CN112784495B - Mechanical structure real-time fatigue life prediction method based on data driving - Google Patents
Mechanical structure real-time fatigue life prediction method based on data driving Download PDFInfo
- Publication number
- CN112784495B CN112784495B CN202110116397.0A CN202110116397A CN112784495B CN 112784495 B CN112784495 B CN 112784495B CN 202110116397 A CN202110116397 A CN 202110116397A CN 112784495 B CN112784495 B CN 112784495B
- Authority
- CN
- China
- Prior art keywords
- crack
- boundary
- representing
- displacement
- component
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000006073 displacement reaction Methods 0.000 claims abstract description 147
- 238000003062 neural network model Methods 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 22
- 238000013528 artificial neural network Methods 0.000 claims abstract description 19
- 230000009977 dual effect Effects 0.000 claims abstract description 18
- 210000002569 neuron Anatomy 0.000 claims description 39
- 239000000243 solution Substances 0.000 claims description 32
- 239000003637 basic solution Substances 0.000 claims description 19
- 238000012360 testing method Methods 0.000 claims description 17
- 239000002585 base Substances 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 12
- 230000003993 interaction Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 3
- 238000004088 simulation Methods 0.000 description 6
- 230000004044 response Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000003754 machining Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 230000009466 transformation Effects 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
- G06F30/27—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
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a real-time fatigue life prediction method for a mechanical structure based on data driving, which comprises the following steps: firstly, obtaining the value ranges of an index m and a coefficient c of a corresponding Paris model in a mechanical structure, and generating a series of indexes and coefficients; second, observation points are randomly acquired on the mechanical structure for a set of (m)q，cq) Circulating, and performing fatigue crack propagation analysis on the observation points by using a dual reciprocity boundary element method to obtain displacement and real-time fatigue life information of the observation points to form a data information pair; until all (m, c) are traversed, obtaining a data set; inputting the data set into a BP neural network for training to obtain a BP neural network model; and finally, acquiring the displacement of the observation point in the mechanical structure, and inputting the displacement of the observation point into the BP neural network model to obtain the real-time fatigue life information of the observation point. The fatigue life of the mechanical structure can be predicted only by the displacement of the observation point, and a large amount of time and cost are saved.
Description
Technical Field
The invention relates to the technical field of detection and prediction of fatigue cracks of mechanical products, in particular to a real-time fatigue life prediction method of a mechanical structure based on data driving.
Background
In the machining process of mechanical materials, defects such as cavities, inclusions, cracks and the like can be generated inevitably, and under the conditions of complex working conditions and cyclic load, stress is excessively concentrated to cause fatigue fracture, and even disastrous results can be caused. The failure caused by fatigue fracture is statistically between 50% and 90% of the total mechanical failure, which is related to the presence of defects in the engineered structure. In order to prevent safety accidents and reduce economic losses, prediction of the remaining fatigue life of mechanical structures is a major concern.
With the development of computational mechanics and computer technology, more and more numerical methods are used to simulate the crack propagation of mechanical structures, and thus predict the fatigue life of the mechanical structures. By using the numerical method, the crack propagation process can be accurately simulated, and the crack propagation rule is mastered, so that the strength analysis and the service life prediction of the mechanical structure are performed. Of these numerical methods, the two most effective methods are the finite element method and the boundary element method, respectively.
When the finite element method is used for carrying out crack propagation simulation, the stress and strain precision are lower than the displacement by one order, a high-density grid needs to be arranged in a high-stress area at the tip of the crack, the crack can be limited on a grid line only, a crack surface is set as a boundary of a unit, a crack tip is set as a node of the unit, and the grid needs to be divided again along with the crack propagation when the crack propagation is simulated. The step function and the crack tip progressive displacement field function which can reflect discontinuity are added into the conventional unit shape function. The grid and the crack are independent from each other, the grid is not required to be reconstructed when the crack propagates, and the problem of strong discontinuity of the crack-containing body can be conveniently analyzed. However, in order to obtain high precision simulation results of crack propagation, a large number of meshes are still required. The boundary element method is used as a semi-analytic numerical method, the utilized surface force and displacement have the advantage of same-order precision, the precision of mechanical structure strain and stress analysis can be improved, and the method has more advantages in crack propagation simulation. The singularity of the basic solution of the boundary element method is more suitable for predicting the stress singularity problem of the real-time fatigue life of the mechanical structure.
Due to the use of the Paris model, uncertainty of mechanical structure parameters, and the fact that coefficients in the Paris model need to be obtained by fitting through experimentally measuring a large number of data points, the coefficients are random. Often, a priori knowledge is applied if perturbation methods are used for the processing. Due to the transformation of parameters in the crack propagation process, the perturbation technology is difficult to be combined with a numerical method and is applied to crack propagation analysis. Therefore, a great challenge is brought to the dual reciprocity boundary element method for predicting the real-time fatigue life of the mechanical structure.
In recent years, machine learning techniques have attracted attention. Back Propagation Neural Networks (BPNN), one of the artificial neural networks, have been successfully applied to engineering prediction problems based on a large number of physical experiments or numerical simulation data sets. The combination of computational methods with neural network techniques has shown great potential in processing crack analysis. However, there is little literature on real-time fatigue life prediction of mechanical structures.
Disclosure of Invention
Aiming at the defects in the background technology, the invention provides a real-time fatigue life prediction method for a mechanical structure based on data driving, and solves the technical problem that the fatigue life of the mechanical structure cannot be predicted in the prior art, so that the detection precision of the mechanical structure is reduced.
The technical scheme of the invention is realized as follows:
a real-time fatigue life prediction method of a mechanical structure based on data driving comprises the following steps:
the method comprises the following steps: obtaining the value ranges of an index m and a coefficient c of a corresponding Paris model in the mechanical structure according to a test means, and generating a series of indexes and coefficients by adopting a priority coefficient in the value ranges of the index m and the coefficient c respectively to obtain Q groups of Paris model parameter pairs (m is a parameter pairq，cq) Wherein Q is 1,2, …, Q;
step two: setting observation points on the mechanical structure, aiming at the q group of Paris model parameter pairs (m)q，cq) Circulating, carrying out fatigue crack propagation analysis on the mechanical structure by using a dual reciprocity boundary element method, obtaining the displacement of the observation point and the real-time fatigue life information of the mechanical structure, and forming a data information pair by using the displacement of the observation point and the real-time fatigue life information of the mechanical structure;
step three: circularly executing the step two until the Q groups of Paris model parameter pairs are traversed to obtain a data set, and dividing the data set into a training set and a testing set, wherein the data set comprises a series of data information pairs;
step four: inputting the training set into a BP neural network for training to obtain a BP neural network model, and verifying the BP neural network model by using a test set;
step five: and acquiring the displacement of an observation point in the mechanical structure, and inputting the displacement of the observation point into the BP neural network model to obtain the real-time fatigue life information of the mechanical structure.
