A lightweight design method and system for automobile rear subframe

Abstract

The present invention discloses a method and system for lightweight design of a rear subframe of an automobile, and relates to the technical field of rear subframe design of an automobile. The method comprises: constructing a natural frequency simulation model of the rear subframe of an automobile, determining the number of design parameters and the design domain of the design parameters, taking the design domain of the design parameters as the design space, generating an optimized population in the design space, and evaluating the frequency constraints and weight target values ​​of the individuals in the population; performing low-dimensional mapping on all evaluated individuals, screening and obtaining the key design parameters of the low-dimensional space, and constructing a second-order polynomial radial basis function machine learning model according to the key design parameters; determining the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model; evaluating the target and constraint values ​​of the optimal solution, and judging whether the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements. The present invention solves the problem in the prior art that efficiency and accuracy cannot be taken into account in the lightweight design of the rear subframe of an automobile.

Description

Technical Field

[0001] The invention relates to the technical field of automobile rear subframe design, and in particular to a method and system for lightweight design of an automobile rear subframe. Background Art

[0002] When designing the performance of the automobile chassis, the main focus is on how to improve the vehicle's handling and output power while reducing fuel consumption and saving materials. As an important load-bearing component of the automobile chassis, the rear subframe of the automobile is mainly used to improve the connection stiffness between the body and the suspension system and isolate the vibration and noise from the powertrain, transmission system and road excitation. During the actual driving of the car, the alternating contact between the road surface and the tires is likely to cause noise. If the noise frequency is close to the natural frequency of the subframe itself, it will produce intense resonance, thereby aggravating the fatigue failure of the rear subframe and further reducing the NVH performance of the car, seriously affecting the overall driving experience and real-time riding experience. To this end, how to reduce fuel consumption by lightweighting the rear subframe of the car while ensuring the NVH performance of the car is one of the keys to the design of the rear subframe of the car.

[0003] At present, the natural frequency and weight of the rear subframe can be obtained by performing multiple modal analyses on the rear subframe of the automobile. However, it is time-consuming to complete a complete modal analysis, which makes it impossible for the existing evolutionary algorithm to obtain the optimal rear subframe structure that meets both lightweight design indicators and NVH performance indicators within a limited design cycle. Therefore, existing research focuses on how to design an effective experimental design method to obtain an ideal rear subframe structure while using only a few modal analyses as much as possible. However, the experimental design method does not involve any evolutionary optimization process, which makes the overall design process blind and unable to stably obtain the design requirements that fully meet the actual operating conditions. Summary of the invention

[0004] In view of this, an object of the present invention is to provide a method and system for lightweight design of a rear subframe of an automobile, aiming to solve the problem in the prior art that both efficiency and accuracy cannot be taken into account when lightweight designing the rear subframe of an automobile.

[0005] The embodiment of the present invention is implemented as follows:

[0006] A method for lightweight design of a rear subframe of an automobile, the method comprising:

[0007] According to the geometric model and modal analysis of the rear subframe of the automobile, a natural frequency simulation model of the rear subframe of the automobile is constructed, and according to the sheet metal connection distribution of the rear subframe of the automobile, the number of design parameters is determined, and the thickness variation range of the sheet metal is used as the design parameter design domain to construct a mathematical optimization model for the lightweight design of the rear subframe of the automobile;

[0008] Taking the design parameter design domain as the design space, generating an optimized population in the design space by using the maximum and minimum distance sampling criteria, and evaluating the natural frequency values ​​and weights of the individuals in the population according to the natural frequency simulation model and the geometric model to obtain the natural frequency constraints and weight target values ​​respectively;

[0009] Affine propagation clustering is used to cluster the currently evaluated individuals. The central individuals used for distributed local exploration are determined based on the distribution characteristics of each category and the potential information of individuals in the category. The local area corresponding to the central individual of distributed exploration is determined based on the kNN model.

[0010] Sammon mapping is used to perform low-dimensional mapping on all evaluated individuals, and key design parameters of the low-dimensional space are screened and obtained, and a second-order polynomial radial basis function machine learning model is constructed according to the key design parameters;

[0011] In the local area corresponding to each central individual, the CoDE algorithm is used to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model;

[0012] The geometric model and the natural frequency simulation model are used to evaluate the target and constraint values ​​of the optimal solution, and the optimization population and key iteration information are updated to determine whether the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements;

[0013] If the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements, the parameter information of the optimized rear subframe of the automobile is output; if the natural frequency and lightweight index corresponding to the optimal solution do not meet the design requirements, the steps of clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model are performed until the design requirements are met.

