CN109918838A - A kind of method and system of prediction large and complex structure military service performance containing splicing - Google Patents
A kind of method and system of prediction large and complex structure military service performance containing splicing Download PDFInfo
- Publication number
- CN109918838A CN109918838A CN201910231729.2A CN201910231729A CN109918838A CN 109918838 A CN109918838 A CN 109918838A CN 201910231729 A CN201910231729 A CN 201910231729A CN 109918838 A CN109918838 A CN 109918838A
- Authority
- CN
- China
- Prior art keywords
- complex structure
- model
- glued
- performance
- adhesive
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 239000000853 adhesive Substances 0.000 claims abstract description 43
- 230000001070 adhesive effect Effects 0.000 claims abstract description 43
- 239000000463 material Substances 0.000 claims abstract description 42
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000010801 machine learning Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 7
- 230000006399 behavior Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 238000012937 correction Methods 0.000 claims description 3
- 239000003292 glue Substances 0.000 claims description 2
- 230000003068 static effect Effects 0.000 abstract description 14
- 238000010586 diagram Methods 0.000 description 19
- 238000012360 testing method Methods 0.000 description 10
- 238000004364 calculation method Methods 0.000 description 8
- 238000013461 design Methods 0.000 description 7
- 230000007613 environmental effect Effects 0.000 description 7
- 238000002474 experimental method Methods 0.000 description 6
- 230000004044 response Effects 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000004088 simulation Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 238000007796 conventional method Methods 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 3
- 238000010521 absorption reaction Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 235000015220 hamburgers Nutrition 0.000 description 2
- 230000035939 shock Effects 0.000 description 2
- 238000012356 Product development Methods 0.000 description 1
- 238000007605 air drying Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 239000013013 elastic material Substances 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000006260 foam Substances 0.000 description 1
- 238000007656 fracture toughness test Methods 0.000 description 1
- 230000003116 impacting effect Effects 0.000 description 1
- 238000005304 joining Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000004080 punching Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 238000012612 static experiment Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Landscapes
- Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)
Abstract
The present invention discloses a kind of method and system of prediction large and complex structure military service performance containing splicing.Method includes: to obtain the material category and the trade mark of two adherends and adhesive;The elasticity modulus of two adherends and adhesive is obtained according to above- mentioned information;Obtain the thickness information of two adherends and glue-line;The meta-model in area is glued in building;According to area's meta-model is glued, the large and complex structure model for not including detailed glue-line is established;The model that previous step obtains is modified with experimental data, obtains large and complex structure performance accurate prediction models;Accurate prediction models are trained, the large and complex structure performance quick predict model after being trained;The parameter for being glued area and load working condition are input to the large and complex structure performance prediction model after training, obtain the detailed forecasts result of large and complex structure.The present invention, which can greatly shorten, calculates the time, state of the quick prediction interface in static, impact, fatigue and hygrothermal environment load whether safety.
Description
Technical field
The present invention relates to field is glued, more particularly to a kind of prediction containing the method for being glued large and complex structure military service performance
And system.
Background technique
Since energy conservation and environmental protection needs, automobile, aircraft, rail traffic and other kinds mechanical equipment are intended to realize lightweight.
According to " suitable material is used in suitable position " this light-weight design theory, various different materials can be applied simultaneously same
It in a large scale structure, and needs to be connected to each other, dissimilar material joining problem is thus caused to burst.Splicing is all materials suitable at present
Expect the sole mode of connection.
In the different phase of product development, in addition to the experimental test procedures of physical model, Numerical-Mode that CAD/CAE is combined
Quasi- technical application obtains more extensive.Such as: in cementing structure design, numerical simulation analysis is generally carried out using FInite Element.
Since cementing structure includes two kinds of connected materials and a kind of adhesive, there are also the mutual interface between three, needed in model
Consider many material parameters and failure criteria, and it is non-linear cause interface calculate be difficult to restrain.If judging whether be glued area
Safety needs local refinement grid, unit and number of nodes, and total number of degrees of freedom, to sharply increase, and the calculating time is too long, loses
Market value, enterprise can not receive.In addition, the determination of these parameters, not only needs largely to test, also to the knowledge of designer
There is very high requirement in face.One such talent of culture needs the study and practical experience in many years.For enterprise, this talent
It is very rare.Generally need to by way of lateral project, in colleges and universities doctor or Master degree candidate cooperate.It not only extends and sets
Meter exploitation and Time To Market, and risk of divulging a secret.
