CN116522797A - Method for predicting RC rectangular column damage mode based on artificial neural network - Google Patents
Method for predicting RC rectangular column damage mode based on artificial neural network Download PDFInfo
- Publication number
- CN116522797A CN116522797A CN202310585872.8A CN202310585872A CN116522797A CN 116522797 A CN116522797 A CN 116522797A CN 202310585872 A CN202310585872 A CN 202310585872A CN 116522797 A CN116522797 A CN 116522797A
- Authority
- CN
- China
- Prior art keywords
- rectangular column
- artificial neural
- neural network
- ratio
- column member
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 230000006378 damage Effects 0.000 title claims abstract description 53
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 49
- 238000005452 bending Methods 0.000 claims abstract description 39
- 239000011150 reinforced concrete Substances 0.000 claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 20
- 230000002787 reinforcement Effects 0.000 claims abstract description 17
- 238000009826 distribution Methods 0.000 claims abstract description 16
- 230000006835 compression Effects 0.000 claims abstract description 10
- 238000007906 compression Methods 0.000 claims abstract description 10
- 238000007619 statistical method Methods 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 5
- 238000010206 sensitivity analysis Methods 0.000 claims abstract description 5
- 230000004913 activation Effects 0.000 claims description 23
- 239000011159 matrix material Substances 0.000 claims description 21
- 238000012360 testing method Methods 0.000 claims description 12
- 238000010586 diagram Methods 0.000 description 18
- 238000004458 analytical method Methods 0.000 description 16
- 238000010008 shearing Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 7
- 238000013459 approach Methods 0.000 description 3
- 239000004567 concrete Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Geometry (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of artificial neural networks, in particular to a method for predicting an RC rectangular column failure mode based on an artificial neural network. The method comprises the following steps: based on the prediction method of the artificial neural network, the failure mode of the rectangular column member is judged by combining Levenberg Marquardt back propagation training algorithm; firstly, setting an artificial neural network structure of a reinforced concrete rectangular column member in a damage mode; processing a data set of the rectangular column member, and performing sensitivity analysis on the hoop distribution rate, the longitudinal reinforcement rate, the effective height ratio of the stirrup spacing to the section, the shear span ratio, the axial compression ratio and the bending shear ratio in the damage mode of the rectangular column member to obtain a prediction result of the damage mode of the rectangular column member; and carrying out statistical analysis on the prediction result of the damage mode of the rectangular column member. The prediction accuracy in the aspect of judging the damage mode of the column member is improved by the method based on the artificial neural network.
Description
Technical Field
The invention relates to the field of artificial neural networks, in particular to a method for predicting an RC rectangular column failure mode based on an artificial neural network.
Background
According to the investigation after earthquake and a large number of reinforced concrete column earthquake resistance tests, different reinforcement characteristics, section geometric characteristics and stress conditions can be found to cause different damage modes of column members. General engineering classifies failure modes into bending failure, bending-shear failure, and shear failure. The damage mode of the reinforced concrete rectangular column member column has direct influence on the ductility and the earthquake resistance of the reinforced concrete rectangular column member column, so that the prediction of the damage mode of the reinforced concrete rectangular column member column is crucial.
The method for predicting the failure mode of the reinforced concrete column member mainly comprises the following steps of: a prediction method based on a shear span ratio, a prediction method based on discriminant analysis and a prediction method based on engineering simplified discriminant analysis. The prediction method based on the shear span ratio only considers the parameter of the shear span ratio; the prediction method based on engineering simplification discriminant analysis considers two parameters, namely a shear-span ratio and a bending-shear ratio; the prediction method based on discriminant analysis considers three parameters of the effective height ratio of the stirrup spacing to the section, the shear span ratio and the bending shear ratio. The common method for predicting the damage mode of the reinforced concrete column member does not consider the influences of the hoop distribution rate, the axial compression ratio and the longitudinal reinforcement rate, and can influence the accuracy of the prediction result of the damage mode of the reinforced concrete column member, so that the accuracy of the prediction result of the damage mode of the reinforced concrete column member is lower.
