Vehicle crash energy absorption box generation method and device, electronic equipment and storage medium

Abstract

Embodiments of the present application provide a method and device for generating a vehicle crash energy absorption box, electronic equipment and a storage medium. The method comprises: obtaining an initial structure model of the vehicle crash energy absorption box, wherein the initial structure model adopts a variable cross-section square tube structure, and the initial structure model contains to-be-optimized geometric parameters corresponding to at least one induced slot arranged on the variable cross-section square tube structure; performing parameter optimization on the to-be-optimized geometric parameters in the initial structure model to obtain a target structure model; and generating the vehicle crash energy absorption box based on the target structure model. The present application solves the technical problem of low energy absorption effect of the vehicle crash energy absorption box generated in the related art.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and more specifically, to a method, apparatus, electronic device, and storage medium for generating a vehicle collision energy-absorbing box. Background Technology

[0002] As a crucial passive safety component in frontal collisions, the performance of a vehicle's collision energy-absorbing box determines its collision energy absorption efficiency and the integrity of the passenger compartment. In related technologies, collision energy-absorbing boxes utilize a square tube with a uniform cross-section, employing a single internal groove to weaken local stiffness and guide crushing. However, this design suffers from the following drawbacks: the simplistic guiding structure makes it prone to asymmetrical folding such as twisting and lateral bending under collision angles or load deviations, resulting in crushing disorder and significantly reducing the effective energy absorption stroke and efficiency, ultimately leading to a lower energy absorption effect in the resulting vehicle collision energy-absorbing box.

[0003] There is currently no good solution to the above problems. Summary of the Invention

[0004] This application provides a method, apparatus, electronic device, and storage medium for generating a vehicle collision energy-absorbing box, so as to at least solve the technical problem that the energy absorption effect of the generated vehicle collision energy-absorbing box in the related art is low.

[0005] According to one aspect of the embodiments of this application, a method for generating a vehicle collision energy-absorbing box is provided, comprising: obtaining an initial structural model of the vehicle collision energy-absorbing box, wherein the initial structural model adopts a variable cross-section square tube structure, and the initial structural model includes the geometric parameters to be optimized corresponding to at least one induction groove provided on the variable cross-section square tube structure; performing parameter optimization on the geometric parameters to be optimized in the initial structural model to obtain a target structural model; and generating a vehicle collision energy-absorbing box based on the target structural model.

[0006] In this embodiment of the application, obtaining the initial structural model of the vehicle collision energy absorption box includes: constructing the axial position parameters, groove width parameters, and groove depth parameters to be optimized, wherein the groove depth parameter is a signed value, a positive value indicates that the corresponding guide groove is a concave structure, and a negative value indicates that the corresponding guide groove is a convex structure; determining the geometric parameters to be optimized based on the axial position parameters, groove width parameters, and groove depth parameters; and constructing the initial structural model based on the geometric parameters to be optimized and the variable cross-section square tube structure.

[0007] In this embodiment of the application, the geometric parameters to be optimized in the initial structural model are optimized to obtain the target structural model, including: obtaining the manufacturing constraint set corresponding to the geometric parameters to be optimized; and optimizing the geometric parameters to be optimized based on the target intelligent agent model and the manufacturing constraint set to obtain the target structural model, wherein the target intelligent agent model takes maximizing the collision energy absorption of the initial structural model as the optimization objective.

[0008] In this embodiment, obtaining the manufacturing constraint set corresponding to the geometric parameters to be optimized includes: obtaining the axial position constraint corresponding to the axial position parameter, the groove width constraint corresponding to the groove width parameter, and the groove depth constraint corresponding to the groove depth parameter; when there is only one induction groove on the variable cross-section square tube structure, the axial position constraint, groove width constraint, and groove depth constraint are integrated to obtain the manufacturing constraint set; when there are multiple induction grooves on the variable cross-section square tube structure, the setting position constraint and setting spacing constraint between the multiple induction grooves are obtained, and the setting position constraint, setting spacing constraint, axial position constraint, groove width constraint, and groove depth constraint are integrated to obtain the manufacturing constraint set.

[0009] In this embodiment, based on a target intelligent agent model and a set of manufacturing constraints, parameter optimization is performed on the geometric parameters to be optimized to obtain a target structural model. This includes: constructing a target penalty function based on the set of manufacturing constraints, wherein the target penalty function is used to apply penalty terms to candidate solutions that do not satisfy the set of manufacturing constraints, and the degree of penalty corresponding to the penalty term is related to the degree to which the candidate solution exceeds the corresponding manufacturing constraint; guiding the target intelligent agent model based on the target penalty function to optimize the geometric parameters to be optimized to obtain the target structural model; preferably, constructing the target penalty function based on the set of manufacturing constraints includes: constructing penalty terms corresponding to different manufacturing constraints based on different manufacturing constraints in the set of manufacturing constraints; integrating the penalty terms corresponding to different manufacturing constraints to obtain the target penalty function.

[0010] In this embodiment of the application, a target intelligent agent model is guided by a target penalty function to optimize the geometric parameters to be optimized and obtain a target structural model. The process includes: inputting the geometric parameters to be optimized into the target intelligent agent model; guiding the target intelligent agent model based on the target penalty function to output collision performance indices corresponding to different geometric parameters; filtering different geometric parameters based on the collision performance indices corresponding to different geometric parameters to obtain optimized geometric parameters; and constructing a target structural model based on the optimized geometric parameters.

[0011] In this embodiment of the application, constructing a target structural model based on optimized geometric parameters includes: performing simulation verification on the optimized geometric parameters to obtain simulation verification results; constructing a target structural model based on the optimized geometric parameters and simulation verification results; preferably, performing simulation verification on the optimized geometric parameters to obtain simulation verification results includes: constructing a finite element analysis model for simulating frontal collision conditions; and performing simulation verification on the optimized geometric parameters based on the finite element analysis model to obtain simulation verification results.

[0012] In this embodiment of the application, the optimized geometric parameters are simulated and verified based on the finite element analysis model to obtain the simulation verification results, including: simulating and verifying the optimized geometric parameters based on the finite element analysis model to obtain the first collision absorption energy corresponding to the optimized geometric parameters; and determining the simulation verification results based on the first collision absorption energy and the second collision absorption energy, wherein the second collision absorption energy is used to represent the collision absorption energy corresponding to the optimized geometric parameters predicted based on the target intelligent agent model.

[0013] In this embodiment of the application, the simulation verification result is determined based on the first collision energy absorption and the second collision energy absorption, including: if the difference between the first collision energy absorption and the second collision energy absorption is less than a preset error threshold and the first collision energy absorption is greater than a preset collision energy absorption threshold, the simulation verification result is determined to be a successful simulation verification; if the difference between the first collision energy absorption and the second collision energy absorption is greater than or equal to the preset error threshold, or the first collision energy absorption is less than or equal to the preset collision energy absorption threshold, the simulation verification result is determined to be a failed simulation verification.

[0014] In this embodiment, a target structural model is constructed based on the optimized geometric parameters and simulation verification results. This includes: if the simulation verification results indicate that the optimized geometric parameters have failed simulation verification, repeatedly reducing the collision absorption energy corresponding to the failed geometric parameters, using the reduced collision absorption energy as a new optimization target, and optimizing the parameters using the new optimization target through a target intelligent agent model, until the simulation verification results indicate that the optimized geometric parameters have passed simulation verification; and if the simulation verification results indicate that the optimized geometric parameters have passed simulation verification, constructing a target structural model based on the passed geometric parameters.

[0015] According to another aspect of the embodiments of this application, a device for generating a vehicle collision energy-absorbing box is also provided, comprising: an acquisition module for acquiring an initial structural model of the vehicle collision energy-absorbing box, wherein the initial structural model adopts a variable cross-section square tube structure, and the initial structural model includes the geometric parameters to be optimized corresponding to at least one induction groove provided on the variable cross-section square tube structure; an optimization module for optimizing the geometric parameters to be optimized in the initial structural model to obtain a target structural model; and a generation module for generating a vehicle collision energy-absorbing box based on the target structural model.

[0016] The acquisition module is used to construct the axial position parameters, groove width parameters, and groove depth parameters to be optimized. The groove depth parameter is a signed value, with a positive value indicating that the corresponding induced groove is a concave structure and a negative value indicating that the corresponding induced groove is a convex structure. Based on the axial position parameters, groove width parameters, and groove depth parameters, the geometric parameters to be optimized are determined. Based on the geometric parameters to be optimized and the variable cross-section square tube structure, an initial structural model is constructed.

[0017] The optimization module is used to obtain the set of manufacturing constraints corresponding to the geometric parameters to be optimized; based on the target intelligent agent model and the set of manufacturing constraints, the geometric parameters to be optimized are optimized to obtain the target structural model, wherein the target intelligent agent model takes maximizing the collision energy absorption of the initial structural model as the optimization objective.

[0018] The optimization module is used to obtain the axial position constraints corresponding to the axial position parameter, the groove width constraints corresponding to the groove width parameter, and the groove depth constraints corresponding to the groove depth parameter. When there is only one induction groove on the variable cross-section square tube structure, the axial position constraints, groove width constraints, and groove depth constraints are integrated to obtain a set of manufacturing constraints. When there are multiple induction grooves on the variable cross-section square tube structure, the setting position constraints and setting spacing constraints between the multiple induction grooves are obtained, and the setting position constraints, setting spacing constraints, axial position constraints, groove width constraints, and groove depth constraints are integrated to obtain a set of manufacturing constraints.

[0019] The optimization module is used to construct a target penalty function based on a set of manufacturing constraints. This target penalty function applies penalty terms to candidate solutions that do not meet the manufacturing constraints, and the degree of penalty corresponding to each term is related to the extent to which the candidate solution exceeds the corresponding manufacturing constraints. The module guides a target intelligent agent model based on the target penalty function to optimize the geometric parameters to obtain the target structural model. Preferably, constructing the target penalty function based on the set of manufacturing constraints includes: constructing penalty terms corresponding to different manufacturing constraints based on the different manufacturing constraints in the set; and integrating the penalty terms corresponding to the different manufacturing constraints to obtain the target penalty function.

[0020] The optimization module is used to input the geometric parameters to be optimized into the target intelligent agent model, guide the target intelligent agent model based on the target penalty function, and output the collision performance index corresponding to different geometric parameters. Based on the collision performance index corresponding to different geometric parameters, different geometric parameters are filtered to obtain optimized geometric parameters, and the target structure model is constructed based on the optimized geometric parameters.

