A prestressed steel cylinder concrete deflection monitoring method under rockfall impact

Abstract

The application discloses a prestressed steel cylinder concrete deflection monitoring method under rockfall impact, and comprises the following steps: step 1, a buried PCCP three-dimensional finite element model is established in ABAQUS, and the strain and deflection change conditions of the buried PCCP under different rockfall parameters are calculated and analyzed through the finite element method; step 2, the strain measurement value is taken as sample data set, and the MSVR model is trained; the BO-MSVR model of the complex nonlinear relationship between the fiber measured PCCP structure strain and deflection is established; step 3, the data of the fiber monitoring arranged on the PCCP in the actual project are input into the BO-MSVR model, the conversion from the measured PCCP strain to the deflection deformation is realized, and then the safety state of the structure under the rockfall impact is evaluated. The application can quickly and accurately monitor the PCCP deflection deformation under the rockfall impact, and has certain practical significance.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of pipeline deflection monitoring method, and particularly relates to a prestressed concrete cylinder pipe deflection monitoring method under rockfall impact. BACKGROUND

[0002] Since the prestressed concrete cylinder pipe (PCCP) was invented, it has been adopted by many water diversion projects due to its strong impermeability, high reliability, good durability, excellent anti-seismic performance and low management cost. The PCCP buried in water conservancy projects has a long transmission distance and passes through various geological units, and is prone to rockfall impact, landslides, settlement and other geological disasters. Once the PCCP is damaged due to large deformation, it will seriously affect the normal operation of the pipeline and cause waste of water resources. At the same time, the high-pressure water jet from the pipeline is prone to cause secondary disasters, which seriously threatens the safety of life and property of the surrounding people. For the buried PCCP passing through mountainous areas, rockfall, as a kind of geological disaster prone to occur in unstable slopes, will exert a short but strong impact load on the soil, which may cause large deformation of the buried PCCP, resulting in cracking of the pipe body and even pipe explosion. Therefore, after a rockfall disaster occurs, it is particularly important to obtain PCCP structural deformation information from the monitoring system and evaluate the structural safety to take appropriate engineering measures in time to reduce the occurrence of engineering accidents.

[0003] The conversion model between the pipe body strain and the deflection is established by using the strain and deflection data calculated by the finite element method under different rockfall conditions through the BO-MSVR model. Then, the pipe body deflection deformation is monitored according to the pipe strain directly monitored by the DOFSS (Distributed Optical Fiber Strain Sensing) technology, which can solve the problem that in real engineering, the load along the long-distance buried PCCP is variable, and the boundary conditions are complex, making it difficult to obtain the relationship between the strain and the pipe deformation through simple stress analysis. Therefore, the efficiency and accuracy of the PCCP deflection deformation monitoring under rockfall impact need to be improved. SUMMARY

[0004] The purpose of the application is to provide a prestressed concrete cylinder pipe deflection monitoring method under rockfall impact, which solves the problem of low efficiency and accuracy of the PCCP deflection deformation monitoring under rockfall impact.

[0005] The technical scheme adopted by the application is as follows:

[0006] A prestressed concrete cylinder pipe deflection monitoring method under rockfall impact is performed according to the following steps:

[0007] Step 1: a three-dimensional finite element model of buried PCCP is established in ABAQUS, and the strain and deflection changes of buried PCCP under different rockfall parameters are calculated and analyzed by the finite element method, and the strain measurement values under different working conditions are obtained;

[0008] Step 2: the strain measurement values obtained in step 1 are taken as a sample data set, a part of the sample data set is selected as a test sample, and the rest is taken as a training sample to train the MSVR model; the value range of the hyperparameters in the MSVR model is set, the optimal value of the hyperparameters is determined in the possible value range of the hyperparameters by using the BO algorithm, thereby establishing a BO-MSVR model of the complex nonlinear relationship between the measured PCCP structure strain and deflection, and the BO-MSVR model takes the fiber strain monitoring data as input and takes the pipe deflection measurement values at each measuring point as output;

[0009] Step 3: the data of the optical fiber monitoring arranged on the PCCP in the actual project are input into the BO-MSVR model obtained in step 2, and then the conversion of the measured PCCP strain to the deflection deformation is realized, and the PCCP structure deflection deformation data are obtained, and the PCCP structure deflection deformation data are used to evaluate the safety state of the structure under the rockfall impact.