Disclosure of Invention
In order to solve the problem of lower accuracy of a prediction result of a reinforced concrete column member failure mode in the existing method, the invention aims to provide a method for predicting an RC rectangular column failure mode based on an artificial neural network, and the adopted technical scheme is as follows:
the invention provides a method for predicting an RC rectangular column damage mode based on an artificial neural network, which comprises the following steps:
based on the prediction method of the artificial neural network, the failure mode of the rectangular column member is judged by combining Levenberg Marquardt back propagation training algorithm;
the method for judging the damage mode of the rectangular column member comprises the following steps: setting an artificial neural network structure of a reinforced concrete rectangular column member destruction mode; processing a data set of the rectangular column member, and performing sensitivity analysis on the hoop distribution rate, the longitudinal reinforcement rate, the effective height ratio of the stirrup spacing to the section, the shear span ratio, the axial compression ratio and the bending shear ratio in the damage mode of the rectangular column member to obtain a prediction result of the damage mode of the rectangular column member; and carrying out statistical analysis on the prediction result of the damage mode of the rectangular column member.
Preferably, parameters of the artificial neural network structure of the reinforced concrete rectangular column member failure mode include: the hidden layer activation function comprises an input layer node number, a hidden layer node number, an output layer node number, an input layer activation function, a hidden layer activation function and an output layer activation function.
Preferably, the input layer has 6 nodes, and the 6 nodes respectively represent hoop distribution rate, longitudinal reinforcement rate, effective height ratio of stirrup spacing to section, shear span ratio, axial compression ratio and bending shear ratio.
Preferably, the expression of the input layer activation function is:
y＝f(Wp+b)
wherein y is the output vector of the activation function; w is a weight matrix; p is the initial input vector; b is the bias vector.
Preferably, the hidden layer comprises two layers, the first layer has 100 nodes, and the second layer has 10 nodes; wherein the hidden layer activation function is a sigmoid function.
Preferably, the number of output layer nodes is 3, wherein each node corresponds to a destruction mode; wherein the output layer activation function is a softmax function.
Preferably, the expression of the Levenberg Marquardt back-propagation training algorithm is:
△x＝-[J T J+μI] -1 J T ε
wherein J isA Jacobian matrix; j (J) T A transpose matrix of the Jacobian matrix J; mu is a positive constant; i is an identity matrix; epsilon is the network error.
Preferably, the bending shear ratio, the shear span ratio are related to the failure mode of the reinforced concrete rectangular column member.
Preferably, the accuracy of the training data of the trained artificial neural network reaches 100%, the accuracy of the test data reaches 95.3%, the minimum mean square error of the training data is 0.0018, and the minimum mean square error of the test data is 0.038.
The invention has the following beneficial effects: compared with the traditional prediction method, the method based on the artificial neural network has better performance in judging the damage mode of the column member than the traditional judging and predicting method. With the increasing number of column member datasets, the accuracy of future artificial neural network models will increase significantly. Along with the development of the integration of civil engineering and artificial neural networks, it is expected that not only the failure mode prediction will use the artificial neural network, but even the deformation limit of the component and the structure of the concrete material can be obtained directly based on the artificial neural network (Artificial Neural Network, ANN) in combination with a large number of test data sets, and the original test data will rejuvenate with the rising of the artificial neural network.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of an ANN structure according to an embodiment of the present invention;
FIG. 2 is a box plot of the effect of the cuff ratio on the failure mode provided by one embodiment of the present invention;
FIG. 3 is a box plot of the impact of longitudinal reinforcement ratio on failure mode provided by one embodiment of the present invention;
FIG. 4 is a box plot of the effect of the effective height ratio of stirrup spacing to cross section on failure mode provided by one embodiment of the present invention;
FIG. 5 is a box diagram of the impact of a cross-cut comparison on a failure mode provided by one embodiment of the present invention;
FIG. 6 is a box diagram of the effect of axle to pressure ratio on failure mode provided by one embodiment of the present invention;
FIG. 7 is a box diagram of the impact of bend-shear ratios on failure modes provided by one embodiment of the present invention;
FIG. 8 is a diagram of a Pearson correlation coefficient matrix according to an embodiment of the present invention;
FIG. 9 is a schematic diagram of data prediction accuracy according to an embodiment of the present invention;
FIG. 10 is a diagram illustrating a minimum mean square error provided by an embodiment of the present invention;
FIG. 11 is a schematic diagram of a confusion matrix according to one embodiment of the present invention;
FIG. 12 is a schematic diagram of failure mode prediction based on bend-to-shear ratio and shear-span ratio provided by an embodiment of the present invention;
fig. 13 is a schematic diagram of a method for predicting an RC rectangular column failure mode based on an artificial neural network according to an embodiment of the invention.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following describes in detail a method for predicting the failure mode of the RC rectangular column based on the artificial neural network according to the present invention with reference to the accompanying drawings and the preferred embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The embodiment of the invention provides a concrete implementation method of a method for predicting an RC rectangular column failure mode based on an artificial neural network, which is suitable for a scene of reinforced concrete rectangular column member failure mode prediction. The method aims to solve the problem that in the existing method, the accuracy of the prediction result of the damage mode of the reinforced concrete column member is low. The invention judges the damage mode of the rectangular column member by combining Levenberg Marquardt back propagation training algorithm through a prediction method based on an artificial neural network.
