CN-115935513-B - Automobile structure optimization method, device and storage medium

CN115935513BCN 115935513 BCN115935513 BCN 115935513BCN-115935513-B

Abstract

The invention discloses an automobile structure optimization method, an automobile structure optimization device and a storage medium, wherein the method comprises the steps of obtaining historical data of automobile structure optimization, wherein the historical data comprise structure parameters and performance simulation test results of corresponding structures, carrying out mathematical modeling on automobile structure optimization problems according to the historical data to determine an optimization target of a model, taking the lowest automobile weight and highest safety as the optimization target, and carrying out optimization processing on the mathematical model by adopting a hierarchical particle swarm optimization algorithm assisted by a classification model to obtain structural design parameters with the lowest automobile weight and highest safety. According to the invention, most of automobile structure performance simulation in the evolution process is replaced by classification model prediction, and the classification result is used for driving the layered particle swarm to evolve, so that the problems of poor automobile structure optimization result and low search efficiency caused by large search space and long evaluation time consumption in the prior art are solved. The method can be widely applied to two fields of evolution calculation and industrial automobile structural design.

Inventors

CHEN WEINENG
WEI FENGFENG
ZHONG JINGHUI

Assignees

华南理工大学

Dates

Publication Date: 20260505
Application Date: 20221206

Claims (9)

1. The automobile structure optimization method is characterized by comprising the following steps of: acquiring historical data of the optimization of the automobile structure, wherein the historical data comprises structural parameters and performance simulation test results of corresponding structures; Carrying out mathematical modeling on the automobile structure optimization problem according to historical data to determine an optimization target of the model, wherein the optimization target is the lowest weight and highest safety of the automobile; Adopting a hierarchical particle swarm optimization algorithm assisted by a classification model to optimize a mathematical model to obtain structural design parameters with minimum automobile weight and highest safety; the hierarchical particle swarm optimization algorithm assisted by the classification model is used for optimizing the mathematical model and comprises the following steps: a1, taking all structural parameters as individuals, encoding the individuals and initializing parameters of a database, wherein the initialized parameters comprise decision variables, corresponding adaptation values and initial speeds; A2, selecting NP individual composition population with optimal adaptation value from the database, wherein the NP individual composition population comprises decision variables Adaptive value Initial velocity Wherein NP is population size; a3, sorting the population according to the adaptation value from good to bad, and dividing the population into four layers; a4, updating and learning individuals in the population to obtain middle learning individuals; a5, performing category prediction on the middle learning individuals by using a classifier; a6, carrying out local information development, and circularly executing the steps A3-A4 until all middle learning individuals are predicted to be of a first type; A7, in the intermediate learning individuals which are all predicted to be the first class, selecting potential individuals by utilizing a geometric relationship to perform automobile structural performance simulation analysis; A8, in order to increase population diversity, except potential individuals, randomly selecting an intermediate learning individual for carrying out automobile structural performance simulation analysis; A9, adding the individuals subjected to the real performance simulation analysis into a database; A10, circularly executing the steps A2 to A7 until the algorithm reaches a termination condition; a11, outputting the individual variable value with the optimal adaptation value in the database as an automobile structure parameter optimization result of global optimization.
2. The method for optimizing the automobile structure according to claim 1, wherein the classification model predicts the performance of the generated structural parameters in the algorithm optimization process, and replaces the performance simulation of the automobile structural parameters in the evolution process to save the algorithm optimization time; and the hierarchical particle swarm optimization algorithm is used as an evolution operator, the hierarchical particle swarm evolution is driven by the performance result predicted by the classification model, and global search optimization is carried out on the automobile structural design.
3. The method for optimizing an automobile structure according to claim 1, wherein the classification model is trained by historical data of automobile structure optimization: Historical data training classification model adopting automobile structure optimization Wherein the structural parameter is input The simulation test result of the weight and the safety of the automobile with the corresponding structure is output ; Generating new structural parameters in an optimization process Afterwards, the performance of the structural parameters is predicted by adopting a classification model ; Selecting structural parameters with better prediction performance Simulation of structural properties of a motor vehicle And utilize And (3) with Updating a training classification model 。
4. The method for optimizing an automobile structure according to claim 1, wherein the step A4 specifically includes: keeping the first layer of individuals in the population unchanged; for each individual in the second layer of the population, two individuals are randomly selected from the first layer And meet the adaptation value Is superior to Wherein Index for marking selected individuals; for each individual in the third tier of the population, randomly selecting an individual from the first tier Randomly selecting an individual from the second layer From the layering, it is known that these two individuals necessarily meet the fitness value Is superior to ; For each individual in the fourth layer in the population, two learning layers are selected randomly at first, and then learning individuals are selected at the two learning layers respectively And meets the adaptation value Is superior to 。
5. The method of optimizing an automotive architecture according to claim 4, wherein the intermediate learning individuals are updated by the following formulas: Wherein, the Represents the algebra of evolution, Representing the structural parameter index to be updated, To take the random number vector with the value space of 0,1, As a parameter for the weight control, And Representing predictive performance ratio of two classification models Good structural parameters, i.e. , ; Represent the first Individual at the first The structural parameters of the generation of the product, Represent the first Individual at the first The structural parameters of the generation of the product, Represent the first Individual at the first The velocity vector of the generation is used, Represent the first Individual at the first And (3) a velocity vector.
6. The method for optimizing an automobile structure according to claim 1, wherein the step A6 specifically includes: Marking the first layer of individuals of the population as The intermediate learning individuals are predicted to be of the first type ; Calculating an intermediate learning individual With the first layer of individuals of the population Euclidean distance of (c); Tagging middle learning individuals Differences from the first layer of the population; And selecting the intermediate learning individuals with the smallest difference as potential individuals to perform automobile structural performance simulation analysis.
7. The method for optimizing an automobile structure according to claim 6, wherein the euclidean distance is calculated by the following formula: the method marks the middle learning individuals Differences from population first layer: Wherein, the Represent the first The individual person is learned in the middle of the study, Represent the first Individuals of the first layer of individuals of the population.
8. An automobile structure optimizing apparatus, comprising: At least one processor; At least one memory for storing at least one program; the at least one program, when executed by the at least one processor, causes the at least one processor to implement the method of any one of claims 1-7.
9. A computer readable storage medium, in which a processor executable program is stored, characterized in that the processor executable program is for performing the method according to any of claims 1-7 when being executed by a processor.

