Mountain area high-fill area bridge damage cause inversion method based on multi-source fusion

Abstract

The invention relates to the technical field of bridge damage analysis, and discloses a mountainous area high-fill area bridge damage cause inversion method based on multi-source fusion. The method comprises the following steps: S1, establishing a parameterized finite element model of a bridge, and acquiring axial strain of each pile foundation under the action of different damage inducements of the bridge; s2, calculating the axial strain of the pile foundation under the same working condition as that of the physical model experiment through the parameterized model, and then correcting the parameterized model by using physical model test data; s3, constructing an inversion model based on a support vector machine, and taking the pile foundation strain sample data set as the input of the model; s4, obtaining pile foundation strain data under multiple working conditions by utilizing the corrected parameterized finite element model; and S5, loading a support vector machine inversion model to obtain an inversion result, and analyzing inversion parameters and model evaluation results under the action of multiple damage inducements.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge damage analysis, and more specifically to a method for inverting bridge damage causes in a mountainous high-fill area based on multi-source fusion. Background Art

[0002] In mountainous areas with multiple sources, bridge construction requires a certain yard area to be obtained by backfilling and land reclamation. However, overhead bridges in high-fill areas have many components and a complex working environment. It is critical to clearly identify which components are used as measurement information carriers and accurately obtain structural mechanical behavior information. Traditional methods for obtaining information carriers of bridge components are mainly through manual inspections, regular inspections, and structural monitoring methods, which are all local and intermittent information acquisition methods. However, bridges are subject to various external loads, and the response information of the structure is continuous. Therefore, traditional information acquisition methods are not comprehensive;

[0003] At present, the method of structural damage inversion is mainly used to accurately obtain the information carrier of bridge components. The optimal model closest to the actual deformation law of the structure is identified mainly by observing the change law of information. That is, the response parameters of the structure before and after damage are compared, and the degree of damage of the structure is calculated according to the degree of change of the response parameters. Due to the difficulty in obtaining damage variables, the evolution law of structural damage is not very clear, and the constitutive equation based on damage needs further study. There are still many difficulties in the research of structural damage inversion. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the present invention provides a damage cause inversion method for bridges in mountainous high-fill areas based on multi-source fusion, which has the advantage of realizing accurate judgment and evaluation of structural damage causes.

[0005] To achieve the above object, the present invention provides the following technical solution: a method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion, comprising the following steps;

[0006] S1: Establish a parametric finite element model of the bridge, determine the action form and inversion index of the damage inducing factors of the bridge structure, and carry out physical model tests to obtain the axial strain of each pile foundation under different damage inducing factors of the bridge;

[0007] S2: The axial strain of the pile foundation under the same working conditions as the physical model experiment is calculated through the parameterized model, and then the parameterized model is corrected using the physical model test data to provide support for the authenticity of the sample data set of the damage-induced inversion model;

[0008] S3: Construct an inversion model based on support vector machine, take the pile foundation strain sample data set as the input of the model, and take the type, intensity and location of the damage cause as the output index. The specific construction steps are as follows:

[0009] (1) Obtain the strain characteristic data of each pile of the parametric finite element model under different working conditions, then fuse and reconstruct these characteristic data, calculate the range and mean of the strain characteristics of each pile, and use these calculation results as a supplement to the sample characteristics. The sample data set Z constructed in this way contains the feature X and the corresponding label Y, where X∈Rm×n, y∈Rm×q, m is the number of samples, n is the number of features, and q=3 is the number of labels including category number, position number and strength value;

[0010] (2) The sample data set is processed by normalization method to make the data have a uniform scale, eliminate the adverse effects caused by singular sample data, and improve the solution speed and accuracy. Then, the data set is randomly split into training set A and test set B in a ratio of 4:1 by the holdout method. At the same time, data samples are randomly selected from the physical model experimental data to form the validation set C.

