CN114894619A - Method for predicting axial load-strain curve of concrete filled steel tubular column based on long-term and short-term memory network - Google Patents
Method for predicting axial load-strain curve of concrete filled steel tubular column based on long-term and short-term memory network Download PDFInfo
- Publication number
- CN114894619A CN114894619A CN202210500867.8A CN202210500867A CN114894619A CN 114894619 A CN114894619 A CN 114894619A CN 202210500867 A CN202210500867 A CN 202210500867A CN 114894619 A CN114894619 A CN 114894619A
- Authority
- CN
- China
- Prior art keywords
- axial
- data
- long
- term memory
- load
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 229910000831 Steel Inorganic materials 0.000 title claims abstract description 53
- 239000010959 steel Substances 0.000 title claims abstract description 53
- 239000004567 concrete Substances 0.000 title claims abstract description 46
- 230000006403 short-term memory Effects 0.000 title claims abstract description 37
- 230000007787 long-term memory Effects 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000013461 design Methods 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 22
- 238000007781 pre-processing Methods 0.000 claims abstract description 13
- 230000015654 memory Effects 0.000 claims description 44
- 230000006835 compression Effects 0.000 claims description 20
- 238000007906 compression Methods 0.000 claims description 20
- 238000013528 artificial neural network Methods 0.000 claims description 19
- 239000000203 mixture Substances 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000010276 construction Methods 0.000 claims description 9
- 239000000463 material Substances 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 125000004122 cyclic group Chemical group 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 238000011160 research Methods 0.000 description 5
- 238000005452 bending Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 239000011150 reinforced concrete Substances 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N3/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N3/08—Investigating strength properties of solid materials by application of mechanical stress by applying steady tensile or compressive forces
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N3/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N3/02—Details
- G01N3/06—Special adaptations of indicating or recording means
-
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/0001—Type of application of the stress
- G01N2203/0003—Steady
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/0014—Type of force applied
- G01N2203/0016—Tensile or compressive
- G01N2203/0019—Compressive
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/0058—Kind of property studied
- G01N2203/0069—Fatigue, creep, strain-stress relations or elastic constants
- G01N2203/0075—Strain-stress relations or elastic constants
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/02—Details not specific for a particular testing method
- G01N2203/06—Indicating or recording means; Sensing means
- G01N2203/067—Parameter measured for estimating the property
- G01N2203/0676—Force, weight, load, energy, speed or acceleration
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/02—Details not specific for a particular testing method
- G01N2203/06—Indicating or recording means; Sensing means
- G01N2203/067—Parameter measured for estimating the property
- G01N2203/0682—Spatial dimension, e.g. length, area, angle
Landscapes
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)
Abstract
The invention provides a method for predicting a steel pipe concrete column axial load-strain curve based on a long-term and short-term memory network. For design parameters, axial load and axial strain data, a design data preprocessing and data constructing method is used for long-term and short-term memory network offline model training; constructing an axial strain data sequence, and inputting the axial strain data sequence and design parameters into a trained off-line model to obtain axial pressure load data output by the model; and restoring an axial pressure load-strain curve by using the axial strain data and the axial pressure load data to realize the prediction of the system axial pressure load-strain curve.
Description
Technical Field
The invention relates to the field of building and civil engineering, in particular to a method for predicting a steel pipe concrete column axial load-strain curve based on a long-term and short-term memory network.
Background
Concrete-filled steel tubular columns are known to be promising substitutes for conventional reinforced concrete and steel columns in high-rise buildings and urban overhead girder bridge construction as structural members that are subject to compression and bending. Compared with the traditional structural column, the steel pipe concrete column has higher bending, axial bearing capacity and shock resistance. In addition, the mutual protection of the steel pipes and the filled concrete gives the concrete-filled steel pipe column better fire resistance, impact resistance and corrosion resistance.
