A reliability optimization design method and system for lightweight structural products

By combining training samples based on finite element model and SVM model and combining genetic algorithm optimization, the problem of low reliability design efficiency of traditional structures is solved, and efficient reliability optimization design of lightweight structures is achieved, and design analysis accuracy is improved. efficiency.

CN115146543BActive Publication Date: 2024-11-12CHINESE PEOPLES LIBERATION ARMY UNIT 32181
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210893148.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-27
Publication Date
2024-11-12
Estimated Expiration
2042-07-27

AI Technical Summary

Technical Problem

The traditional structural reliability design analysis method is inefficient and difficult to meet the current demand for rapid iteration of product design, and has not yet formed an effective lightweight structural reliability optimization design method.

Method used

The training samples are determined by a finite element model based on structural product parameterization, a support vector machine (SVM) model is established, and a genetic algorithm and dual-cycle optimization method are used to solve the reliability-based design optimization model and determine the optimal design results.

Benefits of technology

The efficiency and accuracy of structural reliability design analysis and optimization are improved, and the equipment structure reliability design optimization problem with implicit state equations and extremely small failure probability can be solved while reducing the number of finite element model calls.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115146543B_ABST
    Figure CN115146543B_ABST
Patent Text Reader

Abstract

The present invention relates to a reliability optimization design method and system for lightweight structural products, and belongs to the field of structural design. The method comprises: determining training samples based on a parameterized finite element model of the structural product; establishing an SVM model; training the SVM model through the training samples to obtain a trained SVM model; calculating the maximum VonMises stress of the structure using the trained SVM model; establishing a reliability-based design optimization model according to the maximum VonMises stress of the structure; using a genetic algorithm and a double-loop optimization method to solve the reliability-based design optimization model and determine the optimal design result. The present invention can solve the reliability design optimization problem of equipment structures containing implicit state equations and minimal failure probabilities when the original finite element model is called less times, thereby improving the efficiency and accuracy of structural reliability design analysis and optimization.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of structural design, and in particular to a reliability optimization design method and system for lightweight structural products. Background Art

[0002] As the functional performance indicators of equipment products become higher and higher, the use environment becomes more and more harsh, and the functional structure becomes more and more complex, the equipment structure design becomes more difficult. The lightweight requirements of structural products and the high reliability constraints have become the main contradictions in structural design. The traditional structural reliability analysis method based on Monte Carlo simulation and manual design improvement are inefficient and cannot meet the needs of rapid iteration of current product design. At present, there is no effective lightweight structure reliability optimization design method. Summary of the invention

[0003] The purpose of the present invention is to provide a reliability optimization design method and system for lightweight structural products to solve the problem of low efficiency of traditional structural reliability design and analysis methods.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A reliability optimization design method for lightweight structural products, comprising:

[0006] Determine the training samples based on the parameterized finite element model of the structural product;

[0007] Build the SVM model;

[0008] Training the SVM model using the training samples to obtain a trained SVM model;

[0009] Calculating the VonMises maximum stress of the structure using the trained SVM model;

[0010] According to the VonMises maximum stress of the structure, a reliability-based design optimization model is established;

[0011] The reliability-based design optimization model is solved by using a genetic algorithm and a double-loop optimization method to determine the optimal design result.

[0012] Optionally, determining the training samples based on the parameterized finite element model of the structural product specifically includes:

[0013] Determining the most likely failure point of the finite element model using a first order second moment method;

[0014] Taking the most likely failure point as the mean and σ as the standard deviation, random sampling is performed to obtain sampling points;

[0015] The sampling sample points are used as training samples.

[0016] Optionally, the taking the sampling sample points as training samples further includes:

[0017] The training samples are normalized to obtain normalized training samples.

[0018] Optionally, the training the SVM model to obtain a trained SVM model specifically includes:

[0019] Selecting parameters of the SVM model; the parameters include the order of the polynomial kernel function;

[0020] Using the normalized training samples to train the SVM model to obtain a trained SVM model;

[0021] Determine whether the analysis accuracy of the trained SVM model meets the reliability analysis accuracy, and obtain a determination result;

[0022] If the judgment result is that the analysis accuracy of the trained SVM model meets the reliability analysis accuracy, then outputting the trained SVM model;

[0023] If the judgment result is that the analysis accuracy of the trained SVM model does not meet the reliability analysis accuracy, then return to the step of "selecting the parameters of the SVM model".

