from 5mm to 6mm, increasing the number of reinforcing ribs from 2 to 3; changes to material parameters, such as replacing Q355 steel with 6061 aluminum alloy; or modifications to working boundary conditions, such as adjusting the bolt preload from 8000N to 9000N). Then, these adjustments are mapped to the corresponding modules of the assembly mathematical model through a parameterized interface. For the finite element foundation, the dimensional parameters of the geometric model, the mechanical parameters in the material property library, or the values ​​of constraint loads can be updated. For the deep neural network, if the adjustment involves core model characteristics (e.g., a drastic change in stiffness due to material replacement), supplementary samples can be generated based on the new parameters to retrain the network. If it is a minor adjustment of local parameters (e.g., a small change in preload), the network weights can be quickly updated through transfer learning.

[0062] Through the above-mentioned technical means, the embodiments of this application can enable the assembly mathematical model to reflect the user's design intent in real time through the dynamic update mechanism that accepts the user's adjustment parameters, thereby ensuring that subsequent modal calculations are carried out based on the latest parameters and improving the model's ability to support the design iteration of the suspension bracket.

[0063] like Figure 2 As shown, the modal prediction method for vehicle powertrain suspension brackets proposed in this application will be illustrated below with a specific embodiment. The specific steps can be set as follows: Step S201: Begin; Step S202: Model preprocessing; This step can import the model according to the determined test conditions, use Hypermesh to perform model preprocessing, and output a simplified assembly mathematical model, which can be a finite element model.

[0064] Step S203: Modal calculation; This step can perform modal calculations using the high-performance computing platform Nastran based on the finite element model provided in step S202, and output the results.

[0065] Step S204: Dataset Construction; This step can be done by defining input and output labels for the model after HyperMesh preprocessing in step S202, dividing the model into training and validation sets, and constructing dataset 1 for training deep neural networks; Alternatively, the computational data obtained in step S203 can be processed to define input and output labels, arrange each set of data in vector form, divide the data into training and validation sets, and construct dataset 2 for training deep neural networks.

[0066] Step S205: Model Training; This step can use the Adam algorithm as the model training algorithm. A deep neural network is built using the deep network designer in Matlab. The calculation results are saved as a CSV file, read by Matlab and converted into a numerical matrix. The data is then divided into a training set and a validation set. The validation set is used to predict the model and the test indicators are observed. The model is further optimized based on the prediction results.

[0067] Step S206: Model evaluation; This step determines whether the model needs to continue training by evaluating whether the model's fit meets the standard.

[0068] Step S207: Output model; if the fit meets the standard after evaluation in step S206, the constructed assembly mathematical model is obtained.

[0069] The modal prediction method for vehicle powertrain suspension brackets proposed in this application can characterize the structural features of components and assemblies by setting up a mathematical model of the assembly, and predict the modal calculation results of the suspension bracket by combining input material properties, loads, and other parameters, thereby providing a reference for modal optimization of the suspension bracket. This method is based on deep learning for modal prediction, avoiding a large amount of computation in the early stages, providing a reference for subsequent modal calculations of the suspension bracket, and improving computational efficiency. Therefore, it solves the problems of related technologies' modal prediction methods for powertrain suspension brackets, which struggle to balance the relationship between "prediction accuracy, computational efficiency, and development cost," leading to extended NVH performance development cycles, increased costs, and hindering the rapid optimization of overall vehicle NVH performance.

[0070] Next, refer to the appendix. Figure 3 This application describes a modal prediction device for a vehicle powertrain suspension bracket according to an embodiment of the present application.

[0071] Figure 3 This is a block diagram of the modal prediction device for the vehicle powertrain suspension bracket according to an embodiment of this application.

[0072] like Figure 3 As shown, the modal prediction device 10 for the vehicle powertrain suspension bracket includes: a first building module 100, a second building module 200, and a prediction module 300.

[0073] The first building module 100 is used to build a calculation model of the suspension bracket assembly.

[0074] The second construction module 200 is used to construct a mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions.

[0075] The prediction module 300 is used to perform modal calculations based on the assembly mathematical model to obtain the modal prediction results of the vehicle powertrain suspension bracket.

[0076] Optionally, in one embodiment of this application, the second construction module 200 includes: a processing unit; the processing unit is used to determine the experimental conditions based on the calculation model of the suspension bracket assembly, and to perform model preprocessing using finite element preprocessing components to obtain the processed finite element model.

[0077] Optionally, in one embodiment of this application, the second construction module 200 further includes: a definition unit and a construction unit; wherein, the definition unit is used to define input and output labels for the finite element model; the construction unit is used to generate a training set and a validation set for training a deep neural network based on the input and output labels, and to construct an assembly mathematical model using the training set and the validation set.

[0078] Optionally, in one embodiment of this application, the second construction module 200 is further configured to: obtain the output result of the assembly mathematical model using the test set; and optimize the assembly mathematical model using the output result.

