An intelligent prediction method for the mechanical behavior during the construction process of buried pipelines

Abstract

The present invention provides an intelligent prediction method for the mechanical behavior during the construction process of buried pipelines, belonging to the technical field of water conservancy and hydropower engineering. This method defines the geometric parameters, element types, and material parameters of a refined numerical simulation model, establishes a three-dimensional finite element model, creates the contact relationships between the steel pipe and the backfill soil, and between the backfill soil and the original soil, and conducts multi-stage load step simulations to simulate the actual construction process. The orthogonal test method is used to obtain test data, and combined with the Python platform, a support vector regression and genetic algorithm are used to construct an intelligent prediction model, so as to accurately predict the mechanical behavior during the construction process. By adopting the above intelligent prediction method for the mechanical behavior during the construction process of buried pipelines, the present invention can accurately simulate the construction environment, optimize the construction plan, improve the construction quality, reduce the construction risk, and has significant technical advantages and application values.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy and hydropower engineering, and particularly relates to an intelligent prediction method for the mechanical behavior during the construction process of buried pipelines. Background Art

[0002] Pipelines are the fifth major means of transportation in addition to roads, railways, waterways, and aviation, and play an important role in transporting water, oil, natural gas, etc.

[0003] Buried steel pipes are commonly used pressure pipes in water resource allocation projects. During construction, the trench is first excavated, and then the pipeline is laid, backfilled, and the soil and stone materials are compacted. It has the advantages of simple structure, fast construction, convenient maintenance, and environmental friendliness, and is especially suitable for pipeline projects with large diameters, high internal pressures, and complex external environments. The longer the pipeline is built, the more prominent the comprehensive benefits will be. With the increasing scale of water resource allocation projects, the HD value (the product of water head and pipe diameter) of the pipeline will be very high, and it is more dependent on large buried steel pipes (pipe diameter not less than 1.2m). Compared with industries such as oil and gas, water supply and drainage, heating, and gas, the water conservancy and hydropower industry often uses large buried steel pipes (the maximum diameter can reach 4.0m), while the maximum diameter in the oil and gas industry is only 1.4m. When the pipe diameter becomes larger, the complexity of manufacturing, construction, and deformation control will increase exponentially. Currently, using larger diameter pipes is the common development direction of all industries. The construction process of large diameter pipes is very delicate, and usually supports are set inside the pipes to prevent pipeline deformation during construction.

[0004] However, the current numerical simulation methods for buried pipelines ignore the setting of internal supports in the pipelines and the layered filling of backfill soil, which has a large error compared with the actual situation. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent prediction method for the mechanical behavior during the construction process of buried pipelines, which can accurately simulate the construction environment, optimize the construction plan, improve the construction quality, reduce the construction risk, and has significant technical advantages and application value.

[0006] To achieve the above purpose, the present invention provides an intelligent prediction method for the mechanical behavior during the construction process of buried pipelines, including the following steps:

[0007] Step S1, define the geometric parameters, element types, and material parameters of the refined numerical simulation model for the construction process of buried pipelines;

[0008] Step S2, establish a three-dimensional finite element model of the pipeline, cushion, backfill soil, undisturbed soil, and internal support;

[0009] Step S3, create the contact relationships between the steel pipe and the backfill soil, and between the backfill soil and the undisturbed soil, and set the contact parameters;

[0010] Step S4: Conduct multi-stage load step simulation on the refined numerical simulation model of the buried pipeline construction process, including trench excavation, cushion installation, pipeline and internal support installation, layer-by-layer soil backfilling, and removal of internal supports;

[0011] Step S5: Design an experimental scheme based on the orthogonal test method to obtain experimental data;

[0012] Step S6: Relying on the Python platform, use support vector regression and genetic algorithms to train and test the experimental data, and construct an intelligent prediction model for mechanical behavior;

[0013] Step S7: Use the intelligent prediction model for mechanical behavior to predict the mechanical behavior during the buried pipeline construction process.

[0014] Preferably, in step S1, the geometric parameters include pipeline diameter, pipeline burial depth, central angle of soil arc foundation, cushion thickness, trench width, trench sidewall inclination angle, length and width of the entire model.

[0015] Preferably, in step S1, the element types include plane elements, solid elements, shell elements, and rod elements.