[0024] A reliability optimization design system for lightweight structural products, comprising:

[0025] A training sample determination module, used for determining training samples based on a parameterized finite element model of a structural product;

[0026] SVM model building module, used to build SVM model;

[0027] A model training module, used to train the SVM model through the training samples to obtain a trained SVM model;

[0028] A maximum stress calculation module, used to calculate the Von Mises maximum stress of the structure using the trained SVM model;

[0029] A design optimization model establishment module, used to establish a reliability-based design optimization model according to the VonMises maximum stress of the structure;

[0030] The result determination module is used to solve the reliability-based design optimization model and determine the optimal design result by using a genetic algorithm and a double-loop optimization method.

[0031] Optionally, the training sample determination module includes:

[0032] A most likely failure point determination unit, used to determine the most likely failure point of the finite element model using a first-order second moment method;

[0033] A sampling unit, used to perform random sampling with the most likely failure point as the mean and σ as the standard deviation to obtain sampling sample points;

[0034] The training sample determination unit is used to use the sampling sample points as training samples.

[0035] Optionally, the training sample determination module further includes:

[0036] The normalization unit is used to perform normalization processing on the training samples to obtain normalized training samples.

[0037] Optionally, the model training module includes:

[0038] A parameter selection unit, used to select parameters of the SVM model; the parameters include the order of the polynomial kernel function;

[0039] A training unit, used to train the SVM model using the normalized training samples to obtain a trained SVM model;

[0040] A judgment unit, used to judge whether the analysis accuracy of the trained SVM model meets the reliability analysis accuracy, and obtain a judgment result;

[0041] A model output unit, configured to output the trained SVM model if the judgment result is that the analysis accuracy of the trained SVM model meets the reliability analysis accuracy;

[0042] The return unit is used to return to the "parameter selection unit" if the judgment result is that the analysis accuracy of the trained SVM model does not meet the reliability analysis accuracy.

[0043] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0044] The present invention trains the established SVM model by training samples constructed based on the parameterized finite element model of the structural product to obtain a trained SVM model, and uses the trained SVM model to replace the finite element model to carry out reliability analysis and optimization to obtain the optimal design result. The present invention can solve the equipment structure reliability design optimization problem containing implicit state equations and minimal failure probability when the original finite element model is called less times, thereby improving the efficiency and accuracy of structural reliability design analysis and optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0046] Figure 1 A flowchart of the implementation of a reliability optimization design method for a lightweight structural product provided by the present invention;

[0047] Figure 2 A flow chart of a reliability optimization design method for a lightweight structural product provided by the present invention;

[0048] Figure 3 A flowchart of a specific implementation process of a reliability optimization design method for a lightweight structural product provided by the present invention;

[0049] Figure 4 This is a structural diagram of a reliability optimization design system for a lightweight structural product provided by the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] The purpose of the present invention is to provide a reliability optimization design method and system for lightweight structural products to solve the problem of low efficiency of traditional structural reliability design and analysis methods.

[0052] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0053] Aiming at the problem that it is difficult to carry out structural reliability design optimization efficiently due to the contradiction between the lightweight requirement of the final product and the high reliability constraint, the present invention proposes a structural reliability optimization design method. Firstly, a parametric finite element model of the structural product is used, and the training sample planning technology is adopted to reasonably plan the sampling area and efficiently construct the SVM agent model; secondly, a mathematical model of structural reliability optimization design is established; finally, a genetic algorithm is used to solve the mathematical problem, and an optimal structural design scheme based on reliability is given, such as Figure 1 As shown in the figure, it effectively reduces the number of calls to the computationally complex finite element model, efficiently builds the SVM proxy model, and accurately solves the structural reliability design optimization problem.