The following specifically describes a specific scheme of the method for predicting the RC rectangular column failure mode based on the artificial neural network.
Referring to fig. 13, a method for predicting an RC rectangular column failure mode based on an artificial neural network according to the present embodiment includes the following steps:
based on the prediction method of the artificial neural network, the failure mode of the rectangular column member is judged by combining Levenberg Marquardt back propagation training algorithm. Note that, the Levenberg Marquardt back propagation training algorithm is the marquardt method, which is a well-known technique for those skilled in the art, and will not be described herein.
The method for judging the damage mode of the rectangular column member comprises the following steps: setting an artificial neural network structure of a reinforced concrete rectangular column member destruction mode; processing a data set of the rectangular column member, and performing sensitivity analysis on the hoop distribution rate, the longitudinal reinforcement rate, the effective height ratio of the stirrup spacing to the section, the shear span ratio, the axial compression ratio and the bending shear ratio in the damage mode of the rectangular column member to obtain a prediction result of the damage mode of the rectangular column member; and carrying out statistical analysis on the prediction result of the damage mode of the rectangular column member.
Parameters of the artificial neural network structure of the reinforced concrete rectangular column member failure mode include: the method comprises the steps of inputting the number of nodes of a layer, hiding the number of nodes of a layer, outputting the number of nodes of a layer, inputting an activation function of a layer, hiding the activation function of a layer and outputting the activation function of the layer;
the input layer is provided with 6 nodes, and the 6 nodes respectively represent hoop distribution rate, longitudinal reinforcement rate, effective height ratio of stirrup spacing to section, shear span ratio, axial compression ratio and bending shear ratio;
the expression of the input layer activation function is as follows:
y＝f(Wp+b)
wherein y is the output vector of the activation function; w is a weight matrix; p is the initial input vector; b is the bias vector.
The hidden layer comprises two layers, wherein the first layer is provided with 100 nodes, and the second layer is provided with 10 nodes; wherein the hidden layer activation function is a sigmoid function.
The expression of the sigmoid function is:
wherein f (x) is the output value of the sigmoid function; x is an input variable; e is a natural constant.
The number of the output layer nodes is 3, wherein each node corresponds to a destruction mode; wherein the output layer activation function is a softmax function.
The expression of the softmax function is:
wherein softmax (x) is the output value of the softmax function; e is a natural constant; x is x i Is the i-th input variable; x is x j Is the j-th input variable; n is the number of input variables.
Referring to fig. 1, fig. 1 is a schematic diagram of an artificial neural network structure, that is, an ANN structure schematic diagram.
The expression of the Levenberg Marquardt back propagation training algorithm is:
△x＝-[J T J+μI] -1 J T ε
wherein J is a Jacobian matrix; j (J) T A transpose matrix of the Jacobian matrix J; mu is a positive constant; i is an identity matrix; epsilon is the network error.
When mu is 0, the Levenberg Marquardt back propagation training algorithm is Newton's method; as μ increases in value, the Levenberg Marquardt back propagation training algorithm approaches the gradient descent method.
