CN-122020862-A - Vehicle body structure collaborative optimization method and system based on multi-part intelligent body

CN122020862ACN 122020862 ACN122020862 ACN 122020862ACN-122020862-A

Abstract

The application belongs to the field of intelligent design of vehicle bodies, and particularly discloses a vehicle body structure collaborative optimization method and system based on multi-part intelligent bodies, wherein the method comprises the steps of constructing a multi-part intelligent body system and defining a design action space and a performance state space of each intelligent body; the method comprises the steps of constructing a global rewarding function and a local rewarding function of intelligent design of a vehicle body structure, constructing a mixed rewarding function based on the global rewarding function and the local rewarding function, carrying out iterative multi-stage collaborative training by taking the mixed rewarding function as an optimization target under the coordination of the joint optimization intelligent controller, outputting a trained intelligent agent strategy network, and carrying out parallel collaborative optimization on an initial design scheme of the vehicle body by utilizing the trained intelligent agent strategy network to obtain a coordinated multi-part optimization design result. The application can realize the performance balance of the whole vehicle and improve the design efficiency.

Inventors

Yuan Qiuqi
XU SHIWEI
XIANG XINGCHU
SONG HUA
LU HOUGUO
LI JIANYU
XIAO PEIJIE
KAN HONGGUI
SHU ZHAOKUN

Assignees

湖南大学

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. A vehicle body structure collaborative optimization method based on a multi-part intelligent body is characterized by comprising the following steps: Constructing a multi-part intelligent system and defining a design action space and a performance state space of each intelligent body, wherein the multi-part intelligent system comprises a white body finite element model, intelligent body clusters corresponding to different body parts and a joint optimization intelligent controller; constructing a global rewarding function and a local rewarding function of the intelligent design of the vehicle body structure, and constructing a mixed rewarding function based on the global rewarding function and the local rewarding function; Under the coordination of the combined optimization intelligent controller, performing iterative multi-stage collaborative training by taking the mixed rewarding function as an optimization target, and outputting a trained agent strategy network; and carrying out parallel collaborative optimization on the initial design scheme of the vehicle body by using the trained intelligent agent strategy network to obtain a coordinated multi-part optimal design result.
2. The multi-part agent-based vehicle body structure collaborative optimization method according to claim 1, wherein the agent clusters at least include a threshold beam agent, an integrally die-cast floor agent, and a B-pillar agent; the defining the design action space and the performance state space of each agent comprises: The method comprises the steps of defining a designed action space of a threshold beam intelligent body by taking the thickness of a reinforcing plate as a variable, and defining a performance state space of a threshold Liang Zhineng body by taking the mass, the maximum invasion amount, the maximum energy absorption value and the contact counter force of the junction of the threshold beam, a B column and a floor; The thickness of the reinforcing rib is used as a variable, a design action space of the floor intelligent body after integral die casting is defined, and a performance state space of the floor intelligent body after integral die casting is defined by the self weight, the maximum deformation and a stress concentration coefficient at the joint of the floor and the threshold beam after integral die casting; The action space of the B-pillar intelligent body is defined by taking the sectional thickness of the reinforcing plate as a variable, and the performance state space of the B-pillar intelligent body is defined by the weight of the B-pillar reinforcing plate, the maximum invasion and the local rigidity of the connecting point of the B-pillar and the threshold beam.
3. The multi-part agent-based vehicle body structure collaborative optimization method according to claim 1, wherein the global rewarding function construction process comprises: obtaining the torsional rigidity score, the bending rigidity score, the side collision score, the front collision score and the tail collision score of the whole vehicle body through finite element calculation; Acquiring the total mass of the white automobile body, and calculating compatibility punishment values among all parts to be optimized; And multiplying the torsional rigidity score, the bending rigidity score, the side collision score, the front collision score, the tail collision score, the total body-in-white mass and the compatibility penalty value by the weight coefficients corresponding to the torsional rigidity score, the bending rigidity score, the side collision score, the front collision score, the tail collision score and the total body-in-white mass and the compatibility penalty value respectively, and then performing weighted combination to obtain the global rewarding function value.
4. The method for collaborative optimization of a multi-part intelligent agent-based vehicle body structure according to claim 3, wherein the method for obtaining the compatibility penalty value comprises: Obtaining the maximum equivalent stress on the adjacent parts at the connecting interfaces of the parts and the material yield strength of the adjacent parts; Determining the ratio of the maximum equivalent stress to the material yield strength, calculating the stress concentration value of each connecting interface based on the difference value of the ratio and a preset stress threshold, and accumulating the stress concentration values of all the connecting interfaces to obtain an interface stress concentration punishment subitem; Acquiring the total mass of a specified local area of the vehicle body and the total mass of the white vehicle body to determine a mass ratio, and calculating a mass distribution unbalance penalty subitem according to the difference value of the mass ratio and a preset mass distribution threshold value; and multiplying the interface stress concentration punishment sub-items and the mass distribution unbalance punishment sub-items by the corresponding weight coefficients respectively, and then carrying out weighted combination to obtain the compatibility punishment value.
5. The multi-part agent-based vehicle body structure collaborative optimization method according to claim 1, wherein the process of constructing the hybrid rewards function comprises: acquiring performance rewards and quality penalties corresponding to target part intelligent bodies, wherein the performance rewards at least comprise rigidity, deformation, invasion or connection counter force of the parts; multiplying the performance rewards and the quality penalties by the corresponding weight coefficients respectively, and then carrying out weighted combination to obtain a local rewarding function of the target part intelligent agent; and multiplying the sum of the global rewarding function and all the local rewarding functions by the weight coefficient corresponding to each local rewarding function respectively, and then carrying out weighted summation to obtain the mixed rewarding function.
6. The multi-part agent-based vehicle body structure collaborative optimization method according to claim 1, wherein the agent strategy network is trained through a multi-stage training strategy; the multi-stage training strategy comprises a first training stage and a second training stage; In the first training stage, increasing the weight of the local rewarding function in the mixed rewarding function so as to guide each agent to preferentially promote the performance of each corresponding part; And in the second training stage, increasing the weight of the global rewarding function in the mixed rewarding function so as to guide the intelligent agent to optimize the comprehensive performance of the whole vehicle and realize the cooperative optimization of multiple components.
7. The multi-part agent-based vehicle body structure collaborative optimization method according to claim 1, wherein collaborative strategy training of the joint optimization intelligent controller adopts a multi-agent reinforcement learning algorithm, and a strategy network of each agent is updated according to the local rewarding function and the global rewarding function value; In the process of updating the strategy network, interface information is transmitted among the intelligent agents through a communication channel so as to realize the coordination among the intelligent agents.
8. A car body structure collaborative optimization system based on multiple parts intelligent agent is characterized in that the system comprises: the system construction module is used for constructing a multi-part intelligent system and defining the design action space and the performance state space of each intelligent body, wherein the multi-part intelligent system comprises a white body finite element model, intelligent body clusters corresponding to different body parts and a joint optimization intelligent controller; The function construction module is used for constructing a global rewarding function and a local rewarding function which are intelligently designed for the vehicle body structure, and constructing a mixed rewarding function based on the global rewarding function and the local rewarding function; The training module is used for carrying out iterative multi-stage collaborative training by taking the mixed rewarding function as an optimization target under the coordination of the joint optimization intelligent controller, and outputting a trained intelligent agent strategy network; and the collaborative optimization module is used for carrying out parallel collaborative optimization on the initial design scheme of the vehicle body by utilizing the trained intelligent agent strategy network to obtain a coordinated multi-part optimization design result.
9. An electronic device, comprising: At least one memory for storing a computer program; At least one processor for executing the memory-stored program, which processor is adapted to perform the method according to any of claims 1-7, when the memory-stored program is executed.
10. A computer readable storage medium storing a computer program, characterized in that the computer program, when run on a processor, causes the processor to perform the method according to any one of claims 1-7.