Summary of the invention
The object of the present invention is to provide a kind of predictions containing the method and system for being glued large and complex structure military service performance, can
It greatly shortens and calculates the time, quick predict is glued state of the interface in static, impact, fatigue and environmental load whether safety.
To achieve the above object, the present invention provides following schemes:
A method of prediction large and complex structure military service performance containing splicing, comprising:
Obtain the material category and the trade mark of two adherends and adhesive;
According to the material category and the trade mark of the adherend and adhesive, the springform of two adherends and adhesive is obtained
Amount;
Obtain the thickness information of two adherends and glue-line;
According to the elasticity modulus and thickness information of described two adherends and adhesive, the meta-model in area is glued in building;
According to the meta-model for being glued area, establish based on effective stiffness matrix, hyperelement not comprising detailed glue-line
Large and complex structure model;
The large and complex structure model not comprising detailed glue-line is modified according to experimental data, is obtained large-scale multiple
Miscellaneous structural behaviour accurate prediction models;
The large and complex structure performance accurate prediction models are trained using machine learning method, after being trained
Large and complex structure performance quick predict model;
The parameter for being glued area and load working condition are input to the large and complex structure performance quick predict mould after the training
Type obtains the detailed forecasts result of large and complex structure.
Optionally, the material category and the trade mark for obtaining two adherends and adhesive, specifically includes:
The material category and the trade mark of two adherends and adhesive are directly acquired from Self-built Database.
Optionally, splicing area's meta-model is 4 nodal plane meta-models or 8 Nodes Three-dimensional meta-models.
Optionally, described that the large and complex structure model not comprising detailed glue-line is repaired according to experimental data
Just, large and complex structure performance accurate prediction models are obtained, are specifically included:
The large and complex structure model not comprising detailed glue-line is repaired according to loading demands according to experimental data
Just, large and complex structure performance accurate prediction models are obtained.
Optionally, described that the large and complex structure performance accurate prediction models are instructed using machine learning method
Practice, the large and complex structure performance quick predict model after being trained specifically includes:
Different load working conditions and different splicing areas are inputted into the large and complex structure performance accurate prediction models
Meta-model adjusts the parameter of the large and complex structure performance accurate prediction models, the large and complex structure after being trained
It can quick predict model.
Optionally, described that the parameter for being glued area and load working condition are input to the large and complex structure performance after the training
Quick predict model obtains the detailed forecasts of large and complex structure as a result, specifically including:
The parameter for being glued area and load working condition are input to the large and complex structure performance quick predict mould after the training
Type obtains the strength and stiffness data of large and complex structure;
The failure criteria that adhesive is used according to the strength and stiffness data, determines prediction result safety.
A kind of system of prediction large and complex structure military service performance containing splicing, comprising:
First obtains module, for obtaining the material category and the trade mark of two adherends and adhesive;
Elasticity modulus determining module obtains two for the material category and the trade mark according to the adherend and adhesive
The elasticity modulus of adherend and adhesive;
Second obtains module, for obtaining the thickness information of two adherends and glue-line;
Meta-model constructs module, for the elasticity modulus and thickness information according to described two adherends and adhesive, structure
Build the meta-model for being glued area;
Large and complex structure model construction module, for establishing and being based on equivalent stiffness according to the meta-model for being glued area
Matrix, the hyperelement large and complex structure model not comprising detailed glue-line;
Correction module, for being repaired according to experimental data to the large and complex structure model not comprising detailed glue-line
Just, large and complex structure performance accurate prediction models are obtained;
Training module, for being instructed to the large and complex structure performance accurate prediction models using machine learning method
Practice, the large and complex structure performance quick predict model after being trained;
Prediction result determining module, the large size for the parameter and load working condition of being glued area to be input to after the training are answered
Miscellaneous structural behaviour quick predict model obtains the detailed forecasts result of large and complex structure.