[0014] Furthermore, in the above-mentioned lightweight design method for the rear subframe of an automobile, the expression of the optimization mathematical optimization model is:

[0015] ;

[0016] in, is the weight of the rear subframe of the vehicle; The thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the car; is the number of sheet metal parts; is the sheet metal material density; is the shape function of the rear subframe of the vehicle; is the interpolation function of the thickness of all sheet metal parts of the rear subframe of the car; is the 7th order natural frequency value obtained through modal analysis; are the upper and lower bounds of the thickness variable of the structural sheet metal.

[0017] Furthermore, in the above-mentioned lightweight design method for the rear subframe of an automobile, the steps of taking the design parameter design domain as the design space, generating an optimized population in the design space by using the maximum and minimum distance sampling criterion, and evaluating the natural frequency values ​​and weights of the individuals in the population according to the natural frequency simulation model and the geometric model to obtain the natural frequency constraints and weight target values ​​respectively include:

[0018] According to the space-filling Latin hypercube sampling, the Individuals are used as candidate pools for the population;

[0019] Calculate the minimum distance between each individual in the candidate population and other individuals, arrange the individuals in descending order according to the minimum distance values ​​of all individuals, and select the top individuals from the arranged set. Individuals form an optimized population;

[0020] For each individual in the optimized population, the thickness value of the corresponding sheet metal part is input into the mesh file of the reference rear subframe, and the mesh files corresponding to all the individuals in the optimized population are imported into the preset tool for modal analysis;

[0021] In the preset software, check the natural frequency value and weight of the rear subframe corresponding to each individual in the optimized population.

[0022] Furthermore, in the above-mentioned lightweight design method for the rear subframe of an automobile, the steps of clustering the currently evaluated individuals by using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of the individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model include:

[0023] For all individuals in each category, the minimum Euclidean distance of each individual from other individuals is calculated. The minimum Euclidean distance of each individual and the target value information are used as two goals. The individuals in each category are sorted using fast non-dominated sorting, and the first-level individuals are stored in the non-dominated individual set.

[0024] Taking the position information of each dimension of the individuals in the non-dominated individual set as a reference, the values ​​of each dimension of the distributed central individual are obtained. The neighboring individuals closest to each central individual are obtained through the kNN model, and the hyper-rectangular area surrounded by the neighboring individuals is taken as the local area corresponding to the central individual.

[0025] Furthermore, in the above-mentioned lightweight design method for the rear subframe of an automobile, the expression of the second-order polynomial radial basis function machine learning model is:

[0026] ;

[0027] ;

[0028] in, A second-order polynomial radial basis function machine learning prediction model built for natural frequencies; For the Individual x i The corresponding weight parameters; Indicates Individual x i The basis functions of Represents the number of dimensions of the low-dimensional space obtained by Sammon mapping; represents the weight parameter of the quadratic polynomial; x j represents the jth individual; A second-order polynomial radial basis function machine learning prediction model was constructed for weight; For the Individual x i The corresponding weight parameters; represents the weight parameter of the quadratic polynomial.

[0029] Furthermore, in the above-mentioned lightweight design method for the rear subframe of an automobile, the step of using the CoDE algorithm to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model in the local area corresponding to each central individual includes:

[0030] The CoDE algorithm is used to solve the local optimization problem of the second-order polynomial radial basis function machine learning model mapping and obtain the optimal solution in low-dimensional space;

[0031] Based on Sammon mapping, find the location information of the optimal solution in the low-dimensional space corresponding to the optimal solution in the original design space;

[0032] Among them, the expression of the local optimization problem of the second-order polynomial radial basis function machine learning model mapping is:

[0033] ;

[0034] in, is the thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the automobile. A second-order polynomial radial basis function machine learning prediction model is constructed for weight. A second-order polynomial radial basis function machine learning prediction model built for natural frequency. For the a central individual; For the The local area corresponding to the central individual.

[0035] Furthermore, in the above-mentioned lightweight design method for the rear subframe of an automobile, the step of using the geometric model and the natural frequency simulation model to evaluate the target and constraint values ​​of the optimal solution and updating the optimization population and key iteration information includes:

[0036] The evaluated optimal solution is merged with the current population, and individuals ranked in preset positions are selected from the merged population based on the statistical ranking method to form an updated optimized population, and the current number of iterations and the current optimal individual are updated.

