CN112464541A - Mixed composite material layering method considering multi-scale uncertainty - Google Patents
Mixed composite material layering method considering multi-scale uncertainty Download PDFInfo
- Publication number
- CN112464541A CN112464541A CN202011504349.0A CN202011504349A CN112464541A CN 112464541 A CN112464541 A CN 112464541A CN 202011504349 A CN202011504349 A CN 202011504349A CN 112464541 A CN112464541 A CN 112464541A
- Authority
- CN
- China
- Prior art keywords
- representing
- candidate
- theta
- layer
- parameter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000002131 composite material Substances 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 30
- 239000000463 material Substances 0.000 claims abstract description 218
- 238000009826 distribution Methods 0.000 claims abstract description 76
- 238000003062 neural network model Methods 0.000 claims abstract description 39
- 238000005457 optimization Methods 0.000 claims abstract description 21
- 238000004458 analytical method Methods 0.000 claims abstract description 12
- 230000002068 genetic effect Effects 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 8
- 238000011002 quantification Methods 0.000 claims description 8
- 238000007476 Maximum Likelihood Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 4
- 230000004931 aggregating effect Effects 0.000 claims description 2
- 230000000644 propagated effect Effects 0.000 claims description 2
- 239000000126 substance Substances 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 abstract description 3
- 238000013461 design Methods 0.000 description 3
- 239000000835 fiber Substances 0.000 description 3
- 239000003733 fiber-reinforced composite Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 229930091051 Arenine Natural products 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/26—Composites
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Geometry (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Computer Hardware Design (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Physiology (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a mixed composite material layering method considering multi-scale uncertainty. Selecting an alternative material set, obtaining the microscopic properties of the alternative material set, and obtaining the distribution type of the macroscopic properties of the alternative material set through a multi-scale analysis model; determining a macroscopic finite element model of the candidate material set by using the macroscopic properties and the distribution types thereof; constructing a neural network model of microscopic attributes of the candidate material set and mixed composite material parameters, calculating a Mean Square Error (MSE) of the trained neural network model, judging the accuracy of the trained neural network model according to the MSE, and optimizing the selection of the candidate material set and the selection of the layer stacking sequence by using a genetic algorithm. The method of the invention uses various optimization algorithms, greatly reduces the burden of designers and computers, greatly improves the calculation speed and the optimization accuracy, and simultaneously better conforms to the actual engineering application.
Description
Technical Field
The invention belongs to the field of methods for optimizing stacking sequence and material distribution of fiber reinforced composite materials, and particularly relates to a multi-scale uncertain layering optimization method for a mixed composite material.
Background
Fiber reinforced composites have the advantages of high strength to weight ratio, high stiffness to weight ratio and low cost under various loads and are growing exponentially in automotive, marine, civil, aerospace, and construction applications. In contrast to conventional composites consisting of single fibers and a single matrix, hybrid composites are manufactured by combining two or more fibers into one matrix. The properties of the hybrid composite are a weighted sum of the individual components, and the advantage of one fiber can compensate for the lack of properties of the other types of components in the hybrid composite. Meanwhile, aiming at multi-scale analysis of the composite material, certain research results are provided at home and abroad, and the stability and reliability of the output result of micro-scale analysis can be realized. In addition, in actual engineering application, the multi-scale analysis is more consistent with the practical application of the composite material, and the analysis result is more matched with the actual use condition.
The specified performance parameters are optimized by carrying out proper multi-scale uncertainty engineering optimization design on the mixed composite material, so that the actual application range of the mixed composite material can be better expanded, and the application prospect of the mixed composite material is expanded.
Disclosure of Invention
The invention aims to solve the problem that the design process of a mixed composite material is complex and tedious, and provides a simple multi-scale uncertain mixed composite material layering arrangement method starting from stacking sequence and material distribution.