[0016] Preferably, in step S1, the material parameters include:

[0017] Elastic modulus, Poisson's ratio, density, linear expansion coefficient, and thickness of the steel pipe;

[0018] Elastic modulus, Poisson's ratio, density, cohesion, internal friction angle, swelling angle, and linear expansion coefficient of undisturbed soil, backfill soil, and cushion;

[0019] Section type, cross-sectional area, elastic modulus, Poisson's ratio, density, and linear expansion coefficient of the cross-shaped support.

[0020] Preferably, in step S2, establishing the three-dimensional finite element model includes:

[0021] Create key points of the model according to the coordinates of each position, connect the points into lines, establish a linear profile, and then enclose the lines into a surface to create a two-dimensional plane model of the buried pipeline;

[0022] Perform horizontal and vertical slicing on the plane, divide the two-dimensional plane grid of the buried pipeline, and drag the surface grid to form a three-dimensional finite element model of the buried pipeline;

[0023] Create key points of the cross-shaped support along the pipe circumference, connect them in pairs into lines, divide the cross-shaped support grid, and construct a grid model of the cross-shaped support.

[0024] Preferably, in step S3, creating the contact relationship includes:

[0025] Contact pairs between the steel pipe and the backfill soil, and between the backfill soil and the original soil are established respectively, and parameters such as contact type, friction coefficient, penalty stiffness, and initial gap are set;

[0026] The ANSYS surface-to-surface contact method is adopted, and the surface-to-surface contact elements Target170-Contact174 are used for simulation. By establishing contact pairs on the unit surface and using the method of penetrating the surrounding nodes, the contact stress is calculated.

[0027] Preferably, in step S4, the multi-stage load step simulation includes:

[0028] Load step 1: Kill the elements of the soil in the trench, the steel pipe, and the cross-shaped supports through the EKILL command, and apply full constraints to the nodes of the killed elements. Next, apply the gravity load and solve, and release the constraints of the dead elements after solving;

[0029] Load step 2: Activate the cushion elements using the EALIVE command, apply full constraints to the points of the killed elements, reactivate the axial constraints of the cushion elements, and solve. Release the constraints of the dead elements after solving;

[0030] Load step 3: Activate the pipeline and the cross-shaped supports using the EALIVE command, apply full constraints to the points of the killed elements, reactivate the axial constraints of the pipeline elements, and solve. Release the constraints of the dead elements after solving;

[0031] Load steps 4-9: Activate the first layer of backfill soil using the EALIVE command, apply full constraints to the points of the killed elements, reactivate the axial constraints of the first layer of backfill soil, and solve. Release the constraints of the dead elements after solving; According to the above method, use the loop command to activate the second, third, fourth, fifth, sixth layers of backfill soil in turn and solve layer by layer;

[0032] Load steps 10-15: According to the method of load steps 4-9, use the loop command to activate the seventh, eighth, ninth, tenth, eleventh, twelfth layers of backfill soil in turn and solve layer by layer;

[0033] Load step 16: Activate the 13th layer of backfill soil at the top of the pipe - 0.3m above the top of the pipe, apply full constraints to the points of the killed elements, reactivate the axial constraints of the first layer of backfill soil, and solve. Release the constraints of the dead elements after solving;

[0034] Load step 17: Activate the 14th layer of backfill soil at 0.3m - 0.6m above the top of the pipe, apply full constraints to the points of the killed elements, reactivate the axial constraints of the soil layer, and solve. Release the constraints of the dead elements after solving;

[0035] Load step 18: Activate the 15th layer of backfill soil at 0.6m above the top of the pipe - the ground surface, apply full constraints to the points of the killed elements, reactivate the axial constraints of the soil layer, and solve. Release the constraints of the dead elements after solving;

[0036] Load step 19: Kill the elements of the cross-shaped bracing.

[0037] Preferably, in step S6, constructing the intelligent prediction model of mechanical behavior includes:

[0038] Establish a GA-SVR model, and use the GA algorithm to search for the penalty factor and the width parameter in the SVR model, and then import the optimal solution , into the SVR model to construct a GA-SVR model;

[0039] Establish an OED-GA-SVR coupling algorithm model. Using the experimental scheme obtained by the orthogonal experiment as the main control points in the prediction space, linearly interpolate the distances between the main control points, supplement multiple groups of experimental schemes, list the supplemented experimental schemes as the auxiliary control points in the prediction space, and respectively sort out the independent variables and response quantities under each scheme as the initial input data of the model;

[0040] After normalizing multiple groups of sample data, use them to train and test the proposed GA-SVR model.