The specific embodiment provided according to the present invention, the invention discloses following technical effects: the present invention provides a kind of pre-
Survey containing be glued large and complex structure military service performance method, can quick predict be glued interface whether fail, by analytical Calculation,
The combination of machine learning and finite element establishes between adhesive material geometry and load parameter and glue-line stress strain response amount
Corresponding relationship substantially reduces and calculates the time, has reached prediction and has been glued interface peace in static, impact, fatigue and environmental load
Complete or faulted condition, judges whether the position is able to satisfy the purpose of requirement according to failure criteria.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will be to institute in embodiment
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is present invention prediction containing the method flow diagram for being glued large and complex structure military service performance;
Fig. 2 is traditional three-zone model;
Fig. 3 is single lap joint load transmission schematic diagram of the present invention;
Fig. 4 is single lap joint meta-model schematic diagram of the present invention;
Fig. 5 is that L-type joint load of the present invention transmits schematic diagram;
Fig. 6 is L-type connector meta-model schematic diagram of the present invention;
Fig. 7 is T connector load transmission schematic diagram of the present invention;
Fig. 8 is T connector meta-model schematic diagram of the present invention;
Fig. 9 is the computation model schematic diagram using conventional method;
Figure 10 is computation model schematic diagram of the present invention;
Figure 11 is present invention prediction containing the system construction drawing for being glued large and complex structure military service performance.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The object of the present invention is to provide a kind of predictions containing the method and system for being glued large and complex structure military service performance, can
It greatly shortens and calculates the time, state of the quick prediction interface in static, impact, fatigue and environmental load whether safety.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
In order to which interface local stress and strain regime is described in detail, a large amount of finite element grid and unit need to be used, even if
Single calculation, it is time-consuming also more, if convergence difficulties, also need constantly to debug.When carrying out multiple project plan comparisons, or to structure
When optimizing, calculate that the time is too long, enterprise can not endure.
Classical forecast process are as follows:
(1) detailed labyrinth CAD model, including adherend and glue-line are established.
(2) FEM meshing is carried out.Grid is thinner, as a result more accurate;
(3) in order to obtain material model, the material test specimen of every kind of adherend and adhesive is generally first manufactured respectively, according to phase
It closes regulation to be tested, tests its tensile modulus of elasticity, the basic mechanicals parameter such as intensity and Poisson's ratio.Then, it is tied according to experiment
Fruit obtains the constitutive model of material, identifies undetermined coefficient.
(4) large scale structure including glue-line in the limited splicing area meta-model of all material mode input, will be obtained to be integrated with
Limit is glued area's meta-model;
(5) according to static load operating condition, apply load and boundary condition, select suitable solver, calculated;
(6) material object manufacture and Physical Experiment are carried out, simulation model is corrected;
(7) real load operating condition is re-entered, maximum stress and strain is obtained, is compared with allowable value, judges structure
It is whether safe.
(8) if being unsatisfactory for design requirement, structure or material parameter are remodified, until meeting safety conditions.
(9) if prediction impact loading operating condition, needs making material test specimen in advance, impact experiment is carried out, acquisition is answered
The relevant constitutive model of variability inputs in limited splicing area meta-model;
(10) if prediction fatigue load acts on operating condition, making material test specimen in advance is needed, carries out fatigue experiment, acquisition is answered
Power (strain)-life curve, i.e. S-N curve input in limited splicing area meta-model;
(11) such as predict damp and hot environmental load effect operating condition, then need making material test specimen in advance, in different temperatures and
Under the matching of different humidity, static experiment test is carried out, is compared with air drying environmental consequences, opening relationships is input to
In limited splicing area meta-model.
(12) if judging whether to fail, the judgment criterion that suitably fails also is selected.It is obtained using experiment test critical
Value and simulation result compare, and obtain conclusion whether safety.
Above procedure must be implemented by having received many years mechanics, material, computer software and the personnel of Machine Design training.Cause
This, there is an urgent need to the ease of Use of a kind of " foolproof " (intelligent), and method whether can quickly learn safety, without knowing
Specific formula and numerical value.