After testing the artificial neural network model, it was found that the trained model produced better results without standardized or normalized data. It may be that the input parameter dimensions are relatively close and the amount of data is relatively sparse for one failure mode. Thus, if the dataset is normalized, some sensitivity to this failure mode may be lost.
Referring to fig. 2 to 7, fig. 2 is a box diagram showing the influence of the hoop allocation rate on the damage mode; FIG. 3 is a box plot of the effect of longitudinal reinforcement on failure mode; FIG. 4 is a box plot of the effect of stirrup spacing versus section effective height on failure mode; FIG. 5 is a box plot of the impact of a shear span comparison on a failure mode; FIG. 6 is a box plot of the effect of axle to pressure ratio on failure mode; fig. 7 is a box diagram of the influence of bending-shearing ratio on the damage mode, and fig. 2 to 7 are respectively the qualitative influence of six input parameters, namely hoop distribution rate, longitudinal reinforcement rate, effective height ratio of hooping spacing to section, shearing span ratio, axial compression ratio and bending-shearing ratio, on the damage mode. The horizontal axes of fig. 2 to 7 are all failure modes, wherein 1 in the horizontal axes of fig. 2 to 7 represents a bending failure mode, 2 in the horizontal axes represents a bending-shearing failure mode, and 3 in the horizontal axes represents a shearing failure mode; the vertical axis in fig. 2 is the cuff ratio; the vertical axis in fig. 3 is the longitudinal reinforcement ratio; the vertical axis in fig. 4 is the effective height ratio of the stirrup spacing to the cross section; the vertical axis in fig. 5 is the shear-span ratio; the vertical axis in fig. 6 is the axial pressure ratio; the vertical axis in fig. 7 is the bending shear ratio.
Referring to fig. 8, fig. 8 is a Pearson correlation coefficient matrix, and fig. 8 reflects the quantitative analysis of the influence of each parameter on the destruction mode by the Pearson correlation coefficient matrix; ρ "in fig. 8 is the cuff ratio; ρ l The longitudinal reinforcement ratio is adopted; s/d is the effective height ratio of the stirrup spacing to the section; a/d is the shear span ratio; P/A g *f c ' is the axial pressure ratio; v (V) p /V n Is the bending shear ratio. As can be seen from the lowest data on the coordinate axis in FIG. 8, the bending shear ratio (V p /V n ) And shear span ratio (a/d) is relatively related to the reinforced concrete rectangular column member failure mode. Ratio to bending and shearing (V) p /V n ) The hoop distribution rate (rho) and the longitudinal reinforcement rate (rho) are compared with the shear span ratio (a/d) l ) Ratio of axial pressure (P/A) g *f c '), and the effective height ratio (s/d) of the stirrup spacing to the cross section is not related to the failure modeHas significant impact.
Of the 111 sets of data in table 1, 89 sets were used as model training data, and the remaining 22 sets were used as test data. After model optimization, the accuracy of the training data and the test data respectively reach 100% and 95.3%, namely, the accuracy of the training data of the trained artificial neural network reaches 100%, and the accuracy of the test data reaches 95.3%. Training data minimum Mean Square Error (MSE) was 0.0018 and test data minimum Mean Square Error (MSE) was 0.038. Referring to fig. 9 and 10, fig. 9 is a schematic diagram of data prediction accuracy, and fig. 10 is a schematic diagram of minimum mean square error, wherein the horizontal axes of fig. 9 and 10 are the number of iterations.
As shown in table 1, table 1 is the parameter data and the destruction mode of the rectangular column.
TABLE 1
Wherein FM represents a random variable seismic failure mode.
Referring to fig. 11, fig. 11 is a schematic diagram of an confusion matrix, and according to the confusion matrix shown in fig. 11, among 22 prediction results, 1 prediction error exists, and 1 "bend-shear failure" mode is erroneously determined as "shear failure" mode.
The prediction results are summarized in table 2, and the detailed description of table 2 is as follows:
(1) Of the 13 sets of data in which bend damage occurred, all data were correctly predicted to be bend damage.