[0010] The application also has the characteristics of;

[0011] In step 1, the rockfall parameters include rockfall radius, rockfall height and rockfall position, the X and Y coordinates of the rockfall position, the rockfall radius and the rockfall height parameters are given several different values, and M groups of rockfall parameter combinations are obtained; for each group of rockfall parameters, the strain results of the PCCP finite element model under different rockfall parameter working conditions are calculated by using the buried PCCP three-dimensional finite element model, and the strain results are taken as the distributed fiber strain measurement results.

[0012] In step 2, in the MSVR model, for a series of data (x1, y1), (x2, y2), …, (x n ,y n ), there is a nonlinear relationship between the input x i ∈R r and the output y i ∈R r as follows formula (1);

[0013]

[0014] In the formula, is a nonlinear projection that projects the input to the feature space; W=[w 1 ,...,w k ] and b=[b 1 ,...,b k ]T are parameters of linear projection; the function f(x) is to be guaranteed to make the actual output value y i only a small deviation ε from the predicted value f(x) of the function.

[0015] In step 2, the unconstrained optimization problem corresponding to the MSVR model is defined as follows in formula (2):

[0016]

[0017] In the formula, ||w j || is the L2 norm of the vector w j ; C is a penalty factor; e = y - f(x);

[0018] In iterative solving, in order to derive the next solution (W t , b t ) from the solution (W t+1 , b t+1 ) obtained in the last step, the L P (W, b) can be expanded by a first-order Taylor series around (W t , b t ) as L P (W, b) ≈ L' P (W, b);

[0019]

[0020] Further, L P (W, b) ≈ L" P (W, b);

[0021]

[0022] In the formula, τ is a constant term independent of W or b; the parameter α i can be expressed as:

[0023]

[0024] When (W t , b t ) is known, the optimal solution of L P (W, b) can be transformed into the optimal solution of L" P (W, b), and according to the stationary point condition and , the following can be obtained:

[0025] 2w j - 2Φ T D α [y j-Φw j -1b j ] = 0 (6);

[0026] α T [y j -Φw j -1b j ] = 0 (7);

[0027] The arrangement can obtain:

[0028]

[0029] In the formula, D α is a diagonal matrix composed of parameters α i (i = 1, 2,..., n); Φ is a vector composed of nonlinear projections of input x i , α is a vector composed of parameters α i (i = 1, 2,..., n), α = [α1,..., α n ] T ; y j is n different samples of the jth output, y j = [y j1 ,..., y jn ] T .

[0030] In step 2, according to the Representer's theory, the machine learning problem can be expressed as a linear combination of training samples, that is:

[0031]

[0032] Substituting formula (9) into formula (8) can obtain:

[0033]

[0034] In the formula, K is a kernel function matrix;

[0035] Two hyperparameters of the MSVR model are respectively a penalty factor C and a parameter σ 2 of a radial basis kernel function, in the case of the two hyperparameters, an iterative weighted least square algorithm is used to train the MSVR model to obtain model parameters B = [β 1 ,..., β k ] and b = [b 1 ,..., b k ] T ;

[0036] By setting the value range of two hyperparameters, and using the BO algorithm to optimize the two hyperparameters, a BO-MSVR model for strain to deflection conversion is constructed.

[0037] The pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact has the advantages that the finite element calculation of the PCCP structure under different rockfall impact conditions is carried out by combining actual engineering modeling, and on this basis, the finite element calculation result is taken as a training sample to realize the conversion between the fiber measured strain and the pipe deflection of the PCCP under rockfall impact by the BO-MSVR model, so that the pipe body deflection deformation of the PCCP structure under different rockfall impacts is monitored, and the method has a good application prospect in actual engineering. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 is a flowchart of the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application;

[0039] Figure 2 is a schematic diagram of strain measurement fiber arrangement in the finite element calculation of the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application;

[0040] Figure 3 is a schematic diagram of pipe body deflection measurement point in the finite element calculation of the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application;