The technical scheme of the invention is as follows:
1) selection of multiple different composite materials as a candidate set of hybrid composite materialsWherein the content of the first and second substances,a first alternative material is indicated which,a second alternative material is indicated which,a third alternative material is shown which,representing the nth candidate material, n representing the sum of the candidate materials, and collecting the candidate materialsThe microscopic attributes are subjected to uncertainty quantification, and an alternative material set is obtained after the uncertainty quantificationWherein the set of candidate materials isIncluding microscopic elastic properties and micro-elasticity propertiesAn apparent intensity attribute;
2) set of candidate materials obtained based on step 1)By the type of distribution of the microscopic properties of the candidate material setThe multi-scale analysis model of (2) aggregating the candidate materialsThe uncertainty of the microscopic properties is propagated to a macroscopic level to obtain a set of alternative materialsThe distribution type of the macroscopic elastic property and the macroscopic strength property;
3) obtaining a collection of candidate materialsAfter the distribution types of the macroscopic elastic property and the macroscopic strength property are obtained, a candidate material set is determined by utilizing the macroscopic elastic property, the macroscopic strength property and the distribution types thereofFor a set of candidate materialsAnd set of candidate materialsLayer stacking sequence (theta) of the macroscopic finite element model of (1)1,θ2,θ3、、、θx) After Latin hypercube sampling, alternative materials are gathered by a laminated plate methodThe macroscopic finite element model of (1) is analyzed, and after the analysis, the macroscopic finite element model of (1) is obtainedSet of selected materialsPerformance parameter ofA failure parameter FI and an overall cost parameter C; wherein, theta1Representing the stacking angle of the first layer of material in the hybrid composite; theta2Representing the stacking angle of the second layer material in the hybrid composite; theta3Representing the stacking angle of the third layer of material in the hybrid composite; thetaxRepresents the stacking angle of the x-th layer material in the hybrid composite material, x represents the sum of the stacking sequence of the layers;
4) building a collection of candidate materialsThe constructed neural network model comprises an input layer, an intermediate layer and an output layer, wherein the input layer inputs a required candidate material setRequired set of candidate materialsThe microscopic elastic properties, the microscopic strength properties, and the layer stack sequence (θ) of each candidate material1,θ2,θ3、、、θx),A first alternative material is indicated which,indicating that a second material alternative is desired,indicating that a third alternative material is required,representing the m-th required candidate material, wherein m represents the sum of the required candidate materials; output layer output performance parametersA failure parameter FI and an overall cost parameter C;
5) calculating Mean Square Error (MSE) of the trained neural network model;
6) judging the accuracy of the trained neural network model according to the Mean Square Error (MSE);
if the obtained mean square error MSE is smaller than a preset mean square error threshold value, the constructed neural network model is indicated to be accurate, and the step 7) is carried out;
if the mean square error MSE is larger than or equal to the mean square error threshold, the constructed neural network model is not accurate, parameter learning is carried out on the neural network model based on the performance prediction variance, and then the step 5 is returned to);
7) set of candidate materials using Genetic Algorithm (GA)Selection and layer stacking sequence (theta)1,θ2,θ3、、、θx) And carrying out optimization processing on the selection to obtain the material selection of each layer and the stacking angle of each layer in the mixed composite material laying.
In the step 1), uncertainty quantification is performed on microscopic properties of the candidate material set, specifically:
first, the candidate materials are gatheredThe micro elastic property and the micro strength property are regarded as different sparse variables y, and different distribution types theta are sampled and distributed in the sparse variables ykProcessing statistics under different distribution parameters epsilon to obtain a probability density function f of a sparse variable y represented by the following formulay(y|ε,θk):
Wherein, L (ε, θ)k) For the sparse variable y in the distribution type thetakAnd likelihood estimation values under distribution parameter ∈ L (∈, θ)k) d [ epsilon ] is the integral of the distribution parameter [ epsilon ], theta, of the likelihood estimate value of the sparse variable ykRepresents the kth distribution type;
according to different distribution types theta of sparse variable ykObtaining the maximum likelihood estimated value L of the sparse variable y by counting all likelihood estimated values under different distribution parameters epsilonmax(ε,θk) Then, the minimum information criterion method is adopted to calculate different distribution types thetakInformation entropy AIC ofk:
AICk=2numk-2ln(Lmax(ε,θk))
In the formula, numkDenotes the number, L, of distribution parameters εmax(ε,θk) Representing the distribution type theta of the sparse variable ykAnd maximum likelihood estimate, AIC, of distribution parameter epsilonkRepresenting the information entropy of the kth distribution type, wherein k represents the ordinal number of the distribution type;
then, the probability values P of different distribution types are obtained by the following formulaθ_k:
Pθ_k=exp((AICmin-AICk)/2)
In the formula, AICminRepresenting a minimum information entropy;
finally, the probability value P is determinedθ_kGreater than distribution type probability threshold PΔDistribution type of thetakAnd the distribution type of the sparse variable y is the distribution type of the microscopic property of the candidate material set corresponding to the sparse variable y.
