Method for optimizing aggregate morphology for pervious concrete load bearing guidance

Abstract

The present application relates to a kind of pervious concrete load-bearing guide aggregate morphology preferred method, belong to pervious concrete pavement material optimization and construction quality control technical field.It includes the following steps: collecting the basic data of paving section, establishing construction stage finite element model and discrete element specimen generation rule;Extract discrete element force chain guide index, excavate structure parameter estimation rule base, based on load-bearing guide index similarity propagation generates initial frog group satisfying physical constraint;Through the local evolution of embedded force chain guide and abnormal sparse perception, global templating crossover recombination is completed to complete optimization;Finally output aggregate configuration and construction structure parameter scheme.The present application can realize the binding mapping of micro aggregate morphology and macro mechanical response, guarantee the smooth transition of gradation and stiffness, effectively improve the load-bearing guide performance, reduce stress residual in construction stage.

Description

Technical Field

[0001] This invention belongs to the technical field of permeable concrete pavement material optimization and construction quality control, specifically involving a method for optimizing the aggregate morphology of permeable concrete for load-bearing guidance. Background Technology

[0002] During the continuous paving and compaction of permeable concrete pavement sections, the paving structure is simultaneously subjected to the coupled effects of multiple factors, including the self-weight of the roller, the vibration compaction load, the force of the paver, and the temperature stress during subsequent curing. Its stress state exhibits obvious stage-like and spatial differences along the paving direction. At the same time, the aggregate particles inside the permeable concrete form the main load-bearing path through a contact network. Changes in the angularity of the aggregate, the content of needle-like and flaky particles, the volume fraction of the aggregate, and the particle size distribution directly affect the pore connectivity, particle interlocking ability, and force chain continuity, thereby altering the macroscopic mechanical response and load-bearing guidance performance of the pavement section.

[0003] Existing methods have the following drawbacks: Firstly, they only consider average particle size, porosity, or empirical gradation ranges during aggregate parameter initialization, failing to reflect whether the aggregate particle contact network forms a continuous load-bearing path along the principal compressive stress trace of the pavement segment. This makes it difficult to guarantee load-bearing guidance performance in the optimization results. Secondly, conventional hybrid frog-jumping algorithms randomly generate aggregate morphology and structural parameters for each pavement segment during initialization, without applying continuous constraints on load-bearing guidance, particle size distribution, and structural parameter gradients between adjacent pavement segments. This easily leads to abrupt changes in aggregate gradation and stiffness between pavement segments. Thirdly, existing technologies lack a stable mapping relationship between aggregate morphology and structural parameters, failing to effectively transfer changes in aggregate angularity, needle-like and flaky content, and volume fraction into finite element calculation parameters during construction, such as concrete stiffness reduction factors and interlayer shear correction factors, thus reducing the physical feasibility of optimization. Fourthly, the local search of conventional hybrid frog-jumping algorithms relies solely on fitness sorting and random jump updates, lacking force chain direction guidance and a sparsity sensing mechanism for abnormal pavement segments. This results in unstable search directions, making it difficult to simultaneously reduce stress residuals while concentrating on correcting a few abnormal pavement segments and maintaining load-bearing guidance levels. Summary of the Invention

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete, comprising the following steps: S1. Collect basic data of permeable concrete pavement sections and establish finite element models and discrete element specimen generation rules for the construction stage. S2. By extracting the discrete element force chain guidance index, generating structural parameter association rules, and initializing and mapping the frog swarm constraints based on the similarity propagation of the bearing guidance index, the micro-aggregate morphology is bound to the macro-structural parameter constraints, so that the initial frog swarm individuals meet the requirements of aggregate morphology value range, bearing guidance continuity, structural parameter gradient stability and finite element calculation feasibility in the construction stage during the generation stage, and finally the initial frog swarm that meets the physical constraints is obtained. S3. Divide the initial frog population into multiple subpopulations to perform local evolution; after all subpopulations have completed local evolution, perform global information recombination, merge the frog individuals of each subpopulation into the global candidate pool, screen complementary candidate individuals based on aggregate carrying capacity guidance sequence, particle size distribution continuity and abnormal part differences, generate offspring coding vectors through templated crossover, evaluate and select the fitness of offspring coding vectors to obtain the optimal frog population individuals; S4. Convert the optimal frog swarm individuals into aggregate configuration schemes for permeable concrete pavement sections and structural parameter schemes for the construction stage.

[0005] Furthermore, step S1 specifically includes: The pavement structure to be optimized is divided into paving sections according to the continuous paving and compaction construction sequence. One permeable concrete pavement section; Collect aggregate morphology data for permeable concrete, including average angularity, needle-like and flaky index, aggregate volume fraction, and particle size distribution curve; Establishing a finite element model for the construction phase Finite element model during construction phase The inputs include pavement section geometry data, material reference parameters, construction sequence, roller self-weight, vibration load, paver force, temperature gradient load during curing, and interlayer constraint load; finite element model for the construction stage. The output includes the maximum tensile stress, maximum compressive stress, maximum deflection, and bottom tensile stress of each permeable concrete pavement section at each construction stage; Finite element model from the construction phase Extract the target force transmission direction vector , It refers to the first The average tangent direction of the principal compressive stress trace of a permeable concrete pavement section; Establish discrete element specimen generation rules, including specimen size and target porosity, particle generation method, particle contact parameters, loading boundary conditions, and loading stop conditions.

[0006] Furthermore, in step S2, the extraction of the discrete element force chain guidance index and the generation of structural parameter association rules specifically include: Construct a sample set of aggregate morphology combinations, the first The aggregate morphology parameter vector is ,in, Indicates an index of aggregate morphology combinations. Indicates the first The average edge angle of each aggregate shape combination Indicates the first The flaky and needle-like index of aggregate morphology combinations Indicates the first Aggregate volume fraction for each aggregate morphology combination; particle size distribution curve for each aggregate morphology combination. ; Discrete element specimen models are established based on a set of aggregate morphology combination samples. The input of the discrete element specimen model is the aggregate morphology combination and the bound particle size distribution curve, and the output is the mechanical response, aggregate contact force chain network and peak strength of the specimen under the corresponding aggregate morphology combination. From the Effective force chains are extracted from the aggregate contact force chain network corresponding to each aggregate morphology combination to obtain the first effective force chain. The set of effective force chains corresponding to each aggregate morphology combination; the unit direction vector of each effective force chain is calculated based on the three-dimensional coordinate sequence of the particle contact center. ; Based on the target force transmission direction vector Calculate the effective force chain length weight The load-bearing guidance index is calculated based on the degree of consistency between the effective force chain direction and the target force transmission direction. The load-bearing guidance level is classified according to the load-bearing guidance index, including high guidance level, medium guidance level and low guidance level. Structural parameter vectors corresponding to aggregate morphology combinations and permeable concrete pavement sections were obtained based on finite element calibration. , For the first The aggregate morphology combination is mapped to the first The equivalent pavement mechanical parameter vector obtained when a permeable concrete pavement section is used; The aggregate morphology range, load-bearing guidance level, and structural parameter range are combined into transaction items, and a rule base for structural parameter prediction is mined. , It is a set of queryable mappings composed of association rules.

[0007] Furthermore, in step S2, the initialization of frog swarm constraints and the mapping of structural parameters specifically include: Definition of the first Complete encoding vector of each frog swarm individual , Encode the candidate solution that simultaneously includes the aggregate morphology parameter vector, particle size distribution curve, load-bearing orientation index, and structural parameter vector of all permeable concrete pavement sections; define the first... Among the individuals in the frog pod, the first Structural parameter vector of a permeable concrete pavement section And unify the range of structural parameter values; based on the structural parameter prediction rule base The high-confidence rules initialize the aggregate morphology parameters of the first permeable concrete pavement section; based on the bearing-guiding index similarity propagation, the second to third sections are generated according to the continuous paving and compaction construction sequence. Aggregate morphology parameters of a permeable concrete pavement section; prediction rule base using structural parameters. The aggregate morphology parameters are mapped to structural parameter vectors; the gradient of structural parameter changes between adjacent permeable concrete pavement sections is constrained. The gradient constraint of structural parameter changes is used to prevent unreasonable jumps in the stiffness, compaction response or construction load parameters of local pavement sections; the above process of pavement section generation, structural parameter mapping and gradient constraint correction is repeated to form an initial frog swarm that satisfies physical constraints.

[0008] Furthermore, in step S3, the local evolution of the subpopulation specifically includes: Before performing local evolution, the fitness value of each frog swarm individual was calculated based on the finite element model calculation results during the construction phase, the load-bearing guidance index of each paving section, and the distribution of abnormal paving sections. The initial frog population was sorted and grouped according to its fitness value: After sorting all frogs in the population from best to worst fitness value, they were divided into groups using an alternating sequential allocation method. A subpopulation, in the first From the subpopulation, the frog individual with the best fitness is selected as the optimal individual, and the complete encoding vector of the optimal individual is defined as follows: The frog individual with the worst fitness is selected as the worst individual, and the complete encoding vector of the worst individual is defined as follows: Based on the difference between the complete encoding vectors of the best and worst individuals, and through amplitude truncation and constraint checks using the local constraint maintenance operator, the local update direction from which the worst individual learns to the best individual is obtained. Anomaly indicator vectors are generated based on the median of stress residuals and deflection residuals of each pavement section calculated using the finite element model during the construction phase, as well as the degree of deviation from the structural parameters of adjacent pavement sections. An abnormally sparse penalty function is constructed based on the smoothed absolute deviation between the structural parameters of each pavement segment of the worst individual and the mean of adjacent pavement segments, and the correction direction of the abnormally sparse penalty function is calculated; the gradient direction vector is obtained according to the gradient components of each parameter dimension of the abnormally sparse penalty function. ; The force chain space guidance vector is calculated based on the load-bearing guidance sensitivity of aggregate morphology parameters to the load-bearing guidance index. ; Based on the local update direction, anomaly indication vector, gradient direction vector, and force chain space guidance vector learned from the worst individual to the best individual, a local evolutionary update is performed on the worst individual to obtain the worst individual update encoding vector; the worst individual update encoding vector is then corrected by the feasible region projection operator.

[0009] Furthermore, fitness calculation specifically includes: The first The complete encoding vector for each individual frog in the swarm Input the finite element model of the construction stage, calculate the maximum tensile stress, maximum compressive stress, maximum deflection, and bottom tensile stress of each permeable concrete pavement section in each construction stage according to the continuous paving and compaction construction sequence, and compare the calculation results with the design allowable value, monitoring value, or target control value to obtain the stress residual and deflection residual of the construction stage; combine the mean of the bearing guidance index, the degree of violation of the gradient of structural parameters of adjacent pavement sections, and the number of abnormal pavement sections to form the fitness value.