[0041] Figure 4 is a strain measurement value curve diagram of the measurement line ① at the top of the pipe in the typical working condition in the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application;

[0042] Figure 5 is a strain measurement value curve diagram of the measurement line ② at the bottom of the pipe in the typical working condition in the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application;

[0043] Figure 6 is a strain measurement value curve diagram of the measurement line ③ at the left side of the pipe in the typical working condition in the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application;

[0044] Figure 7 is a strain measurement value curve diagram of the measurement line ④ at the right side of the pipe in the typical working condition in the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application;

[0045] Figure 8 is a comparison diagram of the pipe deflection of the finite element numerical calculation and the BO-MSVR model calculation in the pre-stressed steel cylinder concrete deflection monitoring method under rockfall impact of the present application. DETAILED DESCRIPTION

[0046] The prestressed steel cylinder concrete deflection monitoring method under rockfall impact is specifically implemented according to the following steps.

[0047] As shown in the drawing, Figure 1 The prestressed steel cylinder concrete deflection monitoring method under rockfall impact is specifically implemented according to the following steps.

[0048] Step 1, in the finite element calculation software (ABAQUS), the finite element analysis of PCCP strain and deflection under different rockfall parameters is established, specifically:

[0049] In the finite element calculation software ABAQUS, a three-dimensional finite element model of buried PCCP is established, and the strain and deflection changes of buried PCCP under different rockfall parameters are calculated and analyzed by the finite element method;

[0050] Rockfall parameters are direct factors affecting rockfall impact, including rockfall radius, rockfall height and rockfall position. In the possible value range, the X and Y coordinates of the rockfall position, the rockfall radius and the rockfall height parameters are given several different values, so that M groups of rockfall parameter combinations can be obtained, that is, the strain and displacement of the pipe body structure at the distributed optical fiber arrangement position (which can be set according to actual needs) and the deflection value at the measurement point are calculated by the finite element method under different rockfall parameters. The strain measurement value is taken as a sample data set, and 5% of the sample data set is taken as a test sample, and the rest is taken as a training sample;

[0051] M=N1(rockfall X position value number)*N2(rockfall Y position value number)*N3(rockfall radius value number)*N4(rockfall height value number);

[0052] The establishment process of the three-dimensional finite element model of buried PCCP is as follows:

[0053] The soil around the PCCP pipe adopts the Mohr-Coulomb model, which describes the plastic strain of the soil with the friction angle, cohesion and dilatancy angle of the soil. The material parameters of backfill soil 1, backfill soil 2, backfill soil 3 and medium-coarse sand cushion are shown in Table 1. The undisturbed foundation soil layer adopts a linear elastic model, and the elastic modulus and Poisson's ratio are 4.5 MPa and 0.3, respectively.

[0054] Table 1 Mohr-Coulomb model parameters of overlying soil

[0055] Material Density (kg / m3 3 ) Friction angle (°) Cohesion (MPa) Dilatancy angle (°) Backfill 1 1770 30 0.004 25 Backfill 2 1700 30 0.008 25 Backfill 3 1600 30 0.003 25 Medium-coarse sand cushion 0 25 0 25

[0056] Considering the yield characteristics of steel wire and steel cylinder under load, and the nonlinear characteristics of materials such as concrete and mortar cracking, the constitutive relationship of the pipe body material is determined according to the American AWWA C304 standard in the rockfall impact simulation process. The stress-strain relationship of the prestressed steel wire is shown in formula (11).

[0057]

[0058] where σ is the stress of prestressed steel wire; ε s is the strain of prestressed steel wire; E s is the elastic modulus of prestressed steel wire; f su is the tensile strength of prestressed steel wire.

[0059] The basic parameters of the materials of each part of PCCP are shown in Table 2 in the finite element analysis of the structural response of PCCP under the action of basic load and rockfall impact.

[0060] Table 2 Basic parameters of materials

[0061]

[0062] The discretization of the finite element model mainly adopts three kinds of units: the concrete pipe core, the mortar protective layer and the soil model are mainly discretized by using the eight-node hexahedral linear reduced integration solid element (C3D8R); the steel cylinder and the steel ring of the insert socket are mainly discretized by using the four-node shell element (S4R); and the prestressed steel wire is mainly discretized by using the two-node bar element (T3D2). The details of the finite element grid are shown in Table 3.

