A real-time monitoring inversion method and system for a concrete dam

Abstract

The application discloses a kind of concrete dam real-time monitoring inversion method and system, the method includes that mechanical parameter sample set is calculated by finite element model, the displacement response matrix corresponding to each sample is obtained, then with mechanical parameter sample set composition sample pair set, establish POD-RBF proxy model and objective function, obtain initial extreme point, with initial extreme point or new extreme point of last step as initial value, using Gauss-Newton method carries out local optimization, obtains new extreme point and substitutes into finite element model, whether the single displacement response vector obtained is less than the allowable error set, if not, the current new extreme point and corresponding single displacement response vector are added to sample pair set, update POD-RBF proxy model, repeat calculation until the single displacement response vector obtained and measured displacement response vector is less than the allowable error set, output current new extreme point as concrete dam mechanical parameter inversion value, effectively improve concrete dam monitoring inversion precision.

Description

TECHNICAL FIELD

[0001] The application relates to a concrete dam real-time monitoring inversion method and system and belongs to the technical field of water conservancy and hydropower engineering. BACKGROUND

[0002] Concrete is an important building material and is most widely used in engineering. Especially, under the complex environment, large-volume concrete structures such as concrete dams, cross-sea bridges, offshore concrete platforms and the like, the performance and state of the concrete are closely related to the mechanical parameters thereof, but in actual engineering, it is difficult for people to directly obtain the mechanical parameters of the large-volume concrete, especially the internal mechanical parameters, and therefore a suitable method for understanding the internal mechanical parameters of the large-volume concrete is very important.

[0003] With the development of computer technology and sensor monitoring technology, the monitoring of the displacement field and the temperature field in the concrete is more convenient, which provides a new idea for obtaining the internal mechanical parameters of the large-volume concrete.

[0004] Inverse analysis is an effective method for obtaining the actual mechanical parameters of the concrete from the field monitoring data. Under normal circumstances, the parameter inversion problem in engineering can be converted into a minimum optimization problem, an objective function is constructed according to the inversion purpose, and the minimum value thereof is solved by using a pseudo-inverse method, a gradient method or a meta-heuristic algorithm to obtain the best mechanical parameters. However, in the inversion method, a large amount of forward analysis calculation is required, which is very time-consuming for a large-scale model or a model with many parameters to be inverted, and the safe operation state of the concrete structure cannot be monitored in real time. SUMMARY

[0005] The application aims at overcoming the deficiencies in the background art and providing a concrete dam real-time monitoring inversion method and system to improve the monitoring inversion precision of the concrete dam.

[0006] To achieve the above-mentioned purpose, the technical scheme adopted by the application is as follows:

[0007] In the first aspect, the application provides a concrete dam real-time monitoring inversion method, which comprises the following steps:

[0008] sampling in the range of the mechanical parameters of the concrete dam to obtain a mechanical parameter sample set;

[0009] calculating the mechanical parameter sample set through a finite element model to obtain a displacement response matrix corresponding to each sample;

[0010] grouping the mechanical parameter sample set and the displacement response matrix to form a sample pair set and establishing a POD-RBF surrogate model;

[0011] constructing an objective function according to the POD-RBF surrogate model to obtain an initial extreme point;

[0012] using the initial extreme value point or the new extreme value point of the last step as the initial value, performing local optimization by using the Gauss-Newton method to obtain a new extreme value point;

[0013] substituting the new extreme value point into the finite element model to obtain a single displacement response vector;

[0014] determining whether the single displacement response vector and the measured displacement response vector are less than the set allowable error, if not, adding the current new extreme value point and the corresponding single displacement response vector to the sample pair set, updating the POD-RBF surrogate model; repeating the calculation until the obtained single displacement response vector and the measured displacement response vector are less than the set allowable error, and outputting the current new extreme value point as the mechanical parameter inversion value of the concrete dam.

