Vehicle structure optimization method and device based on random vibration analysis

Abstract

The invention relates to the technical field of vehicle structure engineering, in particular to a vehicle structure optimization method and device based on random vibration analysis, and the method comprises the steps: extracting key feature parameters of a model grid to reconstruct simulation grid data; loading constraint conditions on the reconstruction model to generate a random vibration response result; constructing a target training model by using the key feature parameters; inputting reconstructed geometric data needing to be optimized into the target training model, and outputting root mean square stress of the reconstructed geometric data; and based on the root mean square stress and the reconstructed geometric data, generating a structure lightweight optimization result of the vehicle. Therefore, the problems that related calculation usually depends on high-precision finite element modeling and large-scale numerical calculation, the calculation period is long, the efficiency is low, the dependency on calculation resources and professional experience is high, the practicability is limited, and application and popularization in engineering practice of vehicle structure optimization design are not facilitated are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle structure engineering, and particularly relates to a vehicle structure optimization method and device based on random vibration analysis. BACKGROUND

[0002] During driving, a vehicle is subjected to various random excitations such as road roughness, engine, transmission system and wind load, which will cause random vibration of the vehicle body and its components, thereby affecting the structural strength, fatigue life, ride comfort and noise control effect of the vehicle.

[0003] In the related art, random vibration analysis is a method based on vibration load statistical characteristics, which can predict and evaluate the dynamic response of a structure under random excitation. By establishing a finite element model of the vehicle structure and performing random vibration simulation analysis, the stress distribution, displacement response and high stress concentration position of the vehicle structure under different working conditions can be predicted, thereby providing a basis for structural reinforcement, fatigue life improvement and lightweight design.

[0004] However, in the related art, the random vibration analysis method usually relies on high-precision finite element modeling and large-scale numerical calculation, and the calculation period is long. Especially when performing multi-parameter and multi-objective optimization design, modeling, solving and result analysis need to be repeatedly performed, which leads to low overall calculation efficiency. At the same time, such a method relies highly on computing resources and professional experience, and is difficult to be quickly reused in different vehicle models and different structural schemes, which limits its applicability and practicability, and makes it difficult to guarantee its economy, and is not conducive to its popularization and application in vehicle structure design engineering practice. SUMMARY

[0005] The present application provides a vehicle structure optimization method and device based on random vibration analysis, to solve the problem in the related art that the optimization of a vehicle structure based on random vibration analysis usually relies on high-precision finite element modeling and large-scale numerical calculation, which is time-consuming and inefficient, and relies highly on computing resources and professional experience, which limits its practicability and is not conducive to its popularization and application in vehicle structure optimization design engineering practice.

[0006] The first aspect embodiment of the application provides a vehicle structure optimization method based on random vibration analysis, comprising the following steps: extracting at least one key feature parameter based on a predicted model mesh to reconstruct simulation mesh data based on a reconstruction model; loading at least one constraint condition on the reconstruction model to generate a random vibration response result; constructing a target training model using the at least one key feature parameter; extracting a structure geometry feature key feature parameter that needs to be optimized to obtain reconstruction geometry data; inputting the reconstruction geometry data into the target training model to output a first root mean square stress of the reconstruction geometry data; and generating a structure lightweight optimization result of the vehicle based on the first root mean square stress and the reconstruction geometry data.

[0007] Through the above technical means, the AI training model can be constructed according to the key feature parameters of the vehicle structure, the corresponding root mean square stress can be predicted through the model, and then the reconstruction geometry data can be analyzed and optimized based on the prediction result, so as to realize the structure lightweight design of the vehicle, the random vibration response of different design schemes can be quickly evaluated, high-cost simulation analysis is not needed each time, the acceleration of design iteration and the saving of computing resources are realized, and reference is provided for structure reliability and fatigue life analysis.

[0008] Optionally, in an embodiment of the application, the extraction of at least one key feature parameter based on the predicted model mesh to reconstruct simulation mesh data comprises: discretizing the predicted model mesh to obtain discretized discrete mesh data; extracting the at least one key feature parameter according to the discrete mesh data; and reconstructing the simulation mesh data according to the at least one key feature parameter based on the reconstruction model.

[0009] Through the above technical means, the key feature parameters can be extracted according to the discrete mesh data to reconstruct the simulation mesh data, the main geometry and mechanical characteristics of the structure can be retained, and the reconstructed simulation mesh data can be ensured to be consistent with the initial mesh in the key feature parameters, so as to ensure the authenticity and reliability of the analysis result.

[0010] Optionally, in an embodiment of the application, the loading of at least one constraint condition on the reconstruction model to generate a random vibration response result comprises: loading the at least one constraint condition on the reconstruction model according to an actual constraint state of the structure to obtain a modal calculation result; and loading a unit acceleration load on the modal calculation result according to an actual loading state of the structure to calculate the random vibration response result.

[0011] Through the above technical means, the embodiment of the application can obtain modal calculation results according to constraint conditions, which can be used to analyze the natural frequency and mode shape characteristics of the structure, and further apply a unit acceleration load on the basis of the modal calculation results to simulate the dynamic response of the structure under different frequency excitations, so as to obtain random vibration response results, which can be used to evaluate the performance of the structure under random excitation conditions and provide basic data for structure optimization and lightweight design.