[0021] The optimization module is used to perform simulation verification on the optimized geometric parameters and obtain simulation verification results; based on the optimized geometric parameters and simulation verification results, a target structural model is constructed; preferably, the simulation verification on the optimized geometric parameters and obtain simulation verification results includes: constructing a finite element analysis model for simulating frontal collision conditions; and performing simulation verification on the optimized geometric parameters based on the finite element analysis model and obtaining simulation verification results.

[0022] The optimization module is used to perform simulation verification on the optimized geometric parameters based on the finite element analysis model to obtain the first collision absorption energy corresponding to the optimized geometric parameters; based on the first collision absorption energy and the second collision absorption energy, the simulation verification result is determined, wherein the second collision absorption energy is used to represent the collision absorption energy corresponding to the optimized geometric parameters predicted based on the target intelligent agent model.

[0023] The optimization module is used to determine the simulation verification result as passed when the difference between the energy absorbed by the first collision and the energy absorbed by the second collision is less than a preset error threshold and the energy absorbed by the first collision is greater than a preset collision energy absorption threshold; and to determine the simulation verification result as failed when the difference between the energy absorbed by the first collision and the energy absorbed by the second collision is greater than or equal to the preset error threshold, or when the energy absorbed by the first collision is less than or equal to the preset collision energy absorption threshold.

[0024] The optimization module is used to iteratively reduce the collision absorption energy corresponding to the failed simulation verification geometric parameters when the simulation verification results indicate that the optimized geometric parameters have failed the simulation verification. The reduced collision absorption energy is then used as a new optimization target, and the target intelligent agent model is used to optimize the parameters using the new optimization target. This process continues until the simulation verification results indicate that the optimized geometric parameters have passed the simulation verification. If the simulation verification results indicate that the optimized geometric parameters have passed the simulation verification, a target structural model is constructed based on the geometric parameters that have passed the simulation verification.

[0025] According to another aspect of the embodiments of this application, an electronic device is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0026] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0027] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0028] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the methods in various embodiments of this application.

[0029] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0030] In this embodiment, firstly, an initial structural model of the vehicle collision energy-absorbing box is obtained. The initial structural model adopts a variable cross-section square tube structure and includes the geometric parameters to be optimized corresponding to at least one induction groove set on the variable cross-section square tube structure. Next, the geometric parameters to be optimized in the initial structural model are optimized to obtain a target structural model. Finally, the vehicle collision energy-absorbing box is generated based on the target structural model. This application obtains an initial structural model using a variable cross-section square tube structure and designs the geometric parameters to be optimized for the induction groove. The variable cross-section square tube structure design itself possesses more stable progressive crushing characteristics, improving stress distribution from a geometrical perspective. Intelligent optimization of the geometric parameters to be optimized for the induction groove is performed, accurately seeking the optimal solution, thereby overcoming the limitations of empirical design or single structure in crushing mode control. This closed-loop logic of basic structural optimization and intelligent parameter optimization enables the collision energy-absorbing box to guide orderly and stable plastic folding of the structure during a frontal collision, avoiding ineffective deformations such as twisting or tearing, improving the continuity and efficiency of energy absorption, and thus solving the technical problem of low energy absorption effect of vehicle collision energy-absorbing boxes generated in related technologies. Attached Figure Description

[0031] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0032] Figure 1 This is a flowchart of a method for generating a vehicle collision energy-absorbing box according to an embodiment of the present invention;

[0033] Figure 2 This is a schematic diagram of the generation process of a vehicle collision energy-absorbing box according to an embodiment of the present invention;

[0034] Figure 3 This is a schematic diagram of the generation process of an optional vehicle collision energy-absorbing box according to an embodiment of the present invention;

[0035] Figure 4 This is a schematic diagram of the generation process of another optional vehicle collision energy-absorbing box according to an embodiment of the present invention;

[0036] Figure 5 This is a schematic diagram of the generation process of another vehicle collision energy-absorbing box according to an embodiment of the present invention;

[0037] Figure 6 This is a schematic diagram of a device for generating a vehicle collision energy-absorbing box according to an embodiment of the present invention. Detailed Implementation

[0038] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0039] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0040] According to an embodiment of this application, an embodiment of a method for generating a vehicle collision energy-absorbing box is provided. The steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowcharts, in some cases the steps shown or described may be performed in a different order than that shown here.

[0041] This embodiment provides a method for generating a vehicle collision energy-absorbing box. Figure 1 This is a flowchart of a method for generating a vehicle collision energy-absorbing box according to an embodiment of this application, such as... Figure 1 As shown, the process includes the following steps:

[0042] Step S102: Obtain the initial structural model of the vehicle collision energy absorption box.

[0043] The initial structural model adopts a variable cross-section square tube structure, and the initial structural model includes the geometric parameters to be optimized corresponding to at least one induction groove set on the variable cross-section square tube structure.

[0044] The aforementioned initial structural model can refer to a three-dimensional geometric model of a trapezoidal variable cross-section square tubular energy-absorbing box established before parameter optimization. The initial structural model serves as the foundation for subsequent simulation analysis and optimization, containing the geometric parameters of the induction groove to be optimized. In this application, the initial structural model can be set as a three-dimensional geometric model including axial length, front end width, rear end width, and wall thickness, with at least one induction groove pre-installed on the sidewall, whose geometric parameters can be independently controlled. The initial geometric state of these induction grooves constitutes the starting point for optimization.

[0045] The aforementioned variable cross-section square tube structure refers to the basic cross-sectional shape of the energy-absorbing box body, specifically a trapezoidal variable cross-section square tube. The variable cross-section square tube structure has a varying cross-sectional width along the axial direction, i.e., the collision direction; it is narrower at the front and wider at the rear, or vice versa, depending on the specific installation direction. This design aims to achieve smoother, more gradual crushing during a frontal collision, avoiding stress concentration and deformation instability caused by abrupt changes in the cross-section, thus helping the energy-absorbing box achieve better energy absorption.

[0046] At least one of the aforementioned induction grooves can refer to local geometric features set on the sidewalls of the variable cross-section square tube structure to guide the structural crushing mode. For refined optimization, an induction groove can be arranged on each of the four sidewalls of the initial structural model. These induction grooves serve as carriers for the geometric parameters to be optimized. Their shapes can be concave to weaken stiffness and induce orderly folding, or convex to enhance stiffness and stabilize deformation. By adjusting the position, width, and depth of each induction groove, the local mechanical properties can be changed, thereby controlling the energy absorption path and crushing mode during collision.

[0047] The aforementioned geometric parameters to be optimized can refer to a set of design variables that determine the specific shape and position of the induction groove, serving as the input objects for the intelligent optimization process in this application.

[0048] As an optional implementation, a three-dimensional basic geometric model of the trapezoidal variable cross-section square tube can be established, allowing for the setting of front-end width, rear-end width, wall thickness, and axial length. At least one induction groove is designed on the sidewall of the energy-absorbing box, initializing the set of geometric parameters to be optimized. For a single induction groove, this set includes the groove's axial position, groove width, and groove depth. The axial position represents the distance between the center of the induction groove and the front end of the energy-absorbing box; the groove width represents the circumferential extension of the induction groove along the tube wall; and the groove depth can be defined using signed numerical values, with positive values ​​indicating a concave structure for inducing crushing and negative values ​​indicating a convex structure for enhancing stability. Subsequently, parametric modeling software can be used to map these initial geometric parameters to a specific region of the three-dimensional basic geometric model, generating an initial structural model containing the variable induction groove's characteristic parameters. This initial structural model retains the macroscopic morphology of the trapezoidal variable cross-section and provides an interface for adjusting the induction groove's geometric parameters, offering a standardized input template for subsequent rapid iterative optimization in single-variable or finite-variable spaces. This ensures that the initial structural model maintains the integrity of the topology and the effectiveness of the mesh generation even when parameters change.

[0049] As an alternative implementation, a trapezoidal variable cross-section square tube can be used as the base, with a layout space for multiple induced grooves pre-defined on the sidewall. The set of geometric parameters to be optimized is initialized. For the case of multiple induced grooves, the set of geometric parameters to be optimized is expanded to include the independent axial position, groove width, and groove depth of each induced groove. The geometric parameters of each induced groove are independent of each other, but together constitute the overall optimization variable space. When constructing the initial structural model, according to the preset initial parameter values, such as uniform distribution position, standard width, and initial depth, corresponding grooves or protrusions are excavated or constructed at the corresponding positions on the sidewall of the square tube. This process can initialize the relative positional relationships between multiple grooves, ensuring that the induced grooves do not overlap and meet the basic spacing requirements in the initial state. The generated initial structural model includes variable cross-section features and maps the geometric state of the multi-groove array, providing a foundation for subsequent collaborative optimization under multi-dimensional complex constraints such as spacing constraints and non-overlap constraints. This allows the initial structural model to support switching and expansion from single-groove to multi-groove schemes.

[0050] Through the above settings, the geometric parameters to be optimized in the induction trough using a variable cross-section square tube structure are optimized. The trapezoidal variable cross-section design enables smoother progressive crushing in frontal collisions, reducing stress concentration and improving overall stability. Abstracting the parameters to be optimized independently of the fixed geometric structure gives the design a high degree of versatility and adaptability. Whether in a simple single-trough or a complex multi-trough scenario, the design can be quickly reconstructed by adjusting the same set of parameters, improving the flexibility and coverage of the design method. This provides a standardized and regulated data foundation for subsequent intelligent optimization, thereby improving frontal collision energy absorption efficiency and design iteration speed.

[0051] Step S104: Optimize the geometric parameters to be optimized in the initial structural model to obtain the target structural model.

[0052] The aforementioned target structural model refers to the three-dimensional geometric model corresponding to the optimized result obtained after intelligent optimization, which satisfies the optimization objectives and conforms to engineering manufacturing constraints, such as non-overlap, spacing requirements, and processing size limitations. It is the optimized form of the initial structural model under a specific combination of geometric parameters, serving as a superior and manufacturable energy-absorbing box design scheme under the current design process. The induction groove parameters in the target structural model have been adjusted to optimal values.

[0053] As an optional implementation, the geometric parameters to be optimized in the initial structural model can be used as input variables and fed into a pre-trained intelligent agent model, such as a physical information neural network. The intelligent agent model uses maximizing the total energy absorbed in a frontal collision as its initial optimization objective, quickly predicting the collision performance indicators corresponding to different parameter combinations. To ensure that the optimization results conform to engineering reality, a pre-constructed target penalty function is introduced into the optimization algorithm during the iteration process. When a candidate parameter combination violates the manufacturing constraint set, such as insufficient spacing or depth exceeding the limit, the target penalty function calculates a penalty term based on the degree of violation and adds the penalty term to the optimization objective function, thereby reducing the fitness of the candidate solution. The target penalty function guides the search direction, gradually avoiding infeasible regions and converging towards a feasible region that satisfies the manufacturing constraints and has a large energy absorption. After multiple iterations until the algorithm converges or reaches the upper limit of the number of iterations, the parameter combination that makes the comprehensive objective function better is output, resulting in the parameter set of the target structural model, and thus the target structural model.