[0015] In combination with the first aspect, further, the POD-RBF surrogate model comprises:

[0016] decomposing the displacement response matrix by using the proper orthogonal decomposition:

[0017] Φ = U·V,

[0018] wherein Φ is a POD basis vector matrix, U is an M-row and N-column matrix, and V is an eigenvalue matrix of the positive definite covariance matrix C;

[0019] retaining the first K-order eigenvalues (K < M) of V to obtain obtaining the reduced POD basis vector matrix is:

[0020]

[0021] obtaining the POD-RBF surrogate model according to the radial basis function interpolation:

[0022]

[0023] wherein s represents a single sample in the mechanical parameter sample set,

[0024]

[0025] wherein,

[0026]

[0027]

[0028] wherein r is a constant selected according to engineering experience.

[0029] Further, the positive definite covariance matrix C is defined as:

[0030] C=U T ·U.

[0031] Further, the relative error between the displacement response vector predicted according to the POD-RBF agent model and the measured displacement response vector is taken as a target function, and an initial extreme point is found by using a PSO algorithm.

[0032] Further, the mechanical parameters of the concrete dam include an elastic modulus, a Poisson's ratio and a density.

[0033] Further, the sampling method includes Latin hypercube sampling, orthogonal sampling or uniform sampling.

[0034] In a second aspect, the present application provides a real-time monitoring inversion system for a concrete dam, and the system comprises:

[0035] An acquisition module: sampling according to a range of mechanical parameters of the concrete dam to obtain a sample set of the mechanical parameters;

[0036] A calculation module: calculating the sample set of the mechanical parameters by using a finite element model to obtain a displacement response matrix corresponding to each sample, establishing a POD-RBF agent model and a target function, and finally obtaining a new extreme point and a corresponding single displacement response vector;

[0037] An analysis module: judging whether the single displacement response vector is smaller than a set allowable error from a measured displacement response vector, if yes, outputting the current new extreme point as an inversion value of the mechanical parameters of the concrete dam, and if no, adding the current new extreme point and the corresponding single displacement response vector to a sample pair set and updating the POD-RBF agent model;

[0038] An output module: outputting the inversion value of the mechanical parameters of the concrete dam.

[0039] In a third aspect, the present application further provides a computer device comprising a processor and a storage medium;

[0040] The storage medium is used for storing instructions;

[0041] The processor is used for operating according to the instructions to perform the steps of a real-time monitoring inversion method for a concrete dam.

[0042] Compared with the prior art, the present application has the following beneficial effects:

[0043] The monitoring inversion method of the present application establishes a POD-RBF agent model, which not only has high agent calculation accuracy, but also can resist noise in monitoring data, and is beneficial to improving the robustness of the agent model.

[0044] This invention provides an iterative update inversion framework combining particle swarm optimization and Gauss-Newton algorithm. It uses an initial extreme point or a new extreme point from the previous step as initial values, performs local optimization using the Gauss-Newton method to obtain new extreme points, substitutes these new extreme points into the finite element model to obtain a single displacement response vector, determines whether the single displacement response vector is less than the measured displacement response vector within a set allowable error, and outputs the inverted values ​​of the concrete dam's mechanical parameters or updates the POD-RBF surrogate model. Attached Figure Description

[0045] Figure 1 This is a flowchart of a real-time monitoring and inversion method for concrete dams provided in an embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram of the finite element model of the dam provided in an embodiment of the present invention;

[0047] Figure 3 This is a graph showing the change in relative error with the number of iterations and inversions provided in this embodiment of the invention. Detailed Implementation

[0048] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0049] Example 1

[0050] like Figure 1 The diagram shown is a flowchart of a real-time monitoring and inversion method for concrete dams provided by an embodiment of the present invention, which specifically includes the following steps:

[0051] Step 1: Input basic parameters.

[0052] Based on engineering experience, the approximate range β of the mechanical parameters of the dam to be inverted, the number N of the initial sample set, and the observed displacements of the actual measuring points on the dam are determined.

[0053] Step two, sampling.