[0011] (3) Construct a support vector machine model. This model adopts a one-to-many multi-classification approach to construct a classification hyperplane for different types of damage inducements and different action locations. When constructing the model, the penalty factor C of the inversion model and the bandwidth of the kernel function σ are selected;

[0012] (4) By training the training set A, the test set B is loaded into the model to further evaluate the generalization ability of the model, and the validation set C is used to verify the fit of the model with the actual situation;

[0013] S4: Using the modified parametric finite element model, the pile foundation strain data under multiple working conditions are obtained. Aiming at the needs of the support vector machine, the data are normalized and dimensionally reduced to construct a sample data set.

[0014] S5: Finally, the support vector machine inversion model is loaded, the inversion results are obtained, and the inversion parameters and model evaluation results under various damage inducements are analyzed to evaluate the model prediction accuracy. At the same time, a certain number of strain data are randomly selected from the physical model test data as a verification set to verify the reliability of the inversion model and the effectiveness of the model correction technology.

[0015] As a preferred technical solution of the present invention, in S2, the parameterized model calculation formula of the pile foundation axial strain is:

[0016] [\eps il on=\frac{N}{A\cdotE}];

[0017] Where \eps il on is the axial strain, N is the applied axial load, A is the cross-sectional area of ​​the pile, and E is the elastic modulus of the material.

[0018] As a preferred technical solution of the present invention, the reaction of the surrounding soil needs to be considered when calculating the axial strain of the pile foundation. The interaction between the pile foundation and the soil is accurately simulated by using the foundation reaction method, thereby obtaining an accurate axial strain of the pile foundation.

[0019] As a preferred technical solution of the present invention, the following formula is used when calculating the foundation elastic force using the foundation reaction method:

[0020] R = q\cdotA;

[0021] Where R is the foundation reaction, q is the load per unit area, and A is the area of ​​the foundation.

[0022] As a preferred technical solution of the present invention, in S3, the inversion model of the support vector machine adopts a nonlinear transformation method, that is, the sample data in the low-dimensional space is mapped to the high-dimensional space using a nonlinear transformation, and a linear model is used in the high-dimensional space to model the data.

[0023] As a preferred technical solution of the present invention, in S4, during the construction of the sample data set, the sample data set is linearly separable. The core of SVM is to find the optimal hyperplane in the high-dimensional space to perform classification or regression tasks. If the sample data set is not linearly separable, it is necessary to use a kernel function to map it to the high-dimensional space in order to find the optimal hyperplane.

[0024] As a preferred technical solution of the present invention, the function used when normalizing the data is:

[0025]

[0026] Among them, x i is the stress data, max is the maximum value of the stress data, and min is the minimum value of the stress data.

[0027] As a preferred technical solution of the present invention, the LDA method is used to perform dimensionality reduction processing on the data, and the directional projection of maximizing the difference between classes and minimizing the difference within classes is used to effectively separate samples of different classes.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention calculates the axial strain of the pile foundation under the same working conditions as the physical model experiment through a parameterized model, and then constructs an inversion model of a support vector machine, and uses the pile foundation strain sample data set as the input of the model, and uses the type, intensity and location of the damage cause as indicators as output, so that a theoretical and numerical model close to the actual engineering can be established, thereby realizing accurate judgment and evaluation of the causes of structural damage. DETAILED DESCRIPTION

[0029] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] The present invention provides a method for inverting the damage causes of bridges in mountainous high-fill areas based on multi-source fusion, comprising the following steps:

[0031] S1: Establish a parametric finite element model of the bridge, determine the action form and inversion index of the damage inducing factors of the bridge structure, and carry out physical model tests to obtain the axial strain of each pile foundation under different damage inducing factors of the bridge;

[0032] S2: The axial strain of the pile foundation under the same working conditions as the physical model experiment is calculated through the parameterized model. Then, the parameterized model is corrected using the physical model test data to provide support for the authenticity of the sample data set of the damage-induced inversion model. The parameterized model calculation formula for the axial strain of the pile foundation is:

[0033] [\eps il on=\frac{N}{A\cdotE}];