Fig. 1 is present invention prediction containing the method flow diagram for being glued large and complex structure military service performance.As shown in Figure 1, a kind of
Predict containing be glued large and complex structure military service performance method include:
Step 101: obtaining the material category and the trade mark of two adherends and adhesive;It is directly acquired from Self-built Database
The material category and the trade mark of two adherends and adhesive;
Step 102: according to the material category and the trade mark of the adherend and adhesive, obtaining two adherends and adhesive
Elasticity modulus;
The elastic modulus E of common brand adhesive of the present invention by publishing both at home and abroad and from survey3, shear modulus G3, with
And strain rate, environment (damp and hot) and fatigue S-N diagram, S-N curve result shift to an earlier date input database;It will consider glue-line failure
COHESIVE model curve also inputs in advance.The data of other special-purpose adhesives can be provided by supplier or self testing after expand
It is charged in database.
When static performance analysis, it is proposed that select ideal linear elastic model, σ=E3ε and τ=G3γ.It, can if considering elastoplasticity
Select such as Crushable Foam model.Database can continue to expand new material model.
When tired performance analysis, it is proposed that the rheological model for considering viscous-elastic behaviour is selected, such as Maxwell series model,
Kelvin parallel model.If there is creep and stress relaxation phenomenon, more complicated Burgers model can be selected.Database can
Continue to expand new material model.
When impacting performance analysis, ZWT (Zhu-Wang-Tang) or Burgers constitutive model can be selected.Database can continue
Expand new material model.
Step 103: obtaining the thickness information of two adherends and glue-line;
Step 104: according to the elasticity modulus and thickness information of described two adherends and adhesive, the member in area is glued in building
Model;Splicing area's meta-model is 4 nodal plane meta-models or 8 Nodes Three-dimensional meta-models.
It include three kinds of typical gluded joints: single overlap joint, L-type and T connector in database.By taking two dimensional model as an example, Fig. 3 is
Single lap joint load transmission schematic diagram of the present invention;Fig. 4 is single lap joint meta-model schematic diagram of the present invention;Fig. 5 is L of the present invention
Type joint load transmission schematic diagram;Fig. 6 is L-type connector meta-model schematic diagram of the present invention;Fig. 7 is T connector load of the present invention biography
Pass schematic diagram;Fig. 8 is T connector meta-model schematic diagram of the present invention;It is only wrapped by Fig. 4, Fig. 6 and Fig. 8 it is found that being glued area's meta-model
The adherend and glue-line for being glued area are included.In order to make problem reduction, calculation amount is reduced, it is assumed that load is uniformly distributed in the width direction, glue
The two-dimensional problems that area's meta-model can be reduced to xoy plane are connect, trilaminate material respectively elasticity modulus and thickness need to be only inputted.
Step 105: according to the meta-model for being glued area, establishing based on effective stiffness matrix, hyperelement do not include
The large and complex structure model of detailed glue-line;
Fig. 9 is the computation model schematic diagram using conventional method, and Figure 10 is computation model schematic diagram of the present invention, by that will scheme
Known to 9 and Figure 10 is compared: the present invention is being glued the processing of area's grid and conventional method comparison, can greatly save number of unit.
It in the junction of two adherend finite element models, is handled by conode, is equal to and establishes displacement constraint equation, equation number
Depending on for individual node freedom degree 4 times (plane problems) or 8 times (three-dimensional problem).
Step 106: the large and complex structure model not comprising detailed glue-line being modified according to experimental data, is obtained
To large and complex structure performance accurate prediction models;It specifically includes:
The large and complex structure model not comprising detailed glue-line is modified according to loading demands, is obtained large-scale multiple
Miscellaneous structural behaviour accurate prediction models.
Specific modification rule is as follows:
For static load, maximum distortion and maximum stress value are compared, keeps error minimum;
For fatigue load, cycle-index and load relation are compared, keeps error minimum;
For shock loading, peak acceleration peak value and energy absorption are compared, keeps error minimum;
For damp and hot load, wet heat condition and maximum distortion and maximum stress value relationship are compared, keeps error minimum.