Description

Vehicle body structure collaborative optimization method and system based on multi-part intelligent body Technical Field The application belongs to the field of intelligent design of vehicle bodies, and particularly relates to a vehicle body structure collaborative optimization method and system based on a multi-part intelligent body. Background The body in white is the skeleton of an automobile, and its performance is directly related to the safety, weight saving level and durability of the vehicle. The traditional vehicle body structure optimization usually adopts a serial or isolated mode, namely, key parts such as a threshold beam, an A/B column, a floor and the like are respectively optimized. This approach suffers from inherent drawbacks, firstly the body is a complete force transfer system, optimizing a component alone may result in disrupting the continuity of the overall load path, resulting in a "short plate effect", i.e. the performance of one component is improved at the expense of the performance of the other component. Moreover, the difference in the cooperativity between the different components, the stiffness enhancement of one component may change the distribution of collision force in the whole vehicle, thereby putting new and unforeseen performance requirements on the related components. In addition, serial optimization requires a great deal of manual intervention and iteration, relies on engineers' experience to coordinate conflicts between different components, and is inefficient. Therefore, aiming at the problems of unbalanced performance and low design efficiency of the whole vehicle caused by the isolated design of the components in the traditional vehicle body structure, how to realize the performance balance of the whole vehicle and improve the design efficiency is a technical problem which needs to be solved currently. Disclosure of Invention Aiming at the defects of the prior art, the application aims to realize the performance balance of the whole vehicle and improve the design efficiency. In order to achieve the above object, in a first aspect, the present application provides a method for collaborative optimization of a vehicle body structure based on a multi-component intelligent agent, including: Constructing a multi-part intelligent system and defining a design action space and a performance state space of each intelligent body, wherein the multi-part intelligent system comprises a white body finite element model, intelligent body clusters corresponding to different body parts and a joint optimization intelligent controller; constructing a global rewarding function and a local rewarding function of the intelligent design of the vehicle body structure, and constructing a mixed rewarding function based on the global rewarding function and the local rewarding function; Under the coordination of the combined optimization intelligent controller, performing iterative multi-stage collaborative training by taking the mixed rewarding function as an optimization target, and outputting a trained agent strategy network; and carrying out parallel collaborative optimization on the initial design scheme of the vehicle body by using the trained intelligent agent strategy network to obtain a coordinated multi-part optimal design result. Optionally, the intelligent agent cluster at least comprises a threshold beam intelligent agent, an integrated die-cast floor intelligent agent and a B-pillar intelligent agent; the defining the design action space and the performance state space of each agent comprises: The method comprises the steps of defining a designed action space of a threshold beam intelligent body by taking the thickness of a reinforcing plate as a variable, and defining a performance state space of a threshold Liang Zhineng body by taking the mass, the maximum invasion amount, the maximum energy absorption value and the contact counter force of the junction of the threshold beam, a B column and a floor; The thickness of the reinforcing rib is used as a variable, a design action space of the floor intelligent body after integral die casting is defined, and a performance state space of the floor intelligent body after integral die casting is defined by the self weight, the maximum deformation and a stress concentration coefficient at the joint of the floor and the threshold beam after integral die casting; The action space of the B-pillar intelligent body is defined by taking the sectional thickness of the reinforcing plate as a variable, and the performance state space of the B-pillar intelligent body is defined by the weight of the B-pillar reinforcing plate, the maximum invasion and the local rigidity of the connecting point of the B-pillar and the threshold beam. Optionally, the construction process of the global rewarding function includes: obtaining the torsional rigidity score, the bending rigidity score, the side collision score, the front collision score and the tail collisi