[0037] Another object of the present invention is to provide a lightweight design system for a rear subframe of an automobile, the system comprising:

[0038] Establishing a module for constructing a natural frequency simulation model of the rear subframe of the automobile according to the geometric model and modal analysis of the rear subframe of the automobile, determining the number of design parameters according to the sheet metal connection distribution of the rear subframe of the automobile, taking the thickness variation range of the sheet metal as the design parameter design domain, and constructing a mathematical optimization model for lightweight design of the rear subframe of the automobile;

[0039] A generation module is used to use the design parameter design domain as a design space, generate an optimized population in the design space using a maximum and minimum distance sampling criterion, and evaluate the natural frequency values ​​and weights of individuals in the population according to the natural frequency simulation model and the geometric model to obtain natural frequency constraints and weight target values ​​respectively;

[0040] An evaluation module is used to cluster the currently evaluated individuals using affine propagation clustering, determine the central individual for distributed local exploration based on the distribution characteristics of each class and the potential information of individuals in the class, and determine the local area corresponding to the central individual of the distributed exploration based on the kNN model;

[0041] A mapping module, used to perform low-dimensional mapping on all evaluated individuals using Sammon mapping, screen out key design parameters of the low-dimensional space, and construct a second-order polynomial radial basis function machine learning model according to the key design parameters;

[0042] An exploration module is used to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model using the CoDE algorithm in the local area corresponding to each central individual;

[0043] A judgment module, used to evaluate the target and constraint values ​​of the optimal solution by using the geometric model and the natural frequency simulation model, and to update the optimization population and key iteration information to determine whether the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements;

[0044] The design module is used for outputting the optimized parameter information of the rear subframe of the automobile if the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements; if the natural frequency and lightweight index corresponding to the optimal solution do not meet the design requirements, executing the steps of clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of the individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model until the design requirements are met.

[0045] Another object of the present invention is to provide a readable storage medium having a computer program stored thereon, wherein the program implements the steps of the above method when executed by a processor.

[0046] Another object of the present invention is to provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0047] The present invention uses a second-order polynomial radial basis function machine learning model to approximate the time-consuming natural frequency simulation model, reduces the number of calls of the time-consuming simulation, accelerates the exploration of distributed local areas under the guidance of the machine learning model, provides accurate guidance for quickly locating the optimal structure, uses Sammon mapping to achieve effective dimensionality reduction in high-dimensional space, accelerates the modeling efficiency of the machine learning model, avoids the possibility of blind evolution in high-dimensional space, and accelerates the overall convergence speed of the algorithm, thereby ensuring that a satisfactory optimized structure is obtained within a limited design cycle. The problem that the efficiency and accuracy cannot be taken into account in the lightweight design of the rear subframe of an automobile in the prior art is solved.

[0048] In addition, the present invention has at least the following beneficial effects:

[0049] 1. The present invention generates an optimized population based on space-filling Latin hypercube sampling and maximum and minimum distance criteria, which ensures the rationality of the individual distribution of the optimized population and fundamentally improves the height of the overall optimization starting point;

[0050] 2. The present invention effectively classifies individuals in the population through affine propagation clustering to avoid excessive waste of search resources. By locating distributed local areas based on the kNN model, the integrity of the local areas is ensured, the overlapping probability of different local areas is reduced, and the efficiency of algorithm optimization is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flowchart of a lightweight design method for a rear subframe of an automobile proposed in a first embodiment of the present invention;

[0052] Figure 2 This is a structural block diagram of a lightweight design system for a rear subframe of an automobile proposed in a second embodiment of the present invention.

[0053] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0054] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0055] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0057] The following will describe in detail how to balance efficiency and accuracy when designing a lightweight rear subframe of an automobile in conjunction with specific embodiments and drawings.

[0058] Embodiment 1

[0059] See also Figure 1 , which shows a lightweight design method for a rear subframe of an automobile in a first embodiment of the present invention, and the method includes steps S10 to S16.

[0060] Step S10, constructing a natural frequency simulation model of the rear subframe of the automobile according to the geometric model and modal analysis of the rear subframe of the automobile, determining the number of design parameters according to the sheet metal connection distribution of the rear subframe of the automobile, taking the thickness variation range of the sheet metal as the design parameter design domain, and constructing a mathematical optimization model for the lightweight design of the rear subframe of the automobile.