[0079] Optionally, in one embodiment of this application, it further includes: a receiving module and an updating module; wherein, the receiving module is used to accept the user's adjustment parameters before performing modal calculations based on the assembly mathematical model; and the updating module is used to update the assembly mathematical model based on the adjustment parameters.

[0080] It should be noted that the explanation of the above-described method for modal prediction of vehicle powertrain mounting brackets also applies to the modal prediction device for vehicle powertrain mounting brackets in this embodiment, and will not be repeated here.

[0081] The modal prediction device for vehicle powertrain suspension brackets proposed in this application can characterize the structural features of components and assemblies by setting up a mathematical model of the assembly, and predict the modal calculation results of the suspension bracket by combining input material properties, loads, and other parameters, thereby providing a reference for modal optimization of the suspension bracket. This method is based on deep learning for modal prediction, avoiding a large amount of computation in the early stages, providing a reference for subsequent modal calculations of the suspension bracket, and improving computational efficiency. Therefore, it solves the problems of related technologies' modal prediction methods for powertrain suspension brackets, which struggle to balance the relationship between "prediction accuracy, computational efficiency, and development cost," leading to extended development cycles and increased costs for powertrain suspension bracket NVH performance, thus hindering the rapid optimization of the overall vehicle NVH performance.

[0082] Figure 4 A schematic diagram of the structure of a vehicle provided in an embodiment of this application. The vehicle may include: The memory 401, the processor 402, and the computer program stored on the memory 401 and capable of running on the processor 402.

[0083] When the processor 402 executes the program, it implements the modal prediction method for the vehicle powertrain suspension bracket provided in the above embodiments.

[0084] Furthermore, the vehicle also includes: Communication interface 403 is used for communication between memory 401 and processor 402.

[0085] The memory 401 is used to store computer programs that can run on the processor 402.

[0086] Memory 401 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0087] If the memory 401, processor 402, and communication interface 403 are implemented independently, then the communication interface 403, memory 401, and processor 402 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized into address buses, data buses, control buses, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0088] Optionally, in a specific implementation, if the memory 401, processor 402, and communication interface 403 are integrated on a single chip, then the memory 401, processor 402, and communication interface 403 can communicate with each other through an internal interface.

[0089] Processor 402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0090] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described modal prediction method for vehicle powertrain suspension brackets.

[0091] This application also provides a computer program product storing a computer program that, when executed by a processor, implements the above-described modal prediction method for vehicle powertrain suspension brackets.

[0092] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0093] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0094] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0095] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0096] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or more of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0097] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0098] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0099] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A modal prediction method for a vehicle powertrain suspension bracket, characterized in that, Includes the following steps: Construct a computational model of the suspension bracket assembly; Based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions, a mathematical model of the assembly is constructed. Modal calculations are performed based on the mathematical model of the assembly to obtain the modal prediction results of the vehicle powertrain suspension bracket.

2. The method according to claim 1, characterized in that, The construction of the mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions includes: The experimental conditions are determined based on the calculation model of the suspension bracket assembly, and the model is preprocessed using a finite element preprocessor to obtain the processed finite element model.

3. The method according to claim 2, characterized in that, The construction of the mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions also includes: Define input and output labels for the finite element model; Based on the input and output labels, a training set and a validation set for training a deep neural network are generated, and the mathematical model of the assembly is constructed using the training set and the validation set.

4. The method according to claim 3, characterized in that, The construction of the mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions also includes: The output results of the mathematical model of the assembly are obtained using the test set; The mathematical model of the assembly is optimized using the output results.

5. The method according to claim 1, characterized in that, Before performing modal calculations based on the mathematical model of the assembly, the following steps are also included: Accept user-adjusted parameters; The assembly mathematical model is updated based on the adjusted parameters.

6. A modal prediction device for a vehicle powertrain suspension bracket, characterized in that, include: The first building module is used to build the computational model of the suspension bracket assembly; The second construction module is used to construct a mathematical model of the assembly based on the calculation model of the suspension bracket assembly and the corresponding experimental conditions. The prediction module is used to perform modal calculations based on the mathematical model of the assembly to obtain the modal prediction results of the vehicle powertrain suspension bracket.

7. The apparatus according to claim 6, characterized in that, The second building module includes: The processing unit is used to determine the experimental conditions based on the calculation model of the suspension bracket assembly, and to perform model preprocessing using finite element preprocessing components to obtain the processed finite element model.

8. A vehicle, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, the processor executing the program to implement the modal prediction method for vehicle powertrain suspension brackets as described in any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the modal prediction method for vehicle powertrain suspension brackets as described in any one of claims 1-5.

10. A computer program product, characterized in that, The computer program is executed to implement the modal prediction method for vehicle powertrain suspension brackets as described in any one of claims 1-5.

Images