The axial compressive load-strain curve is one of the most basic and important features in the safety design of structural columns. Once the load-strain curve is known, key design parameters, including modulus of elasticity, ultimate bearing capacity, and ductility of the column, can be readily obtained. Therefore, the axial compressive load-strain curve has been the basic research target for concrete filled steel tubular columns over the last three decades. The theoretical derivation and finite element model analysis by using empirical parameters are two main methods for calculating the load-strain curve of the concrete filled steel tubular column at present. In research based on theoretical methods, the stress-strain relationship between the outer steel pipe and the inner concrete is generally assumed, and then the load-strain curve of the column is gradually deduced according to the equilibrium conditions and the predetermined interaction relationship. However, as materials are updated, the applicability of this method is gradually reduced, and theoretical methods based on empirical parameters are inevitably limited to the strength of the materials and the design concept at the time.
Finite element model analysis is another method for obtaining the axial load-strain curve of the steel tube concrete column. In finite element analysis, concrete and steel pipes should be simulated using appropriate element types, typically concrete solid elements and steel pipe shell elements, while a contact model or spring model is required to define the interaction between the concrete and steel surfaces. By reasonably adopting the constitutive relation and proper grids, the axial load-strain curve of the concrete-filled steel tubular column can be obtained.
In recent years, soft computing methods and intelligent techniques have been remarkably developed and widely used in civil engineering. However, the application of machine learning algorithms to concrete filled steel tubular column research is still in the beginning and exploration phase. Most of the existing research has focused on the prediction of performance metrics (such as axial load and stiffness) based on a collected experimental database. Currently, there is no research on calculating a complete axial compressive load-strain curve of a concrete filled steel tube member using a machine learning method.
Long-short term memory networks are one type of recurrent neural networks, and are commonly used to process time series. The long-short term memory network is characterized in that the long-short term dependency relationship of data is considered. In order to solve the existing problems, it is necessary to provide a technical solution that can regard the points in the axial compressive load-strain curve as a time series with long-short term dependency relationship and realize curve prediction by using a long-short term memory network through a design data construction method.
Disclosure of Invention
Aiming at the problem of predicting the axial load-strain curve of the concrete-filled steel tube column, the invention provides a method for predicting the axial load-strain curve of the concrete-filled steel tube column based on a long-short term memory network, points in the axial load-strain curve can be regarded as a time sequence with long-short term dependency relationship, and the curve prediction is realized by using the long-short term memory network through a design data forming method.
In order to achieve the purpose, the invention adopts the technical scheme that:
the method for predicting the axial load-strain curve of the concrete-filled steel tubular column based on the long-term and short-term memory network is characterized by comprising the following steps of: the method comprises the following steps:
s1: acquiring experimental data;
acquiring axial compression load-strain curve experimental data of concrete-filled steel tubular columns of different materials and sizes, and establishing a load-strain curve and five design parameters, wherein the five design parameters are as follows: diameter D of steel pipe and wall thickness t of steel pipe s Height H of column, strength f of steel s And concrete strength f c Constructing a load-strain curve prediction training sample through design parameters by a corresponding system database;
s2: preprocessing data;
for all the experimental data of the axial compression load-strain curve of the steel tube concrete column, 25000 mu epsilon is set as a limit value for collecting axial strain, if the experimental data curve is shorter and does not reach the axial strain limit value of 25000 mu epsilon, compensation operation is executed, and the axial compression load-strain curve is extended to the set axial strain limit value according to the final slope of the load-strain curve; if the experimental data curve is longer, performing truncation operation, discarding data exceeding a set limit value, and preprocessing all the axial load-strain curve data to have uniform axial strain length;
s3: data composition;
each set of design parameters and the corresponding axial compressive load-strain curve are used as a data sample; averagely dividing experimental data of the axial compression load-strain curve of the steel pipe concrete column subjected to data preprocessing into m points according to strain values, wherein each point comprises an axial compression load value N and an axial strain value epsilon; constructing input data of the neural network of the long and short term memory unit by using the axial strain value and the design parameter; constructing output data of the neural network of the long and short term memory unit by using the axial pressure load value;
s4: training an off-line model;
after all samples are formed by input and output data in S3, performing off-line model training on all samples by using a long-short term memory unit neural network to obtain a long-short term memory unit neural network model which accords with convergence;
s5: predicting an axial pressure load-strain curve;
arranging the design parameters of the concrete-filled steel tube column needing to predict the axial load-strain curve; constructing an axial strain data sequence, and taking a vector with a maximum value of 25000 mu epsilon and m data values which are evenly distributed as the axial strain data sequence; converting concrete filled steel tube sample data needing to be subjected to axial compression load-strain curve prediction into input data composition capable of being used for a long-term and short-term memory network by using the input data composition construction method of S3; inputting the input data number into the offline model obtained in S4 to obtain corresponding axial load output; and taking the average value of all axial strains in each line of the input data and the axial pressure load value in the line corresponding to the output data as a point of an axial pressure load-strain plane, and sequentially calculating all axial pressure load-strain points so as to restore a complete axial pressure load-strain curve and finish the prediction of the axial pressure load-strain curve.