[0010] Furthermore, in step S3, the global information reorganization specifically includes: A global candidate pool was constructed based on all frog individuals in each subpopulation after local evolution. For the global candidate pool Each individual frog in the swarm was re-performed with the finite element calculation of the construction stage to obtain stress residuals, abnormal location indicator vectors, load-bearing guidance index sequences, and particle size distribution curve sequences. The complementary index of aggregate morphology is calculated based on the similarity of bearing guidance sequence, the overlapping area of ​​particle size distribution, and the difference ratio of abnormal parts. Based on aggregate morphology complementary index Determine the optimal complementary candidate individual: For the first... Complete encoding vector of each frog swarm In the global candidate pool The search engine identifies candidate individuals that satisfy the requirements of cosine similarity of the bearing-guiding index sequence, overlap area of ​​particle size distribution curves, and feasible region constraints. The candidate individual with the highest complementary aggregate morphology index is selected and defined as the optimal complementary candidate individual. The complete encoding vector of the optimal complementary candidate individual is defined as follows: ; Generate a template cross-mask vector based on the anomaly indicator vectors of the current individual and the optimal complementary candidate individual. Based on template cross mask vector Templated crossover is performed to obtain the offspring encoding vector. ; Based on the structural parameter prediction rule base and feasible region projection operator, structural parameter mapping and feasibility correction are re-executed on the child encoding vector; The fitness evaluation and selection of the offspring coding vectors are based on a comprehensive evaluation of stress residuals, sparsity of abnormal pavement sections, and load-bearing guidance performance during the construction phase.

[0011] Furthermore, the specific calculation process for the similarity of the bearing guidance sequence, the overlap area of ​​particle size distribution, and the proportion of differences in abnormal parts is as follows: No. The carrying orientation index sequence of each frog swarm is represented as follows: For the first The first frog group and the first The calculation method for the cosine similarity of the bearing guidance index sequence for each candidate individual is expressed as follows: ,in, Indicates the first The first frog group and the first Cosine similarity of the bearing guidance index sequence among candidate individuals; Indicates the first The carrying guidance index sequence of each candidate individual; express Transpose of; This represents the stability constant used in similarity calculations. Represents the Euclidean norm; Area of ​​overlap of particle size distribution curves The calculation method is as follows: It is obtained by summing the minimum volume fractions of two adjacent paving sections in each sieve aperture interval. ,in, Indicates the first The first individual in the frog pod The first permeable concrete pavement section and the first The area of ​​overlap of particle size distribution curves between permeable concrete pavement sections; Indicates the first The first individual in the frog pod The permeable concrete pavement section is in the first The volume fraction of aggregate particles within each sieve aperture interval; Indicates the first The first individual in the frog pod The permeable concrete pavement section is in the first The volume fraction of aggregate particles within each sieve aperture interval; This represents a function that takes the minimum value. The difference ratio of abnormal parts is calculated by the difference ratio between two abnormal part indicator vectors. First, count the number of different values ​​of corresponding elements in the abnormal part indicator vectors of individuals in the two frog groups, and then divide by the total number of structural parameter dimensions to obtain the difference ratio of abnormal parts.

[0012] Furthermore, re-performing structural parameter mapping and feasibility correction on the child encoding vector specifically includes: Extracting the child encoding vector The aggregate morphology parameter vector and load-bearing guidance index of each permeable concrete pavement section are obtained; the structural parameter prediction rule base is queried. The corresponding structural parameter range is obtained; if a high-confidence rule is hit, the structural parameter vector is regenerated; if a high-confidence rule is not hit, the aggregate morphology parameters are corrected to the nearest rule coverage range using the feasible region projection operator; if the correction exceeds 10% of the aggregate morphology parameter value range, it indicates that the sub-encoding vector deviates too much from the rule base coverage area, and the sub-encoding vector is deemed insufficiently feasible and discarded.

[0013] Furthermore, step S4 specifically includes: Decoding the optimal frog swarm individual The optimal aggregate morphology parameter vector, optimal particle size distribution curve, optimal load-bearing guidance index, and optimal structural parameter vector for each permeable concrete pavement section are obtained. The construction mix control range is generated based on the optimal aggregate morphology parameter vector and optimal particle size distribution curve. The optimal structural parameter vector is then input into the finite element model for the construction stage. The verification was carried out; during the construction process, the aggregate morphology optimization plan for the permeable concrete pavement section was updated based on the monitoring data.

[0014] The advantages of this invention are: This invention extracts the target force transmission direction vector of permeable concrete pavement sections and couples it with the direction of the effective chain in discrete element specimens to construct a load-bearing guidance index, thereby establishing a connection between the micro-aggregate force chain network and the macro-pavement load-bearing path. It employs an association rule mining algorithm to generate a structural parameter prediction rule library from a large number of aggregate morphology combinations and finite element calibration structural parameters, achieving a queryable mapping from aggregate angularity, needle-like and flaky content, aggregate volume fraction, and load-bearing guidance level to concrete stiffness reduction factors, interlayer shear correction factors, compaction sensitivity coefficients, and temporary construction load amplification coefficients. Furthermore, it proposes a frog swarm constraint initialization method based on the similarity propagation of the load-bearing guidance index, along with continuous paving and compaction construction. Frog swarm individuals are generated sequentially segment by segment. Through load-bearing guidance continuity constraints, particle size distribution overlap area constraints, and structural parameter gradient constraints, the initial frog swarm individuals are guaranteed to simultaneously satisfy aggregate gradation continuity and mechanical parameter stability. This invention embeds force chain direction field guidance and anomaly sparsity perception mechanisms in the local evolution of hybrid frog jumping. It uses force chain spatial guidance vectors to push aggregate morphological parameters to higher load-bearing guidance regions, and uses anomaly part indicator vectors and sparsity penalty functions to directionally correct structural parameter mutations. At the same time, in global reorganization, templated cross-linking is performed based on aggregate morphology complementary index and anomaly mode exchange, realizing the synergistic optimization of load-bearing guidance improvement, stress residual reduction, and anomalous paving segment sparsity during the search process. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0016] Figure 1 This is a flowchart of the steps of the method of the present invention; Figure 2 Scatter plot of aggregate morphology combination sample distribution; Figure 3 A three-dimensional arrow diagram of the effective force chain direction field of the aggregate contact force chain network; Figure 4 A bubble chart of rules associated with the rule base for structural parameter estimation; Figure 5 A line graph showing the fitness evolution curve of the hybrid frog jumping algorithm; Figure 6 Heatmap of complementary index matrix for aggregate morphology; Figure 7 To analyze the distribution of structural parameters of each pavement section after optimization; Figure 8 To analyze the distribution of aggregate morphological parameters in the optimized permeable concrete pavement section, a line graph was generated. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 In this embodiment, as Figure 1 As shown, the present invention provides a method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete, the specific steps of which include: S1. Data Acquisition and Calculation Model Establishment for Permeable Concrete Pavement Section During the continuous paving and compaction of permeable concrete pavement, the paving structure is simultaneously affected by the self-weight of the roller, the vibration compaction load, the force of the paver, and the temperature stress during subsequent curing. Its stress state has obvious stage and spatial differences along the paving direction. At the same time, the aggregate particles inside the permeable concrete form the main load-bearing path through the contact network. The degree of aggregate angularity, the content of needle-like and flaky particles, the volume fraction of aggregate, and the particle size distribution will affect the pore connectivity, particle interlocking ability, and force chain continuity.

[0019] This step first collects geometric data, construction stage load data, aggregate morphology data, and structural response control data for the permeable concrete pavement section. Then, it establishes a finite element model and discrete element specimen generation rules for the construction stage. This provides a unified data foundation for subsequent extraction of the load-bearing guidance index, generation of a rule base for structural parameter prediction, and frog swarm optimization. The specific steps are as follows: 1) Divide the pavement into sections according to the continuous paving and compaction construction sequence of the road surface, and obtain the following results: One section of permeable concrete pavement. This indicates the total number of permeable concrete pavement sections, used to limit the length of the aggregate morphology parameter chain and structural parameter chain, and can be taken as 8 to 20. This represents the pavement section index, used to identify different permeable concrete pavement sections according to the continuous paving and compaction construction sequence. The value ranges from 1 to... .

[0020] In specific implementation, for the first For each permeable concrete pavement segment, data such as segment length, width, design thickness, cross slope, longitudinal slope, interface friction coefficient with adjacent structural layers, and construction stage number are collected and entered into the geometry and construction sequence input table of the finite element model for each construction stage. The pavement segment division should be consistent with the actual paving and compaction segments or finite element unit grouping, so that each subsequent permeable concrete pavement segment corresponds to an independent aggregate morphology parameter vector and structural parameter vector.

[0021] 2) Collect aggregate morphology data for permeable concrete, including average edge angle, needle-like / flaky index, aggregate volume fraction, and particle size distribution curve. Specifically, the average edge angle is used to characterize the sharpness of aggregate particle boundaries, the needle-like / flaky index is used to characterize the proportion of slender or flaky particles in the aggregate particles, the aggregate volume fraction is used to characterize the volume ratio of aggregate in the permeable concrete specimen, and the particle size distribution curve is used to characterize the distribution of aggregate particle volume fraction within different sieve aperture ranges.

[0022] In one implementation, aggregate morphology data can be obtained through image analysis, sieve analysis, and needle-and-flaky particle gauge testing. The average edge angle can be calculated by combining the particle profile perimeter, equivalent circle circumference, and number of corner points, and normalized to the range of 0 to 1; the needle-and-flaky particle index can be obtained by statistically analyzing the mass or quantity proportion of needle-and-flaky particles according to standard tests; the aggregate volume fraction can be obtained based on the ratio of aggregate volume to total volume in permeable concrete per unit volume; the particle size distribution curve can be obtained by statistically analyzing sieve aperture ranges such as 4.75mm to 9.5mm, 9.5mm to 13.2mm, and 13.2mm to 16.0mm. If other sieve aperture ranges are used, the sieve aperture range division should be kept consistent in the same batch of calculations.

[0023] In one embodiment, as an example, for an image of a certain aggregate particle, its outline perimeter is extracted to be 15.2 mm, the equivalent circle circumference is 12.0 mm, the number of outline corner points is 6, and the normalized baseline value for the number of outline corner points is 8. Then the original value of the edge angle is... After normalization to The average edge angle was then found to be 0.20. 200 particles were randomly selected from a batch of aggregate, and a needle-like / flaky particle index was used to determine that 30 particles were needle-like / flaky. Therefore, the needle-like / flaky particle index for this batch was... Prepare permeable concrete specimens and measure the total volume of the specimens. The volume occupied by aggregate Then the aggregate volume fraction is In the sieve analysis, the aggregate volume fraction was 0.25 for the 4.75mm to 9.5mm range, 0.50 for the 9.5mm to 13.2mm range, and 0.25 for the 13.2mm to 16.0mm range. The remaining sieve sizes were all 0. Therefore, the numerical sequence of the particle size distribution curve is as follows: (For example only, subsequent values ​​are omitted).

[0024] 3) Establish a finite element model for the construction phase. , This refers to a structural analysis model used to calculate stress, maximum deflection, and bottom tensile stress of permeable concrete pavement sections at various construction stages. Specifically, it refers to the finite element model for the construction stage. The inputs include pavement section geometry data, material reference parameters, construction sequence, roller self-weight, vibration load, paver force, temperature gradient load during curing, and interlayer constraint load; the finite element model for the construction stage... The output includes the maximum tensile stress, maximum compressive stress, maximum deflection, and bottom tensile stress of each permeable concrete pavement section at each construction stage.

[0025] In practical implementation, the finite element model during the construction phase The permeable concrete pavement section is discretized by pavement unit. The pavement unit is activated in stages according to the continuous paving and compaction construction sequence. The self-weight of the roller, vibration load, paver force, temperature gradient load during curing, interlayer constraint load and temporary construction load are applied. The overall stiffness matrix is ​​assembled step by step and the nodal displacement is solved. Then, the stress, maximum deflection and bottom tensile stress of each pavement section are obtained by substituting back.