In the step 7), a Genetic Algorithm (GA) is used to assemble the candidate materialsSelection and layer stacking sequence (theta)1,θ2,θ3、、、θx) The optimization treatment is carried out by selecting, specifically: optimizing the variable to a layer Stack sequence (θ)1,θ2,θ3、、、θx) And set of candidate materials
Performance parameters obtained using accurate neural network model outputAnd constructing performance parametersRobust optimization function f1:
θ=(θ1,θ2,θ3、、、θx)
Wherein mean represents taking the mean value, std represents taking the standard deviation, p is the weight coefficient,representing a final solution set of alternative material choices, θ representing a final solution of the layer stack sequence; theta1Representing the stacking angle of the first layer of material in the hybrid composite; theta2Representing the stacking angle of the second layer material in the hybrid composite; theta3Representing the stacking angle of the third layer of material in the hybrid composite; thetaxRepresents the stacking angle of the x-th layer material in the hybrid composite material, x represents the sum of the stacking sequence of the layers;representing the first material in the final recipe set,representing the second material in the final recipe set,representing the third material in the final recipe set,representing the jth material in the final recipe set, j representing the sum of the materials in the final recipe set;
Wherein, lambda represents a punishment parameter, alpha represents a small positive tolerance coefficient, g represents inequality constraint of a failure parameter FI and a total cost parameter C, and h represents stress balance constraint of the mixed composite material;
simultaneously setting constraint conditions of a failure parameter FI and a total cost parameter C; to penalize an objective functionSolving for the maximization target to obtain the final scheme set of the alternative material selectionAnd two parameter results of the final solution θ of the layer stack sequence;
in the step 6), the method for learning the neural network model parameters based on the performance prediction variance specifically comprises the following steps:
6.1) set of candidate materials required in the neural network model input from step 4)The microscopic elastic properties, the microscopic strength properties, and the layer stack sequence (θ) of each candidate material1,θ2,θ3、、、θx) And respectively randomly selecting the required alternative material sets and then establishing an input data set as a sample point; the performance parameters of each sample point are obtained by processing and predicting the sample points through a neural network modelA failure parameter FI and an overall cost parameter C;
6.2) repeating the step 6.1) for a plurality of times to obtain a plurality of sample points and corresponding performance parameters thereofA failure parameter FI and an overall cost parameter C;
6.3) calculating by adopting a formula to obtain the characteristic value gamma of each sample pointi:
γi=ωilog(u(i))
u(i)=rand(0,1)
Wherein, ω isiFor calculating the value of the performance parameter, gamma, obtained at the i-th sample pointiRepresenting the eigenvalue of the ith sample point, rand (0,1) representing the generation of a random number between 0 and 1, u(i)Representing the random number corresponding to the ith sample point;
6.4) selecting the sample point with the maximum characteristic value, and retraining the neural network model by using the input data set of the sample point so as to update the parameters of the neural network model;
6.5) go back to step 5).
The invention has the following beneficial effects:
1) the invention optimizes the stacking sequence and material distribution of the mixed composite material from the micro scale, and compared with the method for optimizing the mixed composite material from the macro scale, the analysis is more accurate, and simultaneously, the method is more in line with the practical engineering application.
2) The single composite material is extremely complex in design and manufacture due to unique composition components, and further, the performance of a computer is tested in the aspect of computer design optimization, but compared with the single composite material, the mixed composite material is more complex, so that various optimization algorithms are used, the burden of designers and computers is greatly reduced, and meanwhile, the optimization accuracy is improved.
3) Compared with the traditional Genetic Algorithm (GA), a new penalty objective function is provided, and the goal is to reduce the optimization variables which do not meet the constraint conditions, so that the computational burden of a computer is further reduced, the computational speed is greatly increased, and the optimization accuracy is improved.
In a word, the method of the invention uses various optimization algorithms, greatly reduces the burden of designers and computers, greatly improves the calculation speed and the optimization accuracy, and simultaneously better conforms to the actual engineering application.
Drawings
FIG. 1 is a schematic representation of a hybrid composite model according to an embodiment of the present invention;
FIG. 2 is an iterative process of stacking order and material patch optimization according to an embodiment of the present invention;
FIG. 3 is a flow chart of finite element analysis of the hybrid composite of the present invention;
FIG. 4 is a flow chart of the hybrid composite material optimization of the present invention.