As a preferred embodiment of the present invention:
in the step of S3, the user is allowed to perform,
s31: the input data composition includes: taking k axial strains as a group, and forming a first row in one data sample by the k axial strains and five design parameters of the sample; then, another group of k axial strains are obtained by adopting a sliding window method, and the k axial strains and five design parameters form another row in one data sample; and so on until all axial strains are taken out, and at the moment, the input data of each data sample form a structure containing m +1-k rows and 5+ k columns;
s32: the output data construction includes: obtaining a group of k axial load values to obtain each corresponding axial load value, averaging the group of axial load values to obtain the corresponding axial load of the row of input data
The data is taken as the output data of the neural network, and the output data of each data sample comprises m +1-k rows and 1 column.
As a preferred embodiment of the present invention:
in the step of S4, the user is allowed to perform,
s41: the long and short term memory unit neural network sequentially comprises: the system comprises an input layer, a long-term and short-term memory unit layer, a full connection layer, a random discarding layer, a full connection layer and a data regression layer; the long-short term memory unit layer comprises G long-short term memory units, namely the length of an input time sequence is G, and the characteristic quantity of input data is C; wherein the long-short term memory unit comprises:
at the current time of t-1, the output at the time of t needs to be calculated through a long-short term memory unit, and the input data is x t ，h t And c t Respectively representing the output and cell states at time t; the long-short term memory cell uses the current state h t-1 And c t-1 To calculate the output h t And updated cell state c t (ii) a Using input gates i t Forgetting door f t Candidate unit g t And an output gate o t To control the refresh of the long/short term memory unit, the calculation formula is as follows：
i t ＝σ(W i x t +R i h t-1 +b i )
f t ＝σ(W f x t +R f h t-1 +b f )
g t ＝σ(W g x t +R g h t-1 +b g )
c t ＝f t ⊙c t-1 +i t ⊙g t
o t ＝σ(W o x t +R o h t-1 +b o )
h t ＝o t ⊙tanh(c t )
Wherein W i ,W f ,W g ,W o Is an input weight matrix, R i ,R f ,R g ,R o Is a cyclic weight matrix, b i ,b f ,b g ,b o Is a bias matrix, and the matrix is adjusted through model training; σ is a sigmoid activation function, which is a dot product operation; directly output the state h t As the output of the long-short term memory unit;
s42: in S4, the offline model training specifically includes:
inputting 5+ k data in each row of the long-short term memory unit neural network input data of S3 as the input data characteristics of S41 into each long-short term memory unit, namely, the characteristic quantity C is 5+ k; and respectively inputting the m +1-k rows in the S3 into G long-short term memory units in the long-short term memory unit layer in the S41, namely, the number G of the long-short term memory units is m +1-k, and training the network by adopting a backward error propagation algorithm.