[0012] Optionally, in an embodiment of the application, the constructing the target training model by using the at least one key feature parameter comprises: performing a Latin hypercube sampling DOE (Design of Experiments) test design based on the at least one key feature parameter to obtain experimental samples; reconstructing a reconstruction model in the experimental samples into non-Euclidean data, and extracting a second root mean square stress of the random vibration response result; and constructing the target training model based on the non-Euclidean data and the second root mean square stress.

[0013] Through the above technical means, the embodiment of the application can perform a Latin hypercube sampling DOE test design to generate experimental samples covering the parameter space, and then construct non-Euclidean data representation based on these samples, and construct a target training model in combination with the corresponding root mean square stress, which is used to predict the random vibration response of the structure under different design parameters, so as to realize structure performance evaluation and optimization design.

[0014] Optionally, in an embodiment of the application, the generating a structure lightweight optimization result of the vehicle based on the first root mean square stress and the reconstructed geometric data comprises: comparing the first root mean square stress of the reconstructed geometric data and a constraint stress to obtain a comparison result, and determining the structure lightweight optimization result in combination with the reconstructed geometric data and the comparison result.

[0015] Through the above technical means, the embodiment of the application can compare the root mean square stress and the constraint stress to evaluate the safety margin of the structure under random excitation conditions, and then perform optimization design on the structure in combination with the reconstructed geometric data and the comparison result, and adjust the geometric parameters of the structure, so as to realize lightweight design under the premise of ensuring the strength requirement.

[0016] The second aspect embodiment of the application provides a vehicle structure optimization device based on random vibration analysis, comprising: a reconstruction module configured to extract at least one key feature parameter based on a predicted model mesh, to reconstruct simulation mesh data based on a reconstruction model; a generation module configured to load at least one constraint condition on the reconstruction model to generate a random vibration response result; a construction module configured to construct a target training model using the at least one key feature parameter; an extraction module configured to extract a structure geometry feature key feature parameter that needs to be optimized to obtain reconstructed geometry data; an output module configured to input the reconstructed geometry data into the target training model to output a first root mean square stress of the reconstructed geometry data; and an optimization module configured to generate a structure lightweight optimization result of a vehicle based on the first root mean square stress and the reconstructed geometry data.

[0017] Through the above technical means, the AI training model can be constructed according to the key feature parameters of the vehicle structure, the corresponding root mean square stress can be predicted through the model, and then the reconstructed geometry data can be analyzed and optimized based on the prediction result, so as to realize the structure lightweight design of the vehicle, the random vibration response of different design schemes can be quickly evaluated, high-cost simulation analysis is not needed each time, the acceleration of design iteration and the saving of computing resources are realized, and reference is provided for structure reliability and fatigue life analysis.

[0018] Optionally, in an embodiment of the application, the reconstruction module comprises: a discretization unit configured to discretize the predicted model mesh to obtain discretized mesh data; an extraction unit configured to extract the at least one key feature parameter according to the discretized mesh data; and a reconstruction unit configured to reconstruct the simulation mesh data according to the at least one key feature parameter based on the reconstruction model.

[0019] Through the above technical means, the key feature parameters can be extracted according to the discretized mesh data to reconstruct the simulation mesh data, the main geometry and mechanical characteristics of the structure can be retained, and the reconstructed simulation mesh data can be ensured to be consistent with the initial mesh in the key feature parameters, so as to ensure the authenticity and reliability of the analysis result.

[0020] Optionally, in an embodiment of the application, the generation module comprises: a loading unit configured to load the at least one constraint condition on the reconstruction model according to an actual constraint state of the structure to obtain a modal calculation result; and a calculation unit configured to load a unit acceleration load on the modal calculation result according to an actual loading state of the structure to calculate the random vibration response result.

[0021] Through the above technical means, the embodiment of the application can obtain modal calculation results according to constraint conditions, which can be used to analyze the natural frequency and mode shape characteristics of the structure, and further apply a unit acceleration load on the basis of the modal calculation results to simulate the dynamic response of the structure under different frequency excitations, so as to obtain random vibration response results, which can be used to evaluate the performance of the structure under random excitation conditions and provide basic data for structure optimization and lightweight design.

[0022] Optionally, in an embodiment of the application, the construction module comprises: a design unit configured to perform Latin hypercube sampling DOE test design based on the at least one key feature parameter to obtain experimental samples; an extraction unit configured to reconstruct the reconstruction model in the experimental samples into non-Euclidean data and extract a second root mean square stress of the random vibration response result; and a construction unit configured to construct the target training model based on the non-Euclidean data and the second root mean square stress.

[0023] Through the above technical means, the embodiment of the application can perform Latin hypercube sampling DOE test design to generate experimental samples covering the parameter space, and then construct non-Euclidean data representation based on these samples and construct a target training model in combination with the corresponding root mean square stress, which is used to predict the random vibration response of the structure under different design parameters, so as to realize structure performance evaluation and optimization design.

[0024] Optionally, in an embodiment of the application, the optimization module comprises: a comparison unit configured to compare the first root mean square stress of the reconstructed geometric data and the constraint stress to obtain a comparison result; and an optimization unit configured to determine the structure lightweight optimization result in combination with the reconstructed geometric data and the comparison result.