The method for carrying out fatigue crack propagation analysis on the mechanical structure by using the dual reciprocity boundary element method to obtain the displacement of an observation point and the real-time fatigue life information comprises the following steps:
s2.1, constructing dual reciprocal boundary integral equations, including a displacement boundary integral equation and a surface force boundary integral equation;
s2.2, respectively carrying out discrete processing on the displacement boundary integral equation and the surface force boundary integral equation to obtain:
wherein H*Integral representing basic solution of displacement, G*Represents the integral of the surface force base solution, S represents the non-crack boundary,
the upper boundary of the crack is represented,
denotes the lower boundary of the crack, uSIndicating the displacement on the non-crack boundary,
indicating the displacement on the upper boundary of the crack,
indicates the displacement at the lower boundary of the crack, tSIndicating the face force at the non-crack boundary,
indicating the face force at the upper boundary of the crack,
representing the face force at the lower boundary of the crack;
s2.3, applying a boundary condition to the formula (5), and finishing to obtain:
wherein A is*Representing a known amount of displacement or surface force, xSRepresenting an unknown displacement or face force at a non-crack boundary,
representing an unknown displacement or surface force on the upper boundary of the crack,
representing an unknown displacement or surface force at the lower boundary of the crack, bSIndicating known displacements or surface forces at non-crack boundariesAnd the resulting components are then combined to form a combined component,
representing the combined resulting components of known displacements or face forces at the upper boundary of the crack,
representing the combined components of known displacements or surface forces at the lower boundary of the crack;
solving the formula (6) by a Gaussian elimination method to obtain the displacement and the surface force of the boundary, and substituting the displacement and the surface force of the boundary into the formula (1) to obtain the displacement of the observation point;
s2.4, calculating a mixed modal stress intensity factor by utilizing an interaction integral algorithm according to the displacement of the boundary:
wherein M is(1,2)The integral of the interaction is represented by,
in order to interact with the strain energy density,
representing the i-direction displacement component in state two,
denotes the i-direction displacement component, x, at state one1Representing the component of the interior point x, Γ representing the path of the interaction integral, AεIs a circle with the center of a crack tip and the radius of R,
which represents the strain in the first state of the state,
which represents the strain in the second state,
the stress in the first state is shown,
representing the stress in state two; the first state represents the real state of the mechanical structure, and the second state represents the auxiliary state;
s2.5, converting the formula (7) into:
wherein E is a Young's model,
representing the type I stress intensity factor in state one,
represents the type I stress intensity factor in the second state,
representing the type II stress intensity factor in state one,
representing the type II stress intensity factor in state two,
representing the type III stress intensity factor in state one,
representing the type III stress intensity factor in the second state;
s2.6, calculating a mixed modal stress intensity factor delta Keq：
Wherein, KIDenotes the type I stress intensity factor, KIIRepresenting a type II stress intensity factor;
s2.7, determining the crack direction by adopting a maximum circumferential stress criterion:
wherein theta is a crack propagation angle, and sign represents a sign function;
s2.8, stress intensity factor delta K according to mixed modeeqDetermining the crack propagation rate, wherein for the Paris model, the crack propagation rate is as follows:
wherein a is the crack length and N is the stress cycle number;
s2.9, depending on the initial length a of the crack0The critical length L of the crack, the crack growth increment Deltaa in the crack growth step, and the total number of growth steps is
During crack propagation, for Δ K in the t stepeqCalculating the corresponding load cycle number DeltaNt：
Wherein T is more than or equal to 1 and less than or equal to T;
s2.10, load cycle times Delta NTAs real-time fatigue life information for the mechanical structure.
The displacement boundary integral equation comprises a displacement boundary integral equation corresponding to a non-crack boundary and a displacement boundary integral equation corresponding to a crack upper boundary; the surface force boundary integral equation refers to a surface force boundary integral equation corresponding to the lower boundary of the crack;
s2.1.1, constructing a displacement boundary integral equation corresponding to the source point falling on the non-crack boundary:
wherein,
representing a source point on a non-crack boundary, x representing a field point on a non-crack boundary, x+Field points, x, representing the upper boundary of the crack-The field point representing the lower boundary of the crack,
the non-crack boundary shape factor is represented,
a fundamental solution of displacement representing a field point falling on a non-crack boundary,
representing the base solution of the face force with the field point falling on the non-crack boundary,
a basic solution of the displacement representing the boundary where the field point falls on the crack,
a basic solution of the displacement representing the field point falling on the lower boundary of the crack,
the base solution of the face force representing the field point falling on the upper boundary of the crack,
the base solution of the face force representing the field point falling at the lower boundary of the crack,
s represents the non-crack boundary and,
the upper boundary of the crack is represented,
the lower boundary of the crack is represented,
representing the j-direction displacement component, u, at the origin point on the non-crack boundaryj(x) Representing the j-direction displacement component, t, of field points at non-crack boundariesj(x) Representing the force component of the j-direction surface at the field point on the non-crack boundary, uj(x+) A j-direction displacement component, u, at a field point representing an upper boundary of the crackj(x-) A j-direction displacement component, t, at a field point representing the lower boundary of the crackj(x+) Representing the force component of the j-direction surface at the upper boundary of the crack, tj(x-) Representing the j-direction surface force component at the lower boundary of the crack,
the basic solution of the displacement is represented,
the basic solution of the surface force is shown,
the field point y' is x, x+Or x-，
Representing a delta function, v representing the Poisson's ratio, r representing the distance of the source point and the field point, n representing a normal vector, niComponent representing the direction of normal vector i, njRepresenting the component in the direction of the normal vector j,
representing a source point
Of i-directional component, y'iThe i-direction component of the field point y' is represented,
representing a source point
Of j-directional component, y'jRepresents the j-direction component of field point y';
the formula (1) is sorted and combined to obtain:
s2.1.2, constructing a displacement boundary integral equation corresponding to the boundary of the source point falling on the crack:
wherein,
indicating the origin of the boundary that falls on the crack,
indicates the source point falling on the lower boundary of the crack, and
the coordinates representing the two source points are the same,
the upper boundary shape factor of the crack is represented,
the crack lower boundary shape factor is expressed,
a j-direction displacement component at the origin of the boundary on the crack,
representing the j-direction displacement component at the origin point representing the upper boundary of the crack,