Step 107: the large and complex structure performance accurate prediction models being trained using machine learning method, are obtained
Large and complex structure performance quick predict model after to training;It specifically includes:
Different load working conditions and different splicing areas are inputted into the large and complex structure performance accurate prediction models
Meta-model adjusts the parameter of the large and complex structure performance prediction model, and the large and complex structure performance after being trained is fast
Fast prediction model.
The step in load working condition be various typical load working conditions, the load working condition includes static, impact, tired
Labor and the operating conditions such as damp and hot, different splicing area's meta-models is by adjusting the thickness of two adherends, adherend and viscous glutinous agent
Elasticity modulus realize adjustment.
The step in output be static problem maximum distortion and maximum stress value;The largest loop time of fatigue problem
Number (service life);The peak acceleration peak value and energy absorption of shock problem;The maximum distortion and maximum stress value of damp and hot problem;
By establishing the relationship of input with output, the large and complex structure after being trained for machine learning and finite element simulation combination
Performance quick predict model.
Step 108: it is fast that the parameter for being glued area and load working condition being input to the large and complex structure performance after the training
Fast prediction model obtains the detailed forecasts result of large and complex structure;It specifically includes:
The parameter for being glued area and load working condition are input to the large and complex structure performance quick predict mould after the training
Type obtains the strength and stiffness data of large and complex structure;The load working condition includes static, impact, fatigue and damp and hot etc.
Operating condition, the parameter for being glued area is various splicing area design parameters to be evaluated;
Prediction result safety is determined using the failure criteria of adhesive according to the strength and stiffness data.
The failure criteria of adhesive mainly has:
(1) based on the failure criteria of intensity, such as maximum principal stress criterion, Von Mises yield criterion and maximum equivalent modeling
Property strain criterion, critical parameter values such as yield strength, strength degree etc. can be obtained by the stretching and shearing test of adhesive bulk
?.
(2) ASTM can refer to if stress intensity theory and crack propagation energy-absorbing are theoretical based on the failure criteria of energy to failure
E1820-2013 does the fracture toughness test of adhesive bulk.
The invention proposes a kind of predictions containing the method for being glued large and complex structure military service performance, being capable of quick predict splicing
Whether interface fails, and by the combination of analytical Calculation, machine learning and finite element, establishes adhesive material geometry and load parameter
It with the corresponding relationship between glue-line stress strain response amount, substantially reduces and calculates the time, reached prediction interface in static, punching
It hits, fatigue and safety or faulted condition when environmental load, judges whether the position is able to satisfy requirement according to failure criteria
Purpose.
Primary ideal, linear elastic materials finite element static solves, and required time depends on all finite elements
The freedom of movement degree Dof of number of nodes N, each node, it is exactly total equation number E that the two, which is seized the opportunity,dof=N*Dof.It is multiple for large size
Miscellaneous structure, N are usually 106Or 107, Dof is usually 6.And it is (material property, several for the nonlinear problem of most of truths
What deformation and interfacial contact cause), a solution step needs to be divided into several sub-steps n, and calculation amount isSuch as
Fruit is related calculation of dynamic response with the time (or frequency, temperature, humidity, cycle-index etc.), is also divided into time T several
Iteration step m, makesM and n is usually 102With 104.It is of the invention in order to solve the problems, such as, if it is desired to guaranteeing knot
The precision of fruit is higher, it is necessary to consider that non-linear, the total calculating time is
If it is glue-line optimization design, then successive ignition is needed.Assuming that iteration K times available optimal solution, then total calculating
Time is exactly
The present invention makes N=4 or 8, calculates the time to greatly shorten by simplifying modeling.It is non-due to omitting detailed modeling
Iteration n=1 caused by linear.Due to using machine learning, m=1, therefore, total calculating time are greatly shortened.