[0025] Through the above technical means, the embodiment of the application can compare the root mean square stress and the constraint stress to evaluate the safety margin of the structure under random excitation conditions, and then perform optimization design on the structure in combination with the reconstructed geometric data and the comparison result to adjust the geometric parameters of the structure, so as to realize lightweight design under the premise of ensuring strength requirements.

[0026] The third aspect embodiment of the application provides a vehicle, comprising 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 vehicle structure optimization method based on random vibration analysis as described in the above embodiments.

[0027] The fourth aspect embodiment of the application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the vehicle structure optimization method based on random vibration analysis as described above.

[0028] The fifth aspect embodiment of the present application provides a computer program product comprising a computer program which, when executed, is configured to implement the vehicle structure optimization method based on random vibration analysis as described above.

[0029] Additional aspects and advantages of the present application will be made apparent by the following description and the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0030] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, including the accompanying drawings. Figure 1 A flow chart of a vehicle structure optimization method based on random vibration analysis according to an embodiment of the present application; Figure 2 A schematic diagram of simulation model mesh feature parameter extraction and reconstruction for an embodiment of the present application; Figure 3 A schematic diagram of model mesh reconstruction by node movement parameters for an embodiment of the present application; Figure 4 A schematic diagram of constraint conditions for an embodiment of the present application; Figure 5 A schematic diagram of volume and stress changes during optimization for an embodiment of the present application; Figure 6 A schematic diagram of structure comparison before and after optimization for an embodiment of the present application; Figure 7 A schematic diagram of the working principle of a vehicle structure optimization method based on random vibration analysis for an embodiment of the present application; Figure 8 A block schematic diagram of a vehicle structure optimization apparatus based on random vibration analysis according to an embodiment of the present application; Figure 9 A structure schematic diagram of a vehicle according to an embodiment of the present application.

[0031] REFERENCE NUMERALS 10 - vehicle structure optimization apparatus based on random vibration analysis; 100 - reconstruction module, 200 - generation module, 300 - construction module, 400 - extraction module, 500 - output module, 600 - optimization module; 901 - memory, 902 - processor, 903 - communication interface. DETAILED DESCRIPTION

[0032] Embodiments of the present application are described below in detail with reference to the accompanying drawings, examples of which are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0033] A vehicle structure optimization method and device based on random vibration analysis of embodiments of the present application are described below with reference to the accompanying drawings. In view of the technical problems in the prior art that the optimization of vehicle structure based on random vibration analysis usually relies on high-precision finite element modeling and large-scale numerical calculation, the calculation period is long, the efficiency is low, and the dependence on computing resources and professional experience is high, the practicability is limited, and the technology is not conducive to popularization and application in the engineering practice of vehicle structure optimization design, the present application provides a vehicle structure optimization method based on random vibration analysis, in which an AI training model is constructed using key feature parameters of the vehicle structure, and the corresponding root mean square stress is predicted through the model, and then the reconstructed geometric data is analyzed and optimized based on the prediction results, thereby realizing lightweight design of the vehicle structure. The AI training model can be used to quickly predict the random vibration response of the model grid node, realize efficient evaluation and optimization of different design schemes, and significantly reduce the design period and computing resource consumption without high-cost simulation analysis each time. Meanwhile, through rapid iterative optimization, a balance between structure performance and lightweight can be achieved, providing a reference for structure reliability and fatigue life analysis, and forming an efficient intelligent structure optimization closed loop. Thus, the problems of related technologies that rely on high-precision finite element modeling and large-scale numerical calculation, long calculation period, low efficiency, high dependence on computing resources and professional experience, limited practicability, and difficulty in popularization and application in the engineering practice of vehicle structure optimization design are solved.

[0034] Specifically, Figure 1 A flowchart of a vehicle structure optimization method based on random vibration analysis provided by an embodiment of the present application is shown in the figure.

[0035] As Figure 1 shown, the vehicle structure optimization method based on random vibration analysis includes the following steps: In step S101, at least one key feature parameter is extracted based on the predicted model grid to reconstruct simulation grid data based on the reconstructed model.

[0036] In embodiments of the present application, the model grid can refer to a geometric model grid containing the vehicle structure and its key feature parameters, which can be used to describe information such as but not limited to the geometric dimensions, material distribution, and constraint conditions of each part including the body, chassis, connecting parts, etc. The predicted model grid can be an initial simulation model grid that needs to be predicted for vibration response in the future.

[0037] The key characteristic parameters can include, but are not limited to, geometric dimensions of each part of the vehicle body, plate thickness, material parameters, etc., and can reflect the key behavior of the vehicle structure under vibration, which can be set by a person skilled in the art according to the actual situation, and is not specifically limited here.

[0038] Further, the embodiment of the present application can reconstruct the simulation grid data according to the extracted key characteristic parameters, to ensure that the reconstructed simulation grid data is consistent with the initial simulation grid data in the key characteristic parameters, so that the dynamic response characteristics of the structure can be accurately reflected when the vehicle structure optimization or random vibration analysis is performed, and the simulation efficiency is improved and the reliability of the prediction result is ensured.