[0054] As an alternative implementation, a dynamic adjustment strategy can be adopted based on the initial optimization objective to address convergence issues under complex constraints. In the initial stage, the algorithm performs a global search to attempt to find theoretically larger energy absorption points. If the initial verification reveals that candidate solutions are difficult to simultaneously satisfy high-precision prediction and strict manufacturing constraints, or if they get trapped in local optima, a constraint relaxation mechanism can be activated. At this point, the algorithm can dynamically adjust the penalty coefficient by monitoring the severity of constraint violations, or temporarily relax the thresholds of some less important constraints, guiding the algorithm to explore a wider solution space. By dynamically balancing performance maximization and constraint satisfaction, the algorithm gradually narrows the search range, seeking an optimal balance between performance and feasibility. When the algorithm finds a parameter combination that results in high energy absorption and the penalty term for manufacturing constraints approaches zero (i.e., the constraints are satisfied), the iteration stops, and the set of parameters is output as the target structural model, ensuring that the obtained solution possesses both high performance and manufacturability.

[0055] Through the above settings, this intelligent parameter optimization method can accurately explore the optimal design of the induction groove in the full parameter space, realize the synergistic optimization of induced orderly crushing and ensuring deformation stability, and improve the energy absorption efficiency and mode stability of the energy absorption box in frontal collision.

[0056] Step S106: Generate a vehicle collision energy absorption box based on the target structure model.

[0057] The aforementioned vehicle collision energy-absorbing box refers to a generated passive safety component installed at the front of a vehicle to absorb frontal collision energy. It is a physical structure generated based on a target structural model, specifically a metal component with a trapezoidal variable cross-section square tube shape and intelligently optimized adjustable guide grooves on its side walls. It can be manufactured through processes such as stamping. The vehicle collision energy-absorbing box aims to improve energy absorption efficiency in frontal collisions, ensure orderly and stable crushing patterns, and protect the integrity of the passenger compartment.

[0058] As an optional implementation, the system can receive a set of optimal geometric parameters corresponding to the target structural model obtained through intelligent optimization, including the axial positions, widths, and depths of the four guide grooves. Next, computer-aided design modeling software or a parametric modeling interface can be invoked, using a trapezoidal variable cross-section square tube as the basic framework. Based on these accurate optimal parameters, concave or convex guide groove features are precisely constructed at the corresponding sidewall positions. For concave grooves with positive depths, material removal is performed to form grooves with reduced stiffness; for convex grooves with negative depths, material addition or local thickening is performed to form convex hulls that enhance stability. Subsequently, the three-dimensional solid data of the model is automatically updated to ensure topological continuity and no interference. During the generation of the three-dimensional model, important dimensional information is automatically extracted to generate engineering drawings containing accurate positioning, width, and depth annotations of the guide grooves. These engineering drawings reflect the intelligently optimized structural morphology and can be directly used as the technical basis for CNC machining or mold manufacturing, realizing the transformation from digital optimization results to physical manufacturing drawings.

[0059] As an alternative implementation, after obtaining the target structural model, a simulation verification closed loop can be executed. Optimal geometric parameters are imported into a high-precision finite element analysis model, and a frontal collision simulation is performed to obtain the energy absorbed in the first collision. The energy absorbed in the first collision can be compared with the predicted value from the surrogate model. If the error is less than a preset threshold and the energy absorbed meets safety requirements, the simulation verification is considered successful. After successful verification, the set of parameters is locked, and a product model of the vehicle collision energy-absorbing box is generated. The product model contains geometric information and also includes a verified performance data package, including collision force-displacement curves, total energy absorbed values, and crush mode cloud maps. These geometric and performance data are packaged and archived to form a product definition data package. If the verification fails, the process reverts to the optimization step, and the target values ​​are adjusted and re-optimized based on feedback until a verified target structural model is generated. This process ensures that the generated vehicle collision energy-absorbing box meets design requirements geometrically and undergoes double verification of physical performance, guaranteeing the safety and reliability of the product and providing a reliable data source for subsequent trial production and mass production.

[0060] Through the above settings, the vehicle collision energy-absorbing box is generated automatically and through a verification-driven approach. This achieves an end-to-end automated process from intelligent optimization algorithms to engineering manufacturing data, eliminating potential errors from manual parameter conversion and improving the accuracy and efficiency of data transmission. The generated target structural model serves the manufacturing process, containing accurate geometric parameters and verified performance data. This provides standardized input for subsequent stamping die design and prototype manufacturing, shortening the cycle from design finalization to mass production and improving the overall response speed and quality control level of vehicle component development.

[0061] In this embodiment, firstly, an initial structural model of the vehicle collision energy-absorbing box is obtained. The initial structural model adopts a variable cross-section square tube structure and includes the geometric parameters to be optimized corresponding to at least one induction groove set on the variable cross-section square tube structure. Next, the geometric parameters to be optimized in the initial structural model are optimized to obtain a target structural model. Finally, the vehicle collision energy-absorbing box is generated based on the target structural model. This application obtains an initial structural model using a variable cross-section square tube structure and designs the geometric parameters to be optimized for the induction groove. The variable cross-section square tube structure design itself possesses more stable progressive crushing characteristics, improving stress distribution from a geometrical perspective. Intelligent optimization of the geometric parameters to be optimized for the induction groove is performed, accurately seeking the optimal solution, thereby overcoming the limitations of empirical design or single structure in crushing mode control. This closed-loop logic of basic structural optimization and intelligent parameter optimization enables the collision energy-absorbing box to guide orderly and stable plastic folding of the structure during a frontal collision, avoiding ineffective deformations such as twisting or tearing, improving the continuity and efficiency of energy absorption, and thus solving the technical problem of low energy absorption effect of vehicle collision energy-absorbing boxes generated in related technologies.

[0062] In this embodiment of the application, obtaining the initial structural model of the vehicle collision energy absorption box includes: constructing the axial position parameters, groove width parameters, and groove depth parameters to be optimized, wherein the groove depth parameter is a signed value, a positive value indicates that the corresponding guide groove is a concave structure, and a negative value indicates that the corresponding guide groove is a convex structure; determining the geometric parameters to be optimized based on the axial position parameters, groove width parameters, and groove depth parameters; and constructing the initial structural model based on the geometric parameters to be optimized and the variable cross-section square tube structure.

[0063] The aforementioned axial position parameter can refer to the coordinate position of the guide groove along the axial direction of the energy-absorbing box, i.e., the vehicle collision direction or the length direction of the energy-absorbing box. In this application, the axial position parameter is expressed as a relative position or a specific coordinate value, based on the axial length of the energy-absorbing box. The range of values ​​can be limited by engineering and manufacturing constraints to ensure that the guide groove is arranged within the effective working area of ​​the energy-absorbing box, neither too far forward leading to early failure nor too far back affecting the overall crushing mode. The axial position parameter determines the distribution strategy of the guide groove along the length direction of the energy-absorbing box, affecting the crushing initiation point and the propagation path of the folded wave.

[0064] The aforementioned groove width parameter can refer to the width of the induction groove extending along the pipe wall in the circumferential or cross-sectional direction. The setting of the axial position parameter aims to balance the induction effect and structural integrity. If the width is too narrow, the induction effect will be insufficient, making it difficult to effectively guide orderly crushing; if the width is too wide, it will excessively weaken the local stiffness, causing the structure to tear or become unstable unexpectedly during the collision. It is one of the important geometric indicators for measuring the degree of weakening of the pipe wall by the induction groove.

[0065] The aforementioned groove depth parameter refers to the degree of concavity or convexity of the induced groove relative to the original wall surface of the energy-absorbing box. This parameter can be a signed value, quantified using a unified variable space: a positive value indicates a concave structure, where the groove is recessed inwards to weaken local stiffness, guiding the structure to undergo plastic folding according to a predetermined pattern, achieving orderly crushing. A negative value indicates a convex structure, where the groove protrudes outwards to increase local stiffness, ensuring structural stability during deformation and preventing asymmetric folding or structural tearing such as twisting or lateral bending. This definition of positive and negative values ​​unifies the concave induction and convex stabilization functions, which were handled separately in the design, within a single variable space. This enables continuous quantitative control of the concave / convex shape, allowing the optimization algorithm to freely explore various configuration combinations, including concave, convex, and zero-depth grooveless configurations, thereby finding a better solution that balances energy absorption efficiency and deformation stability.

[0066] As an optional implementation, a three-dimensional basic geometric framework for a variable cross-section square tube structure, such as a trapezoidal variable cross-section square tube, can be established to determine the axial length, front and rear end widths, and wall thickness. The set of variables to be optimized can be initialized, defined as axial position parameters, trench width parameters, and trench depth parameters. The axial position parameter determines the longitudinal coordinates of the induction trench along the tube wall; the trench width parameter defines the lateral dimension range of the trench; the trench depth parameter uses a signed numerical definition mechanism, with positive values ​​representing concave structures, used to locally weaken stiffness to induce ordered crushing, and negative values ​​representing convex structures, used to locally enhance stiffness to ensure deformation stability. Subsequently, these parameters are mapped to the three-dimensional basic geometric framework, and material removal or addition operations are performed at specified locations to generate an initial structural model containing an induction trench with a specific concave-convex shape. This process achieves parameterized construction from macroscopic structure to microscopic induction features, ensuring that the geometric model can be dynamically updated as parameters change, providing a data foundation for subsequent intelligent optimization.

[0067] By introducing a signed groove depth parameter, the above settings enable a unified quantitative expression of concave and convex induced structures, breaking the limitation of requiring individual processing of grooves or convex slots. This unified variable space allows the optimization algorithm to freely search for concave-convex combinations across the entire parameter domain, expanding design freedom. The parametric modeling method ensures the adjustability and consistency of the initial structural model, allowing the shape, position, and size of the induced groove to be independently and collaboratively controlled. This provides a data foundation for subsequent high-precision prediction and rapid optimization, effectively improving the energy absorption efficiency and design flexibility of the energy-absorbing box in frontal collisions.