The present invention has following advantage compared with prior art:
The method that the present invention exempts the quick predict structural behaviour of detailed finite member simulation calculation, efficiency can be improved 10
N times side;
The present invention does not predict maximum stress and strain not instead of simply, according to suitable failure criteria, immediately arrives at peace
Judgement whether complete;
The present invention is different from other methods based on machine learning, uses the concept of " being glued area's meta-model ";
The present invention is different from other methods based on machine learning, and training pattern is based on the customized stiffness matrix list of user
Member, thus there is specific physical significance;
The present invention is different from the substructure method for large scale structure finite element analysis, and the meta-model of foundation is in entirety and office
The junction in portion, is not only displaced conode, and the generalized force of node is also condensed and transmits;
The present invention is different from the response phase method for simplifying Large Scale Nonlinear Structure Calculation, the load of foundation and the pass of response
System is it is contemplated that the topological pattern of different connections, the independence or immixture operating condition of different loads;
The present invention can predict whether static strength and rigidity meet the requirements, whether predict the relevant impact operating condition of strain rate
It meets the requirements, predict whether fatigue and durability operating condition meet the requirements and predict whether damp and hot operating condition meets the requirements.
Figure 11 is present invention prediction containing the system construction drawing for being glued large and complex structure military service performance.As shown in figure 11, one
Kind is predicted
First obtains module 201, for obtaining the material category and the trade mark of two adherends and adhesive;
Elasticity modulus determining module 202 obtains two for the material category and the trade mark according to the adherend and adhesive
The elasticity modulus of a adherend and adhesive;
Second obtains module 203, for obtaining the thickness information of two adherends;
Meta-model constructs module 204, for the elasticity modulus and thickness information according to described two adherends and adhesive,
The meta-model in area is glued in building;
Large and complex structure model construction module 205, for establishing based on equivalent rigid according to the meta-model for being glued area
Spend matrix, hyperelement the large and complex structure model not comprising detailed glue-line;
Correction module 206, for according to experimental data to the large and complex structure model not comprising detailed glue-line into
Row amendment, obtains large and complex structure performance accurate prediction models;
Training module 207, for the large and complex structure performance accurate prediction models using machine learning method into
Row training, the large and complex structure performance quick predict model after being trained;
Prediction result determining module 208 is big after the training for the parameter and load working condition of being glued area to be input to
Type labyrinth performance quick predict model, obtains the detailed forecasts result of large and complex structure.
The present invention can also be established quiet by combining Orthogonal Experiment and Design (DOE) and the emulation of detailed finite meta-model
Splicing area stress (strain) distribution under state, impact, fatigue, hygrothermal environment load and the response surface between thickness, material parameter
Expression formula.
The present invention can also establish parameterized model, then by artificial neural network training, establish be glued area's load with
The relationship of splicing area stress (strain) distribution and maximum value under static, impact, fatigue, hygrothermal environment load.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For system disclosed in embodiment
For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part
It is bright.
Used herein a specific example illustrates the principle and implementation of the invention, and above embodiments are said
It is bright to be merely used to help understand method and its core concept of the invention;At the same time, for those skilled in the art, foundation
Thought of the invention, there will be changes in the specific implementation manner and application range.In conclusion the content of the present specification is not
It is interpreted as limitation of the present invention.
Claims (7)
1. a kind of prediction is containing the method for being glued large and complex structure military service performance characterized by comprising
Obtain the material category and the trade mark of two adherends and adhesive;
According to the material category and the trade mark of the adherend and adhesive, the elasticity modulus of two adherends and adhesive is obtained;
Obtain the thickness information of two adherends and glue-line;
According to the elasticity modulus and thickness information of described two adherends and adhesive, the meta-model in area is glued in building;
According to the meta-model for being glued area, establish based on effective stiffness matrix, hyperelement big not comprising detailed glue-line
Type labyrinth model;
The large and complex structure model not comprising detailed glue-line is modified according to experimental data, obtains large complicated knot
Structure performance accurate prediction models;
The large and complex structure performance accurate prediction models are trained using machine learning method, it is big after being trained
Type labyrinth performance quick predict model;
The parameter for being glued area and load working condition are input to the large and complex structure performance quick predict model after the training, obtained
To the detailed forecasts result of large and complex structure.
2. prediction according to claim 1 is containing the method for being glued large and complex structure military service performance, which is characterized in that described
The material category and the trade mark for obtaining two adherends and adhesive, specifically include:
The material category and the trade mark of two adherends and adhesive are directly acquired from Self-built Database.