[0041] Therefore, the present invention adopts the above-mentioned intelligent prediction method for the mechanical behavior during the construction process of buried pipelines, and the beneficial technical effects are as follows:

[0042] (1) Accurately simulate complex construction environments:

[0043] By establishing a three-dimensional finite element model including pipelines, cushions, backfill soil, undisturbed soil and internal bracing, the present invention can accurately simulate the complex environments and conditions during the construction process of buried pipelines. This refined modeling method can capture the key mechanical behaviors during the construction process, such as the interaction between the pipeline and the soil mass, the influence of the internal bracing, etc., so as to provide more accurate guidance for the construction.

[0044] (2) Intelligently predict construction mechanical behaviors:

[0045] Using the numerical simulation scheme designed by the orthogonal experiment method and combining the intelligent prediction model constructed by support vector regression (SVR) and genetic algorithm (GA), the present invention can accurately predict the mechanical behaviors during the construction process of buried pipelines. This prediction ability helps to identify potential construction problems in advance, such as pipeline deformation, soil displacement, etc., so as to take preventive measures and reduce construction risks.

[0046] (3) Optimize construction schemes and improve construction quality:

[0047] By using an intelligent prediction model to evaluate the mechanical behaviors under different construction schemes, the present invention can help engineers select the optimal construction scheme, such as determining the appropriate layout of internal supports, the thickness of backfill soil layers, etc. This optimized construction scheme can improve the construction quality, reduce the uncertainties and potential problems during the construction process, thereby enhancing the reliability and durability of the entire project. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of an intelligent prediction method for the mechanical behaviors during the construction process of an underground pipeline according to the present invention;

[0049] Figure 2 It is a three-dimensional model of an underground pipeline; wherein, Figure 2 (a) in it is the overall model; Figure 2 (b) in it is an enlarged view of the trench part;

[0050] Figure 3 It is the simulation process of the construction process of an underground pipeline; wherein, Figure 3 (a) in it is Step 1: Trench excavation; Figure 3 (b) in it is Step 2: Installing the cushion layer; Figure 3 (c) in it is Step 3: Installing the pipeline and internal supports; Figure 3 (d) in it is Step 4: Backfilling the trench with the first layer of soil; Figure 3 (e) in it is Step 5: Backfilling the trench with the second layer of soil; Figure 3 (f) in it is Step 6: Backfilling the trench with the third layer of soil; Figure 3 (g) in it is Step 7: Backfilling the trench with the fourth layer of soil; Figure 3 (h) in it is Step 8: Backfilling the trench with the fifth layer of soil; Figure 3 (i) in it is Step 9: Backfilling the trench with the sixth layer of soil; Figure 3 (j) in it is Step 10: Backfilling the trench with the seventh layer of soil; Figure 3 (k) in it is Step 11: Backfilling the trench with the eighth layer of soil; Figure 3 (l) in it is Step 12: Backfilling the trench with the ninth layer of soil; Figure 3 (m) in it is Step 13: Backfilling the trench with the tenth layer of soil; Figure 3 (n) in it is Step 14: Backfilling the trench with the eleventh layer of soil; Figure 3 (o) in it is Step 15: Backfilling the trench with the twelfth layer of soil; Figure 3 (p) in it is Step 16: Backfilling the trench with the thirteenth layer of soil; Figure 3 (q) in it is Step 17: Backfilling the trench with the fourteenth layer of soil; Figure 3 (r) in it is Step 18: Backfilling the trench with the fifteenth layer of soil; Figure 3 (s) in it is Step 19: Removing the internal supports;

[0051] Figure 4 It is the intelligent prediction flow chart of the pipeline mechanical behavior during the construction process;

[0052] Figure 5 It is the calculation process of the improved SVR model by the GA algorithm. Specific implementation manner

[0053] The technical solution of the present invention will be further described below with reference to the drawings and embodiments.

[0054] Unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those with ordinary skills in the field to which the present invention belongs.

[0055] Embodiment 1

[0056] As Figure 1 shown, it is the flow chart of an intelligent prediction method for the mechanical behavior during the construction process of an underground pipeline of the present invention, specifically including the following steps:

[0057] Step 1: Define the geometric parameters, element types and material parameters of the model.