Detailed Description
The invention will be further described with reference to the following figures and examples, without limiting the scope of the invention thereto, the method of the invention comprising in particular the following steps, as shown in figures 3 and 4:
1. determining the basic information of the mixed composite material to be analyzed: the boundary conditions are nine layers of composite material with four simply-supported sides, length L being 0.508m, width W being 0.406m, and thickness of each layer Δ T being 0.125 mm. The maximum overall cost parameter C is 9.5USD, and the minimum value of the failure probability isWherein the reliability parameter PfThe relationship with the failure parameter FI is: pf=P[1-FI≤0]And P is the distributed force to which the hybrid composite is subjected. Three alternative materials were selected: the material I, the material II and the material III are represented by '1', 2 'and 3' in subsequent calculation; the layer stacking sequence is from an angle [ -45,45,0,90 [ -45]Are selected and recombined. Performance parameterIs set as a natural frequency Ff. An example of creating a hybrid composite model is shown in FIG. 1.
2. Uncertainty quantification of microscopic properties of candidate materials (material i, material ii, material iii): the method comprises the following steps:
firstly, regarding each attribute parameter in the microscopic elastic attribute and the microscopic strength attribute of the candidate materials (material I, material II and material III) as different sparse variables y, and sampling and distributing different distribution types theta in the sparse variables ykProcessing statistics under different distribution parameters epsilon to obtain a probability density function f of a sparse variable y represented by the following formulay(y|ε,θk):
Wherein, L (ε, θ)k) For the sparse variable y in the distribution type thetakAnd likelihood estimation values under distribution parameter ∈ L (∈, θ)k) d [ epsilon ] is the integral of the distribution parameter [ epsilon ], theta, of the likelihood estimate value of the sparse variable ykRepresents the kth distribution type;
according to different distribution types theta of sparse variable ykObtaining the maximum likelihood estimated value L of the sparse variable y by counting all likelihood estimated values under different distribution parameters epsilonmax(ε,θk) Then, the minimum information criterion method is adopted to calculate different distribution types thetakInformation entropy AIC ofk:
AICk=2numk-2ln(Lmax(ε,θk))
In the formula, numkDenotes the number, L, of distribution parameters εmax(ε,θk) Representing the distribution type theta of the sparse variable ykAnd maximum likelihood estimate, AIC, of distribution parameter epsilonkRepresenting the information entropy of the kth distribution type, wherein k represents the ordinal number of the distribution type;
then, the probability values P of different distribution types are obtained by the following formulaθ_k:
PA_k=exp((AICmin-AICk)/2)
In the formula, AICminRepresenting a minimum information entropy;
finally, the probability value P is determinedθ_kGreater than distribution type probability threshold PΔDistribution type of thetakAnd the distribution type of the sparse variable y is the distribution type of the microscopic property of the candidate material set corresponding to the sparse variable y.
And obtaining the distribution type of the microscopic properties of the candidate materials (material I, material II and material III) after uncertainty quantification, wherein the microscopic property quantity of the candidate materials (material I, material II and material III) comprises microscopic elastic properties and microscopic strength properties.