As a preferred embodiment of the present invention:
in the step of S5, the user is allowed to perform,
and transforming concrete filled steel tube sample data needing to be subjected to axial compression load-strain curve prediction into input data composition capable of being used for a long-term and short-term memory network by using the input data composition construction method of S31.
Compared with the prior art, the invention has the beneficial effects that:
the invention designs a data preprocessing and data composing method through design parameters, axial pressure load and axial strain data, and is used for long-short term memory network offline model training; constructing an axial strain data sequence, and inputting the axial strain data sequence and design parameters into a trained off-line model to obtain axial pressure load data output by the model; and restoring an axial pressure load-strain curve by using the axial strain data and the axial pressure load data to realize the prediction of the system axial pressure load-strain curve.
Drawings
Fig. 1 is a graph prediction principle flow.
Fig. 2 shows an input/output data configuration.
FIG. 3 is a layer of long and short term memory cells.
FIG. 4 is a long short term memory cell.
Detailed Description
The invention is described in further detail below with reference to the following figures and embodiments:
the invention provides a method for predicting a steel pipe concrete column axial load-strain curve based on a long-term and short-term memory network. For design parameters, axial load and axial strain data, a design data preprocessing and data constructing method is used for long-term and short-term memory network offline model training; constructing an axial strain data sequence, and inputting the axial strain data sequence and design parameters into a trained off-line model to obtain axial pressure load data output by the model; and restoring an axial pressure load-strain curve by using the axial strain data and the axial pressure load data to realize the prediction of the system axial pressure load-strain curve.
The flow of the curve prediction principle of the present invention is shown in fig. 1.
The invention is realized by the following technical scheme:
s1: and (5) acquiring experimental data.
Acquiring axial compression load-strain curve experimental data of concrete-filled steel tubular columns of different materials and sizes, and establishing a load-strain curve and five design parameters which are respectively as follows: diameter D of steel pipe and wall thickness t of steel pipe s Height H of column, strength f of steel s And concrete strength f c And constructing a load-strain curve prediction training sample by a corresponding system database.
S2: and (4) preprocessing data.
And for all the experimental data of the axial compression load-strain curves of the concrete-filled steel tube columns, setting 25000 mu epsilon as a limit value for acquiring axial strain so as to ensure that all the load-strain curves capture the characteristics of peak load points and post-peak curves. If the experimental data curve is short and does not reach the axial strain limit value of 25000 mu epsilon, the compensation operation is executed, and the axial compressive load-strain curve is extended to the set axial strain limit value according to the final slope of the curve. If the experimental data curve is long, truncation operation is carried out to discard the data exceeding the set limit value, and after the data preprocessing, all the axial load-strain curve data have uniform axial strain length.
S3: and (4) data composition.
Each set of design parameters and corresponding axial compressive load-strain curve is used as a data sample. And averagely dividing the axial compression load-strain curve experimental data of the steel pipe concrete column subjected to data preprocessing into m points according to strain values, wherein each point comprises an axial compression load value N and an axial strain value epsilon. And constructing input data of the neural network of the long and short term memory unit by using the axial strain value and the design parameter. And then constructing output data of the neural network of the long and short term memory unit by using the axial pressure load value. The input and output data of the present invention is constructed as shown in fig. 2, wherein the symbol i represents the ith sample in the database.
S31: the input data composition includes: taking k axial strains as a group, and forming a first row in one data sample by the k axial strains and five design parameters of the sample; then, another group of k axial strains are obtained by adopting a sliding window method, and the k axial strains and five design parameters form another row in one data sample; and so on until all axial strains are taken, at which time the input data for each data sample consists of m +1-k rows and 5+ k columns.
S32: the output data forming includes: obtaining a group of k axial load values to obtain each corresponding axial load value, averaging the group of axial load values to obtain the corresponding row of input dataAxial compressive load of
The data is taken as the output data of the neural network, and the output data of each data sample comprises m +1-k rows and 1 column.
S4: and (5) off-line model training.
After all samples are formed by the input and output data in the S3, the long and short term memory unit neural network is adopted to carry out off-line model training on all samples, and a long and short term memory unit neural network model which accords with convergence is obtained.