Description

Automobile structure optimization method, device and storage medium Technical Field The invention relates to two large fields of evolution calculation and industrial automobile structural design, in particular to an automobile structural optimization method, an automobile structural optimization device and a storage medium. Background In recent years, with the emphasis of national environmental protection and the development of diversified customer demands, the automobile industry is also faced with a plurality of challenges and development opportunities. For example, in terms of optimization of the structure of an automobile, how to reduce the weight of the automobile and improve the collision safety, fuel efficiency, and the like are problems that are in need of optimization. However, the automobile structure optimization is a complex problem, and has the problems of large search space, large optimization difficulty, low efficiency and the like, and the problems are difficult to effectively solve by using a traditional optimization algorithm. Evolution calculation is a common and effective means for solving complex optimization problems, has been widely applied to various industrial optimization problems, and can better find an optimal optimization scheme from a global angle for the automobile structure optimization problem. However, evolution computation implements an optimization process of "superior and inferior elimination" by evaluating candidate solutions several tens of thousands, even hundreds of thousands, millions of fitness values. In the structural optimization design of the automobile, the structural performance needs finite element analysis software for simulation analysis, the safety performance needs collision software for simulation analysis, and the dynamics performance needs computational fluid dynamics software for simulation analysis. The simulation software takes tens of minutes or even hours to perform a simulation analysis, so that the time consumption for evaluating the performance of a candidate solution is huge, and it is difficult to complete tens of thousands of adaptation value evaluations within an acceptable time. Therefore, how to complete global optimization design of an automobile structure within an acceptable time by using an evolution calculation method is still a problem to be solved. Disclosure of Invention In order to solve at least one of the technical problems existing in the prior art to a certain extent, the invention aims to provide an automobile structure optimization method, an automobile structure optimization device and a storage medium. The technical scheme adopted by the invention is as follows: an automobile structure optimization method comprises the following steps: acquiring historical data of the optimization of the automobile structure, wherein the historical data comprises structural parameters and performance simulation test results of corresponding structures; Carrying out mathematical modeling on the automobile structure optimization problem according to historical data to determine an optimization target of the model, wherein the optimization target is the lowest weight and highest safety of the automobile; And optimizing the mathematical model by adopting a hierarchical particle swarm optimization algorithm assisted by the classification model to obtain the structural design parameter with the minimum weight and highest safety of the automobile. Further, the classification model predicts the performance of the generated structural parameters in the algorithm optimization process, replaces the performance simulation of the automobile structural parameters in the evolution process, and saves the algorithm optimization time; and the hierarchical particle swarm algorithm is used as an evolution operator, the hierarchical particle swarm evolution is driven by the performance result predicted by the classification model, and global search optimization is carried out on the automobile structural design. Further, the classification model is trained through historical data of automobile structure optimization: Historical data training classification model adopting automobile structure optimization Wherein the structural parameter is inputThe simulation test result of the weight and the safety of the automobile with the corresponding structure is output Generating new structural parameters in an optimization processAfterwards, the performance of the structural parameters is predicted by adopting a classification model Selecting structural parameters with better prediction performanceSimulation of structural properties of a motor vehicleAnd utilizeAnd (3) withUpdating a training classification model Further, the optimizing process for the mathematical model by adopting the hierarchical particle swarm optimization algorithm assisted by the classification model comprises the following steps: A1, taking all the structural parameters as individuals, encodin