[0039] Optionally, in an embodiment of the present application, based on the predicted model grid, at least one key characteristic parameter is extracted to reconstruct the simulation grid data, comprising: discretizing the predicted model grid to obtain discretized grid data; extracting at least one key characteristic parameter according to the discretized grid data; and reconstructing the simulation grid data according to the at least one key characteristic parameter based on the reconstruction model.

[0040] It can be understood that the embodiment of the present application can discretize the model grid, divide the geometric model grid containing the vehicle structure and the key characteristic parameters into a limited number of nodes, and generate discretized grid data, so as to numerically solve the structure response in random vibration analysis or vehicle structure optimization. The discretization method can select solid, shell or beam elements according to the element type, and combine local encryption, gradual mesh or feature parameter driven methods to ensure that the key geometric dimensions, plate thickness, material properties and other parameters are accurately expressed, and are not specifically limited here.

[0041] As a specific example, as shown in Figure 2 , the embodiment of the present application can discretize the vehicle panel part grid to obtain the key characteristic parameters of the discretized grid data, which can include, but are not limited to, part width , part height , weight reduction hole width , weight reduction hole height , part thickness , etc.

[0042] Further, as shown in Figure 3 , the embodiment of the present application can control the key characteristic parameters by controlling the position of the model grid node, accurately adjust parameters including but not limited to part height, width, weight reduction hole width, height, etc., and then reconstruct the simulation grid data, to ensure that the reconstructed simulation grid data is consistent with the initial simulation grid data in the key characteristic parameters, so that the reconstructed grid data can accurately reflect the dynamic response characteristics of the original vehicle structure.

[0043] In step S102, at least one constraint condition is loaded on the reconstructed model to generate a random vibration response result.

[0044] It should be noted that the constraint condition can be a displacement, rotation or contact restriction applied to the model grid node, which can ensure that the simulation result can accurately reflect the stress and vibration characteristics of the real vehicle structure. The random vibration response result can be the dynamic response data of the structure obtained by calculation under the action of random excitation on the vehicle structure model grid, which can be described by indicators such as RMS (Root Mean Square), PSD (Power Spectral Density), peak value and probability distribution of displacement, velocity and acceleration, and used to evaluate the vibration performance of the structure, verify the design rationality or guide the optimization design of the structure.

[0045] The embodiment of the present application can load the constraint condition on the model to simulate the support, connection and boundary behavior of the structure under actual working conditions, then apply random excitation such as road vibration or engine vibration, and then calculate the dynamic response of the node to generate the random vibration response result, which is used to evaluate the performance of the structure and provide a basis for the optimization of the structure.

[0046] Optionally, in an embodiment of the present application, at least one constraint condition is loaded on the reconstructed model to generate a random vibration response result, comprising: loading at least one constraint condition on the reconstructed model according to the actual constraint state of the structure to obtain a modal calculation result; loading a unit acceleration load on the modal calculation result according to the actual loading state of the structure to calculate a random vibration response result.

[0047] In the embodiment of the present application, the modal is a combination of natural frequency and mode shape of the structure in free vibration, which can be used to analyze and predict the vibration response of the structure under random or deterministic excitation. The modal calculation method can include but is not limited to Lanczos method, direct solution method, subspace iteration method, etc., which is used for modal analysis of different types of structures. The unit acceleration load can be a standardized acceleration excitation with a size of 1 m / s 2 , which can be used to simulate the vibration environment caused by engine vibration, road unevenness or other mechanical equipment operation, as the basic input of random vibration analysis, and provide a standardized reference condition for further calculation of the dynamic response of the structure.

[0048] As a possible implementation manner, the embodiment of the application can load corresponding support, connection and boundary constraints and the like on the reconstructed model according to the actual constraint state of the structure, that is, the real engineering application condition, and perform modal calculation by using the Lanczos method to solve the natural frequency and mode shape of the structure, to obtain the modal calculation result of the structure, and provide basic data support for modal response calculation and random vibration analysis. Exemplarily, the solved frequency can be set to 0-2000Hz.

[0049] Further, the embodiment of the application can load a unit acceleration load unit acceleration sinusoidal sweep load on the key nodes of the structure on the basis of the modal calculation result, analyze the frequency response characteristics of the structure, and then load a PSD spectrum on the basis of the frequency response model to simulate the real dynamic load action of the vehicle structure under random working conditions such as road vibration and engine vibration, so as to obtain the random vibration response result of the vehicle structure. Optionally, the loading position and the constraint position of the unit acceleration load are consistent, the loading direction is consistent with the normal direction of the part, the loading frequency can be 1-2000Hz, the damping parameter can be loaded by using the structural damping G, the damping application range is 1-2000Hz, the structural damping can be 0.15, and the PSD spectrum can be as shown in Table 1. Table 1 is a PSD spectrum table.

[0050] Table 1

[0051] In step S103, a target training model is constructed by using at least one key feature parameter.

[0052] The embodiment of the application can use the key feature parameters extracted after the grid discretization of the vehicle panel structure as input and the random vibration response result as output by using deep learning tools such as GDL (Geometric Deep Learning), PyG (PyTorch Geometric, a graph neural network library based on PyTorch), to perform AI (Artificial Intelligence) model training, establish the mapping relationship between the two, obtain the AI training model, and realize efficient association modeling and prediction of the structure characteristics and the dynamics response.