S41: the long and short term memory unit neural network sequentially comprises: an input layer, a long-short term memory unit layer, a full connection layer, a random discard (dropout) layer, a full connection layer and a data regression layer. The long-short term memory unit layer is shown in fig. 3, and includes G long-short term memory units, i.e. the length of the input time series is G, and the number of input data features is C. Wherein the long-short term memory unit is shown in fig. 4, and comprises:
at the current time of t-1, the output at the time of t needs to be calculated through a long-short term memory unit, and the input data is x t ，h t And c t Respectively representing the output (also called hidden state) and the cell state at time t. The long-short term memory unit uses the current state (h) t-1 And c t-1 ) To calculate the output h t And updated cell state c t . Using input gates i t Forgetting door f t Candidate unit g t And an output gate o t To control the refresh of the long and short term memory cells. The calculation formula is as follows:
i t ＝σ(W i x t +R i h t-1 +b i )
f t ＝σ(W f x t +R f h t-1 +b f )
g t ＝σ(W g x t +R g h t-1 +b g )
c t ＝f t ⊙c t-1 +i t ⊙g t
o t ＝σ(W o x t +R o h t-1 +b o )
h t ＝o t ⊙tanh(c t )
wherein W i ,W f ,W g ,W o Is an input weight matrix, R i ,R f ,R g ,R o Is a cyclic weight matrix, b i ,b f ,b g ,b o Are bias matrices that are adjusted through model training. σ is a sigmoid activation function, and an-is a dot product operation. Directly hiding state h t As the output of the long and short term memory unit.
S42: the offline model training in S4 specifically includes:
inputting 5+ k data in each row of the long-short term memory unit neural network input data of S3 as the input data characteristics of S41 into each long-short term memory unit, namely, the characteristic quantity C is 5+ k; the m +1-k rows in S3 are input into the G long-short term memory cells in the long-short term memory cell layer in S41, i.e., the number G of long-short term memory cells is m + 1-k. And training the network by adopting a backward error propagation algorithm, so that the axial load predicted by the long-short term memory network and the experimental value have the minimum root mean square error.
S5: and predicting an axial pressure load-strain curve.
And (4) finishing the design parameters of the concrete-filled steel tube column needing to predict the axial pressure load-strain curve. An axial strain data series was constructed, taking as the axial strain data series a vector with a maximum value of 25000 μ ∈ containing m evenly distributed data values. And transforming concrete filled steel tube sample data needing to be subjected to axial compression load-strain curve prediction into input data composition capable of being used for a long-term and short-term memory network by using the input data composition construction method of S31. And inputting the input data number into the offline model obtained in the step S4 to obtain the corresponding axial load output. And taking the average value of all axial strains in each line of the input data and the axial pressure load value in the line corresponding to the output data as a point of an axial pressure load-strain plane, and sequentially calculating all axial pressure load-strain points so as to restore a complete axial pressure load-strain curve and finish the prediction of the axial pressure load-strain curve.
According to the invention, through design parameters, axial load and axial strain data, a data preprocessing and data composing method is designed, and the method is used for long-term and short-term memory network offline model training; constructing an axial strain data sequence, and inputting the axial strain data sequence and design parameters into a trained off-line model to obtain axial pressure load data output by the model; and restoring an axial pressure load-strain curve by using the axial strain data and the axial pressure load data to realize the prediction of the system axial pressure load-strain curve.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, but any modifications or equivalent variations made according to the technical spirit of the present invention are within the scope of the present invention as claimed.
