Three-dimensional modeling and finite element integrated pavement structure multi-objective optimization design method

Abstract

The invention discloses a three-dimensional modeling and finite element integrated pavement structure multi-objective optimization design method. The method comprises the following steps: establishing an asphalt pavement road linearity and pavement structure integrated BIM model according to road linear data, a drawing and pavement structure layer information; performing fine grid generation and inspection on the established BIM model in Hypermesh software to obtain a grid model, namely a BIM-FEM automatic numerical simulation framework, and exporting the grid model in a. Inp file form; importing a B IM-FEM automatic numerical simulation framework into ABAQUS software, performing rut deformation numerical simulation on the framework, analyzing the rut deformation condition of the asphalt pavement, and establishing a rut and fatigue numerical simulation automatic preprocessing process; and based on the rut and fatigue numerical simulation automatic preprocessing process, rut and fatigue numerical simulation is carried out by using a BIM-FEM automatic numerical simulation framework, and rut value and fatigue life numerical simulation results of different pavement structures are obtained. According to the method, the road structure design and analysis efficiency of the BIM under complex conditions is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of road engineering design, and in particular to a multi-objective optimization design method for a pavement structure integrating three-dimensional modeling and finite element analysis. Background Art

[0002] Traditional road design mainly adopts two-dimensional and manual calculation design methods. If there is a design change, it requires a lot of work to modify it. At present, BIM (three-dimensional modeling) technology can realize three-dimensional visualization of the design, but the current BIM technology lacks structural verification analysis, which is an important part of asphalt pavement design. Rutting is a permanent deformation of asphalt pavement, which has an important impact on the quality and service life of asphalt pavement. Fatigue cracking is one of the common diseases of asphalt pavement, which seriously affects the service performance and service life of asphalt pavement. The use of numerical simulation can accurately simulate the rutting and fatigue life of asphalt pavement. However, due to the changeable road environment and external conditions, a large amount of data will be generated during the structural design process. It is difficult to realize complex structural parametric modeling and integration of different data formats through the traditional design method combining BIM model with finite element. In addition, the existing research method does not combine the results of data simulation with automatic algorithms to perform multi-objective optimization of pavement structure combination. Therefore, it is necessary to develop a framework and algorithm that can efficiently perform multi-objective optimization design of asphalt pavement structure under complex road conditions.

[0003] In the past, in the field of road engineering, the establishment of a complex integrated model of road linearity and pavement structure required data transmission in multiple software, which seriously affected the efficiency of asphalt pavement structure design and analysis. The traditional BIM and FEM (finite element integration) simulation methods lacked the accuracy and standards of the control grid model during data transmission, resulting in a large deviation between the verification results and the actual results and low accuracy. The existing road BIM-FEM framework lacked actual engineering reference results, and the reliability of the framework could not be verified. In addition, for the multi-objective optimization of pavement structure, there was a lack of accurate data reference and calculation results. The above factors would have a serious impact on the verification results of road structure numerical simulation and the efficiency of pavement structure design, making it difficult to meet the needs of intelligent and automated road design.

[0004] Therefore, in order to solve the above problems, a multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis is provided. Summary of the invention

[0005] The purpose of the present invention is to overcome the existing defects and provide a multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis, which greatly improves the efficiency of BIM in the design and analysis of road structures under complex conditions and more accurately and reliably determines the best design scheme for asphalt pavement.

[0006] The technical solution to achieve the above purpose is:

[0007] A multi-objective optimization design method for pavement structure integrating 3D modeling and finite element analysis, comprising:

[0008] Step S1, establishing an integrated BIM model of asphalt pavement road linearity and pavement structure according to road linearity data, drawings, and pavement structure layer information;

[0009] Step S2, fine mesh generation and inspection of the established BIM model in Hypermesh software to obtain a mesh model, namely, the BIM-FEM automated numerical simulation framework, and export it in the form of a .inp file;

[0010] Step S3, importing the BIM-FEM automated numerical simulation framework into ABAQUS software, performing rutting deformation numerical simulation on the framework, analyzing the rutting deformation of the asphalt pavement, and establishing an automatic preprocessing process for rutting and fatigue numerical simulation;