[0068] In this embodiment of the application, the geometric parameters to be optimized in the initial structural model are optimized to obtain the target structural model, including: obtaining the manufacturing constraint set corresponding to the geometric parameters to be optimized; and optimizing the geometric parameters to be optimized based on the target intelligent agent model and the manufacturing constraint set to obtain the target structural model, wherein the target intelligent agent model takes maximizing the collision energy absorption of the initial structural model as the optimization objective.

[0069] The aforementioned set of manufacturing constraints refers to a series of engineering restrictions that must be met in the structural design of the energy-absorbing box and in subsequent actual production processes. These constraints aim to ensure that the optimized structure exhibits superior performance and manufacturability. The set of manufacturing constraints can consist of the following specific constraints: single-slot geometric constraints, which may include axial position constraints, slot width constraints, and slot depth constraints. The slot depth constraint must balance concave induction, convex stability, and process feasibility. Multi-slot layout constraints, when multiple induction slots exist, may also include non-overlapping position constraints (each induction slot must not exceed the axial range of the energy-absorbing box or overlap with each other), and spacing constraints (the axial spacing between adjacent induction slots must not be less than the corresponding threshold to meet the spacing requirements of the stamping process and avoid processing interference or material breakage). During the optimization process, these constraints are embedded into the algorithm as penalty terms by constructing a target penalty function. This penalizes candidate solutions that violate the constraints, thereby guiding the search process towards the feasible region.

[0070] The aforementioned target intelligent surrogate model refers to a high-precision prediction model built based on a physical information neural network, used to replace time-consuming finite element simulations to accelerate the optimization process. The target intelligent surrogate model receives multiple design variables as input, namely the axial positions of multiple guide grooves, groove width, and groove depth. It outputs predictions of key performance indicators under frontal collision conditions, including collision force, total absorbed energy, and axial displacement. The target intelligent surrogate model incorporates physical constraint information during the collision process, such as mechanical equilibrium equations, into its neural network structure, enabling it to more accurately capture nonlinear mechanical behavior during the collision process, resulting in higher generalization ability and prediction accuracy. In the optimization step, the target intelligent surrogate model can quickly evaluate the collision performance corresponding to different combinations of geometric parameters, guiding the optimization algorithm to find a better solution and shortening the design cycle.

[0071] The aforementioned collision energy absorption refers to the total kinetic energy absorbed by the energy-absorbing box structure through plastic deformation during a frontal vehicle collision. In this application, the optimization objective is to maximize collision energy absorption as the evaluation metric, aiming to ensure that the energy-absorbing box can absorb as much impact energy as possible during a collision, thereby reducing the impact force transmitted to the passenger compartment and protecting the structural integrity of the passenger compartment. The optimization process adjusts the concave-convex shape and geometric parameters of the guide groove to achieve a maximum collision energy absorption value, ensuring the orderliness and stability of the crushing mode.

[0072] As an optional implementation, a set of manufacturing constraints can be obtained, including the range of values ​​for axial position, groove width, and groove depth, as well as physical constraints such as non-overlapping and spacing between induced grooves. Subsequently, an objective function is constructed with the goal of maximizing the total frontal collision energy absorbed by the initial structural model. During optimization, a target intelligent surrogate model is used as a performance predictor, inputting the geometric parameters to be optimized to quickly predict collision energy absorption. A target penalty function is introduced, constructed based on the set of manufacturing constraints. When candidate parameters violate constraints, such as excessively small spacing or exceeding boundaries, a corresponding penalty term is calculated and added to the optimization objective, thereby reducing the fitness of the solution. Optimization algorithms, such as gradient descent or evolutionary algorithms, utilize the predicted values ​​and penalty terms from the target intelligent surrogate model to guide the search direction, iteratively optimizing within the feasible region to gradually approximate parameter combinations that satisfy manufacturing constraints and absorb a large amount of energy, outputting the parameter data as the target structural model.

[0073] By explicitly integrating engineering manufacturing constraints into the intelligent optimization process through the above settings, and transforming complex constraints into mathematical optimization obstacles through a penalty function mechanism, the algorithm is guided to automatically avoid infeasible solutions, ensuring the manufacturability of the optimization results. Combined with an intelligent proxy model aimed at maximizing energy absorption, this process avoids the high computational cost of trial-and-error methods, achieving performance optimization while considering stamping process limitations. This strategy of maximizing performance and hard-filtering constraints shortens the design cycle, solves the problem of optimization results being difficult to implement due to their detachment from actual processes, and enhances the application value and frontal collision safety performance of vehicle collision energy-absorbing boxes in real-world production environments.

[0074] In this embodiment, obtaining the manufacturing constraint set corresponding to the geometric parameters to be optimized includes: obtaining the axial position constraint corresponding to the axial position parameter, the groove width constraint corresponding to the groove width parameter, and the groove depth constraint corresponding to the groove depth parameter; when there is only one induction groove on the variable cross-section square tube structure, the axial position constraint, groove width constraint, and groove depth constraint are integrated to obtain the manufacturing constraint set; when there are multiple induction grooves on the variable cross-section square tube structure, the setting position constraint and setting spacing constraint between the multiple induction grooves are obtained, and the setting position constraint, setting spacing constraint, axial position constraint, groove width constraint, and groove depth constraint are integrated to obtain the manufacturing constraint set.

[0075] The aforementioned axial position constraint refers to the range of coordinate positions that the induction groove must satisfy when arranged along the axial direction of the energy-absorbing box. This constraint stipulates that the axial position of the induction groove must be limited within the effective working range of the energy-absorbing box, specifically ranging from 0.1 to 0.7 times the axial length. This constraint aims to prevent the induction groove from being placed in the ineffective areas at both ends of the energy-absorbing box or exceeding the structural range, ensuring that crushing induction occurs during the expected collision deformation stage.

[0076] The aforementioned channel width constraint refers to the process and functional limitations that the width of the induction channel extending along the pipe wall in the circumferential direction or cross-section must meet. This channel width constraint specifies a range for the channel width, which can be limited to 0.04 to 0.12 times the axial length. This balances the induction effect with structural strength; a width that is too narrow makes it difficult to effectively induce orderly crushing, while a width that is too wide will excessively weaken local stiffness, leading to tearing or instability of the structure during collision. The design feasibility of the stamping die must also be considered.

[0077] The aforementioned groove depth constraint refers to the range of depth requirements that the induction groove must meet relative to the original wall surface of the energy-absorbing box, reflecting a unified quantitative control of the convex and concave shape. This constraint specifies that the groove depth can be limited to between -3 and 4 times the wall thickness of the energy-absorbing box. Positive values ​​indicate inward concavity, and negative values ​​indicate outward convexity. Positive values ​​guide orderly crushing, while negative values ​​ensure stability. This range guarantees sufficient induction or stabilization effects while limiting extreme values ​​to meet the manufacturability requirements of conventional stamping processes.

[0078] The aforementioned positional constraints refer to the spatial relationship restrictions between multiple induction channels when multiple induction channels are installed on the energy-absorbing box. These constraints primarily manifest as non-overlapping constraints, meaning that each induction channel must not exceed the axial range of the energy-absorbing box, and the channels must not overlap or interfere with each other geometrically. This ensures that each induction channel functions independently and maintains the continuity and integrity of the overall energy-absorbing box structure, preventing local structural defects or abnormal stress concentrations caused by overlapping channels.

[0079] The aforementioned spacing constraints refer to the axial distance limit that two adjacent guide grooves must maintain, primarily based on the feasibility of the manufacturing process. The spacing constraints can stipulate that the axial distance between adjacent guide grooves cannot be less than 10mm. This constraint is to meet the spacing requirements of the stamping process, preventing interference, cracking, or compromised processing accuracy due to excessively narrow remaining material during processing, thereby ensuring that the optimized structure can be mass-produced through the stamping process.

[0080] As an optional implementation, geometric constraints can be defined for each individual induction groove, including the range of axial position, the allowable range of groove width, and the extreme limit of groove depth. For the single-groove case, these constraints can be integrated into a manufacturing constraint set. For the multi-groove case, in addition to single-groove constraints, relative positional constraints between induction grooves can be obtained to prevent overlap, and spacing constraints can be implemented to meet stamping process requirements. Subsequently, the single-groove geometric constraints and the mutual constraints of multiple grooves are logically integrated to construct a comprehensive manufacturing constraint set containing restrictions. This process ensures the formation of a complete constraint rule base to guide subsequent optimization algorithms, providing clear boundary conditions and feasibility criteria for parameter optimization.

[0081] Through the above settings, and by classifying and integrating constraints, coverage from single-dimensional to multi-dimensional collaborative constraints is achieved. For single slots, the basic processing boundaries are clearly defined; for multiple slots, constraints on inter-slot interactions are introduced, effectively avoiding the risk of structural interference or stamping fracture caused by densely arranged induced slots. This hierarchical integration mechanism makes the manufacturing constraint set highly flexible and versatile, adaptable to different numbers of induced slot design schemes, ensuring superior optimized structural performance and strict compliance with actual manufacturing process requirements, thus improving the manufacturability and reliability of the design.

[0082] In this embodiment, based on a target intelligent agent model and a set of manufacturing constraints, parameter optimization is performed on the geometric parameters to be optimized to obtain a target structural model. This includes: constructing a target penalty function based on the set of manufacturing constraints, wherein the target penalty function is used to apply penalty terms to candidate solutions that do not satisfy the set of manufacturing constraints, and the degree of penalty corresponding to the penalty term is related to the degree to which the candidate solution exceeds the corresponding manufacturing constraint; guiding the target intelligent agent model based on the target penalty function to optimize the geometric parameters to be optimized to obtain the target structural model; preferably, constructing the target penalty function based on the set of manufacturing constraints includes: constructing penalty terms corresponding to different manufacturing constraints based on different manufacturing constraints in the set of manufacturing constraints; integrating the penalty terms corresponding to different manufacturing constraints to obtain the target penalty function.

[0083] The aforementioned objective penalty function can refer to a mathematical function used in intelligent optimization to transform constrained optimization problems with complex engineering limitations into unconstrained optimization problems that are easier for algorithms to solve. The objective penalty function introduces a penalty mechanism, adding one or more penalty values ​​to the optimization objective, such as maximizing collision energy absorption. When a candidate solution violates the manufacturing constraint set, the objective penalty function calculates the corresponding penalty value, thereby reducing the fitness or score of that candidate solution in the optimization process. This guides optimization algorithms, such as gradient descent or evolutionary algorithms, to avoid infeasible regions, gradually searching and converging to a better solution space that satisfies the manufacturing constraints, achieving a balance between engineering manufacturing feasibility and optimized collision performance.