[0053] Optionally, in an embodiment of the application, the target training model is constructed by using at least one key feature parameter, including: performing Latin hypercube sampling DOE test design based on the at least one key feature parameter to obtain an experimental sample; reconstructing the reconstructed model in the experimental sample into non-Euclidean data, and extracting a second root mean square stress of the random vibration response result; and constructing the target training model based on the non-Euclidean data and the second root mean square stress.

[0054] Those skilled in the art can understand that the DOE is a systematic and multi-factor experimental planning method, which can be used to study the influence of key parameters and their combinations on the performance of the structure; the Latin hypercube sampling is an efficient multi-dimensional sampling strategy, which generates representative experimental samples by uniformly covering the design space, reduces the number of experiments while ensuring the reliability of the results; the non-Euclidean data, which can be in the form of nodes and their topological relationship or graph structure, can preserve the spatial relationship and structural characteristics; the root mean square stress can refer to the square root state of the square mean value of the stress in the structure over time due to external load, temperature change or constraint condition under the action of random vibration or dynamic load, which can reflect the ability of the structure to resist deformation and fatigue.

[0055] As a specific example, the embodiments of the present application can perform DOE test design according to the key feature parameters extracted according to the above steps by Latin hypercube sampling, which can generate test samples covering multi-dimensional design space in a systematic and efficient manner. Wherein, the number of test samples can be 100, and the DOE test design scheme can be shown in Table 2. Wherein, Table 2 is a part of DOE test design scheme table.

[0056] Table 2

[0057] Further, the embodiments of the present application can reconstruct the reconstructed model in the test sample into non-Euclidean data, and extract the RMS stress of the random vibration response result in the above steps, and then build a one-to-one mapping relationship between the above non-Euclidean data and the RMS stress, and perform geometric deep learning (such as GDL), and then construct an AI training model. Optionally, the embodiments of the present application can set the learning width to 30, the learning depth to 3, the iteration step number to 500, and the iteration step length to 0.001.

[0058] The AI model constructed by the embodiments of the present application can realize the rapid prediction between the key feature parameters and the RMS stress, and can be used for subsequent optimization design, thereby improving the efficiency of random vibration analysis and structural optimization design.

[0059] In step S104, the key feature parameters of the structure geometric characteristics to be optimized are extracted to obtain the reconstructed geometric data.

[0060] For example, the embodiments of the present application can extract the key feature parameters of the structure geometric characteristics to be optimized according to methods including but not limited to direct measurement, CAD (Computer-Aided Design) model analysis, etc., which can include but are not limited to part width , part height , weight reduction hole width , weight reduction hole height , part thickness .

[0061] Further, the embodiment of the present application can reconstruct the geometric data according to the extracted structural geometric feature key feature parameters, as the input of the subsequent AI training model, and then obtain the corresponding random vibration response for subsequent structural performance evaluation and optimization design.

[0062] In step S105, the reconstructed geometric data is input into the target training model to output the first root mean square stress of the reconstructed geometric data.

[0063] As a possible implementation manner, the reconstructed geometric data obtained above can be input into the AI training model, and then the RMS stress of the reconstructed geometric data can be quickly predicted by using the AI model, which can significantly save the calculation time and resources.

[0064] In step S106, the structural lightweight optimization result of the vehicle is generated based on the first root mean square stress and the reconstructed geometric data.

[0065] Specifically, based on the reconstructed geometric data of the vehicle structure and the RMS stress predicted by the AI training model, the stress distribution and the key stress position of the structure can be evaluated to determine whether the structural strength meets the design requirements. In combination with the analysis result, the embodiment of the present application can further guide the optimization design to generate the vehicle structure lightweight scheme, so as to reduce the overall vehicle weight and improve the vehicle performance under the premise of ensuring the strength and rigidity.

[0066] Optionally, in an embodiment of the present application, the structural lightweight optimization result of the vehicle is generated based on the first root mean square stress and the reconstructed geometric data, including: comparing the first root mean square stress of the reconstructed geometric data with the constraint stress to obtain a comparison result, and combining the reconstructed geometric data and the comparison result to determine the structural lightweight optimization result.

[0067] It can be explained that the constraint stress can be a local stress generated by the structure under the action of fixed boundary, support or connection constraint, which usually represents the lower limit of design safety. If the RMS stress is close to or exceeds the constraint stress, it means that the region may become a weak point of fatigue or damage, which needs to be reinforced or redesigned. Therefore, the embodiment of the present application can compare the RMS stress with the constraint stress, and consider both weight reduction and safety in lightweight optimization, to ensure that the optimized structure is both light and reliable.

