CN-121980674-A - Intelligent inverse prediction method and system for passenger car body structure design requirement-parameter

CN121980674ACN 121980674 ACN121980674 ACN 121980674ACN-121980674-A

Abstract

The invention discloses a passenger car body structure design demand-parameter intelligent inverse prediction method and system, wherein the method comprises the steps of constructing a parameterized passenger car body finite element model, defining material models, beam section parameters and thicknesses of a plurality of key beam structures as design variables, and defining a plurality of items of bending stiffness, torsional stiffness, bending strength, torsional strength, modal frequency, fatigue life and rollover safety performance of a car body as performance indexes; the method comprises the steps of obtaining a plurality of groups of design variable samples and corresponding performance index data through experimental design to form a data set, training a deep neural network based on the data set to obtain a proxy model capable of predicting performance indexes by design variables, establishing a reverse mapping process, combining the proxy model to output an intelligent reverse-solving model, and solving the intelligent reverse-solving model according to user requirements to output a passenger car body structure design scheme. The invention can quickly generate the optimal structure parameter combination according to different performance requirements.

Inventors

XIA JIANHUA
ZHANG XIAOPENG
QIAO JUNFENG
JIN JUNFENG
YANG XIAOMENG
Shang Hengzhi

Assignees

河南少林客车股份有限公司
河南工业大学

Dates

Publication Date: 20260505
Application Date: 20251230

Claims (10)

1. The intelligent inverse prediction method for the demand-parameter of the body structure of the passenger car is characterized by comprising the following steps of: step 1, constructing a parameterized passenger car body finite element model, defining material models, beam section parameters and thicknesses of a plurality of key beam structures as design variables, and defining a plurality of items of bending rigidity, torsional rigidity, bending strength, torsional strength, modal frequency, fatigue life and rollover safety performance of the passenger car body as performance indexes; Step 2, obtaining a plurality of groups of design variable samples and corresponding performance index data thereof through experimental design to form a data set; step 3, taking the design variable as input and the performance index as output, training a deep neural network based on the data set, and obtaining a proxy model capable of predicting the performance index by the design variable; step 4, establishing a reverse mapping process, and combining the proxy model to output an intelligent reverse model; and 5, according to the requirements of users, solving through the intelligent inverse model, and outputting a passenger car body structure design scheme.
2. The intelligent inverse prediction method for the structural design requirement and parameters of the passenger car body according to claim 1, wherein in the step 2, experimental design sampling is performed by using a Latin hypercube sampling method.
3. The intelligent inverse prediction method for the structural design requirements and parameters of the passenger car body according to claim 1, wherein the step 2 comprises the steps of carrying out statics, modes, fatigue life and rollover simulation analysis on the parameterized finite element model of the passenger car body corresponding to each group of design variable samples, and calculating various corresponding performance index values.
4. The intelligent inverse prediction method for the structural design requirement and parameters of the passenger car body according to claim 1, wherein in the step 4, the inverse mapping process comprises a generation layer, an optimization layer and a verification layer; The generation layer is used for receiving input performance requirements and generating candidate design parameter sets meeting the performance requirements preliminarily through a conditional generation countermeasure network; The optimization layer takes the agent model as an evaluation tool, performs multi-objective optimization on the candidate design parameter set, and outputs a pareto front solution set; And substituting the pareto front solution set into the finite element model by the verification layer to verify, outputting an intelligent inverse model if the verification is passed, otherwise, updating the data set and the model by using a verification result, and re-executing the inverse mapping process.
5. The intelligent inverse prediction method for structural design requirements and parameters of a passenger car body according to claim 4, wherein the generator loss function of the condition generating countermeasure network is as follows: Wherein x gen is a generator generated design variable, y req is an input performance requirement condition vector; The weight super parameter is used for balancing the two losses; scoring x gen and y req for the discriminators; Is the expected value; is the L2 norm; Is a performance predictor for the proxy model pair x gen .
6. The intelligent inverse prediction method for structural design requirements and parameters of a passenger car body according to claim 4, wherein the optimization layer performs multi-objective optimization on the candidate design parameter set by adopting an improved multi-objective genetic algorithm NSGA-III, and a multi-objective optimization mathematical model is established as follows: Wherein the method comprises the steps of Is the total mass of the passenger car body; And Is a bending stiffness performance index and a torsional stiffness performance index; Is the yield limit of the body material; Is that The maximum stress of the vehicle body under the bending working condition is designed; Is a safety coefficient; Is that The design scheme is that the maximum stress of the vehicle body is under the torsion working condition; Representing an intensity safety factor; Is that The fatigue life of the vehicle body under the design proposal, Is a fatigue life safety factor; for the lowest number of fatigue life cycles allowed; Is that The first-order modal frequency of the car body under the design scheme, Is an ideal first-order modal frequency; The distance from the maximum deformation moment of all side wall upright posts of the vehicle body to the living space is set; And The lower and upper limits of the variables, respectively; Is that The ith design variable in the design scheme.
7. The intelligent inverse prediction method for the structural design requirements and parameters of the passenger car body according to claim 4, wherein the verification passing standard of the verification layer is that the deviation between the performance index value calculated through the finite element model and the input performance requirement value does not exceed a preset threshold value, if the deviation does not pass the verification, an online learning mechanism is triggered, a new design parameter-performance index data pair is added into the data set, and the agent model and the intelligent inverse model are finely adjusted.
8. The intelligent inverse prediction method for the design requirements and parameters of the body structure of the passenger car according to claim 1, wherein the key beam structure comprises a roof top beam, a waist beam, a side wall column, a door side column, a front and rear wall column, a roof cross beam, a roof middle longitudinal beam, a front end part of a vehicle bottom framework, a front end longitudinal beam of the vehicle bottom framework, a middle diagonal brace of the vehicle bottom framework, a rear end structure of the vehicle bottom framework and a rear end longitudinal beam of the vehicle bottom framework.
9. Passenger train body structure design demand-parameter intelligence is carried out and is found prediction system, characterized in that includes: The parameterized model construction module is used for constructing a parameterized passenger car body finite element model, defining the material model, beam section parameters and thickness of a plurality of key beam structures as design variables, and defining a plurality of items of bending rigidity, torsional rigidity, bending strength, torsional strength, modal frequency, fatigue life and rollover safety performance of the car body as performance indexes; The data acquisition module is used for acquiring a plurality of groups of design variable samples and corresponding performance index data thereof through experimental design to form a data set; The agent model building module is used for taking the design variable as input and the performance index as output, and training the deep neural network based on the data set to obtain an agent model capable of predicting the performance index by the design variable; the intelligent inverse model construction module is used for establishing an inverse mapping process and outputting an intelligent inverse model by combining the agent model; And the inverse prediction module is used for outputting the structural design scheme of the passenger car body after solving the intelligent inverse model according to the user requirements.
10. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements a passenger vehicle body structure design requirement-parameter intelligent inverse prediction method as claimed in any one of claims 1 to 8.