3. Based on the distribution types of the microscopic attributes of the alternative materials (material I, material II and material III) obtained in the step 2, the uncertainty of the microscopic attributes of the alternative materials (material I, material II and material III) is transmitted to a macroscopic layer through a multi-scale analysis model of the alternative materials (material I, material II and material III), and the distribution types of the macroscopic elastic attributes and the macroscopic strength attributes of the alternative materials (material I, material II and material III) are obtained;
4. after obtaining the distribution types of the macroscopic elastic properties and the macroscopic strength properties of the candidate materials (material I, material II and material III), determining a macroscopic finite element model of the candidate materials (material I, material II and material III) by utilizing the macroscopic elastic properties, the macroscopic strength properties and the distribution types thereof, and stacking the layers of the macroscopic finite element models of the candidate materials (material I, material II and material III) and the candidate materials (material I, material II and material III)(θ1,θ2,θ3、、、Ax) After Latin hypercube sampling, a macroscopic finite element model of the alternative materials (material I, material II and material III) is analyzed by a laminated plate method, and the natural frequency F of the alternative materials (material I, material II and material III) is obtained after analysisfA failure parameter FI and an overall cost parameter C;
5. constructing a neural network model of microscopic properties of the candidate materials (material I, material II and material III) and mixed composite material parameters, wherein the constructed neural network model comprises an input layer, an intermediate layer and an output layer, the input layer inputs the candidate materials (material I, material II and material III), and the microscopic elastic properties, the microscopic strength properties and the layer stacking sequence (theta) of each candidate material in the candidate materials (material I, material II and material III)1,θ2,θ3、、、θx) (ii) a Output layer output natural frequency FfA failure parameter FI and an overall cost parameter C;
6. calculating Mean Square Error (MSE) of the trained neural network model;
7. judging the accuracy of the trained neural network model according to the Mean Square Error (MSE);
if the obtained mean square error MSE is smaller than a preset mean square error threshold value, the constructed neural network model is indicated to be accurate, and the step 8) is carried out;
if the mean square error MSE is larger than or equal to the mean square error threshold, the constructed neural network model is not accurate, parameter learning is carried out on the neural network model based on the performance prediction variance, and then the step 6 is returned to); the method comprises the following specific steps:
7.1) the microscopic elastic properties, the microscopic strength properties and the layer stacking sequence (theta) of each of the candidate materials (material I, material II, material III) in the neural network model input from step 5)1,θ2,θ3、、、θx) And respectively randomly selecting the required alternative material sets and then establishing an input data set as a sample point; the performance parameters of each sample point are obtained by processing and predicting the sample points through a neural network modelA failure parameter FI and an overall cost parameter C;
7.2) repeating the step 7.1) for a plurality of times to obtain a plurality of sample points and the corresponding natural frequencies FfA failure parameter FI and an overall cost parameter C;
7.3) calculating by adopting a formula to obtain the characteristic value gamma of each sample pointi:
γi=ωilog(u(i))
u(i)=rand(0,1)
Wherein, ω isiFor calculating the value of the performance parameter, gamma, obtained at the i-th sample pointiRepresenting the eigenvalue of the ith sample point, rand (0,1) representing the generation of a random number between 0 and 1, u(i)Representing the random number corresponding to the ith sample point;
7.4) selecting the sample point with the maximum characteristic value, and retraining the neural network model by using the input data set of the sample point to update the parameters of the neural network model;
7.5) go back to step 6).
8. Selection of candidate materials (Material I, Material II, Material III) and layer Stacking sequence (θ) Using Genetic Algorithm (GA)1,θ2,A3、、、Ax) And carrying out optimization processing on the selection to obtain the material selection of each layer and the stacking angle of each layer in the mixed composite material laying. The method comprises the following steps: optimizing the variable to a layer Stack sequence (θ)1,θ2,θ3、、、θx) And alternative materials (material I, material II, material III);
performance parameters obtained using accurate neural network model outputAnd constitutes the natural frequency FfRobust optimization function f1:
θ=(θ1,θ2,θ3、、、θx)
Wherein mean represents taking the mean value, std represents taking the standard deviation, p is the weight coefficient,represents the final set of solutions for the choice of candidate materials (material i, material ii, material iii), theta represents the final solution for the layer stacking sequence; theta1Representing the stacking angle of the first layer of material in the hybrid composite; theta2Representing the stacking angle of the second layer material in the hybrid composite; theta3Representing the stacking angle of the third layer of material in the hybrid composite; thetaxRepresents the stacking angle of the x-th layer material in the hybrid composite material, x represents the sum of the stacking sequence of the layers; in order to reduce the choice of an improperly constrained layer stacking sequence (theta)1,θ2,θ3、、、θx) The following penalty objective functions are established by using a penalty function method together with alternative materials (material I, material II and material III)
Wherein, lambda represents a punishment parameter, alpha represents a small positive tolerance coefficient, g represents inequality constraint of a failure parameter FI and a total cost parameter C, and h represents stress balance constraint of the mixed composite material; in the specific implementation, the penalty parameter lambda is set to be-100, and the small positive tolerance coefficient alpha is set to be 10-4Further, the time required by the computer for optimization is shortened;
simultaneously setting constraint conditions of a failure parameter FI and a total cost parameter C; to penalize an objective functionSolving for the maximization target to obtain the final scheme set of the alternative material selectionAnd two parameter results for final scenario a of the layer stack sequence.