[0068] In an implementable embodiment, the embodiments of the present application can compare the obtained RMS stress with the constraint stress, and measure the volume according to the reconstructed geometric data obtained in the above steps, and if the RMS stress > 150 MPa or the reconstructed geometric volume does not converge to a minimum value, it can be indicated that the current structure still has further lightweight space under the premise of ensuring the strength requirement, or the structure exists in the process of reducing the volume. The strength is over limit, then modify the key feature parameters of the structure geometry to be optimized, and then use the AI training model again to obtain the RMS stress of the reconstructed geometric data, until the RMS stress ≤ 150 MPa and the reconstructed geometric volume converges to a minimum value. Wherein, the volume change and stress change of the optimization process of the embodiments of the present application can be as shown in Figure 5 , and the comparison of structure parameters before and after optimization can be as shown in Table 3.

[0069] Table 3

[0070] Further, as shown in Figure 6 , the embodiments of the present application can extract the final key feature parameters of the above steps, and then reconstruct the geometric data to obtain the final optimized structure according to the reconstructed geometric data.

[0071] The working principle of the vehicle structure optimization method based on random vibration analysis of the embodiments of the present application is described below with a specific example, which can include the following steps: (1) Simulation mesh feature parameter extraction and reconstruction: The embodiments of the present application first discretize the model mesh to be predicted, extract the key feature parameters of the discretized mesh data, which can include but not limited to size, thickness, material parameters, etc., and then reconstruct the simulation mesh data according to the extracted key feature parameters, for subsequent modal analysis, random vibration analysis and AI training, which can improve the calculation efficiency and ensure the accuracy of the analysis results.

[0072] (2) Modal calculation: Further, the embodiments of the present application can load the constraint conditions on the reconstructed simulation mesh data in the above steps according to the actual constraint state of the structure, perform static or preloading analysis on the structure to obtain the initial equilibrium state, and then perform modal calculation to obtain the modal calculation results such as the natural frequency and mode shape of the structure.

[0073] (3) Unit acceleration sinusoidal sweep: Then, the embodiment of the present application can load a unit acceleration sine sweep load on the modal calculation result of the above step according to the actual loading state of the structure, to simulate the response characteristics of the structure under different frequency excitations; wherein, the loading frequency is set according to the input PSD spectrum frequency range, the damping parameter is loaded using the structural damping G, and the damping application range is also set according to the input PSD spectrum frequency range.

[0074] (4) Random vibration: Then, the PSD spectrum is loaded on the unit acceleration sine sweep model of the above step to simulate the response of the structure under random vibration conditions, so as to calculate the random vibration response result thereof.

[0075] (5) DOE test design: As a possible implementation manner, the DOE test design can be performed by Latin hypercube sampling according to the key feature parameters extracted above, to systematically generate samples covering the parameter space, for subsequent simulation analysis or AI model training.

[0076] (6) AI model training: The embodiment of the present application can first reconstruct the reconstructed model in the test sample into non-Euclidean data, and extract the RMS stress of the random vibration response result, and then build a one-to-one mapping relationship between the non-Euclidean data and the RMS stress, perform geometric deep learning, and construct an AI training model.

[0077] (7) Structure geometric feature key feature parameter extraction and reconstruction.

[0078] Further, the embodiment of the present application extracts the structure geometric feature key feature parameters that need to be optimized, to reconstruct the geometric data according to the extracted key feature parameters.

[0079] (8) Predicted stress The reconstructed geometric data obtained is taken as input, and the AI training model is used to predict the RMS stress of the reconstructed geometric data.

[0080] (9) Optimized structure data: The RMS stress of the structure is compared with the constraint stress, and the volume thereof is measured based on the reconstructed geometric data, to judge the safety margin and lightweight potential of the current structure, and then the structure is optimized and adjusted according to the analysis result, to obtain the final structure geometric feature key feature parameters meeting the strength requirement and having the smallest volume.

[0081] (10) Model output Finally, the key feature parameters of the structure geometry characteristics are used to reconstruct the geometric data, that is, the final optimized structure is obtained, which is applied to the lightweight design and dynamic performance optimization of the vehicle panel, so as to realize the lightweight of the structure under the premise of meeting the strength and stiffness requirements.

[0082] According to the vehicle structure optimization method based on random vibration analysis provided in the embodiments of the present application, the AI training model is constructed by using the key feature parameters of the vehicle structure, and the corresponding root mean square stress is predicted through the model, and then the reconstructed geometric data is analyzed and optimized based on the prediction result, so as to realize the lightweight design of the vehicle structure. The AI training model can be used to quickly predict the random vibration response of the model grid node, realize efficient evaluation and optimization of different design schemes, and significantly reduce the design cycle and computing resource consumption without high-cost simulation analysis each time. Meanwhile, through rapid iterative optimization, a balance between structure performance and lightweight can be achieved, which provides a reference for structure reliability and fatigue life analysis, and forms an efficient intelligent structure optimization closed loop.

[0083] Secondly, the vehicle structure optimization device based on random vibration analysis according to the embodiments of the present application is described with reference to the accompanying drawings.

[0084] Figure 8 is a block schematic diagram of the vehicle structure optimization device based on random vibration analysis according to the embodiments of the present application.

[0085] As Figure 8 shown, the vehicle structure optimization device based on random vibration analysis 10 includes a reconstruction module 100, a generation module 200, a construction module 300, an extraction module 400, an output module 500 and an optimization module 600.

[0086] The reconstruction module 100 is configured to extract at least one key feature parameter based on the predicted model grid, so as to reconstruct the simulation grid data based on the reconstruction model.