[0011] Step S4, based on the automatic preprocessing process of rutting and fatigue numerical simulation, using the BIM-FEM automatic numerical simulation framework to perform rutting and fatigue numerical simulation, and obtain rutting values ​​and fatigue life numerical simulation results of different pavement structures;

[0012] Step S5, comparing the rutting value and fatigue life numerical simulation results with the test results of the actual project to verify the stability and reliability of the BIM-FEM automated numerical simulation framework;

[0013] Step S6, using the Latin hypercube sampling method to select 1000 groups of asphalt pavement parameter combination samples, simulate through the BIM-FEM automated numerical simulation framework, calculate the results of rutting and fatigue life, and combine CNN and TCN to establish a prediction model CNN-TCN-Attention (convolutional neural network-temporal convolutional network-attention mechanism), use the sample results to train the prediction model, and obtain the relationship between rutting and fatigue life and each input variable;

[0014] Step S7, relying on the improved MOEA / D (multi-decomposition objective evolutionary algorithm) optimization algorithm based on the population evolution degree to optimize the sample data, using TOPSIS to find the best solution in the Pareto optimal solution, and determine the best design solution for the asphalt pavement structure.

[0015] Preferably, in step S1, road data information is used to create road linearity in Dyanmo (an auxiliary tool for parametric design based on Revit), dynamo nodes are developed, and three-dimensional parametric design curves are established. According to given pavement structure parameters, the angle information of each section of the pavement and the position information of the corresponding points are calculated, the position of the pavement structure adaptive family is determined and placed, and an integrated BIM model of asphalt pavement road linearity and pavement structure is established.

[0016] Preferably, create road linearity in Dyanmo, develop Dynamo nodes, and establish three-dimensional parametric design curves, including:

[0017] Use the FilePath node to read the coordinates of the road centerline in the Excel table, which contains the x, y coordinates, design elevation and azimuth of each control point;

[0018] Develop the Control.CenterLineByPoints (define a center line) node to read the coordinate information of Excel to generate a spline curve as the center line of the road. Develop the Control.3D CurveChainagePoint (three-dimensional curve control point) node to read the position information of the road line shape and obtain the coordinates of the point at a fixed length on the curve.

[0019] Create a circle from a given plane and radius, create a surface from a circle, calculate the intersection between a surface and an object, and create control points for line shape control;

[0020] Develop the Control.createThickenAndGetPerimetercurves (get perimeter curves of an object) node to translate the centerline of the road, then use the Loft (a command to create a complex surface) operation to generate a surface and get the boundary curve of the surface;

[0021] Then, a Python script is used in Dynamo to develop the Control.JoinCurves node to group the boundary curves as control lines for the subsequent pavement structure model, namely, three-dimensional parametric design curves.

[0022] Preferably, according to given pavement structure parameters, an integrated BIM model of asphalt pavement road linearity and pavement structure is established, including:

[0023] Use the ReadExcel.ImportExcel (read Excel file) node to read the road surface layer related parameters in the Excel table;

[0024] The required libraries are imported and a function is defined to divide a list into chunks of a specified size;

[0025] Then get the input parameters aligns, segments and lengths;

[0026] Then, each section of the road surface is processed, the angles and positions of the points are calculated, and the results are stored in the corresponding lists;

[0027] Finally, the calculated angles and point positions are returned as output to generate an adaptive pavement structure family with detailed parameter attributes, and parametric design lines are created through NurbsCurve (a mathematical representation method that describes 3D geometric shapes) nodes;

[0028] PolyCurve (polynomial curve) is then used to create parametric cross-sectional profiles of different structural layers at key control nodes, and Solid (geometry) nodes are used to sweep along the path to establish an integrated BIM model of the asphalt pavement linear and pavement structure.

[0029] Preferably, the step S2 comprises:

[0030] Export the established integrated road structure BIM model as a .sat file;

[0031] Record geometric properties, material properties, environmental factors and loading condition data in .txt files;

[0032] The .sat file is imported into Hypermesh (a preprocessing tool in finite element analysis software) for meshing. The mesh size standard and control standard of the mesh shape are defined based on the geometric features. The fine mesh is generated and checked to obtain the mesh model, namely the BIM-FEM automated numerical simulation framework, and exported in the form of a .inp file.