[0061] Among them, the thickness variation range of the sheet metal is taken as the design parameter design domain in combination with the assembly requirements, and a mathematical optimization model for lightweight design of the rear subframe is given under the premise that the 7th order natural frequency of the rear subframe in free vibration (without displacement constraint and force load) is not less than 60Hz.

[0062] Specifically, the expression of the mathematical optimization model is:

[0063] ;

[0064] in, is the weight of the rear subframe of the vehicle; The thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the car; is the number of sheet metal parts; is the sheet metal material density; is the shape function of the rear subframe of the vehicle; is the interpolation function of the thickness of all sheet metal parts of the rear subframe of the car; is the 7th order natural frequency value obtained through modal analysis; are the upper and lower bounds of the thickness variable of the structural sheet metal.

[0065] Step S11, taking the design parameter design domain as the design space, generating an optimized population in the design space using the maximum and minimum distance sampling criteria, and evaluating the natural frequency values ​​and weights of the individuals in the population according to the natural frequency simulation model and the geometric model to obtain the natural frequency constraints and weight target values ​​respectively.

[0066] Specifically, individuals are taken as the population candidate library in the design space according to space-filling Latin hypercube sampling;

[0067] Calculate the minimum distance between each individual and other individuals in the candidate population pool, arrange the individuals in descending order according to the minimum distance values ​​of all individuals, and select the individuals in the first few positions from the arranged individual set to form an optimized population;

[0068] For each individual in the optimized population, the thickness value of the corresponding sheet metal part is input into the mesh file of the reference rear subframe, and the mesh files corresponding to all the individuals in the optimized population are imported into the preset tool for modal analysis;

[0069] In the preset software, check the natural frequency value and weight of the rear subframe corresponding to each individual in the optimized population.

[0070] Among them, for each individual in the optimized population, the corresponding sheet metal thickness value is input into the .bdf mesh file of the reference rear subframe; the .bdf mesh files corresponding to all optimized population individuals are imported into MSC NASTRAN for modal analysis; the natural frequency value of the rear subframe corresponding to each individual in the optimized population is viewed in HyperView; and the weight of the rear subframe corresponding to each individual in the optimized population is viewed in HyperMesh.

[0071] Step S12, clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each class and the potential information of individuals in the class, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model.

[0072] Among them, for all individuals in each category, the minimum Euclidean distance of each individual from other individuals is calculated, and the minimum Euclidean distance of each individual and the target value information are used as two goals. The individuals of each category are sorted by fast non-dominated sorting, and the first-level individuals are stored in the non-dominated individual set;

[0073] Taking the position information of each dimension of the individuals in the non-dominated individual set as a reference, the values ​​of each dimension of the distributed central individual are obtained. The neighboring individuals closest to each central individual are obtained through the kNN model, and the hyper-rectangular area surrounded by the neighboring individuals is taken as the local area corresponding to the central individual.

[0074] Step S13, using Sammon mapping to perform low-dimensional mapping on all evaluated individuals, screening out key design parameters of the low-dimensional space, and constructing a second-order polynomial radial basis function machine learning model based on the key design parameters.

[0075] Among them, the expression of the second-order polynomial radial basis function machine learning model is:

[0076] ;

[0077] ;

[0078] in, A second-order polynomial radial basis function machine learning prediction model built for natural frequencies; For the Individual x i The corresponding weight parameters; Indicates Individual x i The basis functions of Represents the number of dimensions of the low-dimensional space obtained by Sammon mapping; represents the weight parameter of the quadratic polynomial; x j represents the jth individual; A second-order polynomial radial basis function machine learning prediction model was constructed for weight; For the Individual x i The corresponding weight parameters; represents the weight parameter of the quadratic polynomial.

[0079] Specifically, in the Sammon mapping, the loss function of mapping all evaluated individuals to the low-dimensional space is minimized according to the gradient optimization operator to obtain the position information of each evaluated individual in the low-dimensional space; the natural frequency data and weight data of all evaluated individuals in the low-dimensional space are used to construct a second-order polynomial radial basis function machine learning model.

[0080] Step S14, in the local area corresponding to each central individual, the CoDE algorithm is used to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model.