a fundamental solution of displacement representing a field point falling on a non-crack boundary,
representing the base solution of the face force with the field point falling on the non-crack boundary,
a basic solution of the displacement representing the field point falling on the lower boundary of the crack,
a base solution of the face force representing the field point falling on the upper boundary of the crack;
s2.1.3, constructing a surface force boundary integral equation corresponding to the boundary of the source point falling on the lower boundary of the crack:
wherein,
the j-direction surface force component at the origin of the crack lower boundary,
the j-direction surface force component at the origin of the boundary on the crack,
i-directional component, t, at the origin of the lower boundary of the crackk(x) Representing the k-direction surface force component at the field point on the non-crack boundary,
a high order fundamental solution of displacement representing a field point falling on a non-crack boundary,
high order surface force fundamental solution, u, representing a field point falling on a non-crack boundaryk(x) Representing the k-direction displacement component, t, of the field points at the non-crack boundariesk(x+) Representing the surface force component in the k-direction, t, at the boundary field point above the crackk(x-) Represents the surface force component in the k direction at the lower boundary of the crack, uk(x+) Representing the displacement component in the k-direction, u, at the boundary field point above the crackk(x-) Representing the k-direction displacement component at the boundary field point below the crack,
g' represents a shear modulus and a shear modulus,
representing a source point
K-direction component of (2), y'kRepresenting the k-direction component, n, of the field point ykThe component in the direction of the normal vector k,
during the crack propagation, the current crack length is a0Setting the critical length as L, the expansion amount per step as delta a, and totaling the expansion
Step, pair formula
After deformation, carrying out two-end integration to obtain the load cycle times delta N of the first step of expansion1：
Performing the second expansion to obtain the load cycle times Delta N of the second expansion2：
Carrying out third-step expansion to obtain load cycle times delta N of the third-step expansion3：
The same fourth step is expanded until the T-1 expansion, and the corresponding load cycle number is DeltaN4...△NT-1；
Expanding the T step to obtain the load cycle times delta N of the expansion of the T stepT：
Thus, the total life of the mechanical structure is Δ N ═ Δ N1+△N2+...+△NT-1+△NT。
The method for inputting the training set into the BP neural network for training to obtain the BP neural network model comprises the following steps:
s4.1, network initialization: determining the node numbers of the input layer, the hidden layer and the output layer, and initializing the connection weight Wi'j'And Wj'k'And thresholds for the hidden layer and the output layer, i ' ═ 1,2, …, n ', the number of neurons in the input layer, j ' ═ 1,2, …, l, the number of neurons in the hidden layer, k ' ═ 1,2, …, m ', the number of neurons in the output layer;
s4.2, calculating the output of the hidden layer node:
wherein h isj'Representing the output, x, of neuron ji'Representing the input component, ωi'j'∈Wi'j'Represents a weighted weight value, θj'Threshold for neuron j', f (-) is activation function;
s4.3, calculating the output of the output layer node:
wherein, yk'Represents the output of the neuron k', ωj'k'∈Wj'k'Represents a weighted weight value, θk'A threshold for neuron k';
s4.4, calculating the error e between the expected output and the predicted output of the z-th training dataz：
Wherein, yd,k'For the expected output value, y, of the neuron k' at the output layerk'For the predicted output value of neuron k 'at the output layer, the total error of the Z input signals is E':
s4.5, updating the connection weight of the neural network:
ωi'j'(p+1)＝ωi'j'(p)+△ωi'j'(p) (21)；
where p is the number of iterations, Δ ωi'j'(p)＝η×yi'(p)×λj'(p) is the adjustment part of the connection weight in p +1 iterations, η is the learning rate, yi'(p) is the output signal of neuron i',
error gradient, X, for neuron jj'(p) is the weighted input to neuron j', δj'(p) is the error for neuron j';
and S4.6, judging whether the iteration number reaches the maximum iteration number or whether the total error E' of the Z input signals is smaller than an error threshold, if so, outputting the BP neural network model, and if not, returning to the step S4.2 to perform the next iteration.
The beneficial effect that this technical scheme can produce: the method generates a series of structural responses and corresponding fatigue life data sets by using a dual reciprocity boundary element method, trains and tests a BP neural network prediction model by using the data sets, establishes a real-time fatigue life prediction model related to structural responses of observation point positions and fatigue life, and determines real-time fatigue life information by testing response tests of the observation point positions of a mechanical structure in actual engineering, thereby avoiding the problem of detection precision reduction and saving a large amount of time and cost.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 shows the topology of the BP neural network according to the present invention.
FIG. 2 is a flow chart of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
As shown in fig. 2, an embodiment of the present invention provides a data-driven method for predicting a fatigue life of a mechanical structure in real time, where a value range is obtained through a test method because an index m and a coefficient c of a Paris model corresponding to the mechanical structure are uncertain, and a series of data about m and c are produced according to a priority coefficient in the value range and paired (m, c). Aiming at the mechanical structure, observation points are taken at easy-to-test parts, each group (m, c) is respectively taken for circulation, a dual reciprocity boundary element method is used for fatigue crack propagation analysis, a displacement value of the observation points and corresponding real-time fatigue life information are obtained in each step of propagation, a data information pair is formed and stored, and the data information pair is stored in one unit. Until all (m, c) cycles are completed, the unit is divided into a training unit and a testing unit. And then training the BP neural network model by utilizing a training unit. And verifying the trained BP neural network model by using a test unit. Therefore, the relation between the displacement of the observation point and the real-time fatigue life information is completely established, and the corresponding real-time fatigue life information can be obtained by utilizing the established BP neural network model only through the displacement of the observation point. The method comprises the following specific steps:
the method comprises the following steps: obtaining the value ranges of an index m and a coefficient c of a corresponding Paris model in the mechanical structure according to a test means, and generating a series of indexes and coefficients by adopting a priority coefficient in the value ranges of the index m and the coefficient c respectively to obtain Q groups of Paris model parameter pairs (m is a parameter pairq，cq) Wherein Q is 1,2, …, Q; in an actual mechanical structure, the Paris model index m and the coefficient c are uncertain, and the fatigue crack propagation is difficult to directly simulate. Since perturbation methods require a priori knowledge, the distribution function of the model parameters is unknown. For a particular mechanical structure, experimental measurements are taken, and the data fitted,and carrying out fatigue crack propagation analysis according to the index m and the coefficient c of the Paris model obtained by fitting, and predicting the fatigue life. However, batch trial fitting is difficult to perform for a large number of mechanical structures, often lacking time and cost. Therefore, the embodiment of the invention develops a data-driven mechanical structure real-time fatigue life prediction boundary element method.