[0033] Preferably, in step S3, an automatic preprocessing process for rutting and fatigue numerical simulation is established, including:

[0034] Import the BIM-FEM automated numerical simulation framework into ABAQUS (a general finite element analysis software based on the finite element analysis method) finite element software, and use Python (a programming language) to automatically encode the .inp containing geometric information, material properties, and boundary definitions;

[0035] The FORTRAN (a programming language) language is used to define the external temperature and heat flow that change with time, and is used to simulate the temperature field of the road structure under the external environment;

[0036] Modify the initial .inp file, add initial conditions to import the corresponding temperature field, and establish elastic analysis steps and creep analysis steps. The total time of the analysis step is set to the load accumulation time, and the tire ground load is applied to simulate the road rutting after 2 million standard axle cycles.

[0037] Preferably, in step S4, the BIM-FEM automated numerical simulation framework .inp file is re-imported, the material viscoelastic properties are defined, the form of static analysis is selected, the load is applied, the stress intensity factor is output, and the rutting value and fatigue life numerical simulation results of different pavement structures are calculated by the stress-life method.

[0038] Preferably, step S6 comprises:

[0039] The Latin hypercube sampling method was used to select 1000 sets of asphalt pavement parameter combination samples as input parameters of the BIM-FEM automated numerical simulation framework, and the results of the ABAQUS ODB (temperature field calculation results) file were extracted to obtain 1000 sets of rutting and fatigue life results as output parameters.

[0040] Combining CNN and TCN, and introducing the attention mechanism, a prediction model CNN-TCN-Attention is established. By comparing the data of the training set and the test set, the relationship between the rutting and fatigue life and each input variable is obtained; among them, 80% of the samples are set as the training data set in the model training process, and the remaining 20% ​​are used as the test data set.

[0041] Preferably, in step S6, the prediction model CNN-TCN-Attention includes: 6 convolutional layers, 6 pooling layers, 1 fully connected layer, 2 TCN layers and 1 attention layer, wherein the number of convolution kernels of the convolutional layers is 16, 32, 32, 64, 64 and 64 respectively, the area size of the pooling layer is set to 2×1, the dropout ratio of the prediction model is set to 0.25, the learning rate is 0.008, and the number of training rounds is 1000;

[0042] And through the root mean square error RMSE, mean absolute error MAE and goodness of fit R 2 As an evaluation indicator of the prediction model, R 2 It represents the correlation between the model's features and the prediction target. The calculation formula of the evaluation index is as follows:

[0043]

[0044] In the formula, n is the total number of sample data sets, and Represent the predicted value of the prediction model and the actual observed value, respectively. and represent the average predicted value and the average actual value, respectively.

[0045] Preferably, the step S7 comprises:

[0046] Step S71, improving the selection of neighborhoods and replacement domains in the MOEA / D optimization algorithm, proposing a mechanism for dynamically adjusting domains based on the degree of congestion of sub-problems, obtaining an improved MOEA / D optimization algorithm based on the degree of population evolution, and then optimizing the sample data;

[0047] Step S72, fitting the relationship between rutting and fatigue life and the material properties of the pavement surface structure through the prediction model CNN-TCN-Attention regression prediction, and introducing the trained regression function as the optimized fitness function;

[0048] Step S73, using TOPSIS (Topology-Inferior Solution Distance Method) to find the best solution among the Pareto optimal solutions, and obtain the best asphalt pavement structure.

[0049] The beneficial effects of the present invention are:

[0050] 1) Based on Dynamo visual programming technology, the present invention develops a node encapsulation program, efficiently and accurately establishes a three-dimensional BIM model of road linearity and structure integration, and uses pavement linear information, such as pile number coordinates, pavement structure layer information, such as geometric dimensions, and parameterized coordinates to establish a road engineering BIM model that can be quickly updated, providing an efficient method for pavement structure design and optimization of road engineering, and providing a methodological basis for asphalt pavement structure mechanics analysis;

[0051] 2) The present invention proposes a BIM-FEM automatic data transmission and conversion and numerical simulation preprocessing process for asphalt pavement in road engineering. Relying on the TCL scripting language, the Hypermesh software is developed to mesh the BIM road model, provide a mesh model that meets the requirements of numerical simulation calculations, and ensure the fidelity of the model and the feasibility of the framework; in addition, based on Python scripts and FORTRAN languages, a preprocessing process for numerical simulation of rutting of asphalt pavement under continuously variable temperature conditions is written, so that the rutting deformation of asphalt pavement can be calculated under the actual external ambient temperature and load conditions, thereby improving the reliability and automation level of numerical simulation in the field of road engineering;

[0052] 3) The BIM-FEM automated numerical simulation framework in the present invention can control the modeling and simulation of large-scale road projects within one hour, which is much shorter than the traditional modeling time. In addition, the study of the influence of the framework on the pavement structure and the rutting deformation results can provide a certain theoretical reference for the optimization design of asphalt pavement structure and the improvement of road performance.