[0026] Furthermore, in the construction phase, the finite element model The system provides a pre-defined interface for inputting structural parameter vectors for each permeable concrete pavement segment. This interface allows for the input of concrete stiffness reduction factors, inter-layer shear correction factors, compaction sensitivity coefficients, and temporary construction load amplification factors segment by segment, according to the pavement segment index. The structural parameter vectors are used to correct the equivalent elastic stiffness, inter-layer shear stiffness, compaction response characteristics, and temporary construction loads of the permeable concrete pavement segment. Changes in equivalent mechanical parameters caused by aggregate morphology variations can be transferred to the finite element calculations during the construction phase through the structural parameter vectors.

[0027] 4) From the finite element model of the construction stage Extract the target force transmission direction vector , It refers to the first The average tangent direction of the principal compressive stress trace of each permeable concrete pavement section is used to characterize the first... The main compressive stress transfer direction of each permeable concrete pavement section during the construction phase is 3.

[0028] In practical implementation, a finite element model is first developed during the construction phase. Select the control construction stage; then extract the first... Within a permeable concrete pavement section, areas with high principal compressive stress are identified, and several principal compressive stress trajectories are traced along the direction of the principal compressive stress. The tangents of these principal compressive stress trajectories are then averaged and normalized to obtain the target force transmission direction vector. To avoid inconsistencies in directional signs, the target force transmission direction vector... The paving direction should be uniformly aligned according to the construction phase.

[0029] In one embodiment, as an example, for the third permeable concrete pavement section, a certain construction stage is selected as the control construction stage; the area within the pavement section where the principal compressive stress value is greater than 70% of the maximum principal compressive stress is extracted, and five principal compressive stress traces are obtained; the average of the tangent directions of each trace is taken to obtain the vector. After normalization, the target force transmission direction vector is obtained, and the orientation is uniformly set to the paving forward direction along the construction stage.

[0030] 5) Establish discrete element specimen generation rules to convert each set of aggregate morphology parameters into corresponding discrete element specimens. The discrete element specimen generation rules include specimen size and target porosity, particle generation method, particle contact parameters, loading boundary conditions, and loading stop conditions.

[0031] In one implementation, the specimen size can be 150mm×150mm×150mm, and the target porosity can be 15% to 25%. The particle shape can be spherical or polyhedral, allowing control over the degree of concavity and convexity of the aggregates, aspect ratio, average edge angle, and needle-like index. The particle size distribution is generated according to the particle size distribution curve to produce particles of different sizes. The normal contact stiffness, tangential contact stiffness, friction coefficient, rolling resistance coefficient, and bond strength are set according to the empirical range of permeable concrete material tests or literature. The loading method is uniaxial compression loading, which continues until the axial stress of the specimen reaches 95% to 100% of the peak strength, in order to extract the aggregate contact force chain network near the peak strength state.

[0032] S2. Mining of bearing capacity-oriented correlation features of permeable concrete aggregates and initialization of frog swarm constraints The load-bearing and guiding performance of permeable concrete pavement is mainly affected by aggregate gradation, particle angularity, content of needle-like and flaky particles, pore connectivity, and particle contact network. Conventional hybrid frog-jumping algorithms typically initialize aggregate geometric and structural parameters randomly without establishing a stable mapping relationship between aggregate force chain direction, pavement stiffness response, and structural stress state during construction. This easily generates a large number of initial frog swarm individuals that do not meet physical constraints, leading to repeated iterations in the invalid parameter region during subsequent optimization.

[0033] This step involves extracting the discrete element force chain guidance index, generating structural parameter association rules, and initializing and mapping swarm constraints based on the similarity propagation of the bearing guidance index. This binds the micro-aggregate morphology with the macro-structural parameter constraints, ensuring that the initial swarm individuals meet the requirements of aggregate morphology range, bearing guidance continuity, structural parameter gradient stability, and finite element calculation feasibility during the construction phase from the generation stage. The specific steps are as follows: S201. Extraction of Discrete Element Force Chain Guiding Index and Generation of Structural Parameter Association Rules To address the issue that conventional aggregate parameter initialization, which only considers average particle size, porosity, or empirical gradation ranges, is insufficient to reflect whether the aggregate particle contact network forms a continuous load-bearing path along the principal compressive stress trajectory of the pavement section, this step first establishes a discrete element model of permeable concrete specimens under different aggregate morphology combinations using the discrete element method (DEM). Then, under loading conditions, the aggregate contact force chain network near the peak strength state is extracted. Next, the load-bearing guidance index is calculated based on the consistency between the effective force chain direction and the target force transmission direction. Finally, the load-bearing guidance level, aggregate morphology range, and finite element calibration structural parameters are combined into a transaction item. Through association rule mining, a queryable structural parameter prediction rule base is formed. The specific steps are as follows: 1) Construct a sample set of aggregate morphology combinations, the first... The aggregate morphology parameter vector is .in, This represents the aggregate morphology combination index, used to distinguish different discrete element specimens, with a value range of 1 to... ; This represents the total number of aggregate morphology combinations, used to determine the sample size for discrete element parameter scanning; it can be set to 300. Indicates the first The average edge angle of each aggregate shape combination is used to characterize the sharpness of the aggregate particle boundary, and can be taken as 0.25 to 0.85. Indicates the first The needle-like and flaky index of a combination of aggregate morphologies is used to characterize the proportion of slender or flaky particles in the aggregate particles, and can be taken as 0.05 to 0.30. Indicates the first The aggregate volume fraction of each aggregate morphology combination is used to characterize the volume ratio of aggregate in permeable concrete specimens, and can be taken as 0.55 to 0.72.

[0034] In one embodiment, such as Figure 2 As shown, the scatter plot of the aggregate morphology combination sample distribution is analyzed to show the three-dimensional relationship and distribution of the average edge angle, aggregate volume fraction and needle-like index in the aggregate morphology combination; the horizontal axis is the average edge angle (dimensionless), the vertical axis is the aggregate volume fraction (dimensionless), and the scatter plot color from purple to yellow indicates that the needle-like index (dimensionless) increases from small to large; the experiment shows that the average edge angle is mainly concentrated in the range of 0.25 to 0.85, and the aggregate volume fraction is concentrated in the range of 0.55 to 0.72, and is uniformly distributed.

[0035] Furthermore, each aggregate morphology combination is also bound to a particle size distribution curve. , This refers to the particle size distribution curve formed by the volume fraction of aggregate particles in each preset sieve aperture interval, which can be represented as: .in, This represents the sieve aperture interval index, with values ​​ranging from 1 to... ; Indicates the number of sieve aperture intervals; Indicates the first The aggregate morphology combination in the first The aggregate particle volume fraction within each sieve aperture interval, and the aggregate particle volume fraction corresponding to each sieve aperture interval satisfies .

[0036] 2) Establish a discrete element specimen model based on the aggregate morphology combination sample set. The discrete element specimen model refers to the discrete element analysis model that generates permeable concrete numerical specimens based on a set of aggregate morphology parameters and performs loading calculations. Its input is the aggregate morphology combination and the bound particle size distribution curve, and the output is the mechanical response of the specimen under the combination, the aggregate contact force chain network and the peak strength.

[0037] In specific implementation, for the first Each aggregate shape combination is first determined according to the average edge angle. Needle-like index Aggregate volume fraction and particle size distribution curve A set of particles that meet the morphological parameters is generated, and then the particles are filled and compacted according to the target porosity and specimen size. Then, the normal contact stiffness, tangential contact stiffness, friction coefficient, rolling resistance coefficient and bonding strength are set for the particle contact position. Finally, the particles are loaded in a uniaxial compression manner until they stop near the peak strength.

[0038] In practical implementation, if pelletized agglomerates are used to simulate aggregate particles, the particle size distribution curve should be used as the basis for the initial calculation. Spherical sub-particles of different sizes are generated, and then arranged according to the average edge angle. Control the unevenness of the outer contour of the agglomerate particles, and follow the needle-like index. Control the number of particles with a large ratio of major to minor axes; if polyhedral particles are used to simulate aggregate particles, generate particles with different angles and needle-like levels by controlling the number of vertices, boundary sharpness, and aspect ratio of the polyhedra. The loading process continues until the axial stress of the specimen reaches near the peak strength (e.g., in the range of 95% to 100% of the peak strength), and then stops, obtaining the first... The aggregate contact force chain network corresponding to each aggregate morphology combination.

[0039] In one embodiment, for example, the 15th aggregate morphology combination and its bound particle size distribution Aggregate pellets were used to simulate aggregate, and 150mm cubic specimens were generated with a target porosity of 20%; the normal contact stiffness was taken as... Tangential contact stiffness is taken The friction coefficient is 0.5; uniaxial compression is stopped when the axial stress reaches 98% of the peak strength, and the aggregate contact force chain network at this time is extracted, which contains approximately 8500 contact points.

[0040] 3) From the first Effective force chains are extracted from the aggregate contact force chain network corresponding to each aggregate morphology combination. An effective force chain refers to the force transmission path formed by the continuous connection of contact points of particles with large contact forces, which is used to characterize the main load-bearing channels in the aggregate skeleton of permeable concrete.

[0041] In the specific implementation, first count the... The average contact force of all particles in the discrete element specimen corresponds to each aggregate morphology combination. Then set a strong contact threshold. ,in, Indicates the first Strong contact threshold corresponding to each aggregate morphology combination. This indicates the strong contact screening factor, which can be 1.5; then, the contact force should not be less than the strong contact threshold. The particle contact points are identified as strong contact points; then, strong contact points sharing the same particle or with a spatial distance less than a preset distance threshold (e.g., 1.2 times the average radius of the two particles) are connected according to particle contact relationships to form contact paths; finally, contact paths that are too short, have excessive directional fluctuations, or are formed solely by isolated strong contact points are eliminated to obtain the first contact path. The set of effective force chains corresponding to each aggregate morphology combination.

[0042] Furthermore, define For the first The total number of effective force chains corresponding to each aggregate morphology combination Used to limit the number of force chains participating in the load-bearing orientation evaluation; definition This is an effective force chain index used for traversing the first force chain. The effective force chain in each aggregate morphology combination has a value range of 1 to 1. .

[0043] In one embodiment, as an example, the 15th aggregate morphology combination corresponds to the average contact force of all particles in the discrete element specimen. Strong contact threshold Contact points with a contact force of not less than 18.6 N were identified as strong contact points, resulting in approximately 2100 strong contact points. These were then connected according to particle contact relationships to form paths, and paths that were too short or had excessive directional fluctuations were discarded to obtain an effective force chain set. The total number of effective force chains in this set is [not specified]. .

[0044] 4) Calculate the unit direction vector of each effective force chain based on the three-dimensional coordinate sequence of the particle contact center.