(2) Of the 7 sets of data for which a bend-shear failure occurred, 6 sets of data were correctly predicted as a bend-shear failure and 1 set of data were incorrectly predicted as a shear failure.
(3) Of the 2 sets of data where a shear failure occurred, all data were correctly predicted as a shear failure.
TABLE 2
Statistical analysis of the predicted results is shown in Table 3.
TABLE 3 Table 3
From statistical analysis of the prediction results, 95.3% (95.3% = (100% +86% + 100%)/3) of the pillar members were correctly predicted.
In the shear-span-ratio-based approach, the failure mode of the pillar member is related only to the shear span ratio. When the shear span ratio λ=a/d is less than or equal to 2, this failure mode is classified as shear failure; when the shear span ratio λ=a/d is ≡4, this failure mode is classified as bending failure; when 2 < a/d < 4, this failure mode is classified as a bend-shear failure. Where a is expressed as the cross-span length and d is expressed as the effective height of the cross-section.
When the shear span ratio is lambda=a/d is less than or equal to 2, 63% of column members are subjected to shear failure; when the shear span ratio is lambda=a/d is more than or equal to 4', 51% of column members are subjected to bending fracture; at a shear span ratio of 2 < a/d < 4, 92% of the column members experience bend-shear failure.
The results show that 68.7% (68.7% = (63% +51% + 92%)/3) of the column member was correctly predicted. This also means that the shear-span ratio of λ=a/d=2 and λ=a/d=4 is insufficient to distinguish between the different failure modes of the component. Therefore, it is not appropriate to divide the destruction mode by only the shear-span ratio.
The discriminant analysis method includes a parametric method and an nonparametric method. The parametric approach assumes that each class of observations comes from a population of multivariate normal distributions, and that the distribution mean centers of each class may be different. The nonparametric method does not require knowledge of the distribution of the various categories from the population, it uses nonparametric methods for each category and estimates the distribution density of each category, from which a discriminant criterion is then established.
The discriminant analysis method is based on Fisher criterion (Bayer criterion) and is combined with Bayes criterion (Bayes criterion) to judge the failure modes of 111 reinforced concrete rectangular column members, so that three discriminant equations can be obtained, and the three discriminant equations are shown in the following formula. The discriminant equation is a linear statistical equation, and three independent output variable values are calculated using the stirrup spacing to section effective height ratio (χ), shear span ratio (λ), and bend-shear ratio (m): y is 1 、y 2 And y 3 . According to y 1 、y 2 And y 3 And (3) predicting the failure mode of the reinforced concrete column. If y 1 The maximum is the reinforced concrete column failure to bend. If y 2 Maximum, the reinforced concrete column failure is a bend-shear failure. If y 3 The maximum is the shear failure of the reinforced concrete column. The coefficients of the discriminant equation represent weights. It can be seen that the bending shear ratio coefficient is the largest, indicating that the discrimination equation is most sensitive to the value of the bending shear ratio. This illustrates that the bending shear ratio can reflect the failure mode of the reinforced concrete column to some extent. That is why the bending shear ratio can be more accurate than the other parameters when using a single parameter analysis. However, other parameters are not negligible.
y 1 ＝-7.664χ+6.448λ+16.86m-19.455
y 2 ＝-0.168χ+4.942λ+21.232m-20.815
y 3 ＝-2.935χ+4.038λ+25.726m-23.226
Wherein y is 1 、y 2 And y 3 All are independent output variable values; χ is the effective height ratio of the stirrup spacing to the section; lambda is the shear-span ratio; m is the bending shear ratio.
The results of the prediction method based on the discriminant equation are shown in Table 4. The details of table 4 are as follows:
(1) Of the 64 sets of data for which bend corruption occurred, 59 sets of data were correctly predicted to be bend corruption and 5 sets of data were incorrectly predicted to be bend-shear corruption.
(2) Of the 36 sets of data for which a bend-shear failure occurred, 26 sets of data were correctly predicted as a bend-shear failure, 5 sets of data were incorrectly predicted as a bend failure, and 5 sets of data were incorrectly predicted as a shear failure.