[0058] Define the refined numerical simulation model geometric parameters of the underground pipeline construction process: Define the pipeline diameter, pipeline burial depth, central angle of the soil arc foundation, cushion thickness, define the trench width and trench sidewall inclination angle, and define the length and width of the entire model.

[0059] Define element types: plane element PLANE42 (used when creating a two-dimensional grid), solid element SOLID185 (used for undisturbed soil, backfill soil and sand cushion), shell element SHELL181 (used for steel pipes), and rod element LINK180.

[0060] Define material parameters: Set the material parameters of the steel pipe, including elastic modulus, Poisson's ratio, density, linear expansion coefficient, thickness; respectively set the material parameters of undisturbed soil, backfill soil and cushion, including elastic modulus, Poisson's ratio, density, cohesion, internal friction angle, dilation angle, linear expansion coefficient; set the material parameters of the "cross" - shaped support: section type, section area, elastic modulus, Poisson's ratio, density, linear expansion coefficient.

[0061] Step 2: Establish a three - dimensional finite element model of the pipeline, cushion, backfill soil, undisturbed soil and "cross" - shaped support ( Figure 2 ).

[0062] First, according to the coordinates of each position, create the key points of the model; connect the points into lines to establish the linear profile; and then enclose the lines into surfaces to create a two - dimensional plane model of the underground pipeline.

[0063] Next, perform horizontal and vertical slicing on the plane to divide the two-dimensional plane grid of the buried pipeline, and drag the surface grid to form a three-dimensional finite element model of the buried pipeline.

[0064] Finally, create 8 key points with a "rice" - shaped support along the pipe circumference, connect them in pairs to form a line, construct a single "rice" - shaped support, and then use the complex offset command to set a "rice" - shaped support every 2 meters in the three - dimensional finite element model of the buried pipeline and divide the grid.

[0065] Step 3: Create the contact relationships between the steel pipe - backfill soil and the backfill soil - undisturbed soil.

[0066] Establish the relationships between the steel pipe - backfill soil and the backfill soil - undisturbed soil respectively, and then use the ANSYS surface - to - surface contact method to establish the contact pairs between them, and set parameters such as contact type, friction coefficient, penalty stiffness, and initial gap to generate the target surface and the contact surface.

[0067] In the buried pipeline model, accurately simulating the pipe - soil interaction is crucial. Essentially, this interaction involves the contact problem at the pipe - soil interface, accompanied by complex factors such as material, geometry, and contact nonlinearity. Reasonably reproducing the pipe - soil contact behavior is the key to studying the stress and deformation mechanisms of buried pipelines. Therefore, the surface - to - surface contact element (Target170 - Contact174) is used for simulation during modeling. The surface - to - surface contact element calculates the contact stress by establishing contact pairs on the element surface and using the method of penetrating the surrounding nodes to reduce local stress concentration and uneven distribution of contact pressure. The interaction between the contact surfaces includes normal force and tangential force. Among them, the relative sliding state can be determined by the Coulomb friction model, that is:

[0068] (1);

[0069] In the formula, is the equivalent shear stress; is the maximum allowable shear stress; is the friction coefficient; is the normal pressure on the contact surface; is the cohesion.

[0070] Step 4: Settings before solving.

[0071] Set parameters for the model solution, turn on large deformation, use the Newton - Raphson algorithm, turn on automatic time stepping, and apply normal displacement constraints on the top surface of the model freely, and at the bottom, left - right, and front - back positions.

[0072] Step 5: Perform multi - stage load step simulation on the refined numerical simulation model of the buried pipeline construction process, including trench excavation, cushion installation, pipeline and internal support installation, layer - by - layer soil backfilling, and removal of internal supports.

[0073] Load step 1: Trench excavation ( Figure 3 as shown in (a) of

[0074] Use the EKILL command to kill the elements of the soil, steel pipe, and "cross-shaped" supports in the trench, and apply full constraints to the nodes of the killed elements. Next, apply the gravity load and solve, and release the constraints of the dead elements after the solution is completed.

[0075] Load step 2: Bedding installation ( Figure 3 as shown in (b) of

[0076] Use the EALIVE command to activate the bedding elements, apply full constraints to the killed element nodes, reactivate the axial constraints of the bedding elements, solve, and release the constraints of the dead elements after the solution is completed.