[0081] Among them, the expression of the local optimization problem of the second-order polynomial radial basis function machine learning model mapping is:

[0082] ;

[0083] in, is the thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the automobile. A second-order polynomial radial basis function machine learning prediction model is constructed for weight. A second-order polynomial radial basis function machine learning prediction model built for natural frequency. For the a central individual; For the The local area corresponding to the central individual.

[0084] Specifically, the CoDE algorithm is used to solve the local optimization problem of the second-order polynomial radial basis function machine learning model mapping and obtain the optimal solution in low-dimensional space;

[0085] Based on the Sammon mapping, the location information of the optimal solution in the low-dimensional space corresponding to the optimal solution in the original design space is found.

[0086] Step S15, using the geometric model and the natural frequency simulation model to evaluate the target and constraint values ​​of the optimal solution, and updating the optimization population and key iteration information to determine whether the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements.

[0087] Among them, when updating the optimized population and key iteration information, the evaluated optimal solution is merged with the current population, and individuals ranked in preset positions are selected from the merged population based on the statistical ranking method to form the updated optimized population. Specifically, individuals ranked in the top few positions are selected to form the updated optimized population, and the current number of iterations and the current optimal individual are updated to prepare for the next generation.

[0088] Step S16: if the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements, the parameter information of the rear subframe of the vehicle obtained by optimization is output; if the natural frequency and lightweight index corresponding to the optimal solution do not meet the design requirements, the steps of clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of the individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model are performed until the design requirements are met.

[0089] Among them, if the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements, the parameter information of the optimized rear subframe of the vehicle is output; if not, the currently evaluated individuals are clustered using affine propagation clustering, and the central individual for distributed local exploration is determined based on the distribution characteristics of each category and the potential information of individuals in the class. The local area corresponding to the central individual of the distributed exploration is determined based on the kNN model until the design requirements are met.

[0090] In addition, in order to further illustrate the beneficial effects of this embodiment, the data obtained by conducting a design experiment using the lightweight design method for the rear subframe of an automobile proposed in this embodiment is compared with the existing design method. The comparison results are shown in Table 1.

[0091] Table 1

[0092]

[0093] It can be seen that, under the same number of simulation samples, the natural frequency results of all methods are close, but the method of this embodiment obtains the automobile rear subframe structure with the minimum weight.

[0094] In summary, the lightweight design method for the rear subframe of an automobile in the above embodiment of the present invention, by using a second-order polynomial radial basis function machine learning model to approximate the time-consuming natural frequency simulation model, reduces the number of calls for time-consuming simulation, accelerates the exploration of distributed local areas under the guidance of the machine learning model, provides accurate guidance for quickly locating the optimal structure, uses Sammon mapping to achieve effective dimensionality reduction in high-dimensional space, accelerates the modeling efficiency of the machine learning model, avoids the possibility of blind evolution in high-dimensional space, and accelerates the overall convergence speed of the algorithm, thereby ensuring that a satisfactory optimized structure is obtained within a limited design cycle. The problem that efficiency and accuracy cannot be taken into account in the lightweight design of the rear subframe of an automobile in the prior art is solved.

[0095] Embodiment 2

[0096] See also Figure 2 , which shows a lightweight design system for a rear subframe of an automobile proposed in the second embodiment of the present invention, and the system comprises:

[0097] Establishing module 100, for constructing a natural frequency simulation model of the rear subframe of the automobile according to the geometric model and modal analysis of the rear subframe of the automobile, determining the number of design parameters according to the sheet metal connection distribution of the rear subframe of the automobile, taking the thickness variation range of the sheet metal as the design parameter design domain, and constructing a mathematical optimization model for lightweight design of the rear subframe of the automobile;

[0098] A generation module 200 is used to use the design parameter design domain as a design space, generate an optimized population in the design space using a maximum and minimum distance sampling criterion, and evaluate the natural frequency values ​​and weights of individuals in the population according to the natural frequency simulation model and the geometric model to obtain natural frequency constraints and weight target values ​​respectively;

[0099] An evaluation module 300 is used to cluster currently evaluated individuals using affine propagation clustering, determine a central individual for distributed local exploration based on the distribution characteristics of each class and the potential information of individuals in the class, and determine a local area corresponding to the central individual of the distributed exploration based on a kNN model;

[0100] A mapping module 400 is used to perform low-dimensional mapping on all evaluated individuals using Sammon mapping, screen out key design parameters of the low-dimensional space, and construct a second-order polynomial radial basis function machine learning model according to the key design parameters;

[0101] An exploration module 500 is used to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model using the CoDE algorithm in the local area corresponding to each central individual;

[0102] A judgment module 600 is used to evaluate the target and constraint values ​​of the optimal solution by using the geometric model and the natural frequency simulation model, and to update the optimization population and key iteration information to determine whether the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements;

[0103] The design module 700 is used to output the optimized parameter information of the rear subframe of the automobile if the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements. If the natural frequency and lightweight index corresponding to the optimal solution do not meet the design requirements, the steps of clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of the individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model are performed until the design requirements are met.