[0045] In one implementation, for the first... The first in the aggregate morphology combination Each effective force chain is used to sequentially read the three-dimensional coordinates of the contact centers of adjacent particles within that effective force chain. .in, Indicates the first The contact center number in the effective force chain ranges from 1 to ; Indicates the first The total number of contact centers in an effective force chain; It refers to the first In the first aggregate form combination The first effective force chain The three-dimensional coordinates of the contact center are determined. Then, the difference vectors between the coordinates of adjacent contact centers are summed and normalized to obtain the unit direction vector. The calculation method is expressed as follows: ; in, Used to characterize the In the first aggregate form combination The spatial force transmission direction of an effective force chain has a dimension of 3; Indicates the first In the first aggregate form combination The first effective force chain Three-dimensional coordinates of the contact center; This represents the normalized stability constant of the direction vector, used to prevent the denominator from being zero, and can be taken as... ; This represents the Euclidean norm. By accumulating the directions of adjacent contact centers, the influence of local disturbances at a single contact point on the overall direction determination of the force chain can be weakened, making the unit direction vector... It can better characterize the overall force transmission direction of the effective force chain.

[0046] 5) Based on the target force transmission direction vector Calculate the effective force chain length weight .

[0047] In one implementation, for the first... The first permeable concrete pavement section and the first Individual aggregate morphology combinations, vector along the target force transmission direction. Cumulative number The projected length is obtained by projecting the line connecting adjacent contact centers in an effective force chain in the forward direction. , can be represented as: ; in, Indicates the first In the first aggregate form combination The effective force chain relative to the first The continuous positive projection length of the target force transmission direction of a permeable concrete pavement section; This represents the function that takes the maximum value.

[0048] Furthermore, from the first Select from all effective force chains of the aggregate morphology combination, relative to the first The maximum projected length of the target force transmission direction of a permeable concrete pavement section is obtained, which yields the maximum effective force chain projection length. Then, based on the projection length and the maximum effective force chain projection length Calculate the effective force chain length weight ,Right now .in, It refers to the first In the first aggregate form combination The effective force chain relative to the first The length weight of the target force transmission direction of each permeable concrete pavement section is used to improve the influence of long-range continuous force chains in load-bearing orientation evaluation. This represents the stability constant for length weight calculation, used to prevent the denominator from being zero; it can be taken as... .

[0049] In one embodiment, such as Figure 3 As shown, a three-dimensional arrow diagram of the effective force chain direction field of the aggregate contact force chain network (taking pavement segment 5 as an example) is presented, showing the spatial distribution of the effective force chain direction inside the discrete element specimen of the 5th pavement segment. The horizontal axis is the X coordinate (unit: mm), the vertical axis is the Y coordinate (unit: mm), and the vertical axis is the Z coordinate (unit: mm). The orange arrows represent the unit direction vector of each effective force chain, and the thick red arrows represent the target force transmission direction of the pavement segment. The experiment shows that the effective force chain direction has a clustering trend near the target force transmission direction, proving that the load-bearing guidance index needs to comprehensively statistically analyze the consistency of the force chain direction and the length weight in order to accurately evaluate the load-bearing guidance performance of the aggregate contact force chain network.

[0050] 6) Calculate the load-bearing guidance index based on the consistency between the effective force chain direction and the target force transmission direction. , It refers to the first The aggregate morphology combination in the first The evaluation value of the aggregate force chain network bearing capacity under the target force transmission direction of a permeable concrete pavement section is taken from the range of values. The closer the value is to 1, the more concentrated the force chain direction is in the target force transmission direction, and the stronger the load-bearing and guiding performance.

[0051] In practical implementation, for each effective force chain, the unit direction vector is first calculated. Force direction vector of the target The inner product between the two forces is calculated; the larger the inner product value, the smaller the angle between them and the stronger the directional consistency. Then, the effective components with an angle not exceeding 90° to the target force transmission direction are retained using a positive truncation function, and the interference of the reverse force chain on the load-bearing guidance evaluation is weakened. Next, the guidance sensitivity index is used to amplify the penalty effect caused by directional deviation. Finally, a weighted average is performed according to length weights to obtain the load-bearing guidance index, which can be expressed as: ; in, This represents the guidance sensitivity index, used to enhance the penalty effect of directional deviation on the load guidance index, and can be taken as 2; This represents the stability constant used in calculating the load-bearing guidance index, used to prevent the denominator from being zero; it can be taken as... ; This represents the function that takes the maximum value.

[0052] In one embodiment, as an example, 300 aggregate morphology combinations are selected to establish discrete element specimens. After loading to near the peak strength, particle contact paths with contact forces not less than 1.5 times the average contact force are extracted as effective force chains. If the angle between the majority of long-range effective force chains of a certain aggregate morphology combination and the target force transmission direction of the 5th permeable concrete pavement section is small, then the aggregate morphology combination corresponds to... Higher; if the effective force chain direction is dispersed, or the long-range effective force chain deviates significantly from the target force transmission direction, then the aggregate morphology combination corresponds to The value is relatively low. This numerical difference allows for the early elimination of aggregate morphology combinations that only meet the gradation range but lack sufficient load-bearing capacity.

[0053] 7) The carrying orientation level is divided according to the carrying orientation index. The carrying orientation level is used to transform the continuous numerical form of the carrying orientation index into discrete transaction items that can participate in association rule mining.

[0054] In practical implementation, the load-bearing guidance index can be divided into high guidance level, medium guidance level, and low guidance level. The load-bearing guidance index range for the high guidance level can be set to [0.75, 1.00], for the medium guidance level to [0.45, 0.75], and for the low guidance level to [0, 0.45]. These ranges can be adjusted based on the number of discrete element samples, the distribution of the load-bearing guidance index, and the statistical results of finite element stress residuals during the construction phase. However, a unified level classification standard should be used for the training samples of the same batch of structural parameter prediction rule bases.

[0055] 8) Obtain structural parameter vectors corresponding to aggregate morphology combinations and permeable concrete pavement sections based on finite element calibration. , It refers to the first The aggregate morphology combination is mapped to the first The equivalent pavement section mechanical parameter vector obtained when a permeable concrete pavement section is used.

[0056] Specifically, the structure parameter vector It can be represented as .in, Indicates the first The first aggregate morphology combination corresponds to the first The concrete stiffness reduction factor of a permeable concrete pavement section is used to characterize the degree of reduction in the equivalent elastic stiffness of permeable concrete relative to the stiffness of the reference concrete. Indicates the first The first aggregate morphology combination corresponds to the first The interlayer shear correction factor for a permeable concrete pavement section is used to correct the interfacial shear stiffness between the pavement layer and the base layer or adjacent pavement layers, so as to reflect the influence of aggregate interlocking on interlayer shear performance. Indicates the first The first aggregate morphology combination corresponds to the first The compaction sensitivity coefficient of a permeable concrete pavement section is used to describe the ease with which permeable concrete reaches the target porosity and design strength under vibration compaction load. Indicates the first The first aggregate morphology combination corresponds to the first The temporary construction load amplification factor for a permeable concrete pavement section is used to characterize the adverse amplification effects of the roller's self-weight, vibration load, paver force, temperature gradient load during curing, and interlayer constraint load on the load during the construction phase.

[0057] In one implementation, the structural parameter vector can be obtained through finite element calibration. When calibrating, the first... Equivalent elastic modulus, target porosity correction factor, and load-bearing orientation index corresponding to each aggregate morphology combination Input the finite element model for the construction phase The measured road surface deflection, measured stress, or design allowable value are used as the target response, and the structural parameter vector is solved by grid search or least squares fitting. This enables the finite element model to be used during the construction phase. The error between the output response and the target response is minimized. After calibration, each aggregate morphology combination and each permeable concrete pavement segment corresponds to a structural parameter vector, thus providing input samples for association rule mining.

[0058] 9) Combine aggregate morphology range, load-bearing guidance level, and structural parameter range into transaction items, and mine the structural parameter prediction rule base. , It refers to a queryable mapping set composed of association rules, which is used to quickly output the range of structural parameter values ​​based on aggregate morphology parameters and load-bearing orientation index.

[0059] In practical implementation, the average edge angle, needle-like and flaky index, aggregate volume fraction, concrete stiffness reduction factor, interlayer shear correction factor, compaction sensitivity coefficient, and temporary construction load amplification factor are first discretized into several value intervals. Then, the intervals corresponding to the average edge angle, needle-like and flaky index, aggregate volume fraction, load-bearing orientation level, concrete stiffness reduction factor, interlayer shear correction factor, compaction sensitivity coefficient, and temporary construction load amplification factor for each aggregate morphology combination are combined into a transaction sample. Finally, association rule mining algorithms such as the Apriori algorithm or the frequent pattern growth algorithm (FP-Growth algorithm) are used to generate "aggregate morphology intervals and load-bearing orientation levels". Association rules for "structural parameter range".

[0060] Furthermore, to eliminate weak rules with insufficient statistical reliability, the minimum support for association rules can be set to 0.08, the minimum confidence to 0.75, and the minimum lift to 1.20. Support measures the proportion of a rule's occurrence in all transaction samples; confidence measures the probability that the consequent will be true when the antecedent condition is met; and lift measures the reinforcing effect of the antecedent condition on the consequent result. Only rules that simultaneously meet the requirements of minimum support, minimum confidence, and minimum lift are included in the structural parameter prediction rule base. .

[0061] In one embodiment, as an example, when the average edge angle is [0.55, 0.70], the flaky index is [0.08, 0.16], the aggregate volume fraction is [0.62, 0.68], and the load-bearing guidance level is high guidance level, the structural parameter prediction rule base is as follows: It can output the value ranges of concrete stiffness reduction factor [0.86, 0.95], interlayer shear correction factor [0.90, 1.00], compaction sensitivity coefficient [0.96, 1.04], and temporary construction load amplification factor [1.00, 1.10], thereby establishing a queryable mapping relationship between aggregate particle contact structure and pavement section mechanical response.

[0062] In one embodiment, the analysis of the structural parameter prediction rule base association rule bubble chart, such as... Figure 4 As shown, the distribution of association rules across the three dimensions of support, confidence, and lift is illustrated. The horizontal axis represents support (dimensionless), and the vertical axis represents confidence (dimensionless). The size of the bubble indicates the level of support, and the color of the bubble from purple to yellow indicates the increase in lift (dimensionless). The experiment shows that most rules have a support higher than 0.08, a confidence higher than 0.75, and a lift greater than 1.20, meeting the minimum rule selection criteria.

[0063] It should be noted that the structural parameter prediction rule base is not only used to provide the range of structural parameter values, but also to subsequently determine whether individual frogs in the swarm fall into a feasible parameter region with physical basis. When a certain aggregate shape combination does not match any rule that meets the confidence requirement, this aggregate shape combination is not directly used as a valid initialization result. It can be processed through neighboring rule interval correction, local linear interpolation, or resampling, thereby avoiding inputting parameters lacking calibration basis into the finite element model during the construction stage.

[0064] S202, Frog Swarm Constraint Initialization and Structural Parameter Mapping Conventional hybrid frog-jumping algorithms randomly generate parameters for each pavement segment during initialization, which can easily lead to abrupt changes in aggregate gradation, load-bearing guidance index, or structural parameters between adjacent permeable concrete pavement segments, resulting in unreasonable stiffness jumps in the finite element model during construction. This step generates aggregate morphology parameters for each permeable concrete pavement segment segment by segment along the continuous paving and compaction construction sequence, and maps these parameters to structural parameter vectors using a structural parameter prediction rule base. Simultaneously, it constrains the differences in load-bearing guidance index between adjacent permeable concrete pavement segments, the overlap area of ​​particle size distribution curves, and the gradient of structural parameter changes. This ensures that the initial frog swarm individuals simultaneously meet the requirements of continuous aggregate gradation, stable stiffness changes, and finite element calculation feasibility. The specific steps are as follows: 1) Define the first Complete encoding vector of each frog swarm individual , It refers to the candidate solution encoding that simultaneously includes the aggregate morphology parameter vector, particle size distribution curve, load-bearing guidance index, and structural parameter vector of all permeable concrete pavement sections. It is the basic object for fitness evaluation and iterative update of the hybrid frog-jumping algorithm.