Continuously stacking the layers in sequence (theta) by the adaptive genetic algorithm in step 81,θ2,θ3、、、θx) Iteration with the candidate composite materials (Material I, Material II, Material III) resulted in the optimal layer Stack sequence in this example [ -45/-45/45/-45/90/-45/45/-45]And alternative composite distributions [ 1; 1; 2; 1; 1]The corresponding natural frequency value is 36.58Hz, as shown in fig. 2.
The description is given for the sole purpose of illustrating embodiments of the inventive concept and should not be taken as limiting the scope of the invention to the particular forms set forth in the embodiments, but rather as being limited only to the equivalents thereof as may be contemplated by those skilled in the art based on the teachings herein.
Claims (4)
1. A mixed composite material layering method considering multi-scale uncertainty is characterized in that: the method specifically comprises the following steps:
1) selection of multiple different composite materials as a candidate set of hybrid composite materialsWherein the content of the first and second substances,a first alternative material is indicated which,a second alternative material is indicated which,a third alternative material is shown which,denotes the n-th candidate materialN represents the sum of the candidate materials, and is set for the candidate materialsThe microscopic attributes are subjected to uncertainty quantification, and an alternative material set is obtained after the uncertainty quantificationWherein the set of candidate materials isThe microscopic properties of (a) include microscopic elastic properties and microscopic strength properties;
2) set of candidate materials obtained based on step 1)By the type of distribution of the microscopic properties of the candidate material setThe multi-scale analysis model of (2) aggregating the candidate materialsThe uncertainty of the microscopic properties is propagated to a macroscopic level to obtain a set of alternative materialsThe distribution type of the macroscopic elastic property and the macroscopic strength property;
3) obtaining a collection of candidate materialsAfter the distribution types of the macroscopic elastic property and the macroscopic strength property are obtained, a candidate material set is determined by utilizing the macroscopic elastic property, the macroscopic strength property and the distribution types thereofFor a set of candidate materialsAnd set of candidate materialsLayer stacking sequence (theta) of the macroscopic finite element model of (1)1,θ2,θ3、、、θx) After Latin hypercube sampling, alternative materials are gathered by a laminated plate methodAnalyzing the macroscopic finite element model to obtain a candidate material setPerformance parameter ofA failure parameter FI and an overall cost parameter C; wherein, theta1Representing the stacking angle of the first layer of material in the hybrid composite; theta2Representing the stacking angle of the second layer material in the hybrid composite; theta3Representing the stacking angle of the third layer of material in the hybrid composite; thetaxRepresents the stacking angle of the x-th layer material in the hybrid composite material, x represents the sum of the stacking sequence of the layers;
4) building a collection of candidate materialsThe constructed neural network model comprises an input layer, an intermediate layer and an output layer, wherein the input layer inputs a required candidate material setRequired set of candidate materialsThe microscopic elastic properties, the microscopic strength properties, and the layer stack sequence (θ) of each candidate material1,θ2,θ3、、、θx),A first alternative material is indicated which,indicating that a second material alternative is desired,indicating that a third alternative material is required,representing the m-th required candidate material, wherein m represents the sum of the required candidate materials; output layer output performance parametersA failure parameter FI and an overall cost parameter C;
5) calculating Mean Square Error (MSE) of the trained neural network model;
6) judging the accuracy of the trained neural network model according to the Mean Square Error (MSE);
if the obtained mean square error MSE is smaller than a preset mean square error threshold value, the constructed neural network model is indicated to be accurate, and the step 7) is carried out;
if the mean square error MSE is larger than or equal to the mean square error threshold, the constructed neural network model is not accurate, parameter learning is carried out on the neural network model based on the performance prediction variance, and then the step 5 is returned to);
2. The method of laying up a hybrid composite material taking into account multi-scale uncertainty as claimed in claim 1, wherein:
in the step 1), uncertainty quantification is performed on microscopic properties of the candidate material set, specifically:
first, the candidate materials are gatheredThe micro elastic property and the micro strength property are regarded as different sparse variables y, and different distribution types theta are sampled and distributed in the sparse variables ykProcessing statistics under different distribution parameters epsilon to obtain a probability density function f of a sparse variable y represented by the following formulay(y|ε,θk):
Wherein, L (ε, θ)k) For the sparse variable y in the distribution type thetakAnd likelihood estimation values under distribution parameter ∈ L (∈, θ)k) d [ epsilon ] is the integral of the distribution parameter [ epsilon ], theta, of the likelihood estimate value of the sparse variable ykRepresents the kth distribution type;
according to different distribution types theta of sparse variable ykObtaining the maximum likelihood estimated value L of the sparse variable y by counting all likelihood estimated values under different distribution parameters epsilonmax(ε,θk) Then, the minimum information criterion method is adopted to calculate different distribution types thetakIs sent toEntropy of rest AICk:
AICk=2numk-2ln(Lmax(ε,θk))
In the formula, numkDenotes the number, L, of distribution parameters εmax(ε,θk) Representing the distribution type theta of the sparse variable ykAnd maximum likelihood estimate, AIC, of distribution parameter epsilonkRepresenting the information entropy of the kth distribution type, wherein k represents the ordinal number of the distribution type;
then, the probability values P of different distribution types are obtained by the following formulaθ_k:
Pθ_k=exp((AICmin-AICk)/2)
In the formula, AICminRepresenting a minimum information entropy;
finally, the probability value P is determinedθ_kGreater than distribution type probability threshold PΔDistribution type of thetakAnd the distribution type of the sparse variable y is the distribution type of the microscopic property of the candidate material set corresponding to the sparse variable y.