[0104] Furthermore, in the above-mentioned automobile rear subframe lightweight design system, the expression of the optimization mathematical optimization model is:

[0105] ;

[0106] in, is the weight of the rear subframe of the vehicle; The thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the car; is the number of sheet metal parts; is the sheet metal material density; is the shape function of the rear subframe of the vehicle; is the interpolation function of the thickness of all sheet metal parts of the rear subframe of the car; is the 7th order natural frequency value obtained through modal analysis; are the upper and lower bounds of the thickness variable of the structural sheet metal.

[0107] Further, in some optional embodiments of the present invention, the steps of taking the design parameter design domain as the design space, generating an optimized population in the design space using the maximum and minimum distance sampling criteria, and evaluating the natural frequency values ​​and weights of the individuals in the population according to the natural frequency simulation model and the geometric model to obtain the natural frequency constraints and weight target values ​​respectively include:

[0108] According to the space-filling Latin hypercube sampling, the Individuals are used as candidate pools for the population;

[0109] Calculate the minimum distance between each individual in the candidate population and other individuals, arrange the individuals in descending order according to the minimum distance values ​​of all individuals, and select the top individuals from the arranged set. Individuals form an optimized population;

[0110] For each individual in the optimized population, the thickness value of the corresponding sheet metal part is input into the mesh file of the reference rear subframe, and the mesh files corresponding to all the individuals in the optimized population are imported into the preset tool for modal analysis;

[0111] In the preset software, check the natural frequency value and weight of the rear subframe corresponding to each individual in the optimized population.

[0112] Furthermore, in the above-mentioned automobile rear subframe lightweight design system, the steps of clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of the individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model include:

[0113] For all individuals in each category, the minimum Euclidean distance of each individual from other individuals is calculated. The minimum Euclidean distance of each individual and the target value information are used as two goals. The individuals in each category are sorted using fast non-dominated sorting, and the first-level individuals are stored in the non-dominated individual set.

[0114] Taking the position information of each dimension of the individuals in the non-dominated individual set as a reference, the values ​​of each dimension of the distributed central individual are obtained. The neighboring individuals closest to each central individual are obtained through the kNN model, and the hyper-rectangular area surrounded by the neighboring individuals is taken as the local area corresponding to the central individual.

[0115] Furthermore, in the above-mentioned automobile rear subframe lightweight design system, the expression of the second-order polynomial radial basis function machine learning model is:

[0116] ;

[0117] ;

[0118] in, A second-order polynomial radial basis function machine learning prediction model built for natural frequencies; For the Individual x i The corresponding weight parameters; Indicates Individual x i The basis functions of Represents the number of dimensions of the low-dimensional space obtained by Sammon mapping; represents the weight parameter of the quadratic polynomial; x j represents the jth individual; A second-order polynomial radial basis function machine learning prediction model was constructed for weight; For the Individual x i The corresponding weight parameters; represents the weight parameter of the quadratic polynomial.

[0119] Furthermore, in the above-mentioned automobile rear subframe lightweight design system, the step of using the CoDE algorithm to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model in the local area corresponding to each central individual includes:

[0120] The CoDE algorithm is used to solve the local optimization problem of the second-order polynomial radial basis function machine learning model mapping and obtain the optimal solution in low-dimensional space;

[0121] Based on Sammon mapping, find the location information of the optimal solution in the low-dimensional space corresponding to the optimal solution in the original design space;

[0122] Among them, the expression of the local optimization problem of the second-order polynomial radial basis function machine learning model mapping is:

[0123] ;

[0124] in, is the thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the automobile. A second-order polynomial radial basis function machine learning prediction model is constructed for weight. A second-order polynomial radial basis function machine learning prediction model built for natural frequency. For the a central individual; For the The local area corresponding to the central individual.