[0065] Specifically, the complete encoded vector It can be represented as ;in, This represents the individual index of the frog swarm, used to distinguish different candidate solutions in the initial population, with a value ranging from 1 to... ; This represents the total number of frogs in the swarm, used to determine the initial search size; a value of 60 is acceptable. Indicates the first The first individual in the frog pod The aggregate morphology parameter vector of a permeable concrete pavement segment is used to characterize the average edge angle, needle-like index and aggregate volume fraction of the permeable concrete pavement segment. Indicates the first The first individual in the frog pod Particle size distribution curve of a permeable concrete pavement section; Indicates the first The first individual in the frog pod The load-bearing orientation index of a permeable concrete pavement section is used to measure the degree of consistency between the aggregate force chain network of the permeable concrete pavement section and the target force transmission direction. Indicates the first Among the individuals in the frog pod, the first The structural parameter vector of a permeable concrete pavement segment is used to characterize the equivalent mechanical parameters of the permeable concrete pavement segment in the finite element calculation during the construction phase.

[0066] 2) Define the first Among the individuals in the frog pod, the first Structural parameter vector of a permeable concrete pavement section And standardize the range of values ​​for structural parameters.

[0067] Specifically, the structure parameter vector It can be represented as ;in, Indicates the first Among the individuals in the frog pod, the first The concrete stiffness reduction factor for a permeable concrete pavement section is used to characterize the degree of reduction in the equivalent elastic stiffness of permeable concrete relative to the stiffness of the reference concrete, and can be taken as 0.70 to 1.00. Indicates the first Among the individuals in the frog pod, the first The interlayer shear correction factor for a permeable concrete pavement section is used to correct the interfacial shear stiffness between the pavement layer and the base layer or adjacent pavement layers to reflect the influence of aggregate interlocking on interlayer shear performance. It can be taken as 0.75 to 1.10. Indicates the first Among the individuals in the frog pod, the first The compaction sensitivity coefficient of a permeable concrete pavement section is used to describe the ease with which permeable concrete reaches the target porosity and design strength under vibration compaction load, and can be taken as 0.90 to 1.10. Indicates the first Among the individuals in the frog pod, the first The temporary construction load amplification factor for a permeable concrete pavement section is used to characterize the adverse amplification effect of the roller's self-weight, vibration load, paver force, temperature gradient load during curing, and interlayer constraint load on the load during the construction stage. It can be taken as 0.90 to 1.20.

[0068] 3) Based on the structural parameter prediction rule base The high-confidence rules in the initialization process initialize the aggregate morphology parameters of the first permeable concrete pavement section.

[0069] In one implementation, for each individual frog in the swarm, the rule base for predicting structural parameters is first used. High-confidence rules were selected, defined as rules with a confidence level of not less than 0.75 and a lift of not less than 1.20. Then, random sampling was performed within the mean angularity interval, needle-like / flaky index interval, and aggregate volume fraction interval corresponding to the high-confidence rules to obtain the aggregate morphology parameter vector for the first permeable concrete pavement section. .

[0070] In practical implementation, if multiple high-confidence rules can be used for the initialization of the first pavement segment, the rule with a higher load-bearing guidance level, a more stable concrete stiffness reduction factor range, and a lower temporary construction load amplification factor should be prioritized. If it is necessary to maintain the diversity of the initial frog swarm, the rules can be randomly selected based on a comprehensive weight composed of rule confidence and lift. Based on this, the first permeable concrete pavement segment can be generated from the physically feasible area, avoiding random deviations in the parameters of the first segment from affecting the selection and calculation of parameters for subsequent permeable concrete pavement segments.

[0071] 4) Based on the similarity propagation of the load-bearing guidance index, the second to the third are generated according to the continuous paving and compaction construction sequence. Aggregate morphology parameters of a permeable concrete pavement section.

[0072] In one implementation, for the first... For each permeable concrete pavement section, first consider the load-bearing guiding index of the previous permeable concrete pavement section. For reference, candidate aggregate morphology combinations with a load-bearing orientation index difference not exceeding the maximum allowable difference are selected from the aggregate morphology combination sample set; then, candidate aggregate morphology combinations that meet the continuity requirements in terms of average edge angle variation, needle-like / flaky index variation, aggregate volume fraction variation, and particle size distribution curve variation are further selected; finally, the remaining candidate aggregate morphology combinations are randomly selected to obtain the first... Aggregate morphology parameter vector of a permeable concrete pavement section .

[0073] In practical implementation, the continuous constraint of the bearing guidance index between adjacent permeable concrete pavement sections can be expressed as: ;in, This represents the maximum allowable difference in the load-bearing guidance index between adjacent permeable concrete pavement sections. It is used to limit abrupt changes in the load-bearing guidance performance of aggregates along the continuous paving and compaction direction, and can be taken as 0.08. Indicates the first Among the individuals in the frog pod, the first The load-bearing guidance index of a permeable concrete pavement section; This indicates the operation of taking the absolute value.

[0074] Furthermore, the continuity of the particle size distribution curve can be determined by the overlapping area of ​​the particle size distribution curves of adjacent permeable concrete pavement sections. It can be obtained by summing the minimum volume fractions of two adjacent paving sections in each sieve aperture interval. The calculation method is expressed as follows: ; in, Indicates the first The first individual in the frog pod The first permeable concrete pavement section and the first The overlapping area of ​​the particle size distribution curves between permeable concrete pavement sections is used to characterize the smoothness of the gradation transition. Indicates the first The first individual in the frog pod The permeable concrete pavement section is in the first The volume fraction of aggregate particles within each sieve aperture interval; Indicates the first The first individual in the frog pod The permeable concrete pavement section is in the first The volume fraction of aggregate particles within each sieve aperture interval; This represents a function that takes the minimum value. In one implementation, when... When the gradation is not less than 0.70, the transition of gradation between adjacent permeable concrete pavement sections is deemed to meet the initialization requirements.

[0075] In one embodiment, as an example, suppose a frog pod has generated aggregate morphology parameters for the second paving section, and its load-bearing guidance index... , This is currently the third candidate combination for pavement section selection. In the aggregate morphology combination sample set, only combinations with the load-bearing guidance index are retained. The combinations within the specified range are then excluded; based on this, combinations with average edge angle variations exceeding 0.10, needle-like / flaky index variations exceeding 0.06, and aggregate volume fraction variations exceeding 0.05 are removed; finally, the overlapping area of ​​particle size distribution is examined. Finally, seven candidate combinations remained. Combinations with higher load-bearing capacity were selected based on their load-bearing capacity, resulting in the aggregate morphology parameter vector for the third pavement section. .

[0076] 5) Utilize the structural parameter prediction rule base The aggregate morphology parameters are mapped to a structural parameter vector.

[0077] Specifically, for the first The first individual in the frog pod A permeable concrete pavement section, with aggregate morphology parameter vectors Particle size distribution curve and carrying capacity index Input the structure parameter mapping function to obtain the structure parameter vector. The structural parameter mapping function refers to the function derived from the structural parameter prediction rule base. The parameter mapping function is formed together with local linear interpolation.

[0078] In practical implementation, the structural parameter mapping function maps the aggregate morphology parameter vector. Particle size distribution curve and carrying capacity index With structural parameter prediction rule base Perform a matching process. If the aggregate morphology parameter vector and load-bearing guidance level match the structural parameter prediction rule base... If a rule is applied, structural parameters are directly extracted from the corresponding structural parameter interval. When extracting, priority can be given to extracting parameters close to the midpoint of the interval to improve the stability of the initial finite element calculation. If the same aggregate shape parameter vector and load-bearing guidance level match multiple rules, the rule with the highest confidence and the highest lift is selected first. If the aggregate shape parameter vector and load-bearing guidance level do not directly match a rule but fall between the intervals covered by two adjacent rules, the two or more rules closest to each other are first determined, and then local linear interpolation is performed according to the distance of the aggregate shape parameters to obtain the structural parameter estimation results. If the aggregate shape parameters completely deviate from the rule coverage interval, they are guided to be corrected to the nearest rule interval and then remapped.

[0079] In practical implementation, for cases where the aggregate morphology parameter vector and bearing capacity guidance level do not directly match the rules but fall between the coverage areas of two adjacent rules, let... This represents the candidate rule index, with a value range of 1 to... ; Indicates the number of recent candidate rules; This represents the structural parameter category index, used to distinguish between concrete stiffness reduction factor, inter-story shear correction factor, compaction sensitivity coefficient, and temporary construction load amplification factor. This indicates the relationship between the current aggregate morphology parameter vector and the first... The normalized distance between the centers of the antecedent intervals of the candidate rules is calculated by first mapping the average edge angle, needle-like index, and aggregate volume fraction to their respective value ranges for the current aggregate morphology parameter vector. The interval is used to form a normalized eigenvector, and then the eigenvector is compared with the first interval. The Euclidean distance between the normalized feature vectors corresponding to the centers of the antecedent intervals of the candidate rules is used as the normalized distance. ; Indicates the first The first candidate rule The center value of the class structure parameter interval, for the first Candidate rules in the 1st The value range given in the class structure parameter is taken as the arithmetic mean of the upper and lower limits of the range as the center value of the range. Then the first Interpolation results of class structure parameters The calculation method is expressed as follows: ; in, Indicates the first Among the individuals in the frog pod, the first The first permeable concrete pavement section Interpolation results of class structure parameters; This represents the interpolation stability constant, used to prevent denominator anomalies when the distance is 0. It can be taken as... By using inverse distance weighting, rules that are closer to the current aggregate morphology parameters can have a greater impact on the mapping of structural parameters.

[0080] In one embodiment, for example, if the first If the average edge angle of a permeable concrete pavement section is 0.62, the flaky / neat index is 0.12, the aggregate volume fraction is 0.65, and the load-bearing guidance level is high guidance level, then query the structural parameter prediction rule base. Then, the intervals for concrete stiffness reduction factor [0.86, 0.95], inter-layer shear correction factor [0.90, 1.00], compaction sensitivity coefficient [0.96, 1.04], and temporary construction load amplification factor [1.00, 1.10] can be obtained. The structural parameter vector can then be derived from the midpoints of these intervals. .

[0081] Furthermore, define For the first The first individual in the frog pod The first permeable concrete pavement section Class structural parameters. If the aggregate morphology parameter vector and load-bearing guidance level match the structural parameter prediction rule base. The rules in the code are extracted directly from the rule range. If the rule is not directly matched but falls between the coverage areas of two adjacent rules, the interpolation result will be... As the final structural parameters .

[0082] 6) Constrain the gradient of structural parameter changes between adjacent permeable concrete pavement sections. The gradient constraint of structural parameter changes is used to prevent unreasonable jumps in the stiffness, compaction response or construction load parameters of local pavement sections.