Claims (4)
1. The method for predicting the axial load-strain curve of the concrete-filled steel tubular column based on the long-term and short-term memory network is characterized by comprising the following steps of: the method comprises the following steps:
s1: acquiring experimental data;
acquiring axial compression load-strain curve experimental data of concrete-filled steel tubular columns of different materials and sizes, and establishing a load-strain curve and five design parameters, wherein the five design parameters are as follows: diameter D of steel pipe and wall thickness t of steel pipe s Height H of column, strength f of steel s And concrete strength f c Constructing a load-strain curve prediction training sample through design parameters by a corresponding system database;
s2: preprocessing data;
for all the experimental data of the axial compression load-strain curve of the steel tube concrete column, 25000 mu epsilon is set as a limit value for collecting axial strain, if the experimental data curve is shorter and does not reach the axial strain limit value of 25000 mu epsilon, compensation operation is executed, and the axial compression load-strain curve is extended to the set axial strain limit value according to the final slope of the load-strain curve; if the experimental data curve is longer, performing truncation operation, discarding data exceeding a set limit value, and preprocessing all the axial load-strain curve data to have uniform axial strain length;
s3: data composition;
each set of design parameters and the corresponding axial compressive load-strain curve are used as a data sample; averagely dividing the experimental data of the axial compression load-strain curve of the concrete-filled steel tube column subjected to data preprocessing into m points according to strain values, wherein each point comprises an axial compression load value N and an axial strain value epsilon; constructing input data of the neural network of the long and short term memory unit by using the axial strain value and the design parameter; constructing output data of the neural network of the long and short term memory unit by using the axial pressure load value;
s4: training an off-line model;
after all samples are formed by input and output data in S3, performing off-line model training on all samples by using a long-short term memory unit neural network to obtain a long-short term memory unit neural network model which accords with convergence;
s5: predicting an axial pressure load-strain curve;
arranging the design parameters of the concrete-filled steel tube column needing to predict the axial load-strain curve; constructing an axial strain data sequence, and taking a vector with a maximum value of 25000 mu epsilon and m data values which are evenly distributed as the axial strain data sequence; converting concrete filled steel tube sample data needing to be subjected to axial compression load-strain curve prediction into input data composition capable of being used for a long-term and short-term memory network by using the input data composition construction method of S3; inputting the input data number into the offline model obtained in S4 to obtain corresponding axial load output; and taking the average value of all axial strains in each line of the input data and the axial pressure load value in the line corresponding to the output data as a point of an axial pressure load-strain plane, and sequentially calculating all axial pressure load-strain points so as to restore a complete axial pressure load-strain curve and finish the prediction of the axial pressure load-strain curve.
2. The method for predicting the axial load-strain curve of the concrete-filled steel tubular column based on the long and short term memory network as claimed in claim 1, wherein the method comprises the following steps:
in the step of S3, the user is allowed to perform,
s31: the input data composition includes: taking k axial strains as a group, and forming a first row in one data sample by the k axial strains and five design parameters of the sample; then, another group of k axial strains are obtained by adopting a sliding window method, and the k axial strains and five design parameters form another row in one data sample; and so on until all axial strains are taken out, and at the moment, the input data of each data sample form a structure containing m +1-k rows and 5+ k columns;
s32: the output data construction includes: obtaining a group of k axial load values to obtain each corresponding axial load value, averaging the group of axial load values to obtain the corresponding axial load of the row of input data
The data is taken as the output data of the neural network, and the output data of each data sample comprises m +1-k rows and 1 column.