[0063] Table 3 Unit parameters of model section

[0064]

[0065]

[0066] The prestress is applied by using the cooling method. The initial temperature is set in the predefined field by specifying the linear expansion coefficient of the prestressed steel wire, and then the temperature is lowered in the subsequent analysis step. The principle of thermal expansion and contraction is used to simulate the prestress of the steel wire, and the specific temperature lowering value is calculated by using formula (12):

[0067]

[0068] where Δt is the temperature lowering value of the steel wire; f sg is the prestress of the steel wire, which is 1177.5 MPa; and α is the linear expansion coefficient of the steel wire, which is 1×10 -5 ; E s is the elastic modulus of the steel wire, which is 205000 MPa.

[0069] Before simulating the rockfall impact, the whole process of the construction and operation of the PCCP pipeline is simulated, which is specifically divided into four steps of prestress application of steel wire, application of self-weight of pipeline, backfilling of overlying soil and application of internal water pressure of 0.60 MPa. The load application sequence is shown in Table 4.

[0070] Table 4 Load application sequence

[0071]

[0072] Step 2, build the BO-MSVR model of strain-deflection conversion, specifically:

[0073] Take the optical fiber strain monitoring data as input, and take the measured deflection value of each measuring point as output, train the MSVR model according to the training samples obtained in step 1. Set the value range of the hyperparameters (penalty factor C and parameter of radial basis kernel function) of the MSVR model, and determine the optimal value of the MSVR hyperparameters in the possible value range of the hyperparameters by using the BO algorithm, thereby establishing the BO-MSVR model of the complex nonlinear relationship between the measured PCCP structure strain and deflection of the optical fiber;

[0074] Step 3, after the actual engineering rockfall disaster occurs, the deflection deformation of the PCCP is predicted according to the optical fiber measured strain and the BO-MSVR model trained above. The safety state of the structure under the impact of rockfall is evaluated according to the measured strain and the converted PCCP structure deflection deformation;

[0075] The prestressed steel cylinder concrete deflection monitoring method under the impact of rockfall will be further described in detail through specific examples.

[0076] The prestressed steel cylinder concrete deflection monitoring method under the impact of rockfall is specifically implemented according to the following steps:

[0077] Step 1, establish a finite element analysis of PCCP structure damage under different rockfall parameters, specifically:

[0078] The rockfall parameters are as follows: the X coordinate is spaced by 2.0 m, and changes from -4.0 m to 4.0 m; the Y coordinate is spaced by 1.5 m, and changes from 0.0 m to 6.0 m; the rockfall radius r is spaced by 0.2 m, and changes from 0.8 m to 2.0 m; the rockfall height h is spaced by 4.0 m, and changes from 4.0 m to 20.0 m. In this way, there are M=5*5*7*5=875 different working conditions of rockfall parameters. According to the arrangement form of the axial strain measuring optical fiber shown in the figure, the axial (x-direction) strain calculation results of all nodes (the spacing between adjacent nodes is 0.3 m, and the spatial resolution of the optical fiber strain measurement is 0.3 m) on the four measuring lines ①-④ on the top, bottom and left and right sides of the PCCP finite element model along the pipe section are recorded as the distributed optical fiber strain measurement results. Figure 2 The strain measurement value curve diagram of the four strain monitoring optical fibers on the top, bottom, left and right of the pipe under the typical working condition. Figures 4-7 The strain measurement value curve diagram of the four strain monitoring optical fibers on the top, bottom, left and right of the pipe under the typical working condition. The strain measurement value curve diagram of the four strain monitoring optical fibers on the top, bottom, left and right of the pipe under the typical working condition.