3. prediction according to claim 1 is containing the method for being glued large and complex structure military service performance, which is characterized in that described
Being glued area's meta-model is 4 nodal plane meta-models or 8 Nodes Three-dimensional meta-models.
4. prediction according to claim 1 is containing the method for being glued large and complex structure military service performance, which is characterized in that described
The large and complex structure model not comprising detailed glue-line is modified according to experimental data, obtains large and complex structure
Energy accurate prediction models, specifically include:
The large and complex structure model not comprising detailed glue-line is modified according to loading demands according to experimental data, is obtained
To large and complex structure performance accurate prediction models.
5. prediction according to claim 1 is containing the method for being glued large and complex structure military service performance, which is characterized in that described
The large and complex structure performance accurate prediction models are trained using machine learning method, the large size after being trained is multiple
Miscellaneous structural behaviour quick predict model, specifically includes:
Different load working conditions and different splicing Qu Yuanmo are inputted into the large and complex structure performance accurate prediction models
Type adjusts the parameter of the large and complex structure performance accurate prediction models, and the large and complex structure performance after being trained is fast
Fast prediction model.
6. prediction according to claim 1 is containing the method for being glued large and complex structure military service performance, which is characterized in that described
The parameter for being glued area and load working condition are input to the large and complex structure performance quick predict model after the training, obtained big
The detailed forecasts of type labyrinth are as a result, specifically include:
The parameter for being glued area and load working condition are input to the large and complex structure performance prediction model after the training, obtained big
The strength and stiffness data of type labyrinth;
The safety of prediction result is determined using the failure criteria of adhesive according to the strength and stiffness data.
7. a kind of prediction is containing the system for being glued large and complex structure military service performance characterized by comprising
First obtains module, for obtaining the material category and the trade mark of two adherends and adhesive;
Elasticity modulus determining module obtains two and is glued for the material category and the trade mark according to the adherend and adhesive
The elasticity modulus of object and adhesive;
Second obtains module, for obtaining the thickness information of two adherends and glue-line;
Meta-model constructs module, for the elasticity modulus and thickness information according to described two adherends and adhesive, constructs glue
Connect the meta-model in area;
Large and complex structure model construction module, for establishing and being based on effective stiffness matrix according to the meta-model for being glued area
, the large and complex structure model not comprising detailed glue-line of hyperelement;
Correction module, for being modified according to experimental data to the large and complex structure model not comprising detailed glue-line,
Obtain large and complex structure performance accurate prediction models;
Training module, for being trained to the large and complex structure performance accurate prediction models using machine learning method,
Large and complex structure performance quick predict model after being trained;
Prediction result determining module, for the parameter and load working condition of being glued area to be input to the large complicated knot after the training
Structure performance quick predict model, obtains the detailed forecasts result of large and complex structure.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910231729.2A CN109918838A (en) | 2019-03-26 | 2019-03-26 | A kind of method and system of prediction large and complex structure military service performance containing splicing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910231729.2A CN109918838A (en) | 2019-03-26 | 2019-03-26 | A kind of method and system of prediction large and complex structure military service performance containing splicing |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109918838A true CN109918838A (en) | 2019-06-21 |
Family
ID=66966860
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910231729.2A Pending CN109918838A (en) | 2019-03-26 | 2019-03-26 | A kind of method and system of prediction large and complex structure military service performance containing splicing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109918838A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111523268A (en) * | 2020-04-22 | 2020-08-11 | 四川大学 | Material fatigue-resistant optimization design method based on machine learning |