[0083] In specific implementation, define For the first Given the maximum allowable gradient of structural parameters between adjacent pavement sections, the constraint on the gradient of structural parameters between adjacent permeable concrete pavement sections can be expressed as follows: ;in, Indicates the first The first individual in the frog pod The first permeable concrete pavement section Structural parameters; concrete stiffness reduction factor A value of 0.06 can be taken, corresponding to the interlayer shear correction factor. A value of 0.08 can be taken, which corresponds to the compaction sensitivity coefficient. A value of 0.05 can be taken, which corresponds to the temporary construction load amplification factor. 0.10 is acceptable; This indicates the operation of taking the absolute value.

[0084] Furthermore, if a structural parameter exceeds the maximum allowable change gradient of an adjacent pavement section, it will be preferentially included in the structural parameter prediction rule base. Within the interval, the structural parameter is allowed to be corrected to a feasible value close to the structural parameter of the adjacent permeable concrete pavement section; if the correction still fails to meet the rule base constraints, the aggregate morphology parameter of the permeable concrete pavement section is reselected; if reselection is still not feasible after multiple consecutive attempts, the random sampling range is reduced and local linear interpolation is used to generate candidate structural parameters.

[0085] 7) Repeat the above process of generating paving segments, mapping structural parameters, and correcting gradient constraints to form an initial swarm of frogs that meets physical constraints.

[0086] Specifically, the process of generating paving segments, mapping structural parameters, and correcting gradient constraints is repeated for each individual frog in the swarm until the desired result is obtained. A complete encoded vector constitutes the initial frog swarm. , This refers to the initial frog swarm set in the hybrid frog jumping algorithm.

[0087] It should be noted that each individual frog in the initial frog swarm contains a complete chain of aggregate morphology parameters, a chain of particle size distribution curves, a chain of load-bearing guidance indices, and a chain of structural parameters. Furthermore, adjacent permeable concrete pavement sections satisfy continuous constraints on the load-bearing guidance index, particle size distribution, and structural parameter gradients. Therefore, these constraints can be directly input into the finite element model for the construction stage to calculate stress residuals. Based on this, the problem of numerous infeasible frog swarm individuals and abrupt changes in non-physical stiffness caused by conventional random initialization can be avoided, thus improving the convergence stability of the hybrid frog-jump optimization.

[0088] S3, Optimization of Mixed Frog Jumping Aggregate Morphology In the process of optimizing the aggregate shape in permeable concrete pavement sections, the conventional hybrid frog-jumping algorithm mainly relies on fitness sorting and random jump updates, which has problems such as unstable local search direction, inaccurate location of abnormal pavement sections, and low efficiency of global information reorganization.

[0089] This step maps the load-bearing guidance lifting direction of the force chain direction field to the guiding vector of aggregate morphology parameters, and embeds the abnormal sparsity perception into the worst-case individual update and global reorganization process, so that the frog swarm search is simultaneously subject to the joint constraints of load-bearing guidance performance, construction stage stress residuals and abnormal pavement sparsity. The specific steps are as follows: S301, Local Evolution of Subpopulations In the local evolutionary process of the worst individual learning from the best individual in a regular subpopulation, the lack of distinction between stress residual concentration paving sections and normal paving sections can easily lead to problems with simultaneous and significant adjustments to parameters. This step selects the frog individual with the worst fitness within each subpopulation as the update target, and comprehensively utilizes the learning direction of the best individual in the subpopulation, local feasible region constraints, the correction direction of the anomalous sparse penalty function, and the load-bearing guidance direction of the force chain direction field to simultaneously update the aggregate morphological and structural parameters. This guides the search process towards a direction of low stress residuals, high load-bearing capacity, and concentrated correction of a small number of anomalous paving sections. The specific steps are as follows: 1) Before performing local evolution, the fitness value of each frog swarm is calculated based on the finite element model calculation results during the construction phase, the load-bearing guidance index of each paving section, and the distribution of abnormal paving sections.

[0090] Specifically, the first The complete encoding vector for each individual frog in the swarm Input the finite element model for the construction stage, and calculate the maximum tensile stress, maximum compressive stress, maximum deflection, and bottom tensile stress of each permeable concrete pavement segment at each construction stage according to the continuous paving and compaction construction sequence. Compare the calculation results with design allowable values, monitoring values, or target control values ​​to obtain the stress residuals and deflection residuals for each construction stage. Then, combine the mean load-bearing guidance index, the degree of violation of structural parameter gradients between adjacent pavement segments, and the number of abnormal pavement segments to form a fitness value. A lower fitness value indicates a better aggregate morphology and structural parameter combination for the individual frog in the swarm. The calculation method is expressed as follows: ; in, Indicates the first Complete encoding vector of each frog swarm fitness value; Indicates the first Normalized composite value of stress residuals for each individual frog swarm during the construction stage; Indicates the first Normalized composite value of deflection residuals for each individual frog in a swarm; Indicates the first The degree of gradient violation of structural parameters for each individual frog in the swarm; Indicates the first Normalized number of abnormally permeable concrete pavement sections corresponding to each frog swarm. Indicates the first The average load-bearing guidance index of all permeable concrete pavement sections for each frog pod; This represents the stress residual weight, which can be set to 0.40. This represents the weight of the deflection residual, which can be set to 0.20. This indicates a gradient violation weight, which can be set to 0.15. This indicates the weight of the number of abnormal paving sections, which can be set to 0.15. This represents the load-bearing-oriented reward weight, which can be set to 0.10. By adding a load-bearing-oriented reward term to the fitness value, the optimization results can maintain the load-bearing path of the aggregate contact force chain network along the target force transmission direction while pursuing the reduction of finite element residuals.

[0091] 2) Sort and group the initial frog population according to their fitness values.

[0092] In one implementation, all frog individuals in the population are sorted from best to worst fitness value, and then a staggered allocation method is used (i.e., the first-ranked individual is assigned to subpopulation 1, the second-ranked individual to subpopulation 2, ..., and so on). Names are assigned to subpopulations. , No. (The individuals are then redistributed to subpopulation 1, and so on until all individuals have been assigned.) Individual populations. Among them, This represents the total number of subpopulations and is used to control the number of groups for parallel local search; it can be set to 6. This represents the subpopulation index, used to distinguish different local search groups in the hybrid frog leaping algorithm, with a value ranging from 1 to... .

[0093] Furthermore, in the From the subpopulation, the frog individual with the best fitness is selected as the optimal individual, and its complete encoding vector is defined as follows: The frog individual with the worst fitness is selected as the worst individual, and its complete encoding vector is defined as follows: .in, Used to provide the search direction with the best fitness within the current subpopulation; Used for receiving subpopulation learning, anomaly correction, and force chain-guided updates.

[0094] 3) Based on the difference between the complete encoding vectors of the best and worst individuals, and through amplitude truncation and constraint checks using the local constraint maintenance operator, the local update direction for the worst individual to learn from the best individual is obtained.

[0095] Specifically, to prevent the worst-case individual from jumping directly along the global straight line, causing abrupt changes in parameters of adjacent permeable concrete pavement sections, a local constraint maintenance operator is introduced. This local constraint maintenance operator is a modified operator that restricts the basic learning direction vector within a local linear space formed by adjacent feasible individuals. It is used to maintain aggregate morphology continuity constraints, structural parameter gradient constraints, and parameter upper and lower bound constraints. The input to the local constraint maintenance operator is the basic learning direction vector. The basic learning direction vector is used to characterize the changing trend of aggregate morphological parameters and structural parameters of each permeable concrete pavement section as the worst individual moves closer to the best individual. The output is a feasible learning direction vector after being corrected by continuity and boundary constraints.

[0096] In its implementation, the local constraint maintenance operator first sets a maximum single change magnitude for each parameter component in the basic learning direction (for example, it can be 10% of the parameter's value range); then it checks whether this change direction will cause the difference in the bearing capacity guidance index between adjacent permeable concrete pavement sections to exceed [a certain value]. Then check whether changes in structural parameters will cause any If a violation occurs, the corresponding directional component is reduced to the boundary allowable value, or the average parameter of adjacent permeable concrete pavement sections is used for smoothing correction. Based on this, in the process of the worst individual learning the effective trend of the best individual, the local constraint maintenance operator can constrain the parameter update of the worst individual, so that it does not deviate from the feasible parameter domain.

[0097] In one embodiment, as an example, suppose the average edge angle of the 5th paved segment for the worst-performing individual is 0.58, while the average for the best-performing individual is 0.71, with a difference of +0.13. This exceeds the maximum single change range of 0.06 for this parameter. Therefore, the local constraint maintenance operator truncates the component of the basic learning direction in this dimension to +0.06. Simultaneously, it checks the difference in the load-bearing guidance index between adjacent paved segments after the change. If it still exceeds... If the cutoff value is further reduced, the corrected learning direction will be output.

[0098] 4) Generate anomaly indicator vectors based on the median of stress residuals and deflection residuals of each pavement section calculated from the finite element model during the construction phase, as well as the degree of deviation from the structural parameters of adjacent pavement sections. , It refers to the first The anomaly parameter dimension label vector corresponding to the worst individual in the subpopulation is used to mark the permeable concrete pavement section that needs key correction and its corresponding parameter dimension. The dimension and the complete encoded vector are used to mark the anomaly parameter dimension. Consistent, only elements at the corresponding positions of structural parameters are set to 0 or 1, while elements at the corresponding positions of aggregate morphology parameters, particle size distribution, and load-bearing orientation index are all set to 0.

[0099] In the specific implementation, the worst individual's complete encoding vector is first... Input the finite element model for the construction phase Calculate the maximum tensile stress residual, maximum compressive stress residual, maximum deflection residual, and bottom tensile stress residual for each permeable concrete pavement section; then, calculate the median of the comprehensive residual values ​​for all permeable concrete pavement sections corresponding to the worst individual, denoted as [missing information]. If the first The residual composite value of the permeable concrete pavement section exceeded Furthermore, if the deviation of a certain structural parameter of the permeable concrete pavement section relative to the average value of adjacent pavement sections exceeds a deviation threshold, then the first... The structural parameter dimension corresponding to each permeable concrete pavement section is indicated by the vector at the abnormal location. The value is marked as 1 if it is not in the middle, and 0 otherwise; where, This indicates the screening factor for abnormal pavement sections, which can be set to 1.5. Based on this, the search resources can be concentrated on a small number of permeable concrete pavement sections with prominent stress residuals, avoiding convergence oscillations caused by simultaneous large-scale adjustments to all permeable concrete pavement sections.

[0100] In practical implementation, the average value of adjacent paved sections can be determined according to the [further details needed]. The first permeable concrete pavement section and the first Calculation of the average values ​​of similar structural parameters for each permeable concrete pavement section; for the first permeable concrete pavement section and the second... For a permeable concrete pavement section, the structural parameters of the same type of permeable concrete pavement section on the adjacent side can be used as reference values. The deviation threshold for the concrete stiffness reduction factor can be 0.06, the deviation threshold for the interlayer shear correction factor can be 0.08, the deviation threshold for the compaction sensitivity coefficient can be 0.05, and the deviation threshold for the temporary construction load amplification factor can be 0.10.

[0101] 5) Construct an abnormal sparse penalty function based on the smoothed absolute deviation between the structural parameters of each paving segment of the worst individual and the mean of the adjacent paving segments, and calculate the correction direction of the abnormal sparse penalty function.

[0102] Specifically, the anomaly sparsity penalty function is a penalty function used to characterize the degree of abrupt changes in the structural parameters of permeable concrete pavement sections relative to the neighborhood mean, which can concentrate the abnormal parts as much as possible in a small number of permeable concrete pavement sections; the input of the anomaly sparsity penalty function is the worst individual complete encoding vector. The output is a positive scalar penalty function value. The larger the penalty function value, the more significant the abrupt change in structural parameters between adjacent pavement sections.