[0084] The aforementioned penalty terms refer to the basic building blocks of the target penalty function. They are individually calculated measures of the degree of violation for a specific constraint in the set of manufacturing constraints, such as axial position constraints, tank width constraints, and spacing constraints. Each penalty term corresponds to a specific manufacturing constraint. If the candidate solution satisfies the constraint, the penalty term is usually zero; if the candidate solution violates the constraint, the penalty term is greater than zero. The magnitude of the penalty term is related to the degree to which the candidate solution exceeds the corresponding constraint. For example, if the spacing between adjacent induced tanks is less than 10mm, the value of the spacing penalty term will increase as the spacing deficiency increases. The specific functional form can be linear, quadratic, or other nonlinear functions. The target penalty function is composed of multiple penalty terms corresponding to various manufacturing constraints. This method of separate calculation and overall integration allows the optimization algorithm to accurately identify and correct each specific violation, ensuring that the obtained target structural model satisfies multiple manufacturing constraints such as axial position, width, depth, non-overlap, and spacing.

[0085] As an optional implementation, the set of manufacturing constraints can be analyzed, and corresponding penalty functions can be constructed for different constraint types such as axial position, slot width, slot depth, and slot spacing. For example, for cases of axial position exceeding limits or insufficient slot spacing, a quadratic or linear penalty function based on the deviation can be established, making the penalty value increase non-linearly with the degree of violation. Subsequently, the individual penalty terms are mathematically integrated, such as by weighted summation, to form a unified target penalty function. During the optimization process, the geometric parameters to be optimized are input into the target intelligent agent model to obtain the basic energy absorption prediction value, and the total penalty term generated by the violation of constraints for this parameter combination is calculated. The optimization objective can be defined as the basic energy absorption prediction value minus the total penalty term. The algorithm guides the search process away from the high penalty region, i.e., the constraint violation region, through gradient descent or evolutionary strategies, gradually converging to a feasible solution where the penalty term is zero and the collision energy absorption is large, thereby obtaining the target structural model that satisfies the engineering manufacturing constraints.

[0086] By transforming complex discrete constraints into continuous mathematical penalty terms, a smooth transition from constrained optimization to unconstrained optimization is achieved. This mechanism enables the intelligent surrogate model to perform efficient searches in a continuous parameter space, avoiding search interruptions or getting trapped in local optima due to constraint boundaries. The design, where the penalty level is related to the severity of violations, ensures the algorithm's tolerance for minor violations and strong suppression of severe violations, guaranteeing the convergence stability of the optimization process. It also strictly ensures that the results meet the actual manufacturing requirements such as stamping processes, improving the manufacturability and reliability of the design.

[0087] In this embodiment of the application, a target intelligent agent model is guided by a target penalty function to optimize the geometric parameters to be optimized and obtain a target structural model. The process includes: inputting the geometric parameters to be optimized into the target intelligent agent model; guiding the target intelligent agent model based on the target penalty function to output collision performance indices corresponding to different geometric parameters; filtering different geometric parameters based on the collision performance indices corresponding to different geometric parameters to obtain optimized geometric parameters; and constructing a target structural model based on the optimized geometric parameters.

[0088] The aforementioned collision performance indicators refer to a set of important numerical values ​​predicted by the target intelligent agent model based on the input geometric parameters to be optimized, reflecting the mechanical response and energy absorption capacity of the energy-absorbing box under frontal collision conditions. These indicators mainly include: collision force, i.e., axial contact force data that varies with time or displacement during the collision, used to evaluate the structure's resistance to deformation and the stability of the crushing mode; collision energy absorption, i.e., the total kinetic energy absorbed by the energy-absorbing box through plastic deformation during compression, used as the objective function value in the optimization process; and axial displacement, i.e., the amount of compression deformation of the energy-absorbing box during the collision, used to evaluate the effective energy absorption stroke. These collision performance indicators guide the optimization algorithm to judge the performance of the current combination of geometric parameters and, in conjunction with the objective penalty function, screen out candidate solutions that combine high performance and high feasibility.

[0089] The optimized geometric parameters mentioned above refer to a set of optimal design variable values ​​determined through iterative solutions using intelligent optimization algorithms. These values ​​maximize collision performance while satisfying a set of manufacturing constraints. Specifically, the optimized geometric parameters may include the optimal axial position, width, and depth of each induction groove. Based on these optimized geometric parameters, a three-dimensional model of a trapezoidal variable cross-section energy-absorbing box with a specific concave-convex shape induction groove can be reconstructed—the target structural model—to guide the design scheme for engineering production.

[0090] As an optional implementation, the geometric parameters to be optimized can be input in batches into a trained target intelligent agent model. The target intelligent agent model quickly predicts the performance indicators such as collision force, energy absorption, and displacement corresponding to each set of parameters. The degree of violation of manufacturing constraints by each set of parameters is calculated using a target penalty function to generate a penalty value. The optimization algorithm subtracts the penalty value from the predicted energy absorption as a fitness score. Through iterative search, such as genetic algorithms or gradient descent, parameter combinations with high scores and penalty values ​​close to zero are selected, which are the optimized geometric parameters. Subsequently, these optimized geometric parameters are substituted into the parametric modeling interface to dynamically update the geometric features of the trapezoidal variable cross-section square tube, generating a target structural model containing specific concave and convex induced groove morphology.

[0091] By leveraging the high-precision and rapid prediction capabilities of the intelligent surrogate model, the parameter selection and optimization process is accelerated. By incorporating a target penalty function, the selected parameters are ensured to have superior performance and strictly meet engineering manufacturing constraints, avoiding interference from invalid solutions. This closed-loop mechanism of prediction, scoring, and selection achieves a balance between design efficiency and engineering feasibility, shortening the cycle from conceptual design to target model generation, and providing a high-quality foundation for subsequent rapid verification and production.

[0092] In this embodiment of the application, constructing a target structural model based on optimized geometric parameters includes: performing simulation verification on the optimized geometric parameters to obtain simulation verification results; constructing a target structural model based on the optimized geometric parameters and simulation verification results; preferably, performing simulation verification on the optimized geometric parameters to obtain simulation verification results includes: constructing a finite element analysis model for simulating frontal collision conditions; and performing simulation verification on the optimized geometric parameters based on the finite element analysis model to obtain simulation verification results.

[0093] The simulation verification results mentioned above refer to the verification results obtained by simulating a frontal collision of the energy-absorbing box structure with optimized geometric parameters through high-precision finite element simulation. The simulation verification results are mainly used to characterize the accuracy of the target intelligent agent model's predictions and the effectiveness of the optimization scheme. The simulation verification results may include verification conclusions based on data comparison.

[0094] The aforementioned finite element analysis model can refer to a computer numerical simulation model used to simulate the physical process of a frontal vehicle collision, which can be established based on an explicit dynamics solver. The finite element analysis model is a three-dimensional geometric model of a trapezoidal variable cross-section energy-absorbing box reconstructed based on optimized geometric parameters. The finite element analysis model is set to simulate a frontal collision, including appropriate material property definitions, such as the constitutive model of steel, contact relationship definitions, boundary conditions, and collision velocities. The finite element analysis model can perform high-precision verification of the prediction results of the target intelligent proxy model. Due to the high computational cost of finite element simulation, it is used after obtaining a preliminary, relatively optimal solution through intelligent optimization to confirm the physical reality and reliability of the solution, preventing misleading information from the target intelligent proxy model due to overfitting or insufficient generalization ability.

[0095] As an optional implementation, the optimized geometric parameters obtained through intelligent optimization can be imported into finite element analysis software to construct a high-fidelity frontal collision finite element analysis model, such as an explicit dynamic solver, including material properties, boundary conditions, and contact definitions. Subsequently, the finite element analysis model is run to simulate a frontal collision with a rigid barrier, extracting key indicators such as axial collision force, total energy absorption, and crushing mode as simulation verification results. The simulation results are compared with the prediction results of the target intelligent proxy model, and the error rate is calculated. If the error is within a preset threshold and the energy absorption meets the design requirements, the verification is considered successful, and a target structural model is generated based on this set of parameters. If the verification fails or the error exceeds the standard, a feedback mechanism is triggered to lower the target energy absorption threshold and re-enter the optimization process until the optimized parameters that have passed verification are obtained, and the target structural model is constructed.

[0096] Through the above setup, and by introducing high-fidelity finite element simulation for verification, the prediction bias of the target intelligent proxy model under extreme conditions is effectively compensated, ensuring the physical authenticity of the design results. This dual verification mechanism, combining rapid initial screening of the intelligent proxy model with accurate finite element verification, retains the high efficiency of intelligent optimization while ensuring the safety and reliability of the solution. The feedback mechanism ensures the robustness of the design, avoids design failures caused by model errors, improves the confidence level of the energy-absorbing box's frontal collision performance, and provides data support for practical engineering applications.

[0097] In this embodiment of the application, the optimized geometric parameters are simulated and verified based on the finite element analysis model to obtain the simulation verification results, including: simulating and verifying the optimized geometric parameters based on the finite element analysis model to obtain the first collision absorption energy corresponding to the optimized geometric parameters; and determining the simulation verification results based on the first collision absorption energy and the second collision absorption energy, wherein the second collision absorption energy is used to represent the collision absorption energy corresponding to the optimized geometric parameters predicted based on the target intelligent agent model.

[0098] The first collision energy absorption mentioned above can refer to the actual total energy absorption value of the energy-absorbing box structure constructed based on optimized geometric parameters during a frontal collision, calculated by a high-precision finite element analysis model.

[0099] The energy absorbed during the first collision is an output of the physical simulation process, reflecting the energy absorption capacity of a specific geometric structure under real physical laws or a simulation of highly approximate physical laws. The energy absorbed during the first collision can be used to verify whether the performance of the optimized scheme meets the standards and the accuracy of the surrogate model's predictions.

[0100] The aforementioned second collision energy absorption refers to the collision energy absorption value predicted by the target intelligent agent model, based on the same set of optimized geometric parameters. The second collision energy absorption is the prediction result output by the model after quickly inferring from the optimized geometric parameters input into the trained target intelligent agent model. During the validation phase, the second collision energy absorption can be compared with the first collision energy absorption to assess whether the prediction accuracy of the target intelligent agent model is within an acceptable error threshold.