3. The method for predicting the axial load-strain curve of the concrete-filled steel tubular column based on the long and short term memory network as claimed in claim 1, wherein the method comprises the following steps:
in the step of S4, the user is allowed to perform,
s41: the long and short term memory unit neural network sequentially comprises: the system comprises an input layer, a long-term and short-term memory unit layer, a full connection layer, a random discarding layer, a full connection layer and a data regression layer; the long-short term memory unit layer comprises G long-short term memory units, namely the length of an input time sequence is G, and the characteristic quantity of input data is C; wherein the long-short term memory unit comprises: at the current time of t-1, the output at the time of t needs to be calculated through a long-short term memory unit, and the input data is x t ，h t And c t Respectively representing the output and cell states at time t; the long-short term memory cell uses the current state h t-1 And c t-1 To calculate the output h t And updated cell state c t (ii) a Using input gates i t Forgetting door f t Candidate unit g t And an output gate o t To control the update of the long and short term memory unit, the calculation formula is as follows:
i t ＝σ(W i x t +R i h t-1 +b i )
f t ＝σ(W f x t +R f h t-1 +b f )
g t ＝σ(W g x t +R g h t-1 +b g )
c t ＝f t ⊙c t-1 +i t ⊙g t
o t ＝σ(W o x t +R o h t-1 +b o )
h t ＝o t ⊙tanh(c t )
wherein W i ,W f ,W g ,W o Is an input weight matrix, R i ,R f ,R g ,R o Is a cyclic weight matrix, b i ,b f ,b g ,b o Is a bias matrix, and the matrix is adjusted through model training; σ is a sigmoid activation function, which is a dot product operation; directly output the state h t As the output of the long-short term memory unit;
s42: in S4, the offline model training specifically includes:
inputting 5+ k data in each row of the long-short term memory unit neural network input data of S3 as the input data characteristics of S41 into each long-short term memory unit, namely, the characteristic quantity C is 5+ k; and respectively inputting the m +1-k rows in the S3 into G long-short term memory units in the long-short term memory unit layer in the S41, namely, the number G of the long-short term memory units is m +1-k, and training the network by adopting a backward error propagation algorithm.
4. The method for predicting the axial load-strain curve of the concrete-filled steel tubular column based on the long and short term memory network as claimed in claim 2, wherein the method comprises the following steps:
in the step of S5, the user is allowed to perform,
and transforming concrete filled steel tube sample data needing to be subjected to axial compression load-strain curve prediction into input data composition capable of being used for a long-term and short-term memory network by using the input data composition construction method of S31.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210500867.8A CN114894619B (en) | 2022-05-10 | 2022-05-10 | Method for predicting axial compressive load-strain curve of concrete filled steel tube column based on long-short-term memory network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210500867.8A CN114894619B (en) | 2022-05-10 | 2022-05-10 | Method for predicting axial compressive load-strain curve of concrete filled steel tube column based on long-short-term memory network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114894619A true CN114894619A (en) | 2022-08-12 |
| CN114894619B CN114894619B (en) | 2024-04-05 |
Family
ID=82720826
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210500867.8A Active CN114894619B (en) | 2022-05-10 | 2022-05-10 | Method for predicting axial compressive load-strain curve of concrete filled steel tube column based on long-short-term memory network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114894619B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6000277A (en) * | 1998-06-05 | 1999-12-14 | Southwest Research Institute | Method for the prediction of rupture in corroded transmission pipes subjected to combined loads |
| CN111027752A (en) * | 2019-11-19 | 2020-04-17 | 浙江大学 | Crop yield estimation method based on deep spatiotemporal feature joint learning |
| CN111222677A (en) * | 2019-10-22 | 2020-06-02 | 浙江运达风电股份有限公司 | Wind speed prediction method and system based on long-short term memory time neural network |
| RU2744041C1 (en) * | 2019-09-10 | 2021-03-02 | Леонид Сергеевич Чернышев | Method and a system for predicting time series values using an artificial neural network |
| CN114254555A (en) * | 2021-11-18 | 2022-03-29 | 中国空间技术研究院 | Deep learning-based counteraction wheel fault detection and health assessment system and method |