[0054] Based on the range of mechanical parameters for each dam zone, a coarser Latin hypercube sampling (or orthogonal sampling, uniform sampling, etc.) is performed to obtain a mechanical parameter sample set S. S is a P-row N-column matrix, where P is the number of dam mechanical parameter zones, N is the number of samples, and each column of S represents a mechanical parameter sample.

[0055] Step 3: Finite element calculation.

[0056] A finite element model is established and the displacement response matrix U is calculated by calculating the displacement response of each sample in the mechanical parameter sample set S. U is an M by N matrix, M is the number of displacement measurement points (determined according to the actual project), and N is as described in step one. One column of U is a displacement response output, corresponding to S.

[0057] Step four, establish a proxy model.

[0058] According to the mechanical parameter sample set S and the displacement response matrix U, a POD-RBF proxy model is established, and the specific steps include:

[0059] (1) Use the proper orthogonal decomposition (POD) to decompose the displacement response matrix U:

[0060] Φ = U · V

[0061] Wherein, Φ is the POD basis vector matrix, V is the eigenvector matrix of the positive definite covariance matrix C, and C is defined as:

[0062] C = U T ·U

[0063] (2) Reserve the first K eigenvectors of V (K < M) to get Then the reduced POD basis vector matrix is obtained as:

[0064]

[0065] (3) According to the radial basis function (RBF) interpolation, the POD-RBF proxy model is obtained:

[0066]

[0067] Wherein, s represents a single sample in the mechanical parameter sample set, and the values of B and f(s) are:

[0068]

[0069]

[0070] Wherein,

[0071]

[0072]

[0073] Wherein, r is a constant selected according to engineering experience.

[0074] Step five, iterative update inversion.

[0075] For the measured displacement, the actual mechanical parameter value is obtained by using the iterative update inversion algorithm, and the specific steps include:

[0076] (1) Take the relative error of the displacement predicted by the agent model and the measured displacement as the objective function, and use the PSO algorithm to find the initial extreme point;

[0077] (2) Take the extreme point of the previous step (or the initial extreme point) as the initial value, and use the Gauss-Newton method for local optimization to obtain a new extreme point;

[0078] (3) Substitute the new extreme point into the finite element model to calculate the displacement response;

[0079] (4) Determine whether the difference between the displacement response calculated by the finite element method and the measured displacement is less than the pre-set allowable error, if yes, the calculation is completed, and the current latest extreme point is the inverse value of the mechanical parameters of the dam; otherwise, the current extreme point and the corresponding finite element calculation displacement are added to the sample pair set as a new sample pair, the agent model is reconstructed, and step (2) is returned.

[0080] As shown in Figure 2 , it is a dam finite element model schematic diagram provided by the embodiment of the application, the dam height is 80 meters, the dam width is 60 meters, and the dam section thickness is 25 meters. The dam is divided into three mechanical parameter partitions according to the pouring sequence of the concrete along the elevation, and the mechanical parameters in the embodiment only consider the elastic modulus.

[0081] The elastic modulus of each partition is 26-38 GPa, the Poisson's ratio is set as a constant 0.167, and the dam is subjected to the static water pressure of 80 meters in front of the dam and the self-weight action.

[0082] Step one, input basic parameters.

[0083] According to engineering experience, the approximate interval range of the dam mechanical parameters to be inverted is β=[26, 38] GPa, the number of initial sample sets is N=27, and the actual measured point displacement of the dam is

[0084] Step two, sampling.

[0085] According to the range of the elastic modulus of the dam partition, 27 samples are obtained by using Latin hypercube sampling, and the mechanical parameter sample set S is obtained.

[0086] Step three, finite element calculation.

[0087] The samples in the mechanical parameter sample set are respectively brought into the finite element model, and the corresponding displacement response is calculated to form the displacement response set U, and the sample pair set (S, U) is formed.

[0088] Step four, establish an agent model.

[0089] The POD-RBF agent model is established through (S, U).