[0034] Where, \eps il on is the axial strain, N is the applied axial load, A is the cross-sectional area of ​​the pile, and E is the elastic modulus of the material. When calculating the axial strain of the pile foundation, the reaction of the surrounding soil needs to be considered. The interaction between the pile foundation and the soil is accurately simulated by the foundation reaction method, so as to obtain the accurate axial strain of the pile foundation; Where,: When the foundation reaction method is used to calculate the foundation elastic force, the following formula is used:

[0035] R = q\cdotA;

[0036] Where R is the foundation reaction, q is the load per unit area, and A is the area of ​​the foundation;

[0037] S3: Construct an inversion model based on support vector machine. The inversion model of support vector machine adopts nonlinear transformation, that is, it uses nonlinear transformation to map the sample data of low-dimensional space to high-dimensional space, and uses linear model to model the data in high-dimensional space. The pile foundation strain sample data set is used as the input of the model, and the type, intensity and location of the damage inducement are used as indicators as output. The specific construction steps are as follows:

[0038] (1) Obtain the strain characteristic data of each pile of the parametric finite element model under different working conditions, then fuse and reconstruct these characteristic data, calculate the range and mean of the strain characteristics of each pile, and use these calculation results as a supplement to the sample characteristics. The sample data set Z constructed in this way contains the feature X and the corresponding label Y, where X∈Rm×n, y∈Rm×q, m is the number of samples, n is the number of features, and q=3 is the number of labels including category number, position number and strength value;

[0039] (2) The sample data set is processed by normalization method to make the data have a uniform scale, eliminate the adverse effects caused by singular sample data, and improve the solution speed and accuracy. Then, the data set is randomly split into training set A and test set B in a ratio of 4:1 by the holdout method. At the same time, data samples are randomly selected from the physical model experimental data to form the validation set C.

[0040] (3) Construct a support vector machine model. This model adopts a one-to-many multi-classification approach to construct a classification hyperplane for different types of damage inducements and different action locations. When constructing the model, the penalty factor C of the inversion model and the bandwidth of the kernel function σ are selected;

[0041] (4) By training the training set A, the test set B is loaded into the model to further evaluate the generalization ability of the model, and the validation set C is used to verify the fit of the model with the actual situation;

[0042] S4: Using the modified parametric finite element model, the strain data of pile foundation under multiple working conditions are obtained. Aiming at the needs of support vector machine, the data is normalized and dimensionally reduced to construct a sample data set. In the process of constructing the sample data set, the sample data set is linearly separable. The core of SVM is to find the optimal hyperplane in high-dimensional space for classification or regression tasks. If the sample data set is not linearly separable, it is necessary to use a kernel function to map it to a high-dimensional space in order to find the optimal hyperplane; the function used when normalizing the data is:

[0043]

[0044] Among them, x iis stress data, max is the maximum value of stress data, and min is the minimum value of stress data. The LDA method is used to reduce the dimension of the data. By maximizing the difference between classes and minimizing the directional projection of the difference within classes, samples of different classes can be effectively separated.

[0045] S5: Finally, the support vector machine inversion model is loaded, the inversion results are obtained, and the inversion parameters and model evaluation results under various damage inducements are analyzed to evaluate the model prediction accuracy. At the same time, a certain number of strain data are randomly selected from the physical model test data as a verification set to verify the reliability of the inversion model and the effectiveness of the model correction technology.

[0046] The present invention calculates the axial strain of the pile foundation under the same working conditions as the physical model experiment through a parameterized model, and then constructs an inversion model of a support vector machine, and uses the pile foundation strain sample data set as the input of the model, and takes the type, intensity and location of the damage cause as indicators as output. In this way, a theoretical and numerical model close to the actual engineering can be established, and accurate judgment and evaluation of the structural damage causes can be achieved.