[0053] 4) The present invention combines convolutional neural networks with temporal convolutional networks and introduces an attention mechanism to establish a multi-objective prediction model. A large number of data sets are calculated based on the BIM-FEM automated numerical simulation framework. The model is trained based on a training set and a test set. The improved MOEA / D optimization algorithm based on the degree of population evolution is used to optimize the results of the prediction model and select the best asphalt pavement design scheme. This can significantly improve the scientificity, accuracy and economy of the asphalt pavement structure design, greatly extend the service life of the asphalt pavement, reduce the design cost, and provide important technical support for the intelligent transformation of road engineering. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of a multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to the present invention;

[0055] Figure 2 It is a specific flow chart for determining the best design scheme for asphalt pavement structure in the present invention;

[0056] Figure 3 It is a BIM model diagram of the asphalt pavement road linearity and pavement structure integration established in the present invention;

[0057] Figure 4 It is a flow chart of automatic preprocessing of rutting and fatigue numerical simulation in the present invention;

[0058] Figure 5 It is a schematic diagram of numerical simulation results and theoretical formula calculation results in the present invention. DETAILED DESCRIPTION

[0059] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. In the description of the present invention, it should be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside" and the like indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0060] The present invention will be further described below in conjunction with the accompanying drawings.

[0061] In the past, in the field of road engineering, the establishment of a complex integrated model of road linearity and pavement structure required data transmission in multiple software, which seriously affected the efficiency of asphalt pavement structure design and analysis. The traditional BIM and FEM simulation methods lacked the accuracy and standard of the control grid model during data transmission, resulting in a large deviation between the verification results and the actual results and low accuracy. The existing road BIM-FEM framework lacked the results of actual engineering references, and the reliability of the framework could not be verified. In addition, the traditional road structure optimization method was inefficient, and the accuracy could not be guaranteed, resulting in a large cost consumption for road design. Based on the above problems, the present invention provides a multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element.

[0062] like Figure 1 As shown, a multi-objective optimization design method for pavement structure integrating 3D modeling and finite element analysis includes:

[0063] Step S1, establishing an integrated BIM model of asphalt pavement road linearity and pavement structure according to road linearity data, drawings, and pavement structure layer information.

[0064] In the embodiment, the road data information is used to create road linearity in Dyanmo, develop dynamo nodes, establish a three-dimensional parametric design curve, calculate the angle information of each section of the road and the position information of the corresponding point according to the given pavement structure parameters, determine and place the pavement structure adaptive family position, and establish an integrated BIM model of asphalt pavement road linearity and pavement structure, such as Figure 3 shown.

[0065] like Figure 3 As shown in the left part, create road linearity in Dyanmo, develop Dynamo nodes, and establish 3D parametric design curves, including:

[0066] Use the FilePath node to read the coordinates of the road centerline in the Excel table, which contains the x, y coordinates, design elevation and azimuth of each control point;

[0067] Develop the Control.CenterLineByPoints node to read the coordinate information from Excel and generate a spline curve as the center line of the road. Develop the Control.3D CurveChainagePoint node to read the position information of the road line shape and obtain the coordinates of the points on the curve at a fixed length.

[0068] Create a circle from a given plane and radius, create a surface from a circle, calculate the intersection between a surface and an object, and create control points for line shape control;

[0069] Develop the Control.createThickenAndGetPerimetercurves node to translate the centerline of the road, then use the Loft operation to generate a surface and get the boundary curve of the surface;

[0070] Then, a Python script was used in Dynamo to develop the Control.JoinCurves node to group the boundary curves, which served as control lines for the subsequent establishment of the pavement structure model, namely, the three-dimensional parametric design curves.