[0077] Load step 3: Installation of steel pipe and "cross-shaped" supports ( Figure 3 as shown in (c) of

[0078] Use the EALIVE command to activate the pipe and "cross-shaped" supports, apply full constraints to the killed element nodes, reactivate the axial constraints of the pipe elements, solve, and release the constraints of the dead elements after the solution is completed.

[0079] Load steps 4 - 9: Layer-by-layer backfilling of soil in the interval from the top of the bedding to the waist of the pipe ( Figure 3 as shown in (d) of Figure 3 to (i) of

[0080] Use the EALIVE command to activate the first layer of backfill soil, apply full constraints to the killed element nodes, reactivate the axial constraints of the first layer of backfill soil, solve, and release the constraints of the dead elements after the solution is completed.

[0081] Load steps 10 - 15: Layer-by-layer backfilling of soil in the interval from the waist of the pipe to the top of the pipe ( Figure 3 as shown in (g) of Figure 3 to (o) of

[0082] According to the method of load steps 4 - 9, use the loop command to sequentially activate the 7th, 8th, 9th, 10th, 11th, and 12th layers of backfill soil and solve layer by layer.

[0083] Load step 16: Laying of backfill soil from the top of the pipe to 0.3 m above the top of the pipe ( Figure 3 as shown in (p) of

[0084] Activate the 13th layer of backfill soil at the position from the top of the pipe to 0.3 m above the top of the pipe, apply full constraints to the killed element nodes, reactivate the axial constraints of the first layer of backfill soil, solve, and release the constraints of the dead elements after the solution is completed.

[0085] Load step 17: Laying of backfill soil from 0.3 m above the top of the pipe to 0.6 m above the top of the pipeFigure 3 in (q)).

[0086] Activate the 14th layer of backfill soil at the position from 0.3 m above the pipe crown to 0.6 m above the pipe crown, apply full constraints to the killed element points, reactivate the axial constraints of the soil layer, solve, and release the constraints of the dead elements after solving.

[0087] Load step 18: Laying of backfill soil from 0.6 m above the pipe crown to the ground ( Figure 3 in (r)).

[0088] Activate the 15th layer of backfill soil at the position from 0.6 m above the pipe crown to the ground, apply full constraints to the killed element points, reactivate the axial constraints of the soil layer, solve, and release the constraints of the dead elements after solving.

[0089] Load step 19: Kill the elements of the cross-shaped support ( Figure 3 in (s)).

[0090] Step 6: To establish an intelligent prediction model for the mechanical behavior of pipelines during construction, based on the above simulation of the buried pipeline construction process, use the orthogonal experimental design (OED) method to design an experimental scheme that comprehensively considers multiple factors such as pipelines, soil, backfill layers, trench shapes, and burial depths to obtain a large amount of experimental data. Based on the experimental data, relying on the Python platform, use support vector regression (SVR) and genetic algorithms (GA) to train and test a large number of numerical simulation results, verify and correct the prediction model, and the analysis process is shown in Figure 4 .

[0091] Establish a GA-SVR model.

[0092] SVR is a general learning method based on the VC dimension theory and the structural risk minimization theory in statistics, with characteristics such as global optimality and small samples, and has unique advantages in analyzing non-linear mapping relationships. Suppose there are groups of training samples, and the training set , where is the input parameter, is the output parameter, is the parameter number, is the set of real numbers. Then the linear regression function on the high-dimensional space has the expression:

[0093] (2);

[0094] In the formula, is the mapping function reflecting the high-dimensional characteristics of the data set, is the weight vector, is the bias value.

[0095] The GA algorithm can achieve implicit parallel computing and global search capabilities. It uses the rule of "survival of the fittest" in biological theory to optimize nonlinear complex problems. First, genes are formed by encoding chromosomes, and then parameter optimization is achieved through gene expression methods such as selection, crossover, and mutation. Let and be two individuals in the population, and they are encoded with and bit binary respectively to obtain genes and , which can form the genome G.