[0087] The generation module 200 is configured to load at least one constraint condition on the reconstruction model, so as to generate a random vibration response result.

[0088] The construction module 300 is configured to construct a target training model by using at least one key feature parameter.

[0089] The extraction module 400 is configured to extract the key feature parameters of the structure geometry characteristics that need to be optimized, so as to obtain the reconstructed geometric data.

[0090] The output module 500 is configured to input the reconstructed geometric data into the target training model, so as to output the first root mean square stress of the reconstructed geometric data.

[0091] The optimization module 600 is configured to generate a structure lightweight optimization result of the vehicle based on the first root mean square stress and the reconstructed geometry data.

[0092] Optionally, in an embodiment of the present application, the reconstruction module 100 comprises a discretization unit, an extraction unit and a reconstruction unit.

[0093] The discretization unit is configured to discretize the predicted model mesh to obtain discretized mesh data.

[0094] The extraction unit is configured to extract at least one key feature parameter from the discretized mesh data.

[0095] The reconstruction unit is configured to reconstruct simulation mesh data from the at least one key feature parameter based on the reconstruction model.

[0096] Optionally, in an embodiment of the present application, the generation module 200 comprises a loading unit and a calculation unit.

[0097] The loading unit is configured to load at least one constraint condition on the reconstruction model according to the actual constraint state of the structure to obtain a modal calculation result.

[0098] The calculation unit is configured to load a unit acceleration load on the modal calculation result according to the actual loading state of the structure to calculate a random vibration response result. Optionally, in an embodiment of the present application, the construction module 300 comprises a design unit, an extraction unit and a construction unit.

[0099] The design unit is configured to perform a Latin hypercube sampling DOE test design based on the at least one key feature parameter to obtain an experimental sample.

[0100] The extraction unit is configured to reconstruct the reconstruction model in the experimental sample into non-Euclidean data and extract a second root mean square stress of the random vibration response result.

[0101] The construction unit is configured to construct a target training model based on the non-Euclidean data and the second root mean square stress.

[0102] Optionally, in an embodiment of the present application, the optimization module 600 comprises a comparison unit and an optimization unit.

[0103] The comparison unit is configured to compare the first root mean square stress of the reconstructed geometry data and a constraint stress to obtain a comparison result.

[0104] The optimization unit is configured to determine a structure lightweight optimization result in combination with the reconstructed geometry data and the comparison result.

[0105] It should be noted that the foregoing explanation of the embodiment of the vehicle structure optimization method based on random vibration analysis is also applicable to the embodiment of the vehicle structure optimization device based on random vibration analysis, and will not be repeated here.

[0106] The vehicle structure optimization device based on random vibration analysis provided by the embodiment of the application uses the key feature parameters of the vehicle structure to construct an AI training model, and predicts the corresponding root mean square stress through the model, and then analyzes and optimizes the reconstructed geometric data based on the prediction result, so that the lightweight design of the vehicle structure is realized. The AI training model can be used to quickly predict the random vibration response of the model grid node, efficiently evaluate and optimize different design schemes, and significantly reduce the design cycle and the consumption of computing resources without high-cost simulation analysis each time. Meanwhile, through rapid iterative optimization, a balance between structure performance and lightweight can be achieved, and reference for structure reliability and fatigue life analysis is provided, forming an efficient intelligent structure optimization closed loop.

[0107] Figure 9 A vehicle structure diagram is provided for the embodiment of the application. The vehicle can include: The memory 901, the processor 902, and the computer program stored in the memory 901 and executable on the processor 902.

[0108] The processor 902 implements the vehicle structure optimization method based on random vibration analysis provided in the above embodiments when executing the program.

[0109] Further, the vehicle further includes: The communication interface 903 is used for communication between the memory 901 and the processor 902.

[0110] The memory 901 is used to store the computer program executable on the processor 902.

[0111] The memory 901 can include a high-speed RAM memory, and can also include a non-volatile memory, such as at least one disk memory.

[0112] If the memory 901, the processor 902 and the communication interface 903 are implemented independently, the communication interface 903, the memory 901 and the processor 902 can be connected with each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, Figure 9 Only one thick line is used in the figure to represent the bus, but it does not mean that there is only one bus or only one type of bus.

[0113] Optionally, in a specific implementation, if the memory 901, the processor 902 and the communication interface 903 are integrated on a chip, the memory 901, the processor 902 and the communication interface 903 can complete communication between each other through an internal interface.

[0114] The processor 902 can be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the present application.

[0115] The embodiment further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the vehicle structure optimization method based on random vibration analysis.

[0116] The embodiment of the present application further provides a computer program product, which comprises a computer program, and the computer program can run computer instructions, and the computer instructions are executed by a processor to implement the vehicle structure optimization method based on random vibration analysis provided by the embodiment of the present application.

[0117] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that a particular feature, structure, material, or characteristic being described is included in at least one embodiment or example of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment or example. Furthermore, the described specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. In addition, the usage of "N" means at least two, for example, two, three or the like, unless explicitly stated otherwise.

[0118] Furthermore, the terms "first", "second", or the like, are used merely as a designation of certain elements or features of the application, and do not imply or connote relative importance or a specific order of precedence. Thus, features defined with "first", "second", etc. can include at least one of the features, either explicitly or implicitly.