[0090] Step five, iterative update inversion.

[0091] For the measured displacement (The actual monitoring displacement Substitute the finite element calculation value, and the corresponding elastic modulus is: * =[26.148, 30.506, 36.673] T GPa, the actual mechanical parameter value is obtained by using the iterative update inversion algorithm, and the specific steps include:

[0092] (1) The relative error between the displacement predicted by the surrogate model and the measured displacement is used as the objective function, and the PSO algorithm is used to find the initial extreme point s 0 ;

[0093] (2) The extreme point s i (or the initial extreme point s 0 ) is used as the initial value, and the Gauss-Newton method is used for local optimization to obtain a new extreme point s i+1 ;

[0094] (3) Substitute the new extreme point s i+1 into the finite element model to calculate the displacement response u i+1 ;

[0095] (4) Determine whether (The error ∈ in this example is taken as 1e-6) is true, if true, the calculation is finished, and the current latest extreme point s i+1 is the elastic modulus inversion value of the dam; otherwise, the new sample pair (s i+1 , u i+1 ) is added to the sample pair set (S, U), the POD-RBF surrogate model is reconstructed, and returns to (2), and the final inversion result is , which is basically consistent with the true elastic modulus.

[0096] As Figure 3 shown, the relative error changes with the number of iterations of the curve provided by the embodiment of the application, and the figure shows that the method proposed by the application can effectively improve the inversion accuracy of the mechanical parameters of the concrete dam. After the PSO algorithm finds an extreme point, the Gauss-Newton method is called and the sample point is updated to update the mechanical parameter sample set, and after two updates, the relative error is effectively reduced.

[0097] The concrete dam real-time monitoring inversion method provided by the application is based on a POD-RBF agent model, has high agent calculation accuracy, can resist noise in monitoring data, and is beneficial to improving the robustness of the agent model; and the iteration updating inversion framework combining the particle swarm algorithm and the Gauss-Newton algorithm can obtain higher inversion accuracy with as few positive analyses as possible.

[0098] Embodiment two

[0099] The application provides a concrete dam real-time monitoring inversion system, which comprises:

[0100] The acquisition module samples according to the concrete dam mechanical parameter range to obtain a mechanical parameter sample set.

[0101] The calculation module calculates the mechanical parameter sample set through a finite element model to obtain a displacement response matrix corresponding to each sample, establishes a POD-RBF agent model and an objective function, and finally obtains a new extreme point and a corresponding single displacement response vector.

[0102] The judgment module judges whether the single displacement response vector and the measured displacement response vector are less than a set allowable error, and if yes, outputs the current new extreme point as the concrete dam mechanical parameter inversion value; if not, adds the current new extreme point and the corresponding single displacement response vector to a sample pair set and updates the POD-RBF agent model.

[0103] The output module outputs the concrete dam mechanical parameter inversion value.

[0104] Embodiment three

[0105] The application further provides a computer device comprising a processor and a storage medium, wherein the storage medium is used for storing instructions, and the processor is used for operating according to the instructions to perform the steps of the concrete dam real-time monitoring inversion method.

[0106] A computer program is stored on a computer readable storage medium, and the program is executed by a processor to realize the steps of the concrete dam real-time monitoring inversion method.

[0107] Those skilled in the art should understand that the embodiments of the application can be provided as a method, a system or a computer program product. Therefore, the application can adopt a completely hardware embodiment, a completely software embodiment or an embodiment combining software and hardware aspects. Moreover, the application can adopt a computer program product in the form of one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.

[0108] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart Figure 1 one or more flows and / or blocks. Figure 1 one or more flows and / or blocks.

[0109] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart Figure 1 one or more flows and / or blocks. Figure 1 one or more flows and / or blocks.

[0110] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart Figure 1 one or more flows and / or blocks. Figure 1 one or more flows and / or blocks.

[0111] The above only is the preferred embodiment of the present application, it should be pointed out that, for those skilled in the technical field, without departing from the technical principles of the present application, can also make a number of improvements and variations, these improvements and variations should also be considered as the protection scope of the present application.