-
2022
- 2022-05-10 CN CN202210500867.8A patent/CN114894619B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6000277A (en) * | 1998-06-05 | 1999-12-14 | Southwest Research Institute | Method for the prediction of rupture in corroded transmission pipes subjected to combined loads |
| RU2744041C1 (en) * | 2019-09-10 | 2021-03-02 | Леонид Сергеевич Чернышев | Method and a system for predicting time series values using an artificial neural network |
| CN111222677A (en) * | 2019-10-22 | 2020-06-02 | 浙江运达风电股份有限公司 | Wind speed prediction method and system based on long-short term memory time neural network |
| CN111027752A (en) * | 2019-11-19 | 2020-04-17 | 浙江大学 | Crop yield estimation method based on deep spatiotemporal feature joint learning |
| CN114254555A (en) * | 2021-11-18 | 2022-03-29 | 中国空间技术研究院 | Deep learning-based counteraction wheel fault detection and health assessment system and method |
Non-Patent Citations (3)
| Title |
|---|
| XINYU FAN 等: "Prediction of Axial Compressive Load–Strain Curves of Circular Concrete-Filled Steel Tube Columns Using Long Short-Term Memory Network", MATERIALS, 22 April 2023 (2023-04-22) * |
| 王莉萍 等: "冷弯薄壁钢管混凝土柱轴压性能研究", 建筑结构学报, vol. 42, 30 November 2021 (2021-11-30) * |
| 谭继可;张海宾;王宇航;胡少伟;罗伟;: "热轧H型钢组合柱轴压承载性能研究", 建筑结构学报, no. 05, 10 April 2020 (2020-04-10) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114894619B (en) | 2024-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Mehrabi et al. | Seismic response prediction of FRC rectangular columns using intelligent fuzzy-based hybrid metaheuristic techniques | |
| Zhou et al. | Shear capacity estimation of fully grouted reinforced concrete masonry walls using neural network and adaptive neuro-fuzzy inference system models | |
| Li et al. | BP neural network improved by sparrow search algorithm in predicting debonding strain of FRP-strengthened RC beams | |
| CN112528517B (en) | Steel box girder fatigue reliability analysis method based on two-stage convergence criterion | |
| Sun et al. | Machine learning-based seismic fragility analysis of large-scale steel buckling restrained brace frames | |
| Hadidi et al. | A new hybrid algorithm for simultaneous size and semi-rigid connection type optimization of steel frames | |
| Jayalekshmi et al. | Empirical approach for determining axial strength of circular concrete filled steel tubular columns | |
| Pang et al. | Integrated framework for seismic fragility assessment of cable-stayed bridges using deep learning neural networks | |
| CN115952575A (en) | Steel pipe concrete arch bridge linear prediction method based on parameter mapping learning algorithm | |
| CN114894619B (en) | Method for predicting axial compressive load-strain curve of concrete filled steel tube column based on long-short-term memory network | |
| Khorramian et al. | Active Learning Kriging-Based Reliability for Assessing the Safety of Structures Theory and Application | |
| Cheng | Serviceability reliability analysis of prestressed concrete bridges | |
| Kheyroddin et al. | Performance of circular concrete filled steel tube members subjected to axial loading | |
| CN113505422B (en) | Steel concrete member data management method and system and mechanics model optimization method and system | |
| Chaudhary et al. | Neural networks for deflections in continuous composite beams considering concrete cracking | |
| Aghamaleki et al. | Compression behavior of the concrete-filled double skin steel tube columns under hydrostatic pressure: Experimental and modeling study | |
| Mushchanov et al. | Numerical methods in assessing the reliability of spatial metal structures with a high level of responsibility | |
| CN115343784A (en) | Local air temperature prediction method based on seq2seq-attention model | |
| Ghasemi et al. | ANFIS–TLBO Hybrid Approach to Predict Compressive Strength of Rectangular FRP Columns | |
| CN113806839A (en) | Method for calculating integral seismic vulnerability of continuous beam bridge based on multivariate Copula function | |
| Balduk et al. | Calculation of a multi-span frame for stability | |
| Pham et al. | A robust approach in prediction of RCFST columns using machine learning algorithm | |
| Slimani et al. | Contribution to the Evaluation of Steel Structures Resistance to Lateral Displacement’s | |
| Fang et al. | Structural behavior of cold-formed steel channel sections with edge-stiffened and un-stiffened holes under axial compression: numerical simulations, deep learning, and proposed design equations | |
| CN105550479B (en) | A kind of load-bearing part stability loading prediction method considering kinematic hardening behavior |
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 |