[0079] Step 2, taking the fiber strain monitoring data as input and the pipe deflection measurement at each measuring point as output. The strain measurements obtained by finite element calculation under different working conditions in step 1 are taken as sample data sets: after shuffling M = 875 rockfall working conditions, 44 groups of data are selected as the test set to evaluate the accuracy of the model, and the remaining 831 groups of data are used as the training set to train the BO-MSVR conversion model. The value range of the hyperparameters (penalty factor C and radial basis kernel function parameter) of the MSVR model is set, and the optimal value of the MSVR hyperparameters is determined within the possible value range of the hyperparameters by using the BO algorithm, thereby establishing the BO-MSVR model of the complex nonlinear relationship between the measured PCCP structure strain and the deflection. The MAE of the test set is 0.0071 mm, and the RMSE is 0.1013 mm. The pipe body deflection curve under typical working conditions is shown in Figure 8 , and the negative value in the figure is the downward deflection deformation. The deflection comparison results of numerical calculation and MSVR model calculation under typical working conditions are shown in Table 5. Through Figure 8 and Table 5, it can be seen that under 6 working conditions, the maximum absolute value of the pipe body deflection absolute error calculated by the MSVR evaluation model is 0.29 mm, and the maximum relative error absolute value is 7.83%.

[0080] Table 5 Comparison of pipe deflection between numerical calculation and BO-MSVR model calculation

[0081]

[0082]

[0083] In actual engineering, the spatial resolution of distributed optical fiber strain measurement may not reach 0.3 m. Therefore, the upper part of the 4 straight lines ①-④ is selected as the strain result output position to simulate the spatial resolution of strain monitoring at 0.3 m, 0.6 m, 1.2 m, 1.8 m and 2.4 m. The deflection calculation accuracy obtained by the strain monitoring data under these different spatial resolutions is shown in Table 6. As shown in Table 6, the greater the spatial resolution, the higher the accuracy of the BO-MSVR conversion model, but even in the case of low spatial resolution, the conversion accuracy can still meet the engineering application requirements.

[0084] Table 6 Accuracy of BO-MSVR deflection conversion model under different spatial resolutions

[0085] Strain measurement spatial resolution (m) RMSE (mm) MAE (mm) 0.3 0.1013 0.0077 0.6 0.1483 0.0103 1.2 0.1680 0.0118 1.8 0.1805 0.0125 2.4 0.2082 0.0144

[0086] The numerical calculation simulation fiber strain monitoring data does not consider the influence of observation noise, and the observation noise is inevitable in actual engineering, which may affect the accuracy of the BO-MSVR conversion model. In order to study the influence of observation noise on the accuracy of the conversion model, Gaussian white noise with signal-to-noise ratio of 20 dB, 30 dB, 40 dB, 50 dB and 60 dB is added to the simulated fiber strain monitoring data for analysis, and the conversion accuracy of the pipeline deflection obtained is shown in Table 7. As shown in Table 7, when the observation noise intensity increases within a certain range, the accuracy of the BO-MSVR evaluation model decreases, but it is still acceptable, which shows that the strain-deflection conversion model based on BO-MSVR has strong robustness and is suitable for practical engineering application scenarios.

[0087] Table 7 Accuracy of BO-MSVR deflection conversion model under strain observation data signal-to-noise ratio

[0088] Strain observation data signal-to-noise ratio (dB) RMSE (mm) MAE (mm) Noiseless 0.1013 0.0071 60 0.1065 0.0074 50 0.1096 0.0077 40 0.1156 0.0082 30 0.1292 0.0092 20 0.1700 0.0119

[0089] Step 3, after the rockfall disaster occurs in actual engineering, the PCCP deflection deformation is predicted according to the fiber actual measured strain and the BO-MSVR model trained above. The safety state of the structure under rockfall impact is evaluated according to the measured strain and the PCCP structure deflection deformation obtained by conversion.

[0090] The prestressed steel cylinder concrete deflection monitoring method under rockfall impact can monitor the PCCP deflection deformation under rockfall impact relatively quickly and accurately to a certain extent, and has certain practical significance.