[0054] The present invention first establishes a parametric finite element model of a structural product, and then establishes an SVM model based on the model, and then uses the SVM model to carry out reliability optimization design. Figure 2 A flow chart of a reliability optimization design method for a lightweight structural product provided by the present invention, Figure 3 A flowchart of a specific implementation process of a reliability optimization design method for a lightweight structural product provided by the present invention, such as Figure 2 and Figure 3 As shown, the method includes:

[0055] Step 101: Determine training samples based on the parameterized finite element model of the structural product. In practical applications, first establish a parameterized finite element model of the structural product, use the structural strength S as the failure criterion, and clarify the strength failure threshold Random variables such as product geometry and material properties follow normal distribution m~N(μ,σ 2 ).

[0056] Furthermore, the step 101 specifically includes:

[0057] The most probable failure point (MPP) of the finite element model is determined by using a first-order second moment method, and Monte Carlo random sampling is performed with the most probable failure point as the mean and σ as the standard deviation to obtain sampling points, which are used as training samples.

[0058] Step 102: Establish an SVM model. In practical applications, the SVM model is expressed as Represents, where m represents the vector of structural random variables; q represents the vector of structural state parameters; y represents the maximum value of the Von Mises stress of the structure; Indicates the trained SVM model. In the process of establishing the SVM model, the polynomial kernel function of the SVM model is selected, and its form is as follows:

[0059] K(x i ,x)=(x i ·x+1) b ,b=1,2,...,n

[0060] In the formula, x i is the i-th training sample; x is the training sample set; b is the order of the polynomial kernel function. The parameters that need to be determined mainly include the order b of the polynomial kernel function, which is generally 2 according to engineering experience.

[0061] Furthermore, the method of using the sampling sample points as training samples further includes:

[0062] The training samples are normalized to obtain normalized training samples. In practical applications, the training samples are normalized to map the characteristic values ​​of each dimension of the sample data into the same interval;

[0063]

[0064] Among them, x k is the kth training sample; μ is the mean vector of the random variable; σ is the standard deviation vector of the random variable; x k ' is the normalized training sample.

[0065] As a new machine learning method, Support Vector Machines (SVM) has good small sample learning ability and good generalization. Using fewer samples to establish an SVM classification model instead of the real function can effectively solve the structural reliability analysis problem containing implicit or highly nonlinear function.

[0066] Step 103: Train the SVM model using the training samples to obtain a trained SVM model.

[0067] Furthermore, the step 103 specifically includes:

[0068] Parameters of the SVM model are selected; the parameters include the order of the polynomial kernel function.

[0069] The SVM model is trained using the normalized training samples to obtain a trained SVM model.

[0070] Determine whether the analysis accuracy of the trained SVM model meets the reliability analysis accuracy. If so, output the trained SVM model; if not, return to the step of "selecting the parameters of the SVM model".

[0071] Step 104: Calculate the maximum Von Mises stress of the structure using the trained SVM model. In practical applications, the successfully trained SVM model can replace the original finite element model (parameterized finite element model of the structural product) to calculate the maximum Von Mises stress of the structure.

[0072] Step 105: Establish a reliability-based design optimization model according to the VonMises maximum stress of the structure.

[0073] In practical applications, a mathematical model of structural reliability design optimization problem (reliability-based design optimization model) is established to carry out reliability optimization and design of structural products. The optimization goal of this model is to minimize the structural mass, and the optimization constraint is to meet the strength reliability requirements. The established reliability-based design optimization model is as follows:

[0074] min f (x,d)

[0075]

[0076] d L ≤d≤d U

[0077] Among them, d is the structural design variable vector; m is the random variable vector, including the use load factor, etc.; is the maximum Von Mises stress of the structure calculated using the SVM model; is the strength failure threshold; is the specified structural reliability requirement; d L Represents the vector composed of the lower limits of the structural design variables; d U Represents the vector composed of the upper limits of the structural design variables; M f Indicates the total mass of the structural product.

[0078] Step 106: Using genetic algorithm and double loop optimization method, solve the reliability-based design optimization model and determine the optimal design result. In practical applications, genetic algorithm and double loop optimization method are used to solve the reliability-based design optimization model. The inner loop uses Monte Carlo method to calculate the structural reliability of the current design point, and the outer loop uses genetic algorithm to find the optimal design point that meets the reliability constraint.