[0071] like Figure 3 As described in the right part, based on the given pavement structure parameters, an integrated BIM model of asphalt pavement road linearity and pavement structure is established, including:

[0072] Use the ReadExcel.ImportExcel node to read the road surface related parameters in the Excel table;

[0073] The required libraries are imported and a function is defined to divide a list into chunks of a specified size;

[0074] Then get the input parameters aligns, segments and lengths;

[0075] Then, each section of the road surface is processed, the angles and positions of the points are calculated, and the results are stored in the corresponding lists;

[0076] Finally, the calculated angles and point positions are returned as output to generate an adaptive pavement structure family with detailed parameter attributes, and a parametric design line shape is created through the NurbsCurve node;

[0077] PolyCurve then creates parametric cross-sectional profiles of different structural layers at key control nodes, and Solid nodes are used to sweep along the path to establish an integrated BIM model of the asphalt pavement road linearity and pavement structure.

[0078] Step S2, the established BIM model is finely meshed and checked in the Hypermesh software to obtain a mesh model, namely the BIM-FEM automated numerical simulation framework, and exported in the form of an .inp file.

[0079] In the embodiment, step S2 includes:

[0080] Export the established integrated road structure BIM model as a .sat file;

[0081] Record geometric properties, material properties, environmental factors and loading condition data in .txt files;

[0082] The .sat file is imported into Hypermesh for meshing. The mesh size standard and control standard of the mesh shape are defined based on the geometric features. The fine mesh is generated and checked to obtain the mesh model, namely the BIM-FEM automated numerical simulation framework, and exported in the form of a .inp file.

[0083] The uncertainty of external environmental factors and load conditions in the pavement structure model of complex road projects will lead to repeated changes in the design, and in the process of data conversion, there will be risks of data loss and omission, which will seriously affect the accuracy and efficiency of asphalt pavement structure design. Establishing a BIM-FEM automated numerical simulation framework can improve the data interactivity between the BIM model and finite element software, and improve the efficiency and accuracy of road structure numerical simulation.

[0084] Step S3, importing the BIM-FEM automated numerical simulation framework into the ABAQUS software, performing rutting deformation numerical simulation on the framework, analyzing the rutting deformation of the asphalt pavement, and establishing an automatic preprocessing process for rutting and fatigue numerical simulation.

[0085] like Figure 4 As shown in the figure, the automatic preprocessing process of rutting and fatigue numerical simulation is established, including:

[0086] The BIM-FEM automated numerical simulation framework is imported into the ABAQUS finite element software, and the .inp containing geometric information, material properties, and boundary definitions is automatically encoded using the Python language. Figure 4 Define material properties in

[0087] The FORTRAN language is used to define the external temperature and heat flow that change with time, and to simulate the temperature field of the road structure under the external environment;

[0088] Modify the initial .inp file, add initial conditions to import the corresponding temperature field, and establish elastic analysis steps (Static) and creep analysis steps (Vi sco). The total time of the analysis step is set to the load accumulation time, and the tire ground load is applied to simulate the road rutting after 2 million standard axle cycles.

[0089] Step S4, based on the automatic preprocessing process of rutting and fatigue numerical simulation, the BIM-FEM automatic numerical simulation framework is used to perform rutting and fatigue numerical simulation to obtain the rutting values ​​and fatigue life numerical simulation results of different pavement structures.

[0090] like Figure 4As shown, re-import the BIM-FEM automated numerical simulation framework .inp file, define the material viscoelastic properties, select the static analysis form, apply the load, output the stress intensity factor, and calculate the rutting value and fatigue life numerical simulation results of different pavement structures through the stress-life method.

[0091] Step S5, compare the rutting value and fatigue life numerical simulation results with the test results of the actual project to verify the stability and reliability of the BIM-FEM automated numerical simulation framework. The numerical simulation results are compared with the theoretical formula calculation results. Figure 5 As shown in the figure, the rutting deformation results of the asphalt pavement calculated by numerical simulation and theoretical formula are 17.38mm and 17.02mm respectively, and the error of numerical simulation is 2.12%, indicating that the rutting results calculated by the BIM-FEM automated numerical simulation framework are reliable.