[0101] As an optional implementation, a high-fidelity finite element analysis model can be used to perform frontal collision simulation calculations on the optimized geometric parameters obtained from intelligent optimization. Energy absorption data during the simulation process is extracted to obtain the first collision energy absorption, which represents the physically real energy absorption performance. The predicted values ​​for the same set of parameters output by the target intelligent agent model in the previous optimization stage can also be read, i.e., the second collision energy absorption. Subsequently, the absolute difference or relative error rate between the first and second collision energy absorptions is calculated and compared with a preset allowable threshold. By comparing the first and second collision energy absorptions, the accuracy of the target intelligent agent model's predictions and the reliability of the optimization results are quantitatively evaluated, thereby determining the validity of the simulation verification results and providing data basis for confirming the target structural model.

[0102] By comparing the simulated actual values ​​with the surrogate predictions, a quantitative evaluation mechanism was established, effectively verifying the prediction accuracy and reliability of the target intelligent surrogate model near the optimal solution. This dual verification ensures that the generated target structure model performs well at the algorithm level and conforms to the laws of collision mechanics at the physical level, eliminating design risks caused by model deviations. By clarifying the error judgment criteria, the scientific nature of design decisions was improved, ensuring the accurate achievement of the frontal collision safety performance of the energy-absorbing box, and enhancing the controllability of the development process and the safety of the generated product.

[0103] In this embodiment of the application, the simulation verification result is determined based on the first collision energy absorption and the second collision energy absorption, including: if the difference between the first collision energy absorption and the second collision energy absorption is less than a preset error threshold and the first collision energy absorption is greater than a preset collision energy absorption threshold, the simulation verification result is determined to be a successful simulation verification; if the difference between the first collision energy absorption and the second collision energy absorption is greater than or equal to the preset error threshold, or the first collision energy absorption is less than or equal to the preset collision energy absorption threshold, the simulation verification result is determined to be a failed simulation verification.

[0104] The aforementioned preset error threshold can be considered a numerical limit used to measure the allowable range of deviation between the prediction accuracy of the target intelligent surrogate model and the actual value of the finite element simulation. It can serve as a standard for judging the reliability of the surrogate model's prediction results. If the difference between the energy absorbed in the first collision and the energy absorbed in the second collision is less than the preset error threshold, it indicates that the target intelligent surrogate model's prediction of that specific geometric parameter is accurate, and the optimization algorithm can trust the previous screening results. If the difference is greater than or equal to the preset error threshold, it indicates that the target intelligent surrogate model has prediction bias or overfitting issues, requiring re-optimization or adjustment, and is thus judged as failing the simulation verification.

[0105] The aforementioned preset collision energy absorption threshold refers to the energy absorption performance index that the energy absorption box needs to achieve under frontal collision conditions. As a hard requirement to measure whether passive safety performance is qualified, it can be used to evaluate whether the optimized structure has truly improved safety.

[0106] As an optional implementation, the difference between the first collision energy absorbed by the finite element simulation and the second collision energy absorbed by the target intelligent agent model can be calculated, and it can be determined whether the difference is less than a preset error threshold to verify the prediction accuracy. The first collision energy absorbed is checked to see if it is greater than the preset collision energy absorption threshold to confirm that the performance meets the standard. When the accuracy error is within the allowable range and the energy absorption performance meets the preset collision energy absorption threshold, the simulation verification is deemed successful, and the solution is confirmed as valid. If any condition is not met, i.e., the error exceeds the standard or the energy absorption is insufficient, the simulation verification is deemed unsuccessful. At this point, a feedback mechanism can be triggered to return the geometric parameters that failed the verification and the corresponding energy absorption data to the optimization module as new constraints or reference points, guiding the target intelligent agent model to adjust its search direction and re-optimize the parameters until a better structure that passes the verification is generated.

[0107] By establishing rigorous dual criteria for accuracy and performance, the above settings ensure that the output target structural model possesses high predictive reliability while meeting actual safety requirements. This closed-loop verification mechanism effectively avoids design failures caused by local errors in the target intelligent agent model or optimization getting stuck in local optima, thus improving the robustness and engineering applicability of the optimization results. Automated feedback correction reduces the need for manual intervention, achieving an automated closed loop from intelligent design to safety verification, thereby improving the efficiency and quality of vehicle passive safety component development.

[0108] In this embodiment, a target structural model is constructed based on the optimized geometric parameters and simulation verification results. This includes: if the simulation verification results indicate that the optimized geometric parameters have failed simulation verification, repeatedly reducing the collision absorption energy corresponding to the failed geometric parameters, using the reduced collision absorption energy as a new optimization target, and optimizing the parameters using the new optimization target through a target intelligent agent model, until the simulation verification results indicate that the optimized geometric parameters have passed simulation verification; and if the simulation verification results indicate that the optimized geometric parameters have passed simulation verification, constructing a target structural model based on the passed geometric parameters.

[0109] The reduced collision energy absorption mentioned above refers to a lower numerical target than the original optimization objective when the geometric parameters obtained from the initial optimization fail to pass simulation verification—either the surrogate model's prediction error is too large, or the actual energy absorption performance is insufficient or the structure is unstable. In the initial stage, the optimization objective is to maximize collision energy absorption, which may cause the algorithm to search for solutions at the edge of the feasible region, easily leading to verification failure due to manufacturing constraints or model errors. When verification fails, this setting, through a concession strategy, no longer pursues absolute performance maximization, but sets a slightly lower but more robust energy absorption value that is easier to satisfy constraints and accuracy requirements as the new optimization objective. By gradually reducing the pursuit of energy absorption performance, the optimization algorithm is guided to re-search within a more relaxed performance range, thereby finding a set of feasible solutions that meet engineering manufacturing constraints, ensure the surrogate model's prediction accuracy, and still meet the actual energy absorption performance requirements. If such a solution is found and verified, the iteration stops, and the target structural model is constructed.

[0110] As an optional implementation, an iterative correction mechanism can be triggered when simulation verification determines that the optimized geometric parameters have failed. Specifically, the geometric parameter combinations that failed the current verification are extracted, and the corresponding first collision energy absorption is obtained. Then, the optimization objective is appropriately reduced to generate a new optimization objective. Subsequently, the new optimization objective is re-input into the target intelligent agent model, guiding the algorithm to re-search and filter geometric parameters based on the new optimization objective while meeting manufacturing constraints. This process forms a closed-loop re-optimization cycle, with each cycle adjusting the parameters that failed verification in the previous round until the regenerated geometric parameters, after finite element simulation verification, meet the accuracy requirements and reach the performance threshold, thus determining the target structural model.

[0111] By establishing dual criteria for accuracy and performance, the above settings ensure that the output target structural model possesses high predictive reliability while meeting actual safety requirements. This closed-loop verification mechanism effectively avoids design failures caused by local errors in the target intelligent agent model or optimization getting stuck in local optima, improving the robustness and engineering applicability of the optimization results. Through automated feedback correction, the need for manual intervention is reduced, achieving an automated closed loop from intelligent design to safety verification, thus improving the efficiency and quality of vehicle passive safety component development.

[0112] Figure 2 This is a schematic diagram illustrating the generation process of a vehicle collision energy-absorbing box according to an embodiment of the present invention, as shown below. Figure 2 As shown, firstly, the axial position parameters, groove width parameters, and groove depth parameters to be optimized are constructed. The groove depth parameter is a signed value; a positive value indicates a concave structure, and a negative value indicates a convex structure. Based on the axial position parameters, groove width parameters, and groove depth parameters, the geometric parameters to be optimized are determined. Based on the geometric parameters to be optimized and the variable cross-section square tube structure, an initial structural model is constructed. Next, the geometric parameters to be optimized in the initial structural model are optimized to obtain the target structural model. Finally, based on the target structural model, a vehicle collision energy-absorbing box is generated.

[0113] Figure 3 This is a schematic diagram illustrating an optional vehicle collision energy-absorbing box manufacturing process according to an embodiment of the present invention, as shown below. Figure 3As shown, firstly, the axial position parameters, groove width parameters, and groove depth parameters to be optimized are constructed. The groove depth parameter is a signed value; a positive value indicates a concave structure, and a negative value indicates a convex structure. Based on the axial position parameters, groove width parameters, and groove depth parameters, the geometric parameters to be optimized are determined. Based on the geometric parameters to be optimized and the variable cross-section square tube structure, an initial structural model is constructed. Next, the manufacturing constraint set corresponding to the geometric parameters to be optimized is obtained. Based on the target intelligent agent model and the manufacturing constraint set, the geometric parameters to be optimized are optimized to obtain the target structural model. Finally, based on the target structural model, a vehicle collision energy absorption box is generated.

[0114] Figure 4 This is a schematic diagram illustrating the generation process of another optional vehicle collision energy-absorbing box according to an embodiment of the present invention, such as... Figure 4 As shown, firstly, the axial position parameters, trench width parameters, and trench depth parameters to be optimized are constructed; based on the geometric parameters to be optimized and the variable cross-section square tube structure, an initial structural model is constructed. Next, the manufacturing constraint set corresponding to the geometric parameters to be optimized is obtained; a target penalty function is constructed based on the manufacturing constraint set; the target intelligent agent model is guided by the target penalty function to optimize the geometric parameters to be optimized, resulting in the target structural model. Finally, based on the target structural model, a vehicle collision energy absorption box is generated.

[0115] Figure 5 This is a schematic diagram illustrating the generation process of another vehicle collision energy-absorbing box according to an embodiment of the present invention, as shown below. Figure 5 As shown, firstly, the axial position parameters, groove width parameters, and groove depth parameters to be optimized are constructed; based on the geometric parameters to be optimized and the variable cross-section square tube structure, an initial structural model is constructed. Next, the manufacturing constraint set corresponding to the geometric parameters to be optimized is obtained; a target penalty function is constructed based on the manufacturing constraint set; the geometric parameters to be optimized are input into the target intelligent agent model, and the target intelligent agent model is guided by the target penalty function to output collision performance indices corresponding to different geometric parameters; based on the collision performance indices corresponding to different geometric parameters, different geometric parameters are filtered to obtain optimized geometric parameters, and a target structural model is constructed based on the optimized geometric parameters. Finally, a vehicle collision energy absorption box is generated based on the target structural model.

[0116] The technical solution proposed in this application is described below with reference to an optional embodiment. This application proposes an optimization design method for the guide groove of a trapezoidal energy-absorbing box based on an intelligent agent model. Belonging to the field of vehicle passive safety technology, its purpose is to solve problems such as crushing disorder, low energy absorption efficiency, long design cycle, and weak manufacturability in the structural design of energy-absorbing boxes. The proposed intelligent optimization design method for a trapezoidal energy-absorbing box based on an adjustable guide groove quantifies the depth of the guide groove, constructs a frontal collision intelligent agent model, and incorporates engineering manufacturing constraints to achieve rapid and accurate optimization of the energy-absorbing box structure, thereby improving frontal collision energy absorption performance.