Step two: setting observation points on the mechanical structure, aiming at the q group Paris model parameter pair (m)q，cq) And circulating, carrying out fatigue crack propagation analysis on the mechanical structure by using a dual reciprocity boundary element method, obtaining the displacement of the observation point and the real-time fatigue life information of the mechanical structure, and forming a data information pair by using the displacement of the observation point and the real-time fatigue life information of the mechanical structure.
S2.1, constructing dual reciprocal boundary integral equations, including a displacement boundary integral equation and a surface force boundary integral equation; the displacement boundary integral equation comprises a displacement boundary integral equation corresponding to a non-crack boundary and a displacement boundary integral equation corresponding to a crack upper boundary; the surface force boundary integral equation refers to a surface force boundary integral equation corresponding to the lower boundary of the crack.
S2.1.1, constructing a displacement boundary integral equation corresponding to the source point falling on the non-crack boundary:
wherein,
representing a source point on a non-crack boundary, x representing a field point on a non-crack boundary, x+Field points, x, representing the upper boundary of the crack-The field point representing the lower boundary of the crack,
the non-crack boundary shape factor is represented,
a fundamental solution of displacement representing a field point falling on a non-crack boundary,
representing the base solution of the face force with the field point falling on the non-crack boundary,
a basic solution of the displacement representing the boundary where the field point falls on the crack,
a basic solution of the displacement representing the field point falling on the lower boundary of the crack,
the base solution of the face force representing the field point falling on the upper boundary of the crack,
the base solution of the face force representing the field point falling at the lower boundary of the crack,
s represents the non-crack boundary and,
the upper boundary of the crack is represented,
the lower boundary of the crack is represented,
representing the j-direction displacement component, u, at the origin point on the non-crack boundaryj(x) Representing the j-direction displacement component, t, of field points at non-crack boundariesj(x) Representing the force component of the j-direction surface at the field point on the non-crack boundary, uj(x+) A j-direction displacement component, u, at a field point representing an upper boundary of the crackj(x-) Indicating the lower boundary of the crackThe j-direction displacement component, t, at the field pointj(x+) Representing the force component of the j-direction surface at the upper boundary of the crack, tj(x-) Representing the j-direction surface force component at the lower boundary of the crack,
the basic solution of the displacement is represented,
the basic solution of the surface force is shown,
the field point y' is x, x+Or x-，
Representing a delta function, v representing the Poisson's ratio, r representing the distance of the source point and the field point, n representing a normal vector, niComponent representing the direction of normal vector i, njRepresenting the component in the direction of the normal vector j,
representing a source point
Of i-directional component, y'iRepresenting the i-direction component of a field point y', x+Or x-，
Representing a source point
Of j-directional component, y'jRepresenting the j-direction component of field point y'.
The formula (1) is sorted and combined to obtain:
s2.1.2, constructing a displacement boundary integral equation corresponding to the boundary of the source point falling on the crack:
wherein,
indicating the origin of the boundary that falls on the crack,
indicates the source point falling on the lower boundary of the crack, and
(since the crack upper and lower surfaces coincide with each other, the coordinates of the two source points are the same),
the upper boundary shape factor of the crack is represented,
the crack lower boundary shape factor is expressed,
a j-direction displacement component at the origin of the boundary on the crack,
representing the j-direction displacement component at the origin point representing the upper boundary of the crack,
a fundamental solution of displacement representing a field point falling on a non-crack boundary,
representing the base solution of the face force with the field point falling on the non-crack boundary,
a basic solution of the displacement representing the field point falling on the lower boundary of the crack,
the base solution of the face force representing the field point falling on the upper boundary of the crack.
S2.1.3, constructing a surface force boundary integral equation corresponding to the boundary of the source point falling on the lower boundary of the crack:
wherein,
the j-direction surface force component at the origin of the crack lower boundary,
the j-direction surface force component at the origin of the boundary on the crack,
i-directional component, t, at the origin of the lower boundary of the crackk(x) Representing the k-direction surface force component at the field point on the non-crack boundary,
a high order fundamental solution of displacement representing a field point falling on a non-crack boundary,
high order surface force fundamental solution, u, representing a field point falling on a non-crack boundaryk(x) Representing the k-direction displacement component, t, of the field points at the non-crack boundariesk(x+) Representing the surface force component in the k-direction, t, at the boundary field point above the crackk(x-) Represents the surface force component in the k direction at the lower boundary of the crack, uk(x+) Representing the displacement component in the k-direction, u, at the boundary field point above the crackk(x-) Representing the k-direction displacement component at the boundary field point below the crack,
g' represents a shear modulus and a shear modulus,
representing a source point
K-direction component of (2), y'kRepresenting the k-direction component, n, of the field point ykThe component in the direction of the normal vector k,
s2.2, respectively carrying out discrete processing on the displacement boundary integral equation and the surface force boundary integral equation to obtain:
wherein H*Representing the basic solution integral of the displacement, G*Representing the fundamental solution integral of surface force, uSIndicating the displacement on the non-crack boundary,
indicating the displacement on the upper boundary of the crack,
indicates the displacement at the lower boundary of the crack, tSIndicating the face force at the non-crack boundary,
indicating the face force at the upper boundary of the crack,
indicating the face force at the lower boundary of the crack.
S2.3, applying a boundary condition to the formula (5), adjusting the unknown quantity to the left side of the formula, adjusting the known quantity to the right side of the formula, and finishing to obtain:
wherein A is*Indicating a known quantity of displacement or surface force, xSRepresenting an unknown displacement or face force at a non-crack boundary,
representing an unknown displacement or surface force on the upper boundary of the crack,
representing an unknown displacement or surface force at the lower boundary of the crack, bSRepresenting the combined components of known displacements or face forces at non-crack boundaries,
representing the combined resulting components of known displacements or face forces at the upper boundary of the crack,
representing the combined resulting components of known displacement or face forces at the lower boundary of the crack.