[0079] The present invention proposes a structural reliability analysis method, which uses a parametric finite element model of a structural product, adopts training sample planning technology, rationally plans the sampling area, and efficiently constructs an SVM model. Under the premise of ensuring the accuracy of reliability analysis, the efficiency of reliability analysis is improved. In addition, a lightweight structural reliability optimization design framework is established, a reliability-based design optimization model is established based on the SVM model, and a genetic algorithm is used to solve the optimization model, providing technical support for structural reliability optimization design.

[0080] The present invention proposes a lightweight structure reliability optimization design method, which uses training sample planning technology to rationally plan the sampling area, establish an SVM proxy model, and carry out structural product reliability design optimization on this basis. It has been verified that this method can solve the reliability design optimization problem of equipment structures containing implicit state equations and extremely small failure probabilities when the original finite element model is called less times, providing technical support for improving the accuracy and efficiency of structural reliability optimization design.

[0081] Figure 4 A structural diagram of a reliability optimization design system for a lightweight structural product provided by the present invention, such as Figure 4 As shown, the system includes:

[0082] The training sample determination module 401 is used to determine the training sample based on the parameterized finite element model of the structural product.

[0083] The SVM model building module 402 is used to build an SVM model.

[0084] The model training module 403 is used to train the SVM model using the training samples to obtain a trained SVM model.

[0085] The maximum stress calculation module 404 is used to calculate the Von Mises maximum stress of the structure using the trained SVM model.

[0086] The design optimization model establishment module 405 is used to establish a reliability-based design optimization model according to the Von Mises maximum stress of the structure.

[0087] The result determination module 406 is used to solve the reliability-based design optimization model and determine the optimal design result by using a genetic algorithm and a double-loop optimization method.

[0088] Furthermore, the training sample determination module 401 includes:

[0089] The most likely failure point determination unit is used to determine the most likely failure point of the finite element model using a first-order second moment method.

[0090] The sampling unit is used to perform random sampling with the most likely failure point as the mean and σ as the standard deviation to obtain sampling sample points.

[0091] The training sample determination unit is used to use the sampling sample points as training samples.

[0092] Furthermore, the training sample determination module 401 further includes:

[0093] The normalization unit is used to perform normalization processing on the training samples to obtain normalized training samples.

[0094] Furthermore, the model training module 403 includes:

[0095] A parameter selection unit is used to select parameters of the SVM model; the parameters include the order of the polynomial kernel function.

[0096] The training unit is used to train the SVM model using the normalized training samples to obtain a trained SVM model.

[0097] The judgment unit is used to judge whether the analysis accuracy of the trained SVM model meets the reliability analysis accuracy, and obtain a judgment result.

[0098] The model output unit is used to output the trained SVM model if the judgment result is that the analysis accuracy of the trained SVM model meets the reliability analysis accuracy.

[0099] The return unit is used to return to the "parameter selection unit" if the judgment result is that the analysis accuracy of the trained SVM model does not meet the reliability analysis accuracy.