3. The method of laying up a hybrid composite material taking into account multi-scale uncertainty as claimed in claim 1, wherein:
in the step 7), a Genetic Algorithm (GA) is used to assemble the candidate materialsSelection and layer stacking sequence (theta)1,θ2,θ3、、、θx) The optimization treatment is carried out by selecting, specifically: optimizing the variable to a layer Stack sequence (θ)1,θ2,θ3、、、θx) And set of candidate materials
Performance parameters obtained using accurate neural network model outputAnd constructing performance parametersRobust optimization function f1:
θ=(θ1,θ2,θ3、、、θx)
Wherein mean represents taking the mean value, std represents taking the standard deviation, p is the weight coefficient,representing a final solution set of alternative material choices, θ representing a final solution of the layer stack sequence; theta1Representing the stacking angle of the first layer of material in the hybrid composite; theta2Representing the stacking angle of the second layer material in the hybrid composite; theta3Representing the stacking angle of the third layer of material in the hybrid composite; thetaxRepresents the stacking angle of the x-th layer material in the hybrid composite material, x represents the sum of the stacking sequence of the layers;representing the first material in the final recipe set,representing the second material in the final recipe set,representing the third material in the final recipe set,representing the jth material in the final recipe set, j representing the sum of the materials in the final recipe set;
Wherein, lambda represents a punishment parameter, alpha represents a small positive tolerance coefficient, g represents inequality constraint of a failure parameter FI and a total cost parameter C, and h represents stress balance constraint of the mixed composite material;
4. The method of laying up a hybrid composite material taking into account multi-scale uncertainty as claimed in claim 1, wherein:
in the step 6), the method for learning the neural network model parameters based on the performance prediction variance specifically comprises the following steps:
6.1) set of candidate materials required in the neural network model input from step 4)The microscopic elastic properties, the microscopic strength properties, and the layer stack sequence (θ) of each candidate material1,θ2,θ3、、、θx) And respectively randomly selecting the required alternative material sets and then establishing an input data set as a sample point; the performance parameters of each sample point are obtained by processing and predicting the sample points through a neural network modelA failure parameter FI and an overall cost parameter C;
6.2) repeating the step 6.1) for a plurality of times to obtain a plurality of sample points and corresponding performance parameters thereofA failure parameter FI and an overall cost parameter C;
6.3) calculating by adopting a formula to obtain the characteristic value gamma of each sample pointi:
γi=ωilog(u(i))
u(i)=rand(0,1)
Wherein, ω isiFor calculating the value of the performance parameter, gamma, obtained at the i-th sample pointiRepresenting the eigenvalue of the ith sample point, rand (0,1) representing the generation of a random number between 0 and 1, u(i)Representing the random number corresponding to the ith sample point;
6.4) selecting the sample point with the maximum characteristic value, and retraining the neural network model by using the input data set of the sample point so as to update the parameters of the neural network model;
6.5) go back to step 5).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011504349.0A CN112464541B (en) | 2020-12-18 | 2020-12-18 | Multi-scale uncertainty considered mixed composite material layering method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011504349.0A CN112464541B (en) | 2020-12-18 | 2020-12-18 | Multi-scale uncertainty considered mixed composite material layering method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112464541A true CN112464541A (en) | 2021-03-09 |
CN112464541B CN112464541B (en) | 2024-05-24 |
Family
ID=74804791