Solving the formula (6) by a Gaussian elimination method to obtain unknown quantity, configuring the unknown quantity to obtain the displacement and the surface force of the boundary, and substituting the displacement and the surface force of the boundary into the formula (1) to obtain the displacement of the observation point.
And S2.4, in order to evaluate the expansion of the fatigue crack, a dual reciprocity boundary element method is required to be used for calculating a stress intensity factor, and the stress intensity factor is an important parameter for judging whether the crack expands or not. Further, the crack propagation direction and propagation amount are also calculated from the stress intensity factor. Calculating a mixed modal stress intensity factor by utilizing an interaction integral algorithm, wherein the expression of an interaction integral M is as follows:
wherein M is(1,2)The integral of the interaction is represented by,
in order to interact with the strain energy density,
representing the i-direction displacement component in state two,
denotes the i-direction displacement component, x, at state one1Representing the component of the interior point x, Γ representing the path of the interaction integral, AεIs a circle with the center of a crack tip and the radius of R,
which represents the strain in the first state of the state,
which represents the strain in the second state,
the stress in the first state is shown,
representing the stress in state two; the first state represents the real state of the mechanical structure, and the second state represents the auxiliary state (the second state represents the condition of only I-type cracks or II-type fracture cracks); the displacement, stress and strain in the second state can be obtained by an asymptotic displacement field and a stress field of the crack tip under the condition of only type I or type II fracture, and the displacement, stress and strain in the second state can be obtained by a post-treatment process of a dual reciprocity boundary element method.
S2.5, converting the formula (7) into:
wherein E is a Young's model,
representing the type I stress intensity factor in state one,
represents the type I stress intensity factor in the second state,
representing the type II stress intensity factor in state one,
representing the type II stress intensity factor in state two,
representing the type III stress intensity factor in state one,
the type III stress intensity factor in state two is shown.
S2.6 fracture toughness K according to mechanical structure materialICAnd Δ KeqJudging whether the crack is expanded or not according to the relation, and calculating a mixed modal stress intensity factor delta Keq：
Wherein, KIDenotes the type I stress intensity factor, KIIRepresenting the type II stress intensity factor.
S2.7, determining the crack propagation direction in the case of crack propagation. Determining the crack direction using the maximum circumferential stress criterion:
where θ is the crack propagation angle and sign represents the sign function.
S2.8, when the fatigue life is predicted by using a dual boundary element method, a fatigue crack propagation model for describing a crack propagation process needs to be established. According to mixed mode stress intensity factor delta KeqDetermining the crack propagation rate, wherein for the generalized Paris model, the crack propagation rate is as follows:
wherein a is the crack length and N is the number of stress cycles. m is Paris model index, c is Paris model constant, and Delta KeqThe mixed modal stress intensity factor is influenced by loading conditions and material performance, m and c need to be obtained through fatigue crack propagation rate test data fitting, and often only the range of m and c can be obtained.
S2.9, depending on the initial length a of the crack0The critical length L of the crack, the crack growth increment Deltaa in the crack growth step, and the total number of growth steps is
During crack propagation, for Δ K in the t stepeqCalculating the corresponding load cycle number DeltaNt：
Wherein T is more than or equal to 1 and less than or equal to T.
S2.10, load cycle times Delta NTAs real-time fatigue life information for the observation point.
During the crack propagation, the current crack length is a0Setting the critical length as L (failure when the crack propagation length reaches L), the propagation amount per step as Delta a, and the total propagation
Integrating two ends of the formula (11b) to obtain the load cycle number delta N of the first step1：
The same second step of expansion can obtain:
and expanding in the third step to obtain:
the same fourth step is extended until the T-1 step, and the corresponding load cycle times are delta N4...△NT-1；
And T, expanding:
total lifetime N ═ Δ N1+△N2+...+△NT-1。
Observation points are arranged on the non-crack boundary (the observation points are easy to measure by using an experimental instrument), the displacement of the observation points is calculated by using a formula (1) at each step of the expansion, and the displacement of the observation points and the fatigue life composition data of the mechanical structure at the stage are stored.
Step three: and circularly executing the second step until the Q groups of Paris model parameter pairs are traversed to obtain a data set, and dividing the data set into a training set and a testing set, wherein the data set comprises a series of data information pairs.
Step four: inputting the training set into a BP neural network for training to obtain a BP neural network model, and verifying the BP neural network model by using the test set. In order to establish a data-driven prediction model, a series of data of pair model index m and coefficient c is generated according to a priority coefficient according to a data range obtained by an experiment, for each coefficient pair (m, c), a large number of data sets of structural response and fatigue life pairing of an observation point position are calculated by using a dual reciprocity boundary element method, and a back propagation neural network is constructed to learn the data sets. The back propagation neural network is a multi-layer forward neural network based on an error back propagation technology and consists of an input layer, one or more hidden layers and an output layer.
The circles in fig. 1 are referred to as bias nodes. Training a BP neural network by using a data set, and specifically comprising the following steps:
s4.1, network initialization: determining the node numbers of the input layer, the hidden layer and the output layer, and initializing the connection weight Wi'j'And Wj'k'And the thresholds of the hidden layer and the output layer, i ' ═ 1,2, …, n ', the number of neurons in the input layer, j ' ═ 1,2, …, l, the number of neurons in the hidden layer, k ' ═ 1,2, …, m ', the number of neurons in the output layer;
s4.2, calculating the output of the hidden layer node:
wherein h isj'Representing the output, x, of neuron ji'Representing the input component, ωi'j'∈Wi'j'Represents a weighted weight value, θj'Threshold for neuron j', f (-) is activation function;
s4.3, calculating the output of the output layer node:
wherein, yk'Representing the output of k' neurons, ωj'k'∈Wj'k'Represents a weighted weight value, θk'A threshold for neuron k';
s4.4, calculating the error e between the expected output and the predicted output of the z-th training dataz：
Wherein, yd,k'For the expected output value, y, of the neuron k' at the output layerk'For the predicted output value of neuron k 'at the output layer, the total error of the Z input signals is E':
s4.5, updating the connection weight of the neural network:
ωi'j'(p+1)＝ωi'j'(p)+△ωi'j'(p) (21)
where p is the number of iterations, Δ ωi'j'(p)＝η×yi'(p)×λj'(p) is the adjustment part of the connection weight in p +1 iterations, η is the learning rate, yi'(p) is the output signal of neuron i',
error gradient, X, for neuron jj'(p) is the weighted input to neuron j', δj'(p) is the error for neuron j';
and S4.6, judging whether the iteration number reaches the maximum iteration number or whether the total error E' of the Z input signals is smaller than an error threshold, if so, outputting the BP neural network model, and if not, returning to the step S4.2 to perform the next iteration.