(3) Of the 11 sets of data for which a shear failure occurred, 10 sets of data were correctly predicted as a shear failure, and 1 set of data were incorrectly predicted as a bend-shear failure.
TABLE 4 Table 4
It was found that by discriminant analysis, 85.1% (85.1% = (92.2% +72.2% + 90.9%)/3) of the column member was correctly predicted. Its accuracy is far higher than other prediction methods using only a single parameter.
The comprehensive analysis of the discriminant analysis method shows that the hoop distribution rate, the axial pressure ratio, the longitudinal reinforcement rate, the effective height ratio of the hooping spacing and the section have no obvious influence on the destruction mode of the column member. Two of the parameters that have the greatest impact on the failure mode of the column member are the bend-to-shear ratio and the shear-span ratio. Referring to fig. 12, fig. 12 is a schematic diagram of failure mode prediction based on a bending shear ratio and a span ratio, wherein the failure modes of the 111 rectangular column members are divided according to the bending shear ratio and the span ratio, the horizontal axis is the span ratio, and the vertical axis is the reduced bending shear ratio in fig. 12.
Based on fig. 12, a rapid and simple method for manually predicting the failure mode of the column member based on the bending shear ratio and the span ratio, namely an engineering simplification discriminant analysis method, can be developed. According to fig. 12, the failure mode of the reinforced concrete column can be predicted using the following criteria.
(1) For the column component with the shear span ratio lambda less than or equal to 2, when the bending shear ratio V p /V n More than or equal to 0.8, mainly shear damage; when the bending shear ratio is smaller than V p /V n <At 0.8, bend-shear failure predominates.
(2) For column members with shear span ratio 2 < lambda < 4, when bending shear ratio V p /V n <At 0.8, bending fracture is the main; when the bending shear ratio is 0.8 or less V p /V n When the bending and shearing damage is less than or equal to 1.5, the bending and shearing damage is mainly used; when V is p /V n > 1.5, shear failure of the column members occurred.
(3) For the column component with the shear span ratio lambda being more than or equal to 4, when the bending shear ratio V p /V n When the bending damage is less than or equal to 1.2, the bending damage is mainly the bending damage; when the bending shear ratio V p /V n At > 1.2, no conclusions can be drawn at this time due to the lack of samples.
The column member failure mode prediction criteria are summarized in table 5.
TABLE 5
The prediction results based on the prediction criteria in table 5 are listed in table 6. The details of Table 6 are as follows:
(1) Of the 64 sets of data for which bend corruption occurred, 49 sets of data were correctly predicted to be bend corruption and 15 sets of data were incorrectly predicted to be bend-shear corruption.
(2) Of the 36 sets of data where the bend-shear corruption occurred, 31 sets of data were correctly predicted as bend-shear corruption, 2 sets of data were incorrectly predicted as bend corruption, and 3 sets of data were incorrectly predicted as shear corruption.
(3) Of the 11 sets of data for which a shear failure occurred, 10 sets of data were correctly predicted as a shear failure, and 1 set of data were incorrectly predicted as a bend-shear failure.
TABLE 6
From this, it was found that the engineering simplified discriminant analysis method accurately predicts 84.5% (84.5% = (76.6% +86.1% + 90.9%)/3) of the column members, which is slightly lower than the accuracy of the discriminant analysis method.
Compared with the traditional prediction method, the ANN network prediction method has the following advantages:
(1) The accuracy of the ANN network prediction method is higher than that of the traditional prediction method. In the traditional prediction method, the accuracy of the prediction method based on discriminant analysis is highest and reaches 85.1%, while the accuracy of the prediction method of the ANN network reaches 95.3%.
(2) The prediction method of the ANN network takes more comprehensive factors into consideration. For example, the prediction method based on the shear span ratio only considers a single parameter, the prediction method based on the engineering simplified discriminant analysis considers 2 parameters, the prediction method based on the discriminant analysis considers 3 parameters, and the prediction method of the ANN network considers 6 parameters.
(3) The predictive power of the model may be dynamically increased as the database expands.