[0125] Furthermore, in the above-mentioned automobile rear subframe lightweight design system, the step of using the geometric model and the natural frequency simulation model to evaluate the target and constraint values ​​of the optimal solution and updating the optimization population and key iteration information includes:

[0126] The evaluated optimal solution is merged with the current population, and individuals ranked in preset positions are selected from the merged population based on the statistical ranking method to form an updated optimized population, and the current number of iterations and the current optimal individual are updated.

[0127] The functions or operation steps implemented when the above modules are executed are substantially the same as those in the above method embodiments, and will not be repeated here.

[0128] Embodiment 3

[0129] Another aspect of the present invention further provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of the method described in the first embodiment above.

[0130] Embodiment 4

[0131] Another aspect of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in the first embodiment when executing the program.

[0132] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0133] Those skilled in the art will appreciate that the logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be specifically implemented in any computer-readable storage medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For purposes of this specification, a "computer-readable storage medium" may be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0134] More specific examples (a non-exhaustive list) of computer-readable storage media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable storage medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0135] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or a combination thereof: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0136] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0137] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A lightweight design method for a rear subframe of an automobile, characterized in that: The method comprises: According to the geometric model and modal analysis of the rear subframe of the automobile, a natural frequency simulation model of the rear subframe of the automobile is constructed, and according to the sheet metal connection distribution of the rear subframe of the automobile, the number of design parameters is determined, and the thickness variation range of the sheet metal is used as the design parameter design domain to construct a mathematical optimization model for the lightweight design of the rear subframe of the automobile; Taking the design parameter design domain as the design space, generating an optimized population in the design space by using the maximum and minimum distance sampling criteria, and evaluating the natural frequency values ​​and weights of the individuals in the population according to the natural frequency simulation model and the geometric model to obtain the natural frequency constraints and weight target values ​​respectively; Affine propagation clustering is used to cluster the currently evaluated individuals. The central individuals used for distributed local exploration are determined based on the distribution characteristics of each category and the potential information of individuals in the category. The local area corresponding to the central individual of distributed exploration is determined based on the kNN model. Sammon mapping is used to perform low-dimensional mapping on all evaluated individuals, and key design parameters of the low-dimensional space are screened and obtained, and a second-order polynomial radial basis function machine learning model is constructed according to the key design parameters; In the local area corresponding to each central individual, the CoDE algorithm is used to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model; The geometric model and the natural frequency simulation model are used to evaluate the target and constraint values ​​of the optimal solution, and the optimization population and key iteration information are updated to determine whether the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements; If the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements, the parameter information of the rear subframe of the vehicle obtained by optimization is output; if the natural frequency and lightweight index corresponding to the optimal solution do not meet the design requirements, the steps of clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of the individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model are performed until the design requirements are met; The expression of the mathematical optimization model is: ； in, is the weight of the rear subframe of the vehicle; The thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the car; is the number of sheet metal parts; is the sheet metal material density; is the shape function of the rear subframe of the vehicle; is the interpolation function of the thickness of all sheet metal parts of the rear subframe of the car; is the 7th order natural frequency value obtained through modal analysis; are the upper and lower bounds of the thickness variable of the structural sheet metal; The steps of taking the design parameter design domain as the design space, generating an optimized population in the design space by using the maximum and minimum distance sampling criteria, and evaluating the natural frequency values ​​and weights of the individuals in the population according to the natural frequency simulation model and the geometric model to obtain the natural frequency constraints and weight target values ​​respectively include: According to the space-filling Latin hypercube sampling, the Individuals are used as candidate pools for the population; Calculate the minimum distance between each individual in the candidate population and other individuals, arrange the individuals in descending order according to the minimum distance values ​​of all individuals, and select the top individuals from the arranged set. Individuals form an optimized population; For each individual in the optimized population, the thickness value of the corresponding sheet metal part is input into the mesh file of the reference rear subframe, and the mesh files corresponding to all the individuals in the optimized population are imported into the preset tool for modal analysis; Check the natural frequency value and weight of the rear subframe corresponding to each individual in the optimized population in the preset software; The steps of clustering the currently evaluated individuals by using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each class and the potential information of the individuals in the class, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model include: For all individuals in each category, the minimum Euclidean distance of each individual from other individuals is calculated. The minimum Euclidean distance of each individual and the target value information are used as two goals. The individuals in each category are sorted using fast non-dominated sorting, and the first-level individuals are stored in the non-dominated individual set. Taking the position information of each dimension of the non-dominated individual set as a reference, the values ​​of each dimension of the distributed central individual are obtained. The nearest neighboring individuals to each central individual are obtained through the kNN model, and the hyper-rectangular area surrounded by the neighboring individuals is taken as the local area corresponding to the central individual. The expression of the second-order polynomial radial basis function machine learning model is: ； ； in, A second-order polynomial radial basis function machine learning prediction model built for natural frequencies; For the Individual x i The corresponding weight parameters; Indicates Individual x i The basis functions of Represents the number of dimensions of the low-dimensional space obtained by Sammon mapping; represents the weight parameter of the quadratic polynomial; x j represents the jth individual; A second-order polynomial radial basis function machine learning prediction model was constructed for weight; For the Individual x i The corresponding weight parameters; represents the weight parameter of the quadratic polynomial; The steps of using the CoDE algorithm to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model in the local area corresponding to each central individual include: The CoDE algorithm is used to solve the local optimization problem of the second-order polynomial radial basis function machine learning model mapping and obtain the optimal solution in low-dimensional space; Based on Sammon mapping, find the location information of the optimal solution in the low-dimensional space corresponding to the optimal solution in the original design space; Among them, the expression of the local optimization problem of the second-order polynomial radial basis function machine learning model mapping is: ； in, is the thickness variable of the sheet metal parts in the lightweight design of the rear subframe of the automobile. A second-order polynomial radial basis function machine learning prediction model is constructed for weight. A second-order polynomial radial basis function machine learning prediction model built for natural frequency. For the a central individual; For the The local area corresponding to the central individual.