Step five: and acquiring the displacement of an observation point in the mechanical structure, and inputting the displacement of the observation point into the BP neural network model to obtain the real-time fatigue life information of the mechanical structure.
The invention combines the dual reciprocity boundary element method with BP neural network. Firstly, a series of fatigue crack propagation life analysis of mechanical structures is simulated by utilizing a dual reciprocity boundary element method. In each simulation, a series of parameter data sets are generated using a priority coefficient using data taken from the mechanical structure. The obtained data set contains information on the structural response and fatigue life of the fixed observation point. In addition, the data set obtained for each simulation is stored in one unit. Then, the BP neural network model is trained by utilizing a training unit. And verifying the trained BP neural network model by using a test unit. And finally, the trained BP neural network model can be used for realizing the real-time prediction of the fatigue life of the structure when the model parameters are unknown.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (3)
1. A real-time fatigue life prediction method of a mechanical structure based on data driving is characterized by comprising the following steps:
the method comprises the following steps: obtaining the value ranges of an index m and a coefficient c of a corresponding Paris model in the mechanical structure according to a test means, and generating a series of indexes and coefficients by adopting a priority coefficient in the value ranges of the index m and the coefficient c respectively to obtain Q groups of Paris model parameter pairs (m is a parameter pairq，cq) Wherein Q is 1,2, …, Q;
step two: setting observation points on the mechanical structure, aiming at the q group of Paris model parameter pairs (m)q，cq) Circulating, performing fatigue crack propagation analysis on the mechanical structure by using a dual reciprocity boundary element method to obtain the displacement of the observation point and the real-time fatigue life information of the mechanical structure, and positioning the observation pointMoving and combining the real-time fatigue life information of the mechanical structure to form a data information pair;
the method for carrying out fatigue crack propagation analysis on the mechanical structure by using the dual reciprocity boundary element method to obtain the displacement of an observation point and the real-time fatigue life information comprises the following steps:
s2.1, constructing dual reciprocal boundary integral equations, including a displacement boundary integral equation and a surface force boundary integral equation;
the displacement boundary integral equation comprises a displacement boundary integral equation corresponding to a non-crack boundary and a displacement boundary integral equation corresponding to a crack upper boundary; the surface force boundary integral equation refers to a surface force boundary integral equation corresponding to the lower boundary of the crack;
s2.1.1, constructing a displacement boundary integral equation corresponding to the source point falling on the non-crack boundary:
wherein,
representing a source point on a non-crack boundary, x representing a field point on a non-crack boundary, x+Field points, x, representing the upper boundary of the crack-The field point representing the lower boundary of the crack,
the non-crack boundary shape factor is represented,
a fundamental solution of displacement representing a field point falling on a non-crack boundary,
surface force fundamental solution representing field point falling on non-crack boundary，
A basic solution of the displacement representing the boundary where the field point falls on the crack,
a basic solution of the displacement representing the field point falling on the lower boundary of the crack,
the base solution of the face force representing the field point falling on the upper boundary of the crack,
the base solution of the face force representing the field point falling at the lower boundary of the crack,
s represents the non-crack boundary and,
the upper boundary of the crack is represented,
the lower boundary of the crack is represented,
representing the j-direction displacement component, u, at the origin point on the non-crack boundaryj(x) Representing the j-direction displacement component, t, of field points at non-crack boundariesj(x) Representing the force component of the j-direction surface at the field point on the non-crack boundary, uj(x+) A j-direction displacement component, u, at a field point representing an upper boundary of the crackj(x-) A j-direction displacement component, t, at a field point representing the lower boundary of the crackj(x+) Indicating the surface force in the j-direction at the upper boundary of the crackComponent, tj(x-) Representing the j-direction surface force component at the lower boundary of the crack,
the basic solution of the displacement is represented,
the basic solution of the surface force is shown,
the field point y' is x, x+Or x-，
Representing a delta function, v representing the Poisson's ratio, r representing the distance of the source point and the field point, n representing a normal vector, niComponent representing the direction of normal vector i, njRepresenting the component in the direction of the normal vector j,
representing a source point
Of i-directional component, y'iThe i-direction component of the field point y' is represented,
representing a source point
Of j-directional component, y'jRepresents the j-direction component of field point y';
the formula (1) is sorted and combined to obtain:
s2.1.2, constructing a displacement boundary integral equation corresponding to the boundary of the source point falling on the crack:
wherein,
indicating the origin of the boundary that falls on the crack,
indicates the source point falling on the lower boundary of the crack, and
the coordinates representing the two source points are the same,
the upper boundary shape factor of the crack is represented,
the crack lower boundary shape factor is expressed,
a j-direction displacement component at the origin of the boundary on the crack,
representing a j-direction displacement component at a source point representing a boundary on a crack，
A fundamental solution of displacement representing a field point falling on a non-crack boundary,
representing the base solution of the face force with the field point falling on the non-crack boundary,
a basic solution of the displacement representing the field point falling on the lower boundary of the crack,
a base solution of the face force representing the field point falling on the upper boundary of the crack;
s2.1.3, constructing a surface force boundary integral equation corresponding to the boundary of the source point falling on the lower boundary of the crack:
wherein,
the j-direction surface force component at the origin of the crack lower boundary,
the j-direction surface force component at the origin of the boundary on the crack,
i-directional component, t, at the origin of the lower boundary of the crackk(x) Representing the k-direction surface force component at the field point on the non-crack boundary,
a high order fundamental solution of displacement representing a field point falling on a non-crack boundary,