Claims

1. A real-time monitoring and inversion method for concrete dams, characterized in that, Includes the following steps: Based on the sampling of the mechanical parameters of concrete dams within a certain range, a sample set of mechanical parameters is obtained; The displacement response matrix of each sample is obtained by calculating the mechanical parameter sample set using the finite element model. A sample set is formed by combining the mechanical parameter sample set and the displacement response matrix to establish a POD-RBF surrogate model; the establishment of the POD-RBF surrogate model includes: The displacement response matrix is ​​decomposed using eigenorthogonal decomposition: ， in, Let POD be the basis vector matrix. for OK Column matrix The positive definite covariance matrix eigenvector matrix; reserve The former eigenvectors of order ( )get The reduced POD basis vector matrix is ​​obtained. for: ， The POD-RBF surrogate model is obtained based on radial basis function interpolation: ， In the formula, Represents a single sample in a set of mechanical parameter samples. , , in, ， ， in, These are constants selected based on engineering experience; The objective function is constructed based on the POD-RBF surrogate model, and the initial extreme points are obtained; Using the initial extreme point or the new extreme point from the previous step as the initial value, the Gauss-Newton method is used to perform local optimization to obtain the new extreme point; Substituting the new extreme point into the finite element model, we obtain a single displacement response vector; Determine whether the single displacement response vector and the measured displacement response vector are less than the set allowable error. If not, add the current new extreme point and the corresponding single displacement response vector to the sample pair set and update the POD-RBF surrogate model. Repeat the calculation until the obtained single displacement response vector and the measured displacement response vector are less than the set allowable error, and output the current new extreme point as the inversion value of the mechanical parameters of the concrete dam.

2. The real-time monitoring and inversion method for concrete dams according to claim 1, characterized in that, The positive definite covariance matrix Defined as: 。 3. The real-time monitoring and inversion method for concrete dams according to claim 1, characterized in that, The relative error between the displacement response vector predicted by the POD-RBF surrogate model and the measured displacement response vector is used as the objective function, and the PSO algorithm is used to find the initial extreme point.

4. The real-time monitoring and inversion method for concrete dams according to claim 1, characterized in that, The mechanical parameters of the concrete dam include elastic modulus, Poisson's ratio, and density.

5. The real-time monitoring and inversion method for concrete dams according to claim 1, characterized in that, The sampling methods include Latin hypercube sampling, orthogonal sampling, or uniform sampling.

6. A real-time monitoring and inversion system for concrete dams, characterized in that, The system includes: Acquisition module: Samples are taken from the range of mechanical parameters of concrete dams to obtain a mechanical parameter sample set; The calculation module calculates the displacement response matrix for each sample of mechanical parameters using a finite element model, establishes the POD-RBF surrogate model and objective function, and finally obtains the new extreme points and the corresponding individual displacement response vectors. The establishment of the POD-RBF surrogate model includes: The displacement response matrix is ​​decomposed using eigenorthogonal decomposition: ， in, Let POD be the basis vector matrix. for OK Column matrix The positive definite covariance matrix eigenvector matrix; reserve The former eigenvectors of order ( )get The reduced POD basis vector matrix is ​​obtained. for: ， The POD-RBF surrogate model is obtained based on radial basis function interpolation: ， In the formula, Represents a single sample in a set of mechanical parameter samples. , ， in, ， ， in, These are constants selected based on engineering experience; The analysis module determines whether the error between a single displacement response vector and the measured displacement response vector is less than the set allowable error. If so, it outputs the current new extreme point as the inverted value of the mechanical parameters of the concrete dam. If not, it adds the current new extreme point and the corresponding single displacement response vector to the sample pair set and updates the POD-RBF surrogate model. Output module: Outputs the inverted values ​​of mechanical parameters of concrete dams.

7. A computer device, characterized in that, Including processor and storage media; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1 to 5.

Images