[0117] The technical solution of this application includes: constructing a basic model of a trapezoidal variable cross-section energy-absorbing box; determining optimization design variables; constructing a frontal collision simulation dataset; training a target intelligent agent model; setting optimization objectives and engineering constraints; performing intelligent optimization solutions; and performing model verification and output. This application simultaneously sets internal grooves and external convex grooves on the same energy-absorbing box to achieve concave-convex composite induction; this application defines the groove depth as a signed variable, with positive values ​​representing internal concavity and negative values ​​representing external convexity, achieving unified quantification of concave-convex morphology; this application uses a physical information neural network to construct the target intelligent agent model; this application explicitly incorporates engineering manufacturing constraints such as adjacent groove spacing and size limits during the optimization process; this application uses a trapezoidal variable cross-section square tube for frontal collision scenarios.

[0118] The proposed energy-absorbing box optimization design method based on adjustable concave-convex guide grooves includes the following steps: Constructing a basic model of a trapezoidal variable cross-section energy-absorbing box, and establishing a three-dimensional model of a trapezoidal variable cross-section square tubular energy-absorbing box with an axial length of L. This energy-absorbing box adopts a trapezoidal variable cross-section design, with a front-end width of w_top, a rear-end width of w_bot, and a wall thickness of t_wall. The trapezoidal variable cross-section structure can achieve smoother progressive crushing during frontal collisions, effectively avoiding stress concentration and deformation instability caused by abrupt changes in cross-section. Four independent guide grooves are arranged along the axial direction on each of the four side walls of the energy-absorbing box, for a total of four guide grooves. The geometric parameters of each guide groove can be independently controlled, providing ample variable space for subsequent accurate optimization design. The design variables for the concave-convex guide grooves can be determined, with the axial positions x1, x2, x3, x4, groove widths w1, w2, w3, w4, and groove depths d1, d2, d3, d4 of the four guide grooves used as optimization design variables, totaling twelve independent variables. The depth of the groove can be defined using a signed method: a positive depth indicates a concave structure, used to weaken local stiffness and induce the energy-absorbing box to crush in an orderly manner according to a predetermined pattern; a negative depth indicates a convex structure, used to improve local stiffness, ensure deformation stability, and prevent tearing or disordered deformation during collision. This variable definition method unifies the concave and convex grooves, which originally belonged to different design categories, into the same variable space, realizing continuous quantitative control of concave and convex shapes and providing a foundation for subsequent intelligent optimization. A frontal collision simulation dataset can be constructed, and design variable samples can be generated using the Latin hypercube sampling method. For each set of samples, a frontal collision finite element simulation is conducted to extract performance index data such as axial collision force, total energy absorption, and axial displacement at different time steps. The above data is organized into a frontal collision training dataset for subsequent training of the surrogate model. The construction quality of this dataset directly affects the prediction accuracy of the surrogate model; therefore, this step strictly controls the consistency of simulation parameters to ensure data reliability.

[0119] This application trains a target intelligent proxy model. Specifically, a physical information neural network model is first constructed, using twelve design variables—the position, width, and depth of the four guide slots—as the input layer, and collision performance indicators such as frontal collision force, energy absorption, and axial displacement as the output layer. The target intelligent proxy model incorporates physical constraint information on top of the basic neural network, enabling it to better capture nonlinear mechanical behavior during the collision process and exhibiting higher generalization ability and prediction accuracy. The physical information neural network model is trained using a constructed frontal collision dataset. The network weights are continuously optimized through backpropagation until the model loss function converges, resulting in a high-precision target intelligent proxy model capable of quickly and accurately predicting collision performance. Optimization objectives and engineering constraints can be set, with maximizing the total energy absorption in the frontal collision as the optimization objective, ensuring that the energy-absorbing box can absorb as much energy as possible during the collision, protecting the structural integrity of the occupant compartment. To ensure feasibility in actual production, engineering manufacturing constraints are systematically incorporated: axial position constraints, where the axial positions x1, x2, x3, and x4 of the four induction grooves are limited to [0.1, 0.7]L to ensure the induction grooves are positioned within the effective working area of ​​the energy-absorbing box; groove width constraints, where the groove widths w1, w2, w3, and w4 are limited to [0.04, 0.12]L to prevent insufficient induction effect due to excessive narrowness or structural integrity due to excessive width; groove depth constraints, where the groove depths d1, d2, d3, and d4 are limited to [-3, 4]t_wall, with negative values ​​representing convex structures and positive values ​​representing concave structures, balancing induction effect and processing feasibility; position non-overlap constraints, where the induction grooves must not exceed the axial range of the energy-absorbing box to ensure structural integrity; and spacing constraints, where the axial spacing between adjacent induction grooves is not less than 10mm to meet the spacing requirements of the stamping process and avoid interference or material breakage during processing.

[0120] Next, intelligent optimization can be performed using the gradient descent algorithm, iteratively optimizing the solution based on the trained target intelligent agent model. Since the optimization problem includes multiple engineering and manufacturing constraints, a penalty function is applied to individuals violating the constraints during the optimization process, transforming the constrained optimization problem into an unconstrained optimization problem. The penalty coefficient is dynamically adjusted according to the degree of violation, guiding the search process gradually towards the feasible region. After sufficient iteration, an optimal combination of guide groove parameters satisfying the optimization objective and all engineering constraints is obtained. Model verification and output can then be performed. The obtained optimal guide groove parameters are substituted into the finite element model, and frontal collision simulation verification is conducted. The verification results are compared with the predicted values ​​of the target intelligent agent model to confirm that the error is within an acceptable range and the energy absorption performance meets the preset design requirements. After successful verification, the optimal concave-convex guide groove combination parameters and a complete energy-absorbing box structure scheme are output to guide actual engineering production. Through the above steps, this application achieves rapid and accurate optimization design of the trapezoidal energy-absorbing box guide groove, effectively improving the frontal collision energy absorption performance.

[0121] This application distinguishes between concave and convex shapes by quantifying the depth of the guide groove, achieving the dual effect of inducing orderly crushing through concave indentation and ensuring deformation stability through convex outdentation, thereby improving the energy absorption efficiency of the energy-absorbing box in frontal collisions. Based on a physical information neural network model, a dedicated intelligent agent model for frontal collisions is constructed, shortening the design cycle of the energy-absorbing box structure from weeks to hours, reducing design and calculation costs. The optimization process incorporates engineering and manufacturing constraints such as non-overlapping, non-boundary, and size adaptation throughout, and the optimized structure can be directly processed using stamping technology, exhibiting strong manufacturability and adapting to the mass production of vehicle parts. It achieves controllable crushing mode of the energy-absorbing box in frontal collisions, avoiding problems such as deformation disorder and structural tearing, and improving the frontal collision safety performance of the vehicle body.

[0122] The technical solution proposed in this application distinguishes between concave and convex shapes by positive and negative quantification of the depth of the induction groove; it introduces physical information neural networks into the field of energy absorption box optimization design, using design variables as input and frontal collision force and energy absorption as output to construct a high-precision proxy model; and it embeds engineering manufacturing constraints, such as non-overlapping, non-boundary, and size adaptation, into the optimization process in the form of penalty functions.

[0123] According to another aspect of the embodiments of this application, a vehicle collision energy-absorbing box generating apparatus is also provided. This apparatus can execute the vehicle collision energy-absorbing box generating method of the above embodiments. The specific implementation method and preferred application scenarios are the same as those of the above embodiments, and will not be repeated here.

[0124] Figure 6 This is a schematic diagram of a vehicle collision energy-absorbing box generating device according to an embodiment of this application, as shown below. Figure 6 As shown, the device includes the following: an acquisition module 602, an optimization module 604, and a generation module 606.

[0125] The acquisition module 602 is used to acquire the initial structural model of the vehicle collision energy absorption box. The initial structural model adopts a variable cross-section square tube structure and includes the geometric parameters to be optimized corresponding to at least one induction groove set on the variable cross-section square tube structure. The optimization module 604 is used to optimize the geometric parameters to be optimized in the initial structural model to obtain the target structural model. The generation module 606 is used to generate the vehicle collision energy absorption box based on the target structural model.

[0126] The acquisition module is used to construct the axial position parameters, groove width parameters, and groove depth parameters to be optimized. The groove depth parameter is a signed value, with a positive value indicating that the corresponding induced groove is a concave structure and a negative value indicating that the corresponding induced groove is a convex structure. Based on the axial position parameters, groove width parameters, and groove depth parameters, the geometric parameters to be optimized are determined. Based on the geometric parameters to be optimized and the variable cross-section square tube structure, an initial structural model is constructed.

[0127] The optimization module is used to obtain the set of manufacturing constraints corresponding to the geometric parameters to be optimized; based on the target intelligent agent model and the set of manufacturing constraints, the geometric parameters to be optimized are optimized to obtain the target structural model, wherein the target intelligent agent model takes maximizing the collision energy absorption of the initial structural model as the optimization objective.

[0128] The optimization module is used to obtain the axial position constraints corresponding to the axial position parameter, the groove width constraints corresponding to the groove width parameter, and the groove depth constraints corresponding to the groove depth parameter. When there is only one induction groove on the variable cross-section square tube structure, the axial position constraints, groove width constraints, and groove depth constraints are integrated to obtain a set of manufacturing constraints. When there are multiple induction grooves on the variable cross-section square tube structure, the setting position constraints and setting spacing constraints between the multiple induction grooves are obtained, and the setting position constraints, setting spacing constraints, axial position constraints, groove width constraints, and groove depth constraints are integrated to obtain a set of manufacturing constraints.

[0129] The optimization module is used to construct a target penalty function based on a set of manufacturing constraints. This target penalty function applies penalty terms to candidate solutions that do not meet the manufacturing constraints, and the degree of penalty corresponding to each term is related to the extent to which the candidate solution exceeds the corresponding manufacturing constraints. The module guides a target intelligent agent model based on the target penalty function to optimize the geometric parameters to obtain the target structural model. Preferably, constructing the target penalty function based on the set of manufacturing constraints includes: constructing penalty terms corresponding to different manufacturing constraints based on the different manufacturing constraints in the set; and integrating the penalty terms corresponding to the different manufacturing constraints to obtain the target penalty function.

[0130] The optimization module is used to input the geometric parameters to be optimized into the target intelligent agent model, guide the target intelligent agent model based on the target penalty function, and output the collision performance index corresponding to different geometric parameters. Based on the collision performance index corresponding to different geometric parameters, different geometric parameters are filtered to obtain optimized geometric parameters, and the target structure model is constructed based on the optimized geometric parameters.