[0100] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0101] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A reliability optimization design method for lightweight structural products, characterized in that: include: The training samples are determined based on the parameterized finite element model of the structural product, including: Establish parametric finite element model of structural products to determine structural strength S As the failure criterion, set the strength failure threshold , the random variable follows a normal distribution ; Random variables include product geometry and material properties; The most likely failure point of the finite element model is determined by using the first-order second moment method, and the most likely failure point is taken as the mean and the σ is the standard deviation, and Monte Carlo random sampling is performed to obtain sampling points, and the sampling points are used as training samples; Establish SVM model; SVM model uses formula It indicates that, m A vector representing structural random variables; q A vector representing the structural state parameters; y Indicates the maximum value of the Von Mises stress of the structure; Indicates the trained SVM model; in the process of establishing the SVM model, the polynomial kernel function of the SVM model is selected, and the form of the polynomial kernel function is: ; In the formula, x i For the i training samples; x is the training sample set; b is the order of the polynomial kernel function, b =2; The SVM model is trained by the training samples to obtain a trained SVM model, which specifically includes: Selecting parameters of the SVM model; the parameters include the order of the polynomial kernel function; Using the normalized training samples to train the SVM model to obtain a trained SVM model; Determine whether the analysis accuracy of the trained SVM model meets the reliability analysis accuracy. If so, output the trained SVM model; if not, return to the step of "selecting the parameters of the SVM model"; The trained SVM model is used to replace the parameterized finite element model of the structural product to calculate the maximum Von Mises stress of the structure; According to the maximum Von Mises stress of the structure, a reliability-based design optimization model is established; the optimization goal of the reliability-based design optimization model is to minimize the structural mass, and the optimization constraint is to meet the strength reliability requirements; the established reliability-based design optimization model is as follows: ; in, d Design variable vector for the structure; is the maximum VonMises stress of the structure calculated using the trained SVM model; To specify the structural reliability requirements; d L Represents the vector composed of the lower limits of the structural design variables; d U A vector representing the upper limit of the structural design variables; M f Indicates the total mass of the structural product; Genetic algorithm and double loop optimization method are used to solve the reliability-based design optimization model and determine the optimal design result; the inner loop uses Monte Carlo method to calculate the structural reliability of the current design point, and the outer loop uses genetic algorithm to find the optimal design point that meets the reliability constraint.

2. The reliability optimization design method for lightweight structural products according to claim 1 is characterized in that: The method further comprises: taking the sampling sample points as training samples; and then: The training samples are normalized to obtain normalized training samples.

3. A reliability optimization design system for lightweight structural products, characterized in that: include: The training sample determination module is used to determine the training samples based on the parameterized finite element model of the structural product, specifically including: Establish parametric finite element model of structural products to determine structural strength S As the failure criterion, set the strength failure threshold , the random variable follows a normal distribution ; Random variables include product geometry and material properties; The most likely failure point of the finite element model is determined by using the first-order second moment method, and the most likely failure point is taken as the mean and the σ is the standard deviation, and Monte Carlo random sampling is performed to obtain sampling points, and the sampling points are used as training samples; SVM model building module, used to build SVM model; SVM model uses formula It indicates that, m A vector representing structural random variables; q A vector representing the structural state parameters; y Indicates the maximum value of the Von Mises stress of the structure; Indicates the trained SVM model; in the process of establishing the SVM model, the polynomial kernel function of the SVM model is selected, and the form of the polynomial kernel function is: ; In the formula, x i For the i training samples; x is the training sample set; b is the order of the polynomial kernel function, b =2; The model training module is used to train the SVM model through the training samples to obtain a trained SVM model, specifically including: Selecting parameters of the SVM model; the parameters include the order of the polynomial kernel function; Using the normalized training samples to train the SVM model to obtain a trained SVM model; Determine whether the analysis accuracy of the trained SVM model meets the reliability analysis accuracy. If so, output the trained SVM model; if not, return to the step of "selecting the parameters of the SVM model"; A maximum stress calculation module, used to use the trained SVM model to replace the parameterized finite element model of the structural product to calculate the Von Mises maximum stress of the structure; The design optimization model establishment module is used to establish a reliability-based design optimization model according to the Von Mises maximum stress of the structure; the optimization goal of the reliability-based design optimization model is to minimize the structural mass, and the optimization constraint is to meet the strength reliability requirements; the established reliability-based design optimization model is as follows: ; in, d Design variable vector for the structure; is the maximum VonMises stress of the structure calculated using the trained SVM model; To specify the structural reliability requirements; d L Represents the vector composed of the lower limits of the structural design variables; d U A vector representing the upper limit of the structural design variables; M f Indicates the total mass of the structural product; The result determination module is used to solve the reliability-based design optimization model and determine the optimal design result by using a genetic algorithm and a double-loop optimization method; the inner loop uses the Monte Carlo method to calculate the structural reliability of the current design point, and the outer loop uses a genetic algorithm to find the optimal design point that meets the reliability constraint.

4. The reliability optimization design system for lightweight structural products according to claim 3 is characterized in that: The training sample determination module further includes: The normalization unit is used to perform normalization processing on the training samples to obtain normalized training samples.