[0096] The SVR model is sensitive to the penalty factor and the width parameter , but it relies on experience or trial algorithms to determine, with strong instability. While the GA algorithm has strong global search capabilities and can adaptively search for the optimal solution. Therefore, the GA algorithm is used to search and optimize and in the SVR model, and then the optimal solution ( , ) is imported into the SVR model to construct a GA - SVR model with higher accuracy. As shown in Figure 5 , the detailed process is as follows:

[0097] (a) Population initialization: Use and to form chromosomes, and randomly select a chromosome for encoding to generate the initial population of chromosomes;

[0098] (b) Fitness evaluation: Learn from the training data set to determine the fitness value of the individual, and conduct a fitness evaluation on the chromosomes according to the size of the fitness value;

[0099] (c) Genetic operations: Use gene expression methods such as selection, crossover, and mutation to iteratively replicate the chromosomes;

[0100] (d) Determine whether the termination condition is met: If and on the chromosome no longer change or reach the maximum number of iterations, then stop the optimization and output the optimal solution ( , );

[0101] (e) Parameter import: Import the optimal parameters ( , ) into the SVR model for regression prediction.

[0102] Establish an OED - GA - SVR coupled algorithm model.

[0103] During the construction process, large-diameter buried steel pipes are involved in many factors. How to reasonably design an analysis scheme is a major problem. This project intends to establish a pipeline structure prediction model of the OED-GA-SVR coupling algorithm based on the orthogonal experiment (OED) scheme. During the analysis, the calculation schemes obtained from the orthogonal experiment are used as the main control points in the prediction space. Linear interpolation is performed on the distances between the main control points to supplement multiple groups of experimental schemes, and the supplementary schemes are listed as auxiliary control points in the prediction space. The independent variables and response variables under each scheme are sorted out respectively as the initial input data of the model. To improve the stability of the regression model, multiple groups of sample data are normalized:

[0104] (3);

[0105] In the formula, represents the parameter in the th sample value, is the parameter value normalized to the range of [0, 1]. The normalized data samples are used to train and test the proposed GA-SVR model, where the training set samples account for 80% and the test set samples account for 20%.

[0106] To judge the regression effect of the model, the root mean square error RMSE and the coefficient of determination R 2 are used to evaluate the regression fitting effect of the model. The calculation formulas are as follows:

[0107] (4);

[0108] In the formula, and are the measured and predicted values of the prediction index respectively, is the number of samples, is the average value of the measured index.

[0109] It should be noted that the content not elaborated in detail in the present invention is all prior art and is well-known to those skilled in the art.

[0110] Therefore, by adopting the above intelligent prediction method for the mechanical behavior during the construction process of buried pipelines, the present invention can accurately simulate the construction environment, optimize the construction plan, improve the construction quality, reduce the construction risk, and has significant technical advantages and application values.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent prediction method for mechanical behavior during the construction process of a buried pipeline, characterized in that: The following steps are involved: Step S1, defining geometric parameters, unit types and material parameters of a refined numerical simulation model of a buried pipeline construction process; Step S2, establishing a three-dimensional finite element model of the pipeline, cushion layer, backfill soil, original soil and internal support; Step S3, creating contact relationships between the steel pipe and the backfill soil, and between the backfill soil and the original soil, and setting contact parameters; Step S4, performing multi-stage load step simulation on the refined numerical simulation model of the buried pipeline construction process, including trench excavation, cushion layer installation, pipeline and internal support installation, backfilling of soil layer by layer, and removal of internal support; Step S5, designing an experimental scheme based on the orthogonal experimental method to obtain experimental data; Step S6: Relying on the Python platform, support vector regression and genetic algorithm are used to train and test the test data to build an intelligent prediction model for mechanical behavior; Step S7, using the intelligent prediction model for mechanical behavior to predict the mechanical behavior of the buried pipeline during construction; In step S6, constructing a mechanical behavior intelligent prediction model includes: Establish a GA-SVR model and use the GA algorithm to search for the penalty factor in the SVR model and width parameters , and then the optimal solution , Import the SVR model and build the GA-SVR model; , They represent the optimal penalty factor and width parameter obtained by GA algorithm optimization respectively; The OED-GA-SVR coupling algorithm model was established, and the experimental schemes obtained by orthogonal experiments were used as the main control points of the prediction space. The distances between the main control points were linearly interpolated, and multiple groups of experimental schemes were supplemented. The supplemented experimental schemes were listed as auxiliary control points of the prediction space, and the independent variables and response quantities under each scheme were sorted out as the initial input data of the model. After normalization, multiple groups of sample data are used to train and test the proposed GA-SVR model.