[0131] The optimization module is used to perform simulation verification on the optimized geometric parameters and obtain simulation verification results; based on the optimized geometric parameters and simulation verification results, a target structural model is constructed; preferably, the simulation verification on the optimized geometric parameters and obtain simulation verification results includes: constructing a finite element analysis model for simulating frontal collision conditions; and performing simulation verification on the optimized geometric parameters based on the finite element analysis model and obtaining simulation verification results.

[0132] The optimization module is used to perform simulation verification on the optimized geometric parameters based on the finite element analysis model to obtain the first collision absorption energy corresponding to the optimized geometric parameters; based on the first collision absorption energy and the second collision absorption energy, the simulation verification result is determined, wherein the second collision absorption energy is used to represent the collision absorption energy corresponding to the optimized geometric parameters predicted based on the target intelligent agent model.

[0133] The optimization module is used to determine the simulation verification result as passed when the difference between the energy absorbed by the first collision and the energy absorbed by the second collision is less than a preset error threshold and the energy absorbed by the first collision is greater than a preset collision energy absorption threshold; and to determine the simulation verification result as failed when the difference between the energy absorbed by the first collision and the energy absorbed by the second collision is greater than or equal to the preset error threshold, or when the energy absorbed by the first collision is less than or equal to the preset collision energy absorption threshold.

[0134] The optimization module is used to iteratively reduce the collision absorption energy corresponding to the failed simulation verification geometric parameters when the simulation verification results indicate that the optimized geometric parameters have failed the simulation verification. The reduced collision absorption energy is then used as a new optimization target, and the target intelligent agent model is used to optimize the parameters using the new optimization target. This process continues until the simulation verification results indicate that the optimized geometric parameters have passed the simulation verification. If the simulation verification results indicate that the optimized geometric parameters have passed the simulation verification, a target structural model is constructed based on the geometric parameters that have passed the simulation verification.

[0135] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0136] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0137] The aforementioned memory can refer to devices inside a computer used to store data and programs, including RAM, hard disks, etc. RAM can be used to temporarily store running programs and data, while hard disks can be used to store programs and data long-term. Memory enables the computer to read and write data and execute programs. The aforementioned processor is responsible for executing instructions in computer programs and performing data processing. It can also be responsible for controlling and executing various operations, including arithmetic operations, logical operations, and data transmission.

[0138] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0139] The aforementioned computer storage media can refer to the media used in computer memory to store certain discontinuous physical quantities. Computer storage media mainly include semiconductors, magnetic cores, magnetic drums, magnetic tapes, laser discs, etc. Computer-readable storage media include stored programs, which can be a set of instructions that a computer can recognize and execute, running on an electronic computer to meet certain information needs.

[0140] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0141] The aforementioned computer program products can refer to software programs that have been written, tested, and released, and can run on computers or other devices. Computer program products can include application programs, operating systems, utility software, etc., used to achieve specific functions or solve specific problems.

[0142] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the methods in various embodiments of this application.

[0143] The aforementioned non-volatile computer-readable storage medium can refer to a medium for storing data. Non-volatile computer-readable storage media can retain data without loss when power is off and can be used to store long-term data, such as operating systems, applications, and user files. Non-volatile storage media can include hard disk drives, solid-state drives, optical disks, and flash memory storage devices, etc.

[0144] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0145] The aforementioned computer program can refer to a set of instructions used to tell the computer to perform specific tasks or operations. Computer programs can be written by programmers using specific programming languages ​​and can include algorithms, data structures, logic, and control flow. Computer programs can be used for a variety of purposes, including application software, operating systems, etc.

[0146] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0148] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0149] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0150] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0151] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for generating a vehicle collision energy-absorbing box, characterized in that, include: An initial structural model of a vehicle collision energy-absorbing box is obtained, wherein the initial structural model adopts a variable cross-section square tube structure, and the initial structural model includes the geometric parameters to be optimized corresponding to at least one induction groove set on the variable cross-section square tube structure. The geometric parameters to be optimized in the initial structural model are optimized to obtain the target structural model; Based on the target structural model, the vehicle collision energy absorption box is generated.

2. The method for generating a vehicle collision energy-absorbing box according to claim 1, characterized in that, Obtain the initial structural model of the vehicle collision energy absorption box, including: Construct the axial position parameters, groove width parameters, and groove depth parameters to be optimized, wherein the groove depth parameter is a signed value, a positive value indicates that the corresponding induced groove is a concave structure, and a negative value indicates that the corresponding induced groove is a convex structure; Based on the axial position parameter, the groove width parameter, and the groove depth parameter, the geometric parameters to be optimized are determined; Based on the geometric parameters to be optimized and the variable cross-section square tube structure, the initial structural model is constructed.

3. The method for generating a vehicle collision energy-absorbing box according to claim 1, characterized in that, The geometric parameters to be optimized in the initial structural model are optimized to obtain the target structural model, including: Obtain the set of manufacturing constraints corresponding to the geometric parameters to be optimized; Based on the target intelligent agent model and the manufacturing constraint set, the geometric parameters to be optimized are optimized to obtain the target structural model.

4. The method for generating a vehicle collision energy-absorbing box according to claim 3, characterized in that, Obtaining the set of manufacturing constraints corresponding to the geometric parameters to be optimized includes: Obtain the axial position constraints corresponding to the axial position parameters, the groove width constraints corresponding to the groove width parameters, and the groove depth constraints corresponding to the groove depth parameters. When the number of induction grooves provided on the variable cross-section square tube structure is one, the axial position constraint, the groove width constraint, and the groove depth constraint are integrated to obtain the manufacturing constraint set. When there are multiple induction grooves set on the variable cross-section square tube structure, the setting position constraints and setting spacing constraints between the multiple induction grooves are obtained, and the setting position constraints, the setting spacing constraints, the axial position constraints, the groove width constraints, and the groove depth constraints are integrated to obtain the manufacturing constraint set.

5. The method for generating a vehicle collision energy-absorbing box according to claim 3, characterized in that, Based on the target intelligent agent model and the manufacturing constraint set, the geometric parameters to be optimized are optimized to obtain the target structural model, including: A target penalty function is constructed based on the manufacturing constraint set, wherein the target penalty function is used to apply a penalty term to candidate solutions that do not satisfy the manufacturing constraint set, and the degree of penalty corresponding to the penalty term is related to the degree to which the candidate solution exceeds the corresponding manufacturing constraint. Guided by the target penalty function, the target intelligent agent model performs parameter optimization on the geometric parameters to be optimized, thereby obtaining the target structural model; Preferably, constructing a target penalty function based on the set of manufacturing constraints includes: constructing penalty terms corresponding to different manufacturing constraints based on different manufacturing constraints in the set of manufacturing constraints; and integrating the penalty terms corresponding to the different manufacturing constraints to obtain the target penalty function.

6. The method for generating a vehicle collision energy-absorbing box according to claim 5, characterized in that, Guided by the target penalty function, the target intelligent agent model performs parameter optimization on the geometric parameters to be optimized, resulting in the target structural model, including: The geometric parameters to be optimized are input into the target intelligent agent model, and the target intelligent agent model is guided based on the target penalty function to output the collision performance index corresponding to different geometric parameters. Based on the collision performance indices corresponding to the different geometric parameters, the different geometric parameters are filtered to obtain optimized geometric parameters, and the target structure model is constructed based on the optimized geometric parameters.

7. The method for generating a vehicle collision energy-absorbing box according to claim 6, characterized in that, Based on the optimized geometric parameters, the target structural model is constructed, including: The optimized geometric parameters were verified by simulation, and the simulation verification results were obtained. Based on the optimized geometric parameters and the simulation verification results, the target structure model is constructed. Preferably, the optimized geometric parameters are simulated and verified to obtain simulation verification results, including: constructing a finite element analysis model for simulating frontal collision conditions; and based on the finite element analysis model, simulating and verifying the optimized geometric parameters to obtain the simulation verification results.

8. The method for generating a vehicle collision energy-absorbing box according to claim 7, characterized in that, Based on the finite element analysis model, the optimized geometric parameters are simulated and verified to obtain the simulation verification results, including: Based on the finite element analysis model, the optimized geometric parameters are simulated and verified to obtain the first collision energy absorption corresponding to the optimized geometric parameters; Based on the first collision energy absorption and the second collision energy absorption, the simulation verification result is determined, wherein the second collision energy absorption is used to represent the collision energy absorption corresponding to the optimized geometric parameters predicted based on the target intelligent agent model.

9. The method for generating a vehicle collision energy-absorbing box according to claim 8, characterized in that, Based on the energy absorbed by the first and second collisions, the simulation verification results are determined, including: If the difference between the energy absorbed by the first collision and the energy absorbed by the second collision is less than a preset error threshold, and the energy absorbed by the first collision is greater than the preset energy absorbed by the collision threshold, the simulation verification result is determined to be a successful simulation verification. If the difference between the first collision energy absorption and the second collision energy absorption is greater than or equal to a preset error threshold, or if the first collision energy absorption is less than or equal to a preset collision energy absorption threshold, the simulation verification result is determined to be a simulation verification failure.

10. The method for generating a vehicle collision energy-absorbing box according to claim 7, characterized in that, Based on the optimized geometric parameters and the simulation verification results, the target structural model is constructed, including: If the simulation verification results indicate that the optimized geometric parameters have failed the simulation verification, the steps of reducing the collision absorption energy corresponding to the geometric parameters that have failed the simulation verification are repeated, the reduced collision absorption energy is used as a new optimization target, and the target intelligent agent model is used to optimize the parameters using the new optimization target, until the simulation verification results indicate that the optimized geometric parameters have passed the simulation verification. If the simulation verification results indicate that the optimized geometric parameters have passed the simulation verification, the target structural model is constructed based on the geometric parameters that have passed the simulation verification.

11. A device for generating a vehicle collision energy-absorbing box, characterized in that, include: The acquisition module is used to acquire the initial structural model of the vehicle collision energy absorption box. The initial structural model adopts a variable cross-section square tube structure and includes the geometric parameters to be optimized corresponding to at least one induction groove set on the variable cross-section square tube structure. The optimization module is used to optimize the geometric parameters to be optimized in the initial structural model to obtain the target structural model; The generation module is used to generate the vehicle collision energy absorption box based on the target structural model.

12. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, executes the method for generating a vehicle collision energy-absorbing box according to any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the storage medium is located to perform the method for generating a vehicle collision energy-absorbing box according to any one of claims 1 to 10.

Images