[0092] By analyzing the advantages of different neural networks, it is found that combining convolutional neural network (CNN) with temporal convolutional network (TCN) is more effective in predicting rutting and fatigue life of asphalt pavement. In addition, the introduction of attention mechanism can effectively screen and highlight key features, thereby improving the prediction accuracy of the model.

[0093] Step S6, using the Latin hypercube sampling method to select 1000 groups of asphalt pavement parameter combination samples, simulated through the BIM-FEM automated numerical simulation framework, calculated the results of rutting and fatigue life, and combined CNN and TCN to establish a prediction model CNN-TCN-Attent ion, and used the sample results to train the prediction model to obtain the relationship between rutting and fatigue life and each input variable.

[0094] In the embodiment, step S6 includes:

[0095] The Latin hypercube sampling method was used to select 1000 sets of asphalt pavement parameter combination samples as input parameters of the BIM-FEM automated numerical simulation framework, and the results of the ABAQUS ODB file calculation were extracted to obtain 1000 sets of rutting and fatigue life results as output parameters.

[0096] Combining CNN and TCN, and introducing the attention mechanism, a prediction model CNN-TCN-Attention is established. By comparing the data memory of the training set and the test set, the relationship between the rutting and fatigue life and each input variable is obtained. Among them, 80% of the samples are set as the training data set in the model training process, and the remaining 20% ​​is used as the test data set. By comparing the data memory of the training set and the test set, the accuracy of the prediction model is determined, and the relationship between the rutting and fatigue life and each input variable is obtained.

[0097] In the embodiment, the prediction model CNN-TCN-Attention includes: 6 convolutional layers, 6 pooling layers, 1 fully connected layer, 2 TCN layers and 1 attention layer, wherein the number of convolution kernels of the convolutional layer is 16, 32, 32, 64, 64 and 64 respectively, the area size of the pooling layer is set to 2×1, the dropout ratio of the prediction model is set to 0.25, the learning rate is 0.008, and the number of training rounds is 1000 times;

[0098] And through the root mean square error RMSE, mean absolute error MAE and goodness of fit R 2 As an evaluation indicator of the prediction model, R 2 It represents the correlation between the model's features and the prediction target. The calculation formula of the evaluation index is as follows:

[0099]

[0100] In the formula, n is the total number of sample data sets, and Represent the predicted value of the prediction model and the actual observed value, respectively. and represent the average predicted value and the average actual value, respectively.

[0101] After calculation, the R of the training set and test set of the prediction model CNN-TCN-Attention in rutting prediction 2 The R values ​​of the training set and the test set in fatigue life prediction are 0.977 and 0.957 respectively. 2 These are 0.961 and 0.947 respectively, which are accurate enough for optimization use.

[0102] Step S7, optimizing the sample data by using the improved MOEA / D optimization algorithm based on the population evolution degree, using TOPSIS to find the best solution in the Pareto optimal solution, and determining the best design solution for the asphalt pavement structure.

[0103] MOEA / D uses a decomposition method to transform a multi-objective optimization problem into a series of single-objective optimization sub-problems, constructs a problem-oriented neighborhood relationship, and then simultaneously processes all sub-problems in a collaborative manner, finally obtaining a complete Pareto frontier. In the MOEA / D algorithm, each sub-problem corresponds to a weight vector, and its objective function is a continuous function of the weight vector. Close weight vectors usually correspond to similar optimal solutions. Therefore, the concept of neighborhood is introduced. In MOEA / D, two types of neighborhoods are considered: selection neighborhood and replacement neighborhood. The selection neighborhood determines the selection range of the parent solution, and the replacement neighborhood determines which old solutions should be replaced by the new solution. However, the fixed neighborhood design has defects: (1) The fixed selection neighborhood makes the selection range of the parent solution consistent in different search stages, which may limit the exploration of the global optimal solution in the early search stage. (2) The fixed replacement neighborhood leads to a new solution. By replacing all the difference solutions in the neighborhood by aggregating function values, this strategy may produce too many similar sub-solutions, thereby accelerating convergence but reducing diversity and easily falling into local optimality.

[0104] like Figure 2 As shown, step S7 includes:

[0105] Step S71, improve the selection of neighborhoods and replacement areas in the MOEA / D optimization algorithm, propose a mechanism for dynamically adjusting areas based on the degree of congestion of sub-problems, obtain an improved MOEA / D optimization algorithm based on the degree of population evolution, and then optimize the sample data.