2. According to claim 1, the intelligent prediction method for mechanical behavior of buried pipeline construction process is characterized in that: In step S1, the geometric parameters include the pipe diameter, the pipe burial depth, the soil arc foundation center angle, the cushion thickness, the trench width, the trench sidewall inclination angle, and the length and width of the entire model.

3. The intelligent prediction method for mechanical behavior of buried pipeline construction process according to claim 1 is characterized in that: In step S1, the unit types include plane unit, solid unit, shell unit and rod unit.

4. The intelligent prediction method for mechanical behavior of buried pipeline construction process according to claim 1 is characterized in that: In step S1, the material parameters include: Elastic modulus, Poisson's ratio, density, linear expansion coefficient, and thickness of steel pipes; Elastic modulus, Poisson's ratio, density, cohesion, internal friction angle, expansion angle, and linear expansion coefficient of original soil, backfill soil, and cushion layer; Cross-sectional type, cross-sectional area, elastic modulus, Poisson's ratio, density, and linear expansion coefficient of the cross-shaped support.

5. The intelligent prediction method for mechanical behavior of buried pipeline construction process according to claim 1 is characterized in that: In step S2, establishing a three-dimensional finite element model includes: Create key points of the model according to the coordinates of each position, connect the points into lines, establish a linear contour, and then enclose the lines into a surface to create a two-dimensional plane model of the buried pipeline; Divide the plane horizontally and vertically to divide the two-dimensional plane grid of the buried pipeline, and drag the surface grid to form a three-dimensional finite element model of the buried pipeline; The key points of the M-shaped support are created along the circumference of the pipe, and are connected into lines in pairs to divide the M-shaped support grid and construct a grid model of the M-shaped support.

6. The intelligent prediction method for mechanical behavior of buried pipeline construction process according to claim 1 is characterized in that: In step S3, creating a contact relationship includes: Establish contact pairs between steel pipe and backfill soil, and between backfill soil and original soil, and set contact type, friction coefficient, penalty stiffness, and initial gap; The ANSYS surface-to-surface contact method is used to simulate the surface-to-surface contact unit Target170-Contact174. Contact pairs are established on the unit surface, and the contact stress is calculated by penetrating the surrounding nodes.

7. The intelligent prediction method for mechanical behavior of buried pipeline construction process according to claim 1 is characterized in that: In step S4, the multi-stage load step simulation includes: Load step 1: Use the EKILL command to kill the elements of the soil, steel pipes, and cross-shaped supports in the trench, and apply full constraints to the nodes of the killed elements. Then apply gravity load and solve. After solving, release the dead element constraints. Load step 2: Use the EALIVE command to activate the cushion unit, apply full constraints to the killed unit points, reactivate the axial constraints of the cushion unit, solve, and release the dead unit constraints after solving; Load step 3: Use the EALIVE command to activate the pipe and the cross-shaped support, apply full constraints to the killed unit points, reactivate the axial constraints of the pipe unit, solve, and release the dead unit constraints after solving; Load steps 4 to 9: Use the EALIVE command to activate the backfill soil layer 1, apply full constraints to the killed unit points, reactivate the axial constraints of the backfill soil layer 1, and solve. After solving, release the dead unit constraints; according to the above method, use the loop command to activate the backfill soil layers 2, 3, 4, 5, and 6 in turn, and solve them layer by layer; Load step 10-15: According to the method of load step 4-9, use the loop command to activate the 7th, 8th, 9th, 10th, 11th and 12th layers of backfill soil in sequence and solve them layer by layer; Load step 16: Activate the 13th layer of backfill soil at the position of 0.3m above the pipe top and apply full constraints to the killed unit points, reactivate the axial constraints of the 1st layer of backfill soil, solve, and release the dead unit constraints after solving; Load step 17: Activate 14 layers of backfill soil at the position of 0.3m to 0.6m above the top of the pipe, apply full constraints to the killed unit points, reactivate the axial constraints of the soil layer, solve, and release the dead unit constraints after solving; Load step 18: Activate 15 layers of backfill soil from the top of the pipe to the ground at a position of 0.6 m. Apply full constraints to the killed unit points, reactivate the axial constraints of the soil layer, solve, and release the dead unit constraints after solving. Load step 19: Kill the elements of the cross brace.

Images