high order surface force fundamental solution, u, representing a field point falling on a non-crack boundaryk(x) Representing the k-direction displacement component, t, of the field points at the non-crack boundariesk(x+) Representing the surface force component in the k-direction, t, at the boundary field point above the crackk(x-) Represents the surface force component in the k direction at the lower boundary of the crack, uk(x+) Representing the displacement component in the k-direction, u, at the boundary field point above the crackk(x-) Representing the k-direction displacement component at the boundary field point below the crack,
g' represents a shear modulus and a shear modulus,
representing a source point
K-direction component of (2), y'kRepresenting the k-direction component, n, of the field point ykThe component in the direction of the normal vector k,
s2.2, respectively carrying out discrete processing on the displacement boundary integral equation and the surface force boundary integral equation to obtain:
wherein H*Integral representing basic solution of displacement, G*Represents the integral of the surface force base solution, S represents the non-crack boundary,
the upper boundary of the crack is represented,
denotes the lower boundary of the crack, uSIndicating the displacement on the non-crack boundary,
indicating the displacement on the upper boundary of the crack,
indicates the displacement at the lower boundary of the crack, tSIndicating the face force at the non-crack boundary,
indicating the face force at the upper boundary of the crack,
representing the face force at the lower boundary of the crack;
s2.3, applying a boundary condition to the formula (5), and finishing to obtain:
wherein A is*Representing a known amount of displacement or surface force, xSRepresenting an unknown displacement or face force at a non-crack boundary,
representing an unknown displacement or surface force on the upper boundary of the crack,
representing an unknown displacement or surface force at the lower boundary of the crack, bSRepresenting the combined components of known displacements or face forces at non-crack boundaries,
representing the combined resulting components of known displacements or face forces at the upper boundary of the crack,
representing the combined components of known displacements or surface forces at the lower boundary of the crack;
solving the formula (6) by a Gaussian elimination method to obtain the displacement and the surface force of the boundary, and substituting the displacement and the surface force of the boundary into the formula (1) to obtain the displacement of the observation point;
s2.4, calculating a mixed modal stress intensity factor by utilizing an interaction integral algorithm according to the displacement of the boundary:
wherein M is(1,2)The integral of the interaction is represented by,
in order to interact with the strain energy density,
representing the i-direction displacement component in state two,
denotes the i-direction displacement component, x, at state one1Representing the component of the inner point x, and Γ representing the interactionPath of integration, AεIs a circle with the center of a crack tip and the radius of R,
which represents the strain in the first state of the state,
which represents the strain in the second state,
the stress in the first state is shown,
representing the stress in state two; the first state represents the real state of the mechanical structure, and the second state represents the auxiliary state;
s2.5, converting the formula (7) into:
wherein E is a Young's model,
representing the type I stress intensity factor in state one,
represents the type I stress intensity factor in the second state,
representing the type II stress intensity factor in state one,
representing the type II stress intensity factor in state two,
representing the type III stress intensity factor in state one,
representing the type III stress intensity factor in the second state;
s2.6, calculating a mixed modal stress intensity factor delta Keq：
Wherein, KIDenotes the type I stress intensity factor, KIIRepresenting a type II stress intensity factor;
s2.7, determining the crack direction by adopting a maximum circumferential stress criterion:
wherein theta is a crack propagation angle, and sign represents a sign function;
s2.8, stress intensity factor delta K according to mixed modeeqDetermining the crack propagation rate, wherein for the Paris model, the crack propagation rate is as follows:
wherein a is the crack length and N is the stress cycle number;
s2.9, depending on the initial length a of the crack0The critical length L of the crack, the crack growth increment Δ a in the crack growth step, and the total number of growth steps is known to be
During crack propagation, for Δ K in the t stepeqCalculating the corresponding load cycle number DeltaNt：
Wherein T is more than or equal to 1 and less than or equal to T;
s2.10, load cycle times delta NTAs real-time fatigue life information for the mechanical structure;
step three: circularly executing the step two until the Q groups of Paris model parameter pairs are traversed to obtain a data set, and dividing the data set into a training set and a testing set, wherein the data set comprises a series of data information pairs;
step four: inputting the training set into a BP neural network for training to obtain a BP neural network model, and verifying the BP neural network model by using a test set;
step five: and acquiring the displacement of an observation point in the mechanical structure, and inputting the displacement of the observation point into the BP neural network model to obtain the real-time fatigue life information of the mechanical structure.
2. The method for predicting the fatigue life of a mechanical structure in real time based on data driving of claim 1, wherein during the crack propagation process, the current crack length is a0Setting the critical length as L and the expansion amount of each step as delta a, totaling the expansion
Step, pair formula
Performing two-end integration after deformation to obtain the load cycle number delta N of the first step of expansion1：
Performing the second expansion to obtain the load cycle times delta N of the second expansion2：
Carrying out third step expansion to obtain load cycle times delta N of the third step expansion3：
The same fourth step is extended until the T-1 step is extended, and the corresponding load cycle number is delta N4...ΔNT-1；
Expanding the T step to obtain the load cycle times delta N of the T step expansionT：
Therefore, the total life of the mechanical structure is Δ N ═ Δ N1+ΔN2+...+ΔNT-1+ΔNT。
3. The method for predicting the real-time fatigue life of the mechanical structure based on the data driving as claimed in claim 1, wherein the method for inputting the training set into the BP neural network for training to obtain the BP neural network model comprises the following steps:
s4.1, network initialization: determining the node numbers of the input layer, the hidden layer and the output layer, and initializing the connection weight Wi'j'And Wj'k'And thresholds for the hidden layer and the output layer, i ' ═ 1,2, …, n ', the number of neurons in the input layer, j ' ═ 1,2, …, l, the number of neurons in the hidden layer, k ' ═ 1,2, …, m ', the number of neurons in the output layer;
s4.2, calculating the output of the hidden layer node:
wherein h isj'Representing the output, x, of neuron ji'Representing the input component, ωi'j'∈Wi'j'Represents a weighted weight value, θj'Is the threshold value for the neuron j',
is an activation function;
s4.3, calculating the output of the output layer node:
wherein, yk'Represents the output of the neuron k', ωj'k'∈Wj'k'Represents a weighted weight value, θk'A threshold for neuron k';
s4.4, calculating the error e between the expected output and the predicted output of the z-th training dataz：
Wherein, yd,k'For the expected output value, y, of the neuron k' at the output layerk'For the predicted output value of neuron k 'at the output layer, the total error of the Z input signals is E':
s4.5, updating the connection weight of the neural network:
ωi'j'(p+1)＝ωi'j'(p)+Δωi'j'(p) (21)；
where p is the number of iterations, Δ ωi'j'(p)＝η×yi'(p)×λj'(p) is the adjustment part of the connection weight in p +1 iterations, η is the learning rate, yi'(p) is the output signal of neuron i',
error gradient, X, for neuron jj'(p) is the weighted input to neuron j', δj'(p) is the error for neuron j';
and S4.6, judging whether the iteration number reaches the maximum iteration number or whether the total error E' of the Z input signals is smaller than an error threshold, if so, outputting the BP neural network model, and if not, returning to the step S4.2 to perform the next iteration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110116397.0A CN112784495B (en) | 2021-01-28 | 2021-01-28 | Mechanical structure real-time fatigue life prediction method based on data driving |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110116397.0A CN112784495B (en) | 2021-01-28 | 2021-01-28 | Mechanical structure real-time fatigue life prediction method based on data driving |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112784495A CN112784495A (en) | 2021-05-11 |
| CN112784495B true CN112784495B (en) | 2021-09-24 |
Family
ID=75759305
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110116397.0A Active CN112784495B (en) | 2021-01-28 | 2021-01-28 | Mechanical structure real-time fatigue life prediction method based on data driving |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112784495B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115169694B (en) * | 2022-07-06 | 2023-05-30 | 天津大学 | Method for realizing subcritical crack dynamic expansion and life prediction |
| CN115169240A (en) * | 2022-07-26 | 2022-10-11 | 南京理工大学 | TiAl alloy fatigue crack propagation life prediction method based on machine learning |
| CN117649901B (en) * | 2023-11-29 | 2024-07-09 | 哈尔滨工业大学 | Interaction integration method for solving stress intensity factors of cracks of revolving body |
| CN117669390A (en) * | 2024-02-01 | 2024-03-08 | 中国石油大学(华东) | Metal full-stage fatigue crack growth prediction method and system based on neural network |
| CN118095017B (en) * | 2024-04-24 | 2024-07-30 | 中国海洋大学 | Fatigue life prediction method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104062196A (en) * | 2014-01-08 | 2014-09-24 | 中国石油大学(华东) | Corrosion fatigue life prediction method based on damage evolution |
| CN105488328A (en) * | 2015-11-19 | 2016-04-13 | 北京航空航天大学 | Fatigue crack growth rate prediction method based on artificial neuron network |
| CN105956315A (en) * | 2016-05-17 | 2016-09-21 | 北京航空航天大学 | Method capable of carrying out fatigue crack propagation rate estimation and life prediction |
| CN108763841A (en) * | 2018-07-24 | 2018-11-06 | 北京航空航天大学青岛研究院 | A kind of elastic failure emulation mode based on Dual boundary element and strain energy optimization analysis |
| CN112052615A (en) * | 2020-09-07 | 2020-12-08 | 郑州航空工业管理学院 | Micro-motion fatigue performance prediction method based on artificial neural network |
| CN112257197A (en) * | 2020-10-19 | 2021-01-22 | 郑州轻工业大学 | Method for evaluating micro-defect working stress of large-scale casting and forging |
-
2021
- 2021-01-28 CN CN202110116397.0A patent/CN112784495B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104062196A (en) * | 2014-01-08 | 2014-09-24 | 中国石油大学(华东) | Corrosion fatigue life prediction method based on damage evolution |
| CN105488328A (en) * | 2015-11-19 | 2016-04-13 | 北京航空航天大学 | Fatigue crack growth rate prediction method based on artificial neuron network |
| CN105956315A (en) * | 2016-05-17 | 2016-09-21 | 北京航空航天大学 | Method capable of carrying out fatigue crack propagation rate estimation and life prediction |
| CN108763841A (en) * | 2018-07-24 | 2018-11-06 | 北京航空航天大学青岛研究院 | A kind of elastic failure emulation mode based on Dual boundary element and strain energy optimization analysis |
| CN112052615A (en) * | 2020-09-07 | 2020-12-08 | 郑州航空工业管理学院 | Micro-motion fatigue performance prediction method based on artificial neural network |
| CN112257197A (en) * | 2020-10-19 | 2021-01-22 | 郑州轻工业大学 | Method for evaluating micro-defect working stress of large-scale casting and forging |
Non-Patent Citations (1)
| Title |
|---|
| Singularity cancellation method for time-domain boundary element formulation of elastodynamics: A direct approach;Guizhong Xie et al.;《Applied Mathematical Modelling》;20191206;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112784495A (en) | 2021-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112784495B (en) | Mechanical structure real-time fatigue life prediction method based on data driving | |
| Hakim et al. | Structural damage detection of steel bridge girder using artificial neural networks and finite element models | |
| Bilgehan | Comparison of ANFIS and NN models—With a study in critical buckling load estimation | |
| Abouhamze et al. | Multi-objective stacking sequence optimization of laminated cylindrical panels using a genetic algorithm and neural networks | |
| Fei et al. | Vectorial surrogate modeling method for multi-objective reliability design | |
| CN114117840B (en) | Structural performance prediction method based on simulation and test data hybrid drive | |
| Wu et al. | Prediction of stress intensity factors in pavement cracking with neural networks based on semi-analytical FEA | |
| CN111209620B (en) | Method for predicting residual bearing capacity and crack propagation path of crack-containing structure | |
| Gonzalez-Perez et al. | Identification of structural damage in a vehicular bridge using artificial neural networks | |
| Fathnejat et al. | An efficient two-stage approach for structural damage detection using meta-heuristic algorithms and group method of data handling surrogate model | |
| Parhi et al. | Analysis of smart crack detection methodologies in various structures | |
| Sheidaii et al. | Evaluation of compression member buckling and post-buckling behavior using artificial neural network | |
| Sahu et al. | Design and development of 3-stage determination of damage location using Mamdani-adaptive genetic-Sugeno model | |
| Gao et al. | An artificial neural network method for probabilistic life prediction of corroded reinforced concrete | |
| Bilgehan et al. | Buckling load estimation of cracked columns using artificial neural network modeling technique | |
| CN109750150A (en) | The online non-destructive prediction method of laser impact intensified part fatigue life | |
| Rojas et al. | Identification of external loads in mechanical systems through heuristic-based optimization methods and dynamic responses | |
| Meruane | Damage Assessment Algorithms for Structural Health Monitoring | |
| Machavaram et al. | Identification of crack in a structural member using improved radial basis function (IRBF) neural networks | |
| Chen et al. | Delamination detection problems using a combined genetic algorithm and neural network technique | |
| Karamizadeh et al. | An Innovative Forecasting Formula for Axial Compression Capacity of Circular Steel Tubes Filled with Concrete through Neural Networks | |
| Burczyński et al. | Fuzzy-neural and evolutionary computation in identification of defects | |
| CN117171575A (en) | Loosening positioning and loosening degree identification method for bolt node group of power transmission tower | |
| Abdul-Razzak et al. | Artificial neural networks model for predicting thickness of rectangular plates | |
| Takeda et al. | Surrogate Models for Data-inspired Reliability Design |
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 |