[0106] Step S72, the relationship between rutting and fatigue life and the material properties of the pavement surface structure is fitted through the prediction model CNN-TCN-Attention regression prediction, and the trained regression function is introduced as the optimized fitness function.

[0107] Step S73, using TOPSIS to find the best solution in the Pareto optimal solution to obtain the optimal structure of the asphalt pavement.

[0108] After the MOEA / D algorithm optimizes the sample data, the corresponding rutting and fatigue life in the Pareto optimal solution are 9.12 mm and 1.63 million respectively, and the relative closeness corresponding to the optimal solution reaches the maximum value of 0.871. The results show that the solution obtained by the improved MOEA / D algorithm can reduce rutting deformation while achieving the goal of increasing fatigue life, and the optimal design scheme combination of asphalt pavement structure combination can be obtained.

[0109] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some or all of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis, characterized in that: include: Step S1, establishing an integrated BIM model of asphalt pavement road linearity and pavement structure according to road linearity data, drawings, and pavement structure layer information; Step S2, fine mesh generation and inspection of the established BIM model in Hypermesh software to obtain a mesh model, i.e., a BIM-FEM automated numerical simulation framework, and export it in the form of a .inp file; Step S3, importing the BIM-FEM automated numerical simulation framework into ABAQUS software, performing rutting deformation numerical simulation on the framework, analyzing the rutting deformation of the asphalt pavement, and establishing an automatic preprocessing process for rutting and fatigue numerical simulation; Step S4, based on the automatic preprocessing process of rutting and fatigue numerical simulation, using the BIM-FEM automatic numerical simulation framework to perform rutting and fatigue numerical simulation, and obtain rutting values ​​and fatigue life numerical simulation results of different pavement structures; Step S5, comparing the rutting value and fatigue life numerical simulation results with the test results of the actual project to verify the stability and reliability of the BIM-FEM automated numerical simulation framework; Step S6, using the Latin hypercube sampling method to select 1000 groups of asphalt pavement parameter combination samples, simulate through the BIM-FEM automated numerical simulation framework, calculate the results of rutting and fatigue life, and combine CNN and TCN to establish a prediction model CNN-TCN-Attention, use the sample results to train the prediction model, and obtain the relationship between rutting and fatigue life and each input variable; Step S7, optimizing the sample data by using the improved MOEA / D optimization algorithm based on the population evolution degree, using TOPSIS to find the best solution in the Pareto optimal solution, and determining the best design solution for the asphalt pavement structure.

2. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 1 is characterized in that: In step S1, road data information is used to create road linearity in Dyanmo, dynamo nodes are developed, and three-dimensional parametric design curves are established. According to given pavement structure parameters, the angle information of each section of the pavement and the position information of the corresponding points are calculated, the position of the pavement structure adaptive family is determined and placed, and an integrated BIM model of asphalt pavement road linearity and pavement structure is established.

3. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 2 is characterized in that: Create road linearity in Dyanmo, develop Dynamo nodes, and establish 3D parametric design curves, including: Use the FilePath node to read the coordinates of the road centerline in the Excel table, which contains the x, y coordinates, design elevation and azimuth of each control point; Develop the Control.CenterLineByPoints node to read the coordinate information from Excel and generate a spline curve as the center line of the road. Develop the Control.3D CurveChainagePoint node to read the position information of the road line shape and obtain the coordinates of the points on the curve at a fixed length. Create a circle from a given plane and radius, create a surface from a circle, calculate the intersection between a surface and an object, and create control points for line shape control; Develop the Control.createThickenAndGetPerimetercurves node to translate the centerline of the road, then use the Loft operation to generate a surface and get the boundary curve of the surface; Then, a Python script was used in Dynamo to develop the Control.JoinCurves node to group the boundary curves, which served as control lines for the subsequent establishment of the pavement structure model, namely, the three-dimensional parametric design curves.

4. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 2 is characterized in that: According to the given pavement structure parameters, the BIM model of asphalt pavement road linearity and pavement structure integration is established, including: Use the ReadExcel.ImportExcel node to read the road surface related parameters in the Excel table; The required libraries are imported and a function is defined to divide a list into chunks of a specified size; Then get the input parameters aligns, segments and lengths; Then, each section of the road surface is processed, the angles and positions of the points are calculated, and the results are stored in the corresponding lists; Finally, the calculated angles and point positions are returned as output to generate an adaptive pavement structure family with detailed parameter attributes, and a parametric design line shape is created through the NurbsCurve node; PolyCurve then creates parametric cross-sectional profiles of different structural layers at key control nodes, and Solid nodes are used to sweep along the path to establish an integrated BIM model of the asphalt pavement road linearity and pavement structure.

5. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 4 is characterized in that: The step S2 comprises: Export the established integrated road structure BIM model as a .sat file; Record geometric properties, material properties, environmental factors and loading condition data in .txt files; The .sat file is imported into Hypermesh for meshing. The mesh size standard and control standard of the mesh shape are defined based on the geometric features. The fine mesh is generated and checked to obtain the mesh model, namely the BIM-FEM automated numerical simulation framework, and exported in the form of a .inp file.

6. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 1 is characterized in that: In step S3, an automatic preprocessing process for rutting and fatigue numerical simulation is established, including: Import the BIM-FEM automated numerical simulation framework into the ABAQUS finite element software, and use Python to automatically encode the .inp file containing geometric information, material properties, and boundary definitions; The FORTRAN language is used to define the external temperature and heat flow that change with time, and to simulate the temperature field of the road structure under the external environment; Modify the initial .inp file, add initial conditions to import the corresponding temperature field, and establish elastic analysis steps and creep analysis steps. The total time of the analysis step is set to the load accumulation time, and the tire ground load is applied to simulate the road rutting after 2 million standard axle cycles.

7. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 1 is characterized in that: In step S4, the BIM-FEM automated numerical simulation framework .inp file is re-imported, the material viscoelastic properties are defined, the form of static analysis is selected, the load is applied, the stress intensity factor is output, and the rutting value and fatigue life numerical simulation results of different pavement structures are calculated by the stress-life method.

8. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 1 is characterized in that: The step S6 comprises: The Latin hypercube sampling method was used to select 1000 sets of asphalt pavement parameter combination samples as input parameters of the BIM-FEM automated numerical simulation framework, and the results of the ABAQUS ODB file calculation were extracted to obtain 1000 sets of rutting and fatigue life results as output parameters. Combining CNN and TCN, and introducing the attention mechanism, a prediction model CNN-TCN-Attention is established. By comparing the data of the training set and the test set, the relationship between the rutting and fatigue life and each input variable is obtained; among them, 80% of the samples are set as the training data set in the model training process, and the remaining 20% ​​are used as the test data set.

9. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 8, characterized in that: In step S6, the prediction model CNN-TCN-Attention includes: 6 convolutional layers, 6 pooling layers, 1 fully connected layer, 2 TCN layers and 1 attention layer, wherein the number of convolution kernels of the convolutional layers is 16, 32, 32, 64, 64 and 64 respectively, the area size of the pooling layer is set to 2×1, the dropout ratio of the prediction model is set to 0.25, the learning rate is 0.008, and the number of training rounds is 1000; And through the root mean square error RMSE, mean absolute error MAE and goodness of fit R 2 As an evaluation indicator of the prediction model, R 2 It represents the correlation between the model's features and the prediction target. The calculation formula of the evaluation index is as follows: In the formula, n is the total number of sample data sets, and Represent the predicted value of the prediction model and the actual observed value, respectively. and represent the average predicted value and the average actual value, respectively.

10. The multi-objective optimization design method for pavement structure integrating three-dimensional modeling and finite element analysis according to claim 1, characterized in that: The step S7 comprises: Step S71, improving the selection of neighborhoods and replacement domains in the MOEA / D optimization algorithm, proposing a mechanism for dynamically adjusting domains based on the degree of congestion of sub-problems, obtaining an improved MOEA / D optimization algorithm based on the degree of population evolution, and then optimizing the sample data; Step S72, fitting the relationship between rutting and fatigue life and the material properties of the pavement surface structure through the prediction model CNN-TCN-Attention regression prediction, and introducing the trained regression function as the optimized fitness function; Step S73, using TOPSIS to find the best solution in the Pareto optimal solution to obtain the optimal structure of the asphalt pavement.

Images