[0047] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0048] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion, characterized by: The method comprises the following steps: S1: Establish a parametric finite element model of the bridge, determine the action form and inversion index of the damage inducing factors of the bridge structure, and carry out physical model tests to obtain the axial strain of each pile foundation under different damage inducing factors of the bridge; S2: The axial strain of the pile foundation under the same working conditions as the physical model experiment is calculated through the parameterized model, and then the parameterized model is corrected using the physical model test data to provide support for the authenticity of the sample data set of the damage-induced inversion model; S3: Construct an inversion model based on support vector machine, take the pile foundation strain sample data set as the input of the model, and take the type, intensity and location of the damage cause as the output index. The specific construction steps are as follows: (1) Obtain the strain characteristic data of each pile of the parametric finite element model under different working conditions, then fuse and reconstruct these characteristic data, calculate the range and mean of the strain characteristics of each pile, and use these calculation results as a supplement to the sample characteristics. The sample data set Z constructed in this way contains the feature X and the corresponding label Y, where X∈Rm×n, y∈Rm×q, m is the number of samples, n is the number of features, and q=3 is the number of labels including category number, position number and strength value; (2) The sample data set is processed by normalization method to make the data have a uniform scale, eliminate the adverse effects caused by singular sample data, and improve the solution speed and accuracy. Then, the data set is randomly split into training set A and test set B in a ratio of 4:1 by the holdout method. At the same time, data samples are randomly selected from the physical model experimental data to form the validation set C. (3) Construct a support vector machine model. This model adopts a one-to-many multi-classification approach to construct a classification hyperplane for different types of damage inducements and different action locations. When constructing the model, the penalty factor C of the inversion model and the bandwidth of the kernel function σ are selected; (4) By training the training set A, the test set B is loaded into the model to further evaluate the generalization ability of the model, and the validation set C is used to verify the fit of the model with the actual situation; S4: Using the modified parametric finite element model, the pile foundation strain data under multiple working conditions are obtained. Aiming at the needs of the support vector machine, the data are normalized and dimensionally reduced to construct a sample data set. S5: Finally, the support vector machine inversion model is loaded, the inversion results are obtained, and the inversion parameters and model evaluation results under various damage inducements are analyzed to evaluate the model prediction accuracy. At the same time, a certain number of strain data are randomly selected from the physical model test data as a verification set to verify the reliability of the inversion model and the effectiveness of the model correction technology.

2. The method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion according to claim 1 is characterized by: In S2, the parameterized model calculation formula of the pile foundation axial strain is: [\epsilon=\frac{N}{A\cdotE}]; Where, \epsilon is the axial strain, N is the applied axial load, A is the cross-sectional area of ​​the pile, and E is the elastic modulus of the material.

3. The method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion according to claim 2 is characterized by: When calculating the axial strain of the pile foundation, the reaction of the surrounding soil needs to be considered. The interaction between the pile foundation and the soil can be accurately simulated by using the foundation reaction method to obtain the accurate axial strain of the pile foundation.

4. The method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion according to claim 3 is characterized by: The following formula is used when calculating foundation elastic force using the foundation reaction method: R = q\cdotA; Where R is the foundation reaction, q is the load per unit area, and A is the area of ​​the foundation.

5. The method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion according to claim 1 is characterized by: In S3, the inversion model of the support vector machine adopts a nonlinear transformation method, that is, the sample data in the low-dimensional space is mapped to the high-dimensional space using a nonlinear transformation, and a linear model is used in the high-dimensional space to model the data.

6. The method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion according to claim 1 is characterized by: In S4, during the construction of the sample data set, the sample data set is linearly separable. The core of SVM is to find the optimal hyperplane in the high-dimensional space to perform classification or regression tasks. If the sample data set is not linearly separable, a kernel function is needed to map it to the high-dimensional space in order to find the optimal hyperplane.

7. The method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion according to claim 6 is characterized by: The function used to normalize the data is: Among them, x i is the stress data, max is the maximum value of the stress data, and min is the minimum value of the stress data.

8. The method for inversion of bridge damage causes in mountainous high-fill areas based on multi-source fusion according to claim 7 is characterized by: The LDA method is used to reduce the dimensionality of the data. By maximizing the difference between classes and minimizing the directional projection of the difference within classes, samples of different classes can be effectively separated.

Images