Description

Intelligent inverse prediction method and system for passenger car body structure design requirement-parameter Technical Field The invention relates to the technical field of automobile body manufacturing, in particular to an intelligent inverse prediction method and system for parameters of the structural design requirement of a passenger car body. Background Many inventors innovate the body structure as an important research object of the bus structure design, but do not perform performance-based structural optimization on the body, for example, a bus body structure and a bus are designed in the patent number CN202421341700.2, the body structure replaces steel skins fixed on a front wall framework and a rear wall framework with carbon fiber composite boards, a light-weight body structure of a pure electric city bus is disclosed in the CN202111593179.2, the body adopts a full-load structure, so that the dead weight of the bus is greatly reduced, and the CN201911088129.1 discloses a full-load pure electric business bus structure, wherein the full-stamping front wall framework and the full-stamping rear wall framework are connected into a whole through a full-stamping top cover arranged above the full-stamping side wall framework, rectangular pipe side wall bodies arranged on two sides of the full-stamping side wall framework and a truss type underframe assembly arranged below the full-stamping side wall framework to form a box type framework body assembly of the whole bus. Some inventors carry out structural optimization design on a bus body, but the designed bus body structure is difficult to meet the requirements of all road driving conditions due to fewer considered performances, meanwhile, the design process is not simplified and verified, for example, CN202110599842.3 discloses a pure electric bus body structure optimization design method based on target split flow, a pure electric bus body skeleton light-weight cooperative mechanism based on target cascade connection is constructed according to the structural characteristics of the pure electric bus body skeleton, the light-weight problem of the pure electric bus body skeleton structure is disassembled into a father system optimization problem and a subsystem optimization problem, CN201720596028.5 discloses a low-wind-resistance bus skeleton structure imitating Caribbean seal, and a finite element structure optimization technology is used for carrying out integral optimization on the whole bus skeleton according to bus load so as to form a bionic bus space skeleton with uniform integral bearing and reasonable rigidity distribution. Disclosure of Invention Aiming at the problems that only single or few performance indexes are focused in the existing passenger car body structure design, the comprehensive road driving condition requirement cannot be fully met, and the design process is not simplified and verified, the invention provides a passenger car body structure design requirement-parameter intelligent inverse prediction method and system, by establishing an inverse mapping process, according to the method, the material model, the beam section parameters and the thickness design variables of the vehicle body structure can be directly and reversely calculated according to the index requirements of the vehicle body such as strength, rigidity, mode, fatigue life, rollover and the like, and the efficient and accurate reverse mapping design from the user requirements to the structural parameters is realized, so that the design efficiency and accuracy of the passenger vehicle body are improved, and the vehicle body research and development period is shortened. In order to achieve the above purpose, the present invention adopts the following technical scheme: the invention provides an intelligent inverse prediction method for the structural design requirement-parameter of a passenger car body, which comprises the following steps: step 1, constructing a parameterized passenger car body finite element model, defining material models, beam section parameters and thicknesses of a plurality of key beam structures as design variables, and defining a plurality of items of bending rigidity, torsional rigidity, bending strength, torsional strength, modal frequency, fatigue life and rollover safety performance of the passenger car body as performance indexes; Step 2, obtaining a plurality of groups of design variable samples and corresponding performance index data thereof through experimental design to form a data set; step 3, taking the design variable as input and the performance index as output, training a deep neural network based on the data set, and obtaining a proxy model capable of predicting the performance index by the design variable; step 4, establishing a reverse mapping process, and combining the proxy model to output an intelligent reverse model; and 5, according to the requirements of users, solving through the intelligent inverse model, and outputt