2. The lightweight design method for the rear subframe of an automobile according to claim 1, characterized in that: The step of using the geometric model and the natural frequency simulation model to evaluate the target and constraint values ​​of the optimal solution and updating the optimization population and key iteration information includes: The evaluated optimal solution is merged with the current population, and individuals ranked in preset positions are selected from the merged population based on the statistical ranking method to form an updated optimized population, and the current number of iterations and the current optimal individual are updated.

3. A lightweight design system for a rear subframe of an automobile, characterized in that: The system is used to implement the lightweight design method of the rear subframe of an automobile according to any one of claims 1 to 2, and comprises: Establishing a module for constructing a natural frequency simulation model of the rear subframe of the automobile according to the geometric model and modal analysis of the rear subframe of the automobile, determining the number of design parameters according to the sheet metal connection distribution of the rear subframe of the automobile, taking the thickness variation range of the sheet metal as the design parameter design domain, and constructing a mathematical optimization model for lightweight design of the rear subframe of the automobile; A generation module is used to use the design parameter design domain as a design space, generate an optimized population in the design space using a maximum and minimum distance sampling criterion, and evaluate the natural frequency values ​​and weights of individuals in the population according to the natural frequency simulation model and the geometric model to obtain natural frequency constraints and weight target values ​​respectively; An evaluation module is used to cluster the currently evaluated individuals using affine propagation clustering, determine the central individual for distributed local exploration based on the distribution characteristics of each class and the potential information of individuals in the class, and determine the local area corresponding to the central individual of the distributed exploration based on the kNN model; A mapping module, used to perform low-dimensional mapping on all evaluated individuals using Sammon mapping, screen out key design parameters of the low-dimensional space, and construct a second-order polynomial radial basis function machine learning model according to the key design parameters; An exploration module is used to explore the optimal solution of the local optimization problem mapped by the second-order polynomial radial basis function machine learning model using the CoDE algorithm in the local area corresponding to each central individual; A judgment module, used to evaluate the target and constraint values ​​of the optimal solution by using the geometric model and the natural frequency simulation model, and to update the optimization population and key iteration information to determine whether the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements; The design module is used for outputting the optimized parameter information of the rear subframe of the automobile if the natural frequency and lightweight index corresponding to the optimal solution meet the design requirements; if the natural frequency and lightweight index corresponding to the optimal solution do not meet the design requirements, executing the steps of clustering the currently evaluated individuals using affine propagation clustering, determining the central individual for distributed local exploration based on the distribution characteristics of each category and the potential information of the individuals in the category, and determining the local area corresponding to the central individual of the distributed exploration based on the kNN model until the design requirements are met.

4. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are implemented.

5. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 2 when executing the program.

Images