[0103] In practical implementation, the abnormal sparsity penalty function for the first... The first permeable concrete pavement section For structural parameters, first calculate the deviation between the structural parameter and the mean of adjacent pavement sections; then use a smoothing absolute value function to measure the magnitude of the deviation; finally, sum the values ​​for all permeable concrete pavement sections and all structural parameter categories to obtain the abnormal sparsity penalty function value, which is calculated as follows: ; in, Represents the complete encoding vector of the worst individual. The corresponding abnormally sparse penalty function value; Indicates the first The worst individual in the subpopulation was in the [number]th [period]. The first permeable concrete pavement section Class structure parameters; Indicates the first The worst individual in the subpopulation was in the [number]th [period]. The first permeable concrete pavement section The average value of adjacent paved sections corresponding to the structural parameters; This represents the smoothing stability constant, used to make the absolute value metric calculate smoothly near 0. It can be taken as... .

[0104] Furthermore, the correction direction of the anomaly sparsity penalty function with respect to structural parameters can be determined by the deviation sign. If a structural parameter is greater than the average of adjacent pavement segments, the correction direction decreases the structural parameter; if a structural parameter is less than the average of adjacent pavement segments, the correction direction increases the structural parameter; if a structural parameter is close to the average of adjacent pavement segments, the correction direction approaches 0. This is achieved by comparing the anomaly location indicator vector... Element-wise multiplication results in the correction direction of the anomalous sparsity penalty function, which only applies to the dimension of the marked anomalous parameter.

[0105] 6) Obtain the gradient direction vector based on the gradient components of each parameter dimension of the anomalous sparsity penalty function. .

[0106] In one implementation, an abnormally sparse penalty function is used to encode the complete vector of the worst individual. The Middle The first permeable concrete pavement section Class structure parameters Taking the partial derivative, we can obtain the gradient components along the dimension of this parameter. The calculation method is expressed as follows: ; in, The abnormally sparse penalty function represents the first... The worst individual in the subpopulation was in the [number]th [period]. On the first permeable concrete pavement section The partial derivatives of a class structure parameter are used to indicate the direction and relative strength of correction when the parameter deviates from the neighborhood mean.

[0107] Furthermore, the gradient components of all permeable concrete pavement sections and all structural parameter categories are correlated with the complete encoded vector. Assemble using a consistent dimensional structure and encode the complete vector. The gradient components corresponding to the dimensions of the structural parameters in the non-agricultural part (i.e., the corresponding positions of aggregate morphology parameters, particle size distribution, and load-bearing orientation index) are set to... This yields the gradient direction vector. , Refers to the abnormally sparse penalty function Complete encoding vector of the worst individual The gradient direction vector, whose dimension is the same as the complete encoded vector. Consistent, only taking non-zero values ​​at the corresponding positions of structural parameters.

[0108] 7) Calculate the spatial guiding vector of the force chain based on the load-bearing guidance sensitivity of aggregate morphological parameters according to the load-bearing guidance index. , It refers to the first The load-bearing orientation enhancement direction vector corresponding to the worst individual in the subpopulation is used to push the low load-bearing orientation aggregate morphological parameter combination to the parameter region corresponding to a higher load-bearing orientation index. Dimensions and complete encoding vectors are also included. Consistent.

[0109] In one implementation, for each permeable concrete pavement segment in the worst-case individual, small perturbations are applied to the average edge angle, flaky index, and aggregate volume fraction, respectively. The perturbation amplitude can be 1% of the corresponding parameter value range. After each perturbation, the change in the load-bearing guidance index after the perturbation is estimated using a load-bearing guidance index interpolation function. Then, the change in the load-bearing guidance index is divided by the perturbation amplitude to obtain the load-bearing guidance sensitivity of the load-bearing guidance index to this aggregate morphology parameter. The direction that can improve the sensitivity of the load-bearing guidance index is then taken as the guiding direction of the aggregate morphology parameter.

[0110] In specific implementation, let's assume Indicates the category of aggregate morphology parameters. These correspond to the average angularity, needle-like and flaky index, and aggregate volume fraction, respectively. Indicates the first The amplitude of disturbance of aggregate morphological parameters; Indicates the first The unit perturbation direction corresponding to the aggregate morphological parameters; definition This represents the load-bearing guidance index interpolation function, used to estimate the load-bearing guidance index of samples that are not directly calculated based on existing discrete element samples. Its inputs are the aggregate morphology parameter vector of a pavement segment and the pavement segment index. The output is the estimated load-bearing guidance index of that parameter vector under the target force transmission direction of that pavement segment. The load-bearing guidance index interpolation function uses the load-bearing guidance index of the discrete element samples as data points. It can use Kriging interpolation (a method of interpolation that performs optimal unbiased estimation at unknown points based on spatially distributed data, calculating the estimated value by weighting the spatial correlation between known points) to quickly estimate any new combination of aggregate morphology parameters, thus obtaining the load-bearing guidance index of the new parameter combination. Then the... The first permeable concrete pavement section Load-bearing orientation sensitivity of aggregate morphological parameters It can be represented as: ; in, Indicates the first The first permeable concrete pavement section Local sensitivity of aggregate morphology parameters to load-bearing orientation index; Indicates the first The worst individual in the subpopulation was in the [number]th [period]. The aggregate morphology parameter vector of a permeable concrete pavement section. If A positive value indicates that increasing the morphological parameters of this type of aggregate is beneficial to improving the load-bearing orientation index; if... A negative value indicates that reducing the morphological parameters of this type of aggregate is beneficial to improving the load-bearing orientation index.

[0111] Furthermore, the guiding direction of the structural parameter components can be synchronously obtained through the structural parameter mapping function. That is, when the aggregate morphology parameters change along the load-bearing guidance lifting direction, the structural parameter prediction rule base is re-queried. Alternatively, local linear interpolation can be performed to obtain a new structural parameter vector, and the difference between the new structural parameter vector and the original structural parameter vector can be used as the guiding direction of the structural parameter components.

[0112] It should be noted that the force chain spatial guiding vector does not simply increase or decrease a single aggregate morphology parameter, but rather guides parameters differently based on the local sensitivity of the load-bearing guidance index. When an increase in the needle-like and flaky index of a permeable concrete pavement section leads to a decrease in the load-bearing guidance index, the force chain spatial guiding vector indicates a decreasing direction in the needle-like and flaky index dimension of that permeable concrete pavement section; conversely, when appropriately increasing the average edge angle can enhance contact and interlocking and improve the load-bearing guidance index, the force chain spatial guiding vector indicates an increasing direction in the average edge angle dimension of that permeable concrete pavement section.

[0113] 8) Perform local evolutionary updates on the worst individual based on the learning direction of the best individual in the subpopulation, the abnormal sparsity correction direction, and the force chain space guidance direction.

[0114] In one implementation, the worst-case individual's complete encoding vector is updated based on the optimal individual learning direction, the abnormal sparsity correction direction, and the force chain space guidance direction, resulting in the worst-case individual's updated encoding vector. The local evolutionary update method is represented as: ; in, It refers to the first The candidate solutions formed by the worst individual in the subpopulation after local evolution are used to participate in the next round of fitness evaluation. This represents the adaptive evolutionary step size, used to control the update magnitude of the worst individual learning from the best individual, and can be taken as 0.15 to 0.65; This represents the abnormal sparse perception coefficient, used to adjust the influence of the abnormal paving section parameter correction term on the update result, and can be taken from 0.05 to 0.25; This represents the force chain guiding strength, used to adjust the influence of the force chain direction field on parameter updates, and can be taken as 0.02 to 0.20; This represents the element-wise multiplication operator, used to ensure that the correction direction of the anomaly sparsity penalty function applies only to the parameter dimension marked by the anomaly indicator vector. The feasible parameter domain is determined by the upper and lower bounds of aggregate morphology parameters, the upper and lower bounds of structural parameters, the continuous constraint of load-bearing guidance, the gradient constraint of structural parameters of adjacent pavement sections, and the matching conditions of the rule base. This represents the feasible region projection operator.

[0115] It should be noted that, The term is a fundamental term that maintains the current solution, representing the term of the first term. The current complete encoding vector of the worst individual in each subpopulation provides a starting point for the update, ensuring that the update is a correction based on the existing solution rather than a complete replacement. The term is an evolutionary term within a subpopulation that learns from the best individual. By calculating the direction of difference between the worst and best individuals and correcting it with a local constraint maintenance operator, it guides the aggregate morphology and structural parameters of the worst individual to move in a direction with better fitness. Used to control the learning step size of this item; The term is an anomaly sparse perception correction term, which utilizes the anomaly location indicator vector. The outlier parameter dimensions that require key correction are identified, and then these dimensions are adjusted along the negative gradient direction of the outlier sparsity penalty function to make the structural parameters converge towards the mean of adjacent pavement segments. This suppresses abrupt changes in local parameters and makes the distribution of outlier pavement segments sparser. Used to control the strength of the correction for this item; The term is the force chain direction field guiding term. Based on the sensitivity analysis results of the load-bearing guidance index on aggregate morphology parameters, the aggregate morphology parameters are pushed towards the parameter region that is more conducive to forming an effective force chain load-bearing path. At the same time, the corresponding structural parameters are updated in conjunction, so that the search process takes into account the improvement of load-bearing guidance performance. This is used to control the guiding strength of this item; after completing the weighted superposition of each item, the updated encoding vector is subjected to parameter upper and lower bound truncation, load-bearing guiding continuity constraint correction, structural parameter gradient constraint correction, and rule base matching correction to ensure that the output candidate solution is corrected. Completely within the feasible parameter domain Inside.

[0116] 9) The worst individual is corrected and the encoding vector is updated by the feasible region projection operator.

[0117] Specifically, the feasible region projection operator is used to correct the parameters updated by local evolution to the feasible parameter domain. Within this framework, the worst-performing individual is guaranteed to update the encoding vector. Finite element models that can be used during the construction phase Direct computation takes the code vector to be corrected as input and outputs the corrected version that satisfies the feasible parameter domain. The encoded vector.

[0118] In practical implementation, the feasible region projection operator first performs upper and lower bound truncation on the average edge angle, needle-like index, aggregate volume fraction, and various structural parameters; then it recalculates or queries the load-bearing guidance index of each permeable concrete pavement section and checks whether the difference in load-bearing guidance index between adjacent permeable concrete pavement sections exceeds [the specified threshold]. Next, check whether the gradient of structural parameter changes exceeds the corresponding... If the load-bearing guidance continuity constraint is violated, the aggregate morphology parameters of the adjacent permeable concrete pavement section will be incorporated into the structural parameter prediction rule base. The nearest feasible aggregate morphology range is shifted; if the structural parameter gradient constraint is violated, the average structural parameter of adjacent pavement sections and the structural parameter prediction rule base are used. Allow intervals to smoothly correct structural parameters; finally, re-execute the structural parameter mapping function to ensure that the aggregate morphology parameters are consistent with the structural parameters.