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011504349.0A Active CN112464541B (en) | 2020-12-18 | 2020-12-18 | Multi-scale uncertainty considered mixed composite material layering method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112464541B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106991212A (en) * | 2017-03-07 | 2017-07-28 | 西安交通大学 | A kind of blade root Strength Safety characteristic prediction method for optimizing GRNN network algorithms based on GA_PSO |
US20180068216A1 (en) * | 2015-03-13 | 2018-03-08 | Institute Of Acoustics, Chinese Academy Of Sciences | Big data processing method based on deep learning model satisfying k-degree sparse constraint |
CN110083900A (en) * | 2019-04-17 | 2019-08-02 | 大连理工大学 | A kind of fast synergistic optimization method towards fiber hybrid composite plate and shell structure |
-
2020
- 2020-12-18 CN CN202011504349.0A patent/CN112464541B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180068216A1 (en) * | 2015-03-13 | 2018-03-08 | Institute Of Acoustics, Chinese Academy Of Sciences | Big data processing method based on deep learning model satisfying k-degree sparse constraint |
CN106991212A (en) * | 2017-03-07 | 2017-07-28 | 西安交通大学 | A kind of blade root Strength Safety characteristic prediction method for optimizing GRNN network algorithms based on GA_PSO |
CN110083900A (en) * | 2019-04-17 | 2019-08-02 | 大连理工大学 | A kind of fast synergistic optimization method towards fiber hybrid composite plate and shell structure |
Non-Patent Citations (1)
Title |
---|
李洪波;徐元铭;黄英兰;: "试验设计法与BP-GA算法在复合材料结构优化中的应用", 飞机设计, no. 03, 15 June 2007 (2007-06-15) * |
Also Published As
Publication number | Publication date |
---|---|
CN112464541B (en) | 2024-05-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111985796B (en) | Method for predicting concrete structure durability based on random forest and intelligent algorithm | |
CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
CN110544011B (en) | Intelligent system combat effectiveness evaluation and optimization method | |
CN107886161A (en) | A kind of global sensitivity analysis method for improving Complex Information System efficiency | |
WO2021027153A1 (en) | Method and apparatus for constructing traffic flow data analysis model | |
CN115374995A (en) | Distributed photovoltaic and small wind power station power prediction method | |
CN113033105B (en) | Method for predicting fatigue performance of steel material based on mechanical theory guiding transfer learning | |
CN113761748B (en) | Industrial process soft measurement method based on federal incremental random configuration network | |
CN111832839B (en) | Energy consumption prediction method based on sufficient incremental learning | |
CN112926893A (en) | Horizontal well profile control effect evaluation method based on fuzzy comprehensive evaluation and hierarchical analysis | |
CN114706840A (en) | Load perception-oriented method for optimizing parameters of ArangoDB of multi-mode database | |
CN116842459A (en) | Electric energy metering fault diagnosis method and diagnosis terminal based on small sample learning | |
CN113887694A (en) | Click rate estimation model based on characteristic representation under attention mechanism | |
CN115982141A (en) | Characteristic optimization method for time series data prediction | |
CN112905436B (en) | Quality evaluation prediction method for complex software | |
CN114707692A (en) | Wetland effluent ammonia nitrogen concentration prediction method and system based on hybrid neural network | |
CN112464541A (en) | Mixed composite material layering method considering multi-scale uncertainty | |
CN116911459A (en) | Multi-input multi-output ultra-short-term power load prediction method suitable for virtual power plant | |
CN116467934A (en) | Carbon fiber preform parameter optimization method | |
CN114611389A (en) | Artificial intelligence-based efficient composite material failure simulation method | |
Yang et al. | Power demand forecast based on optimized neural networks by improved genetic algorithm | |
CN112116238A (en) | Satisfaction evaluation method based on index weight system design | |
CN111080053B (en) | EHF-WAS-based power grid development strategy efficiency and benefit evaluation method | |
CN110941767A (en) | Network community detection countermeasure enhancement method based on multi-similarity integration | |
CN110942149B (en) | Feature variable selection method based on information change rate and condition mutual information |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |