CN114528741A - Large-span steel structure index change prediction method based on finite element analysis method - Google Patents
Large-span steel structure index change prediction method based on finite element analysis method Download PDFInfo
- Publication number
- CN114528741A CN114528741A CN202210312596.3A CN202210312596A CN114528741A CN 114528741 A CN114528741 A CN 114528741A CN 202210312596 A CN202210312596 A CN 202210312596A CN 114528741 A CN114528741 A CN 114528741A
- Authority
- CN
- China
- Prior art keywords
- node
- model
- finite element
- steel structure
- similarity
- 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
- 229910000831 Steel Inorganic materials 0.000 title claims abstract description 58
- 239000010959 steel Substances 0.000 title claims abstract description 58
- 238000000034 method Methods 0.000 title claims abstract description 52
- 230000008859 change Effects 0.000 title claims abstract description 33
- 238000004458 analytical method Methods 0.000 title claims abstract description 29
- 238000013528 artificial neural network Methods 0.000 claims description 31
- 238000004364 calculation method Methods 0.000 claims description 28
- 238000005452 bending Methods 0.000 claims description 12
- 230000035945 sensitivity Effects 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 9
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000006073 displacement reaction Methods 0.000 claims description 6
- 238000012805 post-processing Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 3
- 238000012821 model calculation Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 238000009827 uniform distribution Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 2
- 238000009826 distribution Methods 0.000 abstract description 4
- 230000035882 stress Effects 0.000 description 14
- 230000006870 function Effects 0.000 description 9
- 238000004422 calculation algorithm Methods 0.000 description 4
- 239000002699 waste material Substances 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000004643 material aging Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Structural Engineering (AREA)
- Civil Engineering (AREA)
- Architecture (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a large-span steel structure index change prediction method based on a finite element analysis method, which comprises the following steps of: step 100, acquiring original index data of a large-span steel structure, and storing the original index data in an Excel table; step 200, establishing a function model according to the index data parameter range, and establishing a finite element node model by adopting large finite element software ABAQUS; step 300, importing an original index data table into an ABAQUS finite element node model, and automatically outputting node rigidity; step 400, analyzing influence weights of all factors according to node rigidity parameters, obtaining data error prediction indexes by using a non-complete similarity error prediction method, establishing an accurate node finite element model, analyzing the consistency of the model of the end plate connecting node and the stress distribution of a prototype, obtaining corresponding main influence factors, and distinguishing the similarity error prediction problem according to specific conditions by constructing a node model database which is not completely similar, so that the complexity of the model is reduced, and the prediction precision is improved.
Description
Technical Field
The embodiment of the invention relates to the technical field of large-span steel structure index prediction, in particular to a large-span steel structure index change prediction method based on a finite element analysis method.
Background
The service life of the large-span steel structure is long, in the using process of the large-span steel structure, the influences of the erosion, the fatigue effect, the material aging, the natural disaster and the like of the external environment and the reasons that the actual state of the structure is inconsistent with the design, the construction defects and the like can be caused, the structure can generate the reactions of resistance reduction, damage accumulation and the like, and the reactions can cause the engineering accident to be sudden to a certain degree.
According to the stress characteristics of the large-span steel structure, the stability of the structure is controlled, the large-span steel structure cannot continuously bear load after elastic-plastic instability, and a small external load can cause rapid increase of structural deformation, the instability belongs to brittle failure and is very unfavorable for the bearing of the structure, in addition, the instability of the large-span steel structure is caused by the bearing capacity of the structure, the instability of the large-span steel structure is also related to a support system and boundary conditions of a space structure, designers often presume the support and boundary conditions of the large-span steel structure in the design, the presumption is only approximately consistent with the stress state of the structure, the presumed difference with the stress state of the structure per se can cause great influence on the safety and the service life of the large-span steel structure in the construction and use processes, and the damage to the large-span steel structure can not be avoided, how to accurately know the actual stress state of the structure also plays a key role in judging the critical load after the structure is damaged and analyzing the current safety state of the structure.
The current prediction method for the index change of the large-span steel structure can apply an accurate finite element analysis method and detection and identification procedures of buildings to the safety evaluation of the structure, but has the following defects:
(1) according to the existing large-span steel structure index change prediction, due to the fact that a plurality of factors influencing steel structure indexes are provided, the influence of each factor is different in size and form, and meanwhile, the final overall error of an incomplete similarity model is formed under the comprehensive action of each factor, the existing large-span steel structure index change prediction model is relatively complex, the influence of each factor on the model error cannot be simply linearly superposed or accumulated, the steel structure index change prediction complexity is high, and the influence weight of each factor cannot be clearly determined;
(2) the existing large-span steel structure index change prediction method does not pay attention to the problem of similarity errors, and incomplete similar factors are not studied due to excessive details, so that a uniform large-span steel structure index change prediction error estimation value is lacked, overall analysis cannot be performed, and resource waste is easily caused.
Disclosure of Invention
Therefore, the embodiment of the invention provides a large-span steel structure index change prediction method based on a finite element analysis method, and aims to solve the problems that in the prior art, the steel structure index change prediction complexity is high, a uniform large-span steel structure index change prediction error estimation value is lacked, overall analysis cannot be performed, and resource waste is easily caused.
In order to achieve the above object, an embodiment of the present invention provides the following:
a large-span steel structure index change prediction method based on a finite element analysis method comprises the following steps:
step 100, obtaining original index data of a large-span steel structure, and storing the original index data in an Excel table;
step 200, establishing a function model according to the index data parameter range, and establishing a finite element node model by adopting large finite element software ABAQUS;
step 300, importing an original index data table into an ABAQUS finite element node model, and automatically outputting node rigidity;
and 400, analyzing influence weights of all factors according to the node rigidity parameters, and acquiring a data error prediction index by using a non-complete similarity error prediction method.
In a preferred embodiment of the present invention, in step 200, a linear reduction integration unit C3D8R in ABAQUS is used, and a node stiffness parameter of a steel structure is calculated by using a finite element, and the step of measuring the node stiffness in the finite element calculation model is as follows:
firstly, acquiring the distance L from a steel structure column end to a beam top and column head load F, and calculating a column end node bending moment M
M=F×L;
Secondly, calculating a node rotation angle theta according to the displacement of the left end and the right end of the steel structure column end in the vertical direction
θ=(s1-s2)/H
Wherein s is1、s2Respectively displacement in the vertical direction of the left end and the right end of the steel structure column, and H is the height of the section of the end column;
finally, calculating the initial rotational rigidity K of the node according to the column end node bending moment M and the node corner theta
K=M/θ。
As a preferred scheme of the present invention, a bending moment corner curve of a node is simulated in a finite element calculation model according to the node stiffness, a difference between a prototype and the simulation model is obtained, a similar error of the node is defined, and a non-complete similar node database of an end plate node model is established, wherein a similar error calculation formula is as follows:
wherein, delta1For independent variable similarity error, δ2Is the target quantity similarity error, phi is the independent variable value of the incomplete similarity model, phifIs the independent variable value of the complete similarity model, K is the initial rotational rigidity of the incomplete similarity model, KfIs the initial rotational stiffness of a completely similar model.
As a preferred scheme of the invention, a sensitivity coefficient is introduced into the similarity error, the variation trend between node beam end variables is analyzed, a reduced scale model and a non-complete similarity reduced scale model with a basic geometric reduced scale ratio of 1:2 are respectively established, and non-complete similarity factors including end plate thickness, beam section height, column section height and bolt specification are set.
As a preferred embodiment of the present invention, the incomplete similarity factor is preferably selected by using a latin hypercube sampling method, and the latin hypercube sampling method comprises the following steps:
firstly, determining the number d of influencing factors of similar error calculation and the total number N of model calculation;
secondly, dividing the value range of each factor into N non-overlapping intervalsThe probability of each interval is the same, and the length of each interval is the same under the condition of uniform distribution;
furthermore, in each interval of the value range of each similarity factorIn which a sample point is extracted according to the corresponding probability density function
And finally, randomly combining all the extracted sample points into a sample vector to generate N Latin hypercube sample points x(i)。
As a preferred embodiment of the invention, the method is based on a series of Latin hypercube sample points x(i)And establishing a corresponding accurate analysis finite element model by using the parameter values of the representative points so as to calculate the response values of the performance function to be approximated at the N sample points.
As a preferred scheme of the invention, the corresponding precise analysis finite element model is automatically modeled according to two types of beam column end plate connecting nodes and beam column T-shaped part connecting nodes, and the implementation steps are as follows:
firstly, counting key parameters of all sample points by using excel software, wherein each row in a table represents a model, each column represents a parameter, and the scale of the table is [ i x j ], wherein i is the row number representing the number of the models, and j is the total number of the parameters, including sensitivity calculation parameters and invariant parameters;
secondly, automatically reading the key parameters of the model in the table according to rows, and calling a modeling module in the ABAQUS to complete the establishment of an operation flow based on influence factors;
then, automatically entering a model post-processing module, extracting stress and deformation data of key points, and calculating to obtain rotational stiffness data of the nodes;
and finally, creating an output table and outputting a data table of stress, deformation, support reaction force and node rigidity of the key points.
According to the optimal scheme of the invention, the rigidity is predicted by adopting a neural network prediction method according to the beam column end plate connecting node and the beam column T-shaped part connecting node model and the similarity error.
As a preferred scheme of the present invention, parameters of the beam column end plate connection node and the beam column T-shaped connection node model are used as input data of a neural network, and a parameterized automatic modeling technology is used to obtain neural network training data, wherein the neural network prediction method specifically comprises the following steps:
firstly, acquiring node key parameters according to an ABAQUS parameterized modeling structure, using the node key parameters as input parameters of a neural network, and freely setting data calculation groups according to precision requirements and calculation time requirements;
secondly, defining an index population with a fixed scale, wherein individuals in the index population are weight thresholds to be optimized, the coding length is the sum of the weights and the number of the thresholds, and an initial population, a winner sub-population and a temporary sub-population are sequentially generated;
and finally, inputting different indexes of the parameters representing nodes, and inputting the indexes after normalization processing into a neural network to analyze the index change trend.
As a preferred scheme of the present invention, the number of hidden nodes is determined in the neural network to determine a linear prediction function, and the calculation formula of the number of hidden nodes is:
wherein n is the number of nodes of the input layer, m is the number of nodes of the output layer, and alpha is the sensitivity coefficient
The embodiment of the invention has the following advantages:
(1) the method establishes an accurate node finite element model, analyzes the consistency of the model of the full similar end plate connecting node and the stress distribution of a prototype, obtains corresponding main influence factors, distinguishes layers of similar error prediction problems according to specific conditions by establishing an incomplete similar node model database, predicts the similar error of a large-span steel structure by using a neural network method, analyzes error influence factors and weight, and reduces the complexity of the model;
(2) according to the invention, a model database with different similarity degrees and taking main incomplete factors as cores is established through a large number of finite element modeling, and a non-complete similar model can be established for a plurality of similar targets, so that the estimated value of the prediction error can be comprehensively analyzed and predicted, and the resource waste is reduced.
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 should be apparent that the drawings in the following description are merely exemplary, and that other embodiments can be derived from the drawings provided by those of ordinary skill in the art without inventive effort.
Fig. 1 is a schematic flow diagram of a method for predicting index change of a large-span steel structure in an embodiment of the present invention.
Detailed Description
The present invention is described in terms of particular embodiments, other advantages and features of the invention will become apparent to those skilled in the art from the following disclosure, and it is to be understood that the described embodiments are merely exemplary of the invention and that it is not intended to limit the invention to the particular embodiments disclosed. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in FIG. 1, the invention provides a method for predicting index change of a large-span steel structure based on a finite element analysis method, which realizes automatic modeling, calculation and data processing of ABAQUS by inputting key parameters of a plurality of nodes in a table, determines similarity relation ratio of each parameter by using completely similar conditions, to determine the parameter range, establish an accurate node finite element model, analyze the consistency of the model of the connecting node of the fully similar end plate and the stress distribution of the prototype, obtain corresponding main influence factors, by constructing an incomplete similar node model database, the similar error prediction problem is classified according to specific conditions, and the neural network method is utilized to predict the similar error of the large-span steel structure, the error influence factor and the weight are analyzed, the complexity of the model is reduced, and the model is constructed by adopting the finite element method, so that the effect close to or even better can be realized with lower cost.
The method comprises the following steps:
step 100, obtaining original index data of a large-span steel structure, and storing the original index data in an Excel table;
step 200, establishing a function model according to the index data parameter range, and establishing a finite element node model by adopting large finite element software ABAQUS;
step 300, importing an original index data table into an ABAQUS finite element node model, and automatically outputting node rigidity;
and 400, analyzing influence weights of all factors according to the node rigidity parameters, and acquiring a data error prediction index by using a non-complete similarity error prediction method.
In this embodiment, the change conditions of each similarity factor are listed in a table form, the initial stiffness calculation results of the nodes of all the incomplete similarity models are obtained, and the influence factors of each parameter and the target variable are quantized to obtain the influence weight.
In step 200, a linear reduction integral unit C3D8R in ABAQUS is adopted, and the node rigidity parameter of the steel structure is calculated by a finite element, and the measurement steps of the node rigidity in the finite element calculation model are as follows:
firstly, acquiring the distance L from a steel structure column end to a beam top and column head load F, and calculating a column end node bending moment M
M=F×L;
Secondly, calculating a node rotation angle theta according to the displacement of the left end and the right end of the steel structure column end in the vertical direction
θ=(s1-s2)/H
Wherein s is1、s2Respectively displacement in the vertical direction of the left end and the right end of the steel structure column, and H is the height of the section of the end column;
finally, calculating the initial rotational rigidity K of the node according to the column end node bending moment M and the node corner theta
K=M/θ。
In this embodiment, the node stiffness is used as a core performance index, the node stiffness is used as a target quantity for measuring the similarity degree of the incomplete similar node, and the node stiffness in the model is measured by using the ratio of the column end node bending moment M to the node rotation angle θ.
According to the bending moment corner curve of the node rigidity simulated in the finite element calculation model, obtaining the difference between a prototype and the simulation model, defining the similar error of the node, and establishing a non-complete similar node database of the end plate node model, wherein the similar error calculation formula is as follows:
wherein, delta1For independent variable similarity error, δ2Is the target quantity similarity error, phi is the independent variable value of the incomplete similarity model, phifIs the independent variable value of the complete similarity model, K is the initial rotational rigidity of the incomplete similarity model, KfIs the initial rotational stiffness of a completely similar model.
In this embodiment, the node stress and the node initial stiffness are used as similar targets, error analysis is performed on key parameters affecting the performance of the two nodes, and quantification and formulation of the influence of each variable on the similar error are sought, so that the prediction accuracy of the similar error is improved.
In this embodiment, a non-complete similar node database of the end plate node model is established, and the initial rotational stiffness prediction accuracy of the node is improved by defining similar errors of the node, including independent variable errors, dependent variable errors and the like.
Introducing sensitivity coefficients into the similarity errors, analyzing the variation trend between the variables of the node beam ends, respectively establishing a reduced scale model and a non-complete similarity reduced scale model with a basic geometric reduced scale ratio of 1:2, and setting non-complete similarity factors including end plate thickness, beam section height, column section height and bolt specification.
In the embodiment, based on the sensitivity calculation method, the factors affecting the complete similarity degree of the nodes are screened, the similar error is introduced, the bending moment and corner curve comparison value of the prototype node and the completely similar model node is obtained, the influence of various influencing factors and the target variable is reduced, the value of each parameter of the model can be accurately calculated within the error allowable range, and the rotational stiffness value of the corresponding node is obtained.
In the embodiment, by establishing the incomplete similar node model, parameters such as the width, the height and the plate thickness of a beam-column component of the node, the beam length, the column length, the width and the thickness of an end plate, the layout and the spacing of bolt holes, the equivalent diameter of bolts, the bolt hole diameter, the bolt pretightening force and the like are comprehensively analyzed, and nonparametric correlation coefficients are introduced to obtain four factors which have the greatest influence on the node performance, so that a theoretical basis is provided for finite element fine analysis.
In the embodiment, finite element software ABAQUS is used as a computing platform to respectively perform fine modeling on the large-span steel structure beam column node and the large-span steel structure frame, so that the computing assumption of various models is determined, the nonlinear characteristic of similar errors is obtained by analyzing the bending moment and corner curve contrast value of the prototype node and the completely similar model node, and different influences of different parameters on the target quantity are obtained.
The incomplete similarity factor is preferably selected by adopting a Latin hypercube sampling method, and the Latin hypercube sampling step is as follows:
firstly, determining the number d of influencing factors of similar error calculation and the total number N of model calculation;
secondly, dividing the value range of each factor into N non-overlapping intervalsThe probability of each interval is the same, and the length of each interval is the same under the condition of uniform distribution;
furthermore, in each interval of the value range of each similarity factorIn which a sample point is extracted according to the corresponding probability density function
Finally, extracting allRandomly combining the sample points into a sample vector to generate N Latin hypercube sample points x(i)。
According to a series of Latin hypercube sample points x(i)And establishing a corresponding accurate analysis finite element model by using the parameter values of the representative points so as to calculate the response values of the performance function to be approximated at the N sample points.
In this embodiment, according to the theorem of the majority and the theorem of the central limit, the latin hypercube sampling method is used to perform random sampling under the condition that the sample points are enough, the general law can be approximately estimated by the law of the sample, and the sample can be infinitely close to the true solution under the condition that the samples are infinite, and in order to reduce the number of the samples, when the number of the parameters is the same as the horizontal number of the sample points, the latin hypercube sampling method is used to optimize the number of the samples.
Automatically modeling the corresponding accurate analysis finite element model according to two types of beam column end plate connecting nodes and beam column T-shaped part connecting nodes, wherein the method comprises the following implementation steps:
firstly, counting key parameters of all sample points by using excel software, wherein each row in a table represents a model, each column represents a parameter, and the scale of the table is [ i x j ], wherein i is the row number representing the number of the models, and j is the total number of the parameters, including sensitivity calculation parameters and invariant parameters;
secondly, automatically reading the key parameters of the model in the table according to rows, and calling a modeling module in the ABAQUS to complete the establishment of an operation flow based on influence factors;
then, automatically entering a model post-processing module, extracting stress and deformation data of key points, and calculating to obtain rotational stiffness data of the nodes;
and finally, creating an output table and outputting a data table of stress, deformation, support reaction force and node rigidity of the key points.
In the embodiment, an automatic entry model post-processing module in the ABAQUS is called, stress and deformation data of key points are extracted, and rotational stiffness data of the nodes are calculated; and (3) creating an output table, outputting post-processing data including key point stress, deformation, support reaction force, node rigidity and the like, automatically completing the whole process by a program, and automatically calculating and automatically acquiring an accurate solution set of a node model library only by preparing a key data table of a sample point at the early stage.
And predicting the rigidity by adopting a neural network prediction method according to the beam column end plate connecting node and the beam column T-shaped part connecting node model in combination with the similarity error.
In this embodiment, two types of automatic modeling programs, namely a beam column end plate connecting node and a beam column T-shaped part connecting node, are compiled by using a Python language, and an accurate solution at a sample point is calculated by using finite element software ABAQUS and used as a database of a node similarity error proxy model.
Parameters of the beam column end plate connecting node and the beam column T-shaped part connecting node model are used as input data of a neural network, neural network training data are obtained by using a parameterized automatic modeling technology, and the neural network prediction method specifically comprises the following implementation steps:
firstly, acquiring node key parameters according to an ABAQUS parameterized modeling structure, using the node key parameters as input parameters of a neural network, and freely setting data to calculate group number according to precision requirements and calculation time requirements;
secondly, defining an index population with a fixed scale, wherein individuals in the index population are weight thresholds to be optimized, the coding length is the sum of the weights and the number of the thresholds, and an initial population, a winner sub-population and a temporary sub-population are sequentially generated;
and finally, inputting different indexes of the parameters representing nodes, and inputting the indexes after normalization processing into a neural network to analyze the index change trend.
Determining the number of nodes in the neural network to determine a linear prediction function, wherein the calculation formula of the number of hidden nodes is as follows:
wherein n is the number of nodes of the input layer, m is the number of nodes of the output layer, and alpha is the sensitivity coefficient.
In this embodiment, the input layer and the output layer are mapped by using a three-layer neural network through a continuous function, and the complexity of the mapping function is reduced by reducing the number of hidden layers, so as to improve the calculation rate.
In this embodiment, an ABAQUS parameterized modeling structure of Python programming language is adopted, and by using a key parameter table, batch modeling of finite element software and automatic output of a node key parameter calculation result are realized, so that necessary data are provided for training of a neural network, and the generalization capability of a neural network mapping model is improved.
In the embodiment, a thought evolution algorithm is adopted to optimize the neural network structure, the algorithm can improve the convergence speed and precision of the neural network, improve the fault diagnosis rate, avoid falling into a local minimum value, facilitate finding of a global optimal solution, solve the problems of precocity, low convergence speed and the like of the traditional evolution algorithms such as a genetic algorithm and the like, improve the generalization capability and the prediction precision of the neural network, and enable the neural network to accurately simulate a complex nonlinear function relationship between a node incomplete similarity condition and a similarity error, thereby being used as an error analysis method when considering multi-factor correlation.
Therefore, in the embodiment, by establishing an accurate node finite element model, the consistency of the model of the fully similar end plate connecting node and the stress distribution of the prototype is analyzed, corresponding main influence factors are obtained, by establishing an incomplete similar node model database, the similar error prediction problem is classified according to specific situations, the similar error of the large-span steel structure is predicted by using a neural network method, the error influence factors and the weight are analyzed, and the complexity of the model is reduced;
therefore, as another innovative point of the present invention, the implementation method establishes model databases with different degrees of similarity, which take the main incomplete factors as the core, through a large number of finite element modeling, and can establish incomplete similar models for a plurality of similar targets, thereby comprehensively analyzing the prediction error estimation value and reducing the resource waste.
Although the invention has been described in detail above with reference to a general description and specific examples, it will be apparent to one skilled in the art that modifications or improvements may be made thereto based on the invention. Accordingly, such modifications and improvements are intended to be within the scope of the invention as claimed.
Claims (10)
1. A large-span steel structure index change prediction method based on a finite element analysis method is characterized by comprising the following steps:
step 100, obtaining original index data of a large-span steel structure, and storing the original index data in an Excel table;
step 200, establishing a function model according to the index data parameter range, and establishing a finite element node model by adopting large finite element software ABAQUS;
step 300, importing an original index data table into an ABAQUS finite element node model, and automatically outputting node rigidity;
and 400, analyzing influence weights of all factors according to the node rigidity parameters, and acquiring a data error prediction index by using a non-complete similarity error prediction method.
2. The method for predicting index change of large-span steel structure based on finite element analysis of claim 1, wherein in step 200, a linear reduction integration unit C3D8R in ABAQUS is used to calculate the node stiffness parameter of the steel structure in the finite element calculation model, and the steps of measuring the node stiffness in the finite element calculation model are as follows:
firstly, acquiring the distance L from a steel structure column end to a beam top and column head load F, and calculating a column end node bending moment M
M=F×L;
Secondly, calculating a node rotation angle theta according to the vertical displacement of the left end and the right end of the steel structure column end
θ=(s1-s2)/H
Wherein s is1、s2Respectively displacement in the vertical direction of the left end and the right end of the steel structure column, and H is the height of the section of the end column;
finally, calculating the initial rotational stiffness K of the node according to the bending moment M of the column end node and the node rotation angle theta
K=M/θ。
3. The method for predicting the index change of the large-span steel structure based on the finite element analysis method as claimed in claim 2, wherein the difference between a prototype and a simulation model is obtained according to the bending moment corner curve of the node simulated in the finite element calculation model based on the node rigidity, the similar error of the node is defined, and a non-complete similar node database of the end plate node model is established, wherein the similar error calculation formula is as follows:
wherein, delta1For independent variable similarity error, δ2Is the target quantity similarity error, phi is the independent variable value of the incomplete similarity model, phifIs the independent variable value of the complete similarity model, K is the initial rotational rigidity of the incomplete similarity model, KfIs the initial rotational stiffness of a completely similar model.
4. The method for predicting the index change of the large-span steel structure based on the finite element analysis method as claimed in claim 3, wherein sensitivity coefficients are introduced into the similarity errors, the change trend between the variables of the node beam ends is analyzed, a reduced scale model and a non-complete-similarity reduced scale model with a basic geometric reduced scale ratio of 1:2 are respectively established, and the non-complete-similarity factors are set to include end plate thickness, beam section height, column section height and bolt specification.
5. The finite element analysis method-based large-span steel structure index change prediction method according to claim 4, wherein the incomplete similarity factor is preferably selected by adopting a Latin hypercube sampling method, and the Latin hypercube sampling step is as follows:
firstly, determining the number d of influencing factors of similar error calculation and the total number N of model calculation;
secondly, dividing the value range of each factor into N non-overlapping intervalsThe probability of each interval is the same, and the length of each interval is the same under the condition of uniform distribution;
furthermore, in each interval of the value range of each similarity factorIn which a sample point is extracted according to the corresponding probability density function
And finally, randomly combining all the extracted sample points into a sample vector to generate N Latin hypercube sample points x(i)。
6. The finite element analysis method-based large-span steel structure index change prediction method as claimed in claim 5, wherein the index change prediction method is characterized in that the index change prediction method is carried out according to a series of Latin hypercube sample points x(i)And establishing a corresponding accurate analysis finite element model by using the parameter values of the representative points so as to calculate the response values of the performance function to be approximated at the N sample points.
7. The method for predicting the index change of the large-span steel structure based on the finite element analysis method as claimed in claim 6, wherein the corresponding precise analysis finite element model is automatically modeled according to two types of beam-column end plate connecting nodes and beam-column T-shaped part connecting nodes, and the method comprises the following implementation steps:
firstly, counting key parameters of all sample points by using excel software, wherein each row in a table represents a model, each column represents a parameter, and the scale of the table is [ i x j ], wherein i is the row number representing the number of the models, and j is the total number of the parameters, including sensitivity calculation parameters and invariant parameters;
secondly, automatically reading the key parameters of the model in the table according to rows, and calling a modeling module in the ABAQUS to complete the establishment of an operation flow based on influence factors;
then, automatically entering a model post-processing module, extracting stress and deformation data of key points, and calculating to obtain rotational stiffness data of the nodes;
and finally, creating an output table and outputting a data table of stress, deformation, support reaction force and node rigidity of the key points.
8. The finite element analysis method-based large-span steel structure index change prediction method according to claim 7, wherein a neural network prediction method is adopted to predict the rigidity according to the beam-column end plate connecting node and the beam-column T-shaped part connecting node model in combination with the similarity error.
9. The method for predicting the index change of the large-span steel structure based on the finite element analysis method according to claim 8, wherein parameters of the beam-column end plate connecting node model and the beam-column T-shaped part connecting node model are used as input data of a neural network, and neural network training data are obtained by using a parameterized automatic modeling technology, and the neural network prediction method is implemented by the following specific steps:
firstly, acquiring node key parameters according to an ABAQUS parameterized modeling structure, using the node key parameters as input parameters of a neural network, and freely setting data to calculate group number according to precision requirements and calculation time requirements;
secondly, defining an index population with a fixed scale, wherein individuals in the index population are weight thresholds to be optimized, the coding length is the sum of the weights and the number of the thresholds, and an initial population, a winner sub-population and a temporary sub-population are sequentially generated;
and finally, inputting different indexes of the parameters representing nodes, and inputting the indexes after normalization processing into a neural network to analyze the index change trend.
10. The method for predicting the index change of the large-span steel structure based on the finite element analysis method as claimed in claim 9, wherein the number of hidden nodes is determined in the neural network to determine a linear prediction function, and the calculation formula of the number of hidden nodes is as follows:
wherein n is the number of nodes of the input layer, m is the number of nodes of the output layer, and alpha is the sensitivity coefficient.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210312596.3A CN114528741A (en) | 2022-03-28 | 2022-03-28 | Large-span steel structure index change prediction method based on finite element analysis method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210312596.3A CN114528741A (en) | 2022-03-28 | 2022-03-28 | Large-span steel structure index change prediction method based on finite element analysis method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114528741A true CN114528741A (en) | 2022-05-24 |
Family
ID=81626471
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210312596.3A Pending CN114528741A (en) | 2022-03-28 | 2022-03-28 | Large-span steel structure index change prediction method based on finite element analysis method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114528741A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115292792A (en) * | 2022-09-26 | 2022-11-04 | 北京云庐科技有限公司 | Monte Carlo sampling simulation-based large-span spatial structure monitoring and optimizing point distribution method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105956216A (en) * | 2016-04-15 | 2016-09-21 | 东南大学 | Finite element model correction method for large-span steel bridge based on uniform temperature response monitoring value |
CN109670252A (en) * | 2018-12-25 | 2019-04-23 | 中南大学 | A kind of head vehicle contracting mould construction method and head vehicle contracting mould based on power and stiffnes s equivalent |
-
2022
- 2022-03-28 CN CN202210312596.3A patent/CN114528741A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105956216A (en) * | 2016-04-15 | 2016-09-21 | 东南大学 | Finite element model correction method for large-span steel bridge based on uniform temperature response monitoring value |
CN109670252A (en) * | 2018-12-25 | 2019-04-23 | 中南大学 | A kind of head vehicle contracting mould construction method and head vehicle contracting mould based on power and stiffnes s equivalent |
Non-Patent Citations (1)
Title |
---|
赵东卓: "《钢结构半刚性连接及框架非完全相似误差分析方法的研究》", 《中国博士论文全文库》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115292792A (en) * | 2022-09-26 | 2022-11-04 | 北京云庐科技有限公司 | Monte Carlo sampling simulation-based large-span spatial structure monitoring and optimizing point distribution method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111209708B (en) | Machine learning-based pile-soil interaction prediction analysis method | |
CN116448419A (en) | Zero sample bearing fault diagnosis method based on depth model high-dimensional parameter multi-target efficient optimization | |
CN113627098B (en) | CFD model confirmation method and product design method | |
CN114490065A (en) | Load prediction method, device and equipment | |
CN112347537B (en) | Calibration method and device for engineering structure numerical model, electronic equipment and medium | |
CN111428419A (en) | Suspended sediment concentration prediction method and device, computer equipment and storage medium | |
CN114528741A (en) | Large-span steel structure index change prediction method based on finite element analysis method | |
CN115495965A (en) | Method for analyzing time-varying reliability of complex aviation structure under mixed uncertainty | |
CN115982141A (en) | Characteristic optimization method for time series data prediction | |
CN110188399B (en) | Dam safety monitoring single-measuring-point evaluation method based on multiple correlation sequences | |
Zhang et al. | Automatic identification of structural modal parameters based on density peaks clustering algorithm | |
CN111985845B (en) | Node priority optimization method of heterogeneous Spark cluster | |
Zhang et al. | Long‐term bridge performance assessment using clustering and Bayesian linear regression for vehicle load and strain mapping model | |
CN117076887A (en) | Pump station unit running state prediction and health assessment method and system | |
Oh et al. | Artificial intelligence-based damage localization method for building structures using correlation of measured structural responses | |
CN109766637B (en) | Bridge crane structure reliability optimization method based on Krigng agent model | |
CN116842722A (en) | Method, system and electronic equipment for determining confidence coefficient of simulation model of gas turbine | |
CN116011071A (en) | Method and system for analyzing structural reliability of air building machine based on active learning | |
CN110705106A (en) | Mechanical reliability analysis method based on probability design | |
CN114611768A (en) | Power distribution network industry expansion matching capacity time sequence construction scale prediction method | |
CN107292045B (en) | Complex modal evaluation method of finite element model with damping structure | |
CN112365022A (en) | Engine bearing fault prediction method based on multiple stages | |
CN112488282A (en) | Method, system, equipment and storage medium for predicting gas concentration | |
CN111062118A (en) | Multilayer soft measurement modeling system and method based on neural network prediction layering | |
CN110543724A (en) | Satellite structure performance prediction method for overall 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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220524 |