The embodiment is based on a prediction method of an artificial neural network and combines Levenberg Marquardt back propagation training algorithm to judge the damage mode of the rectangular column member. Firstly, setting an artificial neural network structure of a reinforced concrete rectangular column member failure mode; processing a data set of the rectangular column member, and performing sensitivity analysis on the hoop distribution rate, the longitudinal reinforcement rate, the effective height ratio of the stirrup spacing to the section, the shear span ratio, the axial compression ratio and the bending shear ratio in the damage mode of the rectangular column member to obtain a prediction result of the damage mode of the rectangular column member; and carrying out statistical analysis on the prediction result of the damage mode of the rectangular column member.
It should be noted that: the foregoing description of the preferred embodiments of the present invention is not intended to be limiting, but rather, any modifications, equivalents, improvements, etc. that fall within the principles of the present invention are intended to be included within the scope of the present invention.
Claims (9)
1. A method for predicting an RC rectangular column failure mode based on an artificial neural network, the method comprising the steps of:
based on the prediction method of the artificial neural network, the failure mode of the rectangular column member is judged by combining Levenberg Marquardt back propagation training algorithm;
the method for judging the damage mode of the rectangular column member comprises the following steps: setting an artificial neural network structure of a reinforced concrete rectangular column member destruction mode; processing a data set of the rectangular column member, and performing sensitivity analysis on the hoop distribution rate, the longitudinal reinforcement rate, the effective height ratio of the stirrup spacing to the section, the shear span ratio, the axial compression ratio and the bending shear ratio in the damage mode of the rectangular column member to obtain a prediction result of the damage mode of the rectangular column member; and carrying out statistical analysis on the prediction result of the damage mode of the rectangular column member.
2. The method for predicting an RC rectangular column failure mode based on an artificial neural network of claim 1, wherein the parameters of the artificial neural network structure of the reinforced concrete rectangular column member failure mode include: the hidden layer activation function comprises an input layer node number, a hidden layer node number, an output layer node number, an input layer activation function, a hidden layer activation function and an output layer activation function.
3. The method for predicting the failure mode of the RC rectangular column based on the artificial neural network according to claim 2, wherein the input layer is provided with 6 nodes, and the 6 nodes respectively represent a hoop distribution rate, a longitudinal reinforcement rate, a ratio of the effective heights of the hoops and the section, a shear span ratio, an axial compression ratio and a bending shear ratio.
4. The method for predicting an RC rectangular column failure mode based on an artificial neural network of claim 2, wherein the expression of the input layer activation function is:
y＝f(Wp+b)
wherein y is the output vector of the activation function; w is a weight matrix; p is the initial input vector; b is the bias vector.
5. The method for predicting an RC rectangular column failure mode based on an artificial neural network of claim 2, wherein the hidden layer comprises two layers, the first layer having 100 nodes and the second layer having 10 nodes; wherein the hidden layer activation function is a sigmoid function.
6. The method for predicting an RC rectangular column failure mode based on an artificial neural network of claim 2, wherein the number of output layer nodes is 3, wherein each node corresponds to a failure mode; wherein the output layer activation function is a softmax function.
7. The method for predicting an RC rectangular column failure mode based on an artificial neural network of claim 1, wherein the expression of the Levenberg Marquardt back propagation training algorithm is:
△x＝-[J T J+μI] -1 J T ε
wherein J is a Jacobian matrix; j (J) T A transpose matrix of the Jacobian matrix J; mu is a positive constant; i is an identity matrix; epsilon is the network error.
8. The method for predicting failure modes of an RC rectangular column based on an artificial neural network of claim 1, wherein the bend-to-shear ratio and the shear-span ratio are related to failure modes of a reinforced concrete rectangular column member.