CN112765734A (en) * | 2021-01-27 | 2021-05-07 | 同济大学 | Method for predicting curing deformation and residual internal stress of door cover part of adhesive heterogeneous vehicle body |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090192766A1 (en) * | 2008-01-30 | 2009-07-30 | Airbus Espana, S.L. | Method for simulating the behavior of a bonded joint of two parts |
CN105069241A (en) * | 2015-08-19 | 2015-11-18 | 山东大学 | Step-by-step analysis and prediction method for dynamic performances of rubber material structure |
CN108256281A (en) * | 2018-03-26 | 2018-07-06 | 中国矿业大学 | A kind of intensity prediction method for considering overlap joint interface topography and overlapping object graded properties |
-
2019
- 2019-03-26 CN CN201910231729.2A patent/CN109918838A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090192766A1 (en) * | 2008-01-30 | 2009-07-30 | Airbus Espana, S.L. | Method for simulating the behavior of a bonded joint of two parts |
CN105069241A (en) * | 2015-08-19 | 2015-11-18 | 山东大学 | Step-by-step analysis and prediction method for dynamic performances of rubber material structure |
CN108256281A (en) * | 2018-03-26 | 2018-07-06 | 中国矿业大学 | A kind of intensity prediction method for considering overlap joint interface topography and overlapping object graded properties |
Non-Patent Citations (4)
Title |
---|
王跃 等: "胶层Ⅰ/Ⅱ型断裂破坏内聚单元参数的确定和应用", 《玻璃钢/复合材料》 * |
程起有 等: "ANN技术在复合材料胶接修理分析中的应用", 《直升机技术》 * |
程起有 等: "基于神经网络的复合材料胶接修补参数优化", 《计算机仿真》 * |
赵波 等: "一种考虑胶瘤的胶接简化有限元模型:应力与刚度分析", 《机械强度》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111523268A (en) * | 2020-04-22 | 2020-08-11 | 四川大学 | Material fatigue-resistant optimization design method based on machine learning |
CN111523268B (en) * | 2020-04-22 | 2021-06-01 | 四川大学 | Material fatigue-resistant optimization design method based on machine learning |
CN112765734A (en) * | 2021-01-27 | 2021-05-07 | 同济大学 | Method for predicting curing deformation and residual internal stress of door cover part of adhesive heterogeneous vehicle body |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Sanayei et al. | Structural element stiffness identification from static test data | |
Sanayei et al. | Damage assessment of structures using static test data | |
CN104573392B (en) | A kind of welding spot fatigue Forecasting Methodology | |
Wang et al. | A computationally efficient dynamical model of fluidic soft actuators and its experimental verification | |
CN106096257B (en) | A kind of non-linear cable elements analysis method and system | |
Stapleton et al. | Design of functionally graded joints using a polyurethane-based adhesive with varying amounts of acrylate | |
CN104091033B (en) | Bridge static(al) correction method for finite element model based on hyperelement combined with virtual deformation method | |
CN109918838A (en) | A kind of method and system of prediction large and complex structure military service performance containing splicing | |
Overgaard et al. | A methodology for the structural analysis of composite wind turbine blades under geometric and material induced instabilities | |
Madenci et al. | Advances in peridynamics | |
Aliabadi et al. | Fracture mechanics analysis of cracking in plain and reinforced concrete using the boundary element method | |
CN110287593A (en) | One kind being bolted model interface parameter identification method | |
Neumayer et al. | An explicit cohesive element combining cohesive failure of the adhesive and delamination failure in composite bonded joints | |
Shen et al. | Experimental path-following of equilibria using Newton’s method. Part I: Theory, modelling, experiments | |
Nie et al. | Cable anchorage system modeling methods for self-anchored suspension bridges with steel box girders | |
Yang et al. | Hybrid simulation of a zipper‐braced steel frame under earthquake excitation | |
Nonaka et al. | Dynamic response of half-through steel arch bridge using fiber model | |
Zgoul | Use of artificial neural networks for modelling rate dependent behaviour of adhesive materials | |
Kłosowski et al. | Finite element description of nonlinear viscoelastic behaviour of technical fabric | |
AsadiGorgi et al. | Effects of all-over part-through cracks on the aeroelastic characteristics of rectangular panels | |
Wang et al. | Nonlinear 3D numerical computations for the square membrane versus experimental data | |
Ingraffea et al. | Representation and probability issues in the simulation of multi-site damage | |
Nardin et al. | Application of artificial neural network for identification of parameters of a constitutive law for soils | |
Fallah et al. | Analytical–numerical study of interfacial stresses in plated beams subjected to pulse loading | |
Jirasek et al. | Model-based active vibration control for next generation bridges using reduced finite element models |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190621 |
|
RJ01 | Rejection of invention patent application after publication |