[0119] Any process or method descriptions or blocks in flow charts or otherwise described herein represent embodiments of modules, segments, or portions of code which include one or more executable instructions for implementing specific logic functions or steps, and alternate implementations are possible. The preferred embodiments of this application are preferably practiced in conjunction with a computer system capable of carrying out the functions described herein, although the application is not limited to being implemented by any particular computer system. The machine-executable instructions can be stored on one or more machine-readable media, which can include any available storage media or memory element. Any of the machine-readable media can be a computer- readable storage medium or memory element. Some examples of such computer- readable storage media or memory elements include primary storage, secondary storage, removable storage, and non-removable storage.

[0120] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of executable instructions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a product of the manufacturing and / or processing. The computer-readable medium can include, but is not limited to, the following: an electronic connection (an electronic device with one or N wires), a portable computer diskette (a magnetic device), a RAM (random access memory), a ROM (read-only memory), an EPROM (erasable programmable ROM) or a Flash memory, an optical fiber, and a portable CD ROM. In addition, the computer-readable medium can even be paper or other suitable medium upon which the program can be printed, as the program can be electronically captured, via the optically scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and stored in the computer memory.

[0121] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware and in another embodiment, the hardware can be implemented using any or a combination of the following technologies, which are all well known in the art: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0122] Those of skill in the art would understand that the steps carried out in the above-mentioned embodiment methods can be implemented by programs instructing relevant hardware to complete all or part of the steps, and the programs can be stored in a computer-readable storage medium. When the programs are executed, they include one or a combination of the steps of the method embodiments.

[0123] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software functional module. When the integrated module is realized in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0124] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for optimizing a vehicle structure based on random vibration analysis, characterized by, The method comprises the following steps: extracting at least one key feature parameter based on a predicted model mesh to reconstruct simulation mesh data based on a reconstruction model; loading at least one constraint condition on the reconstruction model to generate a random vibration response result; constructing a target training model by using the at least one key feature parameter; extracting a key feature parameter of a structure geometric feature to be optimized to obtain reconstruction geometric data; inputting the reconstruction geometric data into the target training model to output a first root mean square stress of the reconstruction geometric data; generating a structure lightweight optimization result of a vehicle based on the first root mean square stress and the reconstruction geometric data.

2. The method of claim 1, wherein, The step of extracting at least one key feature parameter based on a predicted model mesh to reconstruct simulation mesh data comprises: discretizing the predicted model mesh to obtain discretized mesh data; extracting the at least one key feature parameter according to the discretized mesh data; reconstructing the simulation mesh data according to the at least one key feature parameter based on the reconstruction model.

3. The method of claim 1, wherein, The step of loading at least one constraint condition on the reconstruction model to generate a random vibration response result comprises: loading the at least one constraint condition on the reconstruction model according to an actual constraint state of a structure to obtain a modal calculation result; loading a unit acceleration load on the modal calculation result according to an actual load state of the structure to calculate the random vibration response result.

4. The method of claim 1, wherein, The step of constructing a target training model by using the at least one key feature parameter comprises: performing Latin hypercube sampling DOE test design based on the at least one key feature parameter to obtain an experimental sample; reconstructing a reconstruction model in the experimental sample into non-Euclidean data and extracting a second root mean square stress of the random vibration response result; constructing the target training model based on the non-Euclidean data and the second root mean square stress.

5. The method of claim 1, wherein, The step of generating a structure lightweight optimization result of a vehicle based on the first root mean square stress and the reconstruction geometric data comprises: comparing the first root mean square stress of the reconstruction geometric data with a constraint stress to obtain a comparison result; determining the structure lightweight optimization result in combination with the reconstruction geometric data and the comparison result.

6. A vehicle structure optimization device based on random vibration analysis, characterized in that, The method comprises: a reconstruction module configured to extract at least one key feature parameter based on a predicted model mesh to reconstruct simulation mesh data based on a reconstruction model; a generation module configured to load at least one constraint condition on the reconstruction model to generate a random vibration response result; a construction module configured to construct a target training model by using the at least one key feature parameter; an extraction module configured to extract a key feature parameter of a structure geometric feature to be optimized to obtain reconstruction geometric data; an output module configured to input the reconstruction geometric data into the target training model to output a first root mean square stress of the reconstruction geometric data; an optimization module configured to generate a structure lightweight optimization result of a vehicle based on the first root mean square stress and the reconstruction geometric data.

7. The apparatus of claim 6, wherein, The reconstruction module comprises: a discretization unit configured to discretize the predicted model mesh to obtain discretized mesh data; an extraction unit configured to extract the at least one key feature parameter from the discretized mesh data; a reconstruction unit configured to reconstruct the simulation mesh data based on the reconstruction model and the at least one key feature parameter.

8. A vehicle characterized by comprising: comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor executing the program to implement the method for vehicle structure optimization based on random vibration analysis according to any one of claims 1-5.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the method for vehicle structure optimization based on random vibration analysis according to any one of claims 1-5.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed to implement the method for vehicle structure optimization based on random vibration analysis according to any one of claims 1-5.

Images