9. The method for predicting RC rectangular column failure modes based on artificial neural network of claim 1,
the accuracy of the training data of the trained artificial neural network reaches 100 percent, the accuracy of the test data reaches 95.3 percent,
wherein the minimum mean square error of the training data is 0.0018, and the minimum mean square error of the test data is 0.038.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310585872.8A CN116522797A (en) | 2023-05-23 | 2023-05-23 | Method for predicting RC rectangular column damage mode based on artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310585872.8A CN116522797A (en) | 2023-05-23 | 2023-05-23 | Method for predicting RC rectangular column damage mode based on artificial neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116522797A true CN116522797A (en) | 2023-08-01 |
Family
ID=87406393
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310585872.8A Pending CN116522797A (en) | 2023-05-23 | 2023-05-23 | Method for predicting RC rectangular column damage mode based on artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116522797A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102032146B1 (en) * | 2018-04-11 | 2019-10-15 | 경희대학교 산학협력단 | Artificial neural network using piecewise linear rectification unit for compensating defect of element |
| CN112883463A (en) * | 2021-01-18 | 2021-06-01 | 华北水利水电大学 | Reinforced concrete column earthquake failure mode probability model and probability prediction method thereof |
| CN115688242A (en) * | 2022-11-02 | 2023-02-03 | 山东建筑大学 | Multi-strategy fused RC column earthquake damage mode identification method |
-
2023
- 2023-05-23 CN CN202310585872.8A patent/CN116522797A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102032146B1 (en) * | 2018-04-11 | 2019-10-15 | 경희대학교 산학협력단 | Artificial neural network using piecewise linear rectification unit for compensating defect of element |
| CN112883463A (en) * | 2021-01-18 | 2021-06-01 | 华北水利水电大学 | Reinforced concrete column earthquake failure mode probability model and probability prediction method thereof |
| CN115688242A (en) * | 2022-11-02 | 2023-02-03 | 山东建筑大学 | Multi-strategy fused RC column earthquake damage mode identification method |
Non-Patent Citations (1)
| Title |
|---|
| HAITAO WAN ET.AL: "Prediction of column failure modes based on artificial neural network", 《EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION》, vol. 22, 25 April 2023 (2023-04-25), pages 481 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Prediction of triaxial behavior of recycled aggregate concrete using multivariable regression and artificial neural network techniques | |
| Beck et al. | Bayesian updating of structural models and reliability using Markov chain Monte Carlo simulation | |
| Oommen et al. | Validation and application of empirical liquefaction models | |
| Lignos et al. | Development and utilization of structural component databases for performance-based earthquake engineering | |
| Lagaros et al. | Fragility assessment of steel frames using neural networks | |
| Gondia et al. | Mechanics-guided genetic programming expression for shear-strength prediction of squat reinforced concrete walls with boundary elements | |
| Mergos et al. | Loading protocols for European regions of low to moderate seismicity | |
| Qian et al. | Multi‐criteria decision making for seismic intensity measure selection considering uncertainty | |
| CN112365885B (en) | Training method and device of wake-up model and computer equipment | |
| Rahimi et al. | Collapse assessment of steel moment frames using endurance time method | |
| CN107784191A (en) | Anisotropic rock joint peak shear strength Forecasting Methodology based on neural network model | |
| CN109101717A (en) | Solid propellant rocket Reliability Prediction Method based on reality with the study of fuzzy data depth integration | |
| CN112528517A (en) | Steel box girder fatigue reliability analysis method based on two-stage convergence criterion | |
| CN111058840A (en) | Organic carbon content (TOC) evaluation method based on high-order neural network | |
| Allahvirdizadeh et al. | Probabilistic comparative investigation on introduced performance-based seismic design and assessment criteria | |
| CN110458341B (en) | Ultra-short-term wind power prediction method and system considering meteorological characteristics | |
| CN115688242A (en) | Multi-strategy fused RC column earthquake damage mode identification method | |
| Barros et al. | Adaptive approximate Bayesian computation by subset simulation for structural model calibration | |
| CN108763164B (en) | Evaluation method for coal and gas outburst inversion similarity | |
| Banimahd et al. | A multi‐level approach to predict the seismic response of rigid rocking structures using artificial neural networks | |
| CN116522797A (en) | Method for predicting RC rectangular column damage mode based on artificial neural network | |
| Zou et al. | Surrogate modeling for the seismic response estimation of residential wood frame structures | |
| Peng-hui et al. | Structural damage identification based on genetically trained ANNs in beams | |
| Ayyub et al. | Selective sampling in simulation-based reliability assessment | |
| Almustafa et al. | Data-driven reliability framework for qualitative damage states of reinforced concrete beams under blast loading |
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 |