Claims

1. A method for monitoring deflection of a pre-stressed steel cylinder concrete under rockfall impact, characterized in that, The following steps are specifically followed: Step 1: A three-dimensional finite element model of buried PCCP is established in ABAQUS, and the strain and deflection changes of buried PCCP under different rockfall parameters are calculated and analyzed by finite element method, and the strain measurement values under different working conditions are obtained; Step 2: The optical fiber strain monitoring data is taken as the input, and the pipeline deflection measurement values at each measuring point are taken as the output; the strain measurement values obtained in step 1 are taken as the sample data set, a part of the sample data set is selected as the test sample, and the rest is taken as the training sample to train the MSVR model; The value range of the hyperparameters in the MSVR model is set, and the optimal value of the hyperparameters is determined in the possible value range of the hyperparameters by using the BO algorithm, thereby establishing the BO-MSVR model of the complex nonlinear relationship between the measured PCCP structure strain and the deflection, and the BO-MSVR model takes the optical fiber strain monitoring data as the input and takes the pipeline deflection measurement values at each measuring point as the output; Step 3: The data of the optical fiber monitoring arranged on the PCCP in the actual project is input into the BO-MSVR model obtained in step 2, and then the conversion of the measured PCCP strain to the deflection deformation is realized, and the PCCP structure deflection deformation data is obtained, and the PCCP structure deflection deformation data is used to evaluate the safety state of the structure under the action of rockfall impact.

2. The method of claim 1, wherein the method is characterized by, In step 1, the rockfall parameters include rockfall radius, rockfall height and rockfall position, the X and Y coordinates of the rockfall position, the rockfall radius and the rockfall height parameters are given several different values, and M groups of rockfall parameter combinations are obtained; for each group of rockfall parameters, the strain results of the PCCP finite element model under different rockfall parameter conditions are calculated by using the buried PCCP three-dimensional finite element model, and the strain results are taken as the distributed optical fiber strain measurement results.

3. The method of claim 1, wherein the method is characterized by: In step 2, in the MSVR model, for a series of data (x1, y1), (x2, y2), …, (x n ,y n ), where there is a nonlinear relationship between the input x i ∈R r of the system and the output y i ∈R r of the system as follows formula (1); wherein is a nonlinear projection that projects the input into the feature space; W = [w 1 ,...,w k ] and b = [b 1 ,...,b k ] T are parameters of the linear projection; the function f(x) is to be guaranteed such that the actual output value y i and the predicted value f(x) of this function only have a small deviation ε from each other.

4. The method of claim 3, wherein the method is characterized by, In step 2, the unconstrained optimization problem corresponding to the MSVR model is defined as formula (2): where ||w j || is the L2 norm of the vector w j ; C is a penalty factor; In the iterative solution, in order to deduce the next step solution (W t ,b t ) according to the solution (W t+1 ,b t+1 ) obtained in the last step, the first order Taylor series expansion of L P (W,b) around (W t ,b t ) can be used, that is, L P (W,b)≈L' P (W,b). Further, L P (W,b)≈L" P (W,b); where τ is a constant term independent of W or b; the parameter a i may be expressed as: When (W t , b t ) is known, the optimal solution of L P (W,b) can be transformed into finding the optimal solution of L" P (W,b), according to the stationary point condition and can be obtained: 2w j -2Φ T D α [y j -Φw j -1b j ] = 0 (6); a T [y j -Φw j -1b j ]=0 (7) After sorting, we get: where D α is a diagonal matrix composed of parameters α i (i = 1, 2,..., n); Φ is a vector composed of nonlinear projections of input x i , α is a vector composed of parameters α i (i = 1, 2,..., n), α = [α1,...,α n ] T ; y j is n different samples of the jth output, y j = [y j1 ,...,y jn ] T .

5. The method of claim 4, wherein the method is characterized by: In step 2, according to the Representer's theory, the machine learning problem can be expressed as a linear combination of training samples, that is: Substituting formula (9) into formula (8) can obtain: In the formula, K is the kernel function matrix; The two hyperparameters of the MSVR model are penalty factor C and parameter σ of the radial basis kernel function 2 With the two hyperparameters, the MSVR model is trained by using an iterative weighted least square algorithm to obtain model parameters B = [β 1 ,..., β k ] and b = [b 1 ,..., b k ] T ; By setting the value range of the two hyperparameters and optimizing the two hyperparameters by using the BO algorithm, the BO-MSVR model for converting strain to deflection is constructed.

Images