[0119] In one embodiment, as an example, suppose that after an individual update, the interlayer shear correction factor for the 6th pavement segment is 1.15, which exceeds the upper limit of 1.10. Therefore, it is first truncated to 1.10; then, the difference between this parameter and that of the 5th pavement segment is checked, which is 0.12, exceeding... If so, adjust the parameter of the 6th paving section to 1.02; then check its load-bearing guidance index and rule base matching, fine-tune the aggregate morphology parameters and remap them until all constraints are met.

[0120] It should be noted that by jointly weighting and updating the optimal individual learning direction, the abnormal sparsity correction direction, and the force chain space guidance direction, the update of the worst individual can simultaneously consider the optimal individual of the subpopulation, the abnormal paving segment sparsity correction, and the force chain bearing guidance direction, thereby reducing invalid searches and improving the mechanical rationality and convergence stability of local evolution.

[0121] In one embodiment, the fitness evolution curve of the hybrid frog-jumping algorithm is analyzed using a broken line graph, such as... Figure 5 As shown, the trend of the optimal fitness value and the average fitness value of the frog swarm during the iterative optimization process is displayed; the horizontal axis is the number of iterations (dimensionless), and the vertical axis is the fitness value (dimensionless). The experiment shows that the optimal fitness value basically decreases and slows down after about 80 generations. The hybrid frog jumping algorithm with embedded force chain guidance and abnormal sparse perception has good convergence stability and global search ability.

[0122] S302, Global Information Reassembly After all subpopulations have completed local evolution, simply reordering and randomly mixing them based on fitness can easily lead to the loss of locally effective correction fragments already formed in different subpopulations. This step merges individuals from each subpopulation into a global candidate pool and selects complementary candidate individuals based on aggregate bearing capacity guidance sequence, particle size distribution continuity, and differences in anomalous locations. Then, templated crossover generates offspring encoding vectors, enabling the global search to introduce differentiated anomaly correction patterns while maintaining the synergy of paving section bearing capacity guidance. The specific steps are as follows: 1) Construct a global candidate pool based on all frog individuals from each subpopulation after local evolution. Specifically, the updated encoding vectors of all frog individuals from each subpopulation after local evolution, the high-fitness frog individuals retained before local evolution, and the worst individual satisfying the feasible region constraint are all merged into the global candidate pool. , Used to provide a source of parameter fragments for screening complementary candidate individuals and templated crossover.

[0123] In the specific implementation, the global candidate pool Each individual frog in the swarm was re-performed with finite element analysis for the construction phase, yielding stress residuals, anomaly indicator vectors, a load-bearing guidance index sequence, and a particle size distribution curve sequence. The anomaly indicator vectors characterize the locations where stress residuals and structural parameter mutations occur in concentrated bursts; the load-bearing guidance index sequence is a set of load-bearing guidance indices for each permeable concrete pavement segment arranged in the continuous paving and compaction construction sequence, used to characterize the spatial variation of the load-bearing path organization capacity of each pavement segment; the particle size distribution curve sequence is a set of particle size distribution curves for each permeable concrete pavement segment arranged in the continuous paving and compaction construction sequence, used to characterize the transition relationship of aggregate gradation along the construction direction.

[0124] 2) Calculate the aggregate morphology complementary index based on the similarity of the bearing-guiding sequence, the overlap area of ​​particle size distribution, and the proportion of differences in anomalous locations. Specifically, the aggregate morphology complementary index is used to select from the global candidate pool. In the process of screening, candidate individuals with complementary relationships to the current frog population are selected. The complementary relationship is not simply a matter of the greater the difference, but requires that the two frog populations maintain coordination in the overall carrying capacity, have smooth transition conditions in particle size distribution, and have exchangeable correction fragments in abnormal parts.

[0125] In one implementation, the first The carrying orientation index sequence of each frog swarm is represented as follows: ;in, Indicates the first The bearing guidance index sequence of individual frogs arranged according to the continuous paving and compaction construction sequence, with dimension [missing information]. ; This represents the transpose of a vector. For the ... The first frog group and the first The calculation method for the cosine similarity of the bearing guidance index sequence for each candidate individual is expressed as follows: ; in, Indicates the first The first frog group and the first Cosine similarity of the bearing guidance index sequence among candidate individuals; Indicates the first The carrying guidance index sequence of each candidate individual; express Transpose of; The stability constant for similarity calculation can be taken as... ; This represents the Euclidean norm.

[0126] Furthermore, the overlapping area of ​​the particle size distribution curves is used to determine whether the gradation transition between the two frog groups in the paving section to be exchanged is smooth. The difference ratio of abnormal parts can be calculated by the difference ratio between the two abnormal part indicator vectors. That is, first count the number of corresponding elements with different values ​​in the abnormal part indicator vectors of the two frog groups, and then divide by the total number of structural parameter dimensions. The resulting ratio represents the proportion of differences in abnormal parts. To avoid selecting candidate individuals with completely inconsistent overall mechanical states, basic screening conditions for complementary candidate individuals can be set as follows: the cosine similarity of the load-bearing guidance index sequence is not less than 0.85, and the overlap area of ​​the particle size distribution curves of the pavement section to be exchanged is not less than 0.70.

[0127] In one implementation, a complementary aggregate morphology index is calculated based on the similarity of the load-bearing guidance sequence, the overlapping area of ​​particle size distribution, and the proportion of differences in abnormal parts. This index can then be used to screen candidate individuals that do not disrupt the overall load-bearing guidance consistency and can provide correction segments for abnormal pavement sections. The calculation method is expressed as follows: ; in, Indicates the first The first frog group and the first Complementary index of aggregate morphology among candidate individuals; Indicates the first The first frog group and the first The average overlap area of ​​the particle size distribution curves of the candidate individuals on the paving section to be exchanged and its adjacent transition paving section; Indicates the first The first frog group and the first The proportion of abnormal sites differing among candidate individuals; This represents the similarity weight of the bearing orientation, which can be taken as 0.45; This represents the transition weight of the gradation, which can be set to 0.30. This represents the weight of the difference in abnormal parts, which can be set to 0.25.

[0128] In one embodiment, the heatmap of the complementary index matrix of aggregate morphology is analyzed, such as... Figure 6 As shown, the quantitative evaluation results of the complementary relationship of aggregate morphology among individuals in the global candidate pool of the frog swarm are presented. The horizontal axis is the individual index of the candidate frog swarm (dimensionless), and the vertical axis is the individual index of the current frog swarm (dimensionless). The color from light yellow to dark red indicates that the complementary index value (dimensionless) is from low to high. The experiment shows that there are significant differences in complementarity among different individuals, and some individuals have high complementary index values, which proves the rationality of templated cross-pollination based on the complementary index to select the optimal complementary candidate individuals.

[0129] 3) Based on aggregate morphology complementary index Determine the optimal complementary candidate individual.

[0130] In one implementation, for the first... Complete encoding vector of each frog swarm In the global candidate pool The search process identifies candidate individuals that satisfy the requirements of cosine similarity of the bearing-guiding index sequence, overlap area of ​​particle size distribution curves, and feasible region constraints. The candidate individual with the highest complementary aggregate morphology index is selected from these and defined as the optimal complementary candidate individual. Its complete encoding vector is defined as... .in, Represents the index of the best complementary candidate individual, used to identify individuals from the global candidate pool. Complementary candidate individuals obtained through screening; Used to provide with the first Complementary parameter fragments of individual frogs in terms of aggregate morphology and anomaly patterns.

[0131] In one embodiment, for example, if the first If an individual frog in a swarm exhibits anomalous markers in the 4th and 5th permeable concrete pavement sections, while a candidate individual shows no anomalous markers in the same sections, and the overlap area of ​​their particle size distribution curves in the relevant pavement sections is greater than 0.70, and their cosine similarity of the bearing capacity guidance index sequence is greater than 0.85, then this candidate individual can be considered a complementary candidate individual. If multiple candidate individuals simultaneously meet the conditions, the candidate individual with a lower fitness value, fewer anomalous pavement sections, and higher rule base matching confidence is prioritized.

[0132] 4) Generate a template cross-mask vector based on the anomaly indicator vectors of the current individual and the optimal complementary candidate individual. , It refers to the first Individual frogs in the swarm and their optimal complementary candidate. The parameter segment selection vector is used to control the parameters of each permeable concrete pavement segment to come from the current frog swarm individual or the best complementary candidate individual, and the element value is 0 or 1.

[0133] In specific implementation, for the first If, among the individuals in a frog swarm, the permeable concrete pavement segment marked by the anomaly indicator vector has no anomaly marker, and the aggregate morphology parameters of that pavement segment satisfy the continuity of particle size distribution and load-bearing guidance with those of adjacent pavement segments, then the template cross-mask vector... In this paving section, the corresponding dimension is preferentially set to 0, indicating that the parameter fragment of the optimal complementary candidate individual is adopted; for the th In a permeable concrete pavement section where no abnormal markers were found in any frog individuals and fitness was stable, the template cross-mask vector was used. In this paving section, the corresponding dimension is preferentially set to 1, indicating that the parameter fragments of the current frog swarm individuals are retained.

[0134] Furthermore, to avoid abrupt changes in adjacent permeable concrete pavement sections due to segment replacement, a buffer judgment can be set for the transition pavement section adjacent to the abnormal pavement section. Directly replacing the abnormal pavement section would cause the difference in the load-bearing guidance index between adjacent permeable concrete pavement sections to exceed [a certain threshold]. If so, the adjacent transition paving section will also be included in the template intersection range.

[0135] 5) Generate offspring encoding vectors through templated crossover. , It refers to the first New candidate solutions generated from individual frogs in a swarm through global information recombination are used to participate in the next generation of global fitness evaluation.

[0136] In one implementation, it is based on the template cross mask vector. Templated crossover is performed to obtain the offspring encoding vector. The calculation method is expressed as follows: ; in, Indicates the first The offspring encoding vectors generated by recombining global information from individual frogs in a swarm; This represents a single vector with the same dimension as the complete encoded vector. It is used in conjunction with the template cross mask vector to complete the replacement of complementary parameter fragments. This represents the element-wise multiplication operator, used to combine the encoding parameters of two candidate individuals dimension by dimension according to the template cross-mask vector.

[0137] 6) Based on the structural parameter prediction rule base and feasible region projection operator, re-execute structural parameter mapping and feasibility correction on the child encoding vector. Since templated crossing may change the aggregate morphology parameters of some permeable concrete pavement sections, the structural parameters before crossing cannot be directly used; it is necessary to re-apply the structural parameter prediction rule base. Perform structural parameter mapping.

[0138] In the specific implementation, the child encoding vector is extracted first. The aggregate morphology parameter vector and load-bearing guidance index of each permeable concrete pavement segment are obtained; then the structural parameter prediction rule base is queried. The corresponding structural parameter range is obtained; if a high-confidence rule is hit, the structural parameter vector is regenerated; if a high-confidence rule is not hit, the aggregate morphology parameters are corrected to the nearest rule coverage range using the feasible region projection operator; if the correction exceeds 10% of the aggregate morphology parameter value range, it indicates that the sub-encoding vector deviates too much from the rule base coverage area, and the sub-encoding vector is deemed insufficiently feasible and discarded.

[0139] In one embodiment, for example, if the first Offspring encoding vectors of individual frogs After the complementary individual segments were introduced, the aggregate volume fraction in the third pavement section became 0.72, exceeding the upper limit of 0.68 for the aggregate volume fraction corresponding to this load-bearing guidance level in the rule base. The feasible region projection operator corrected it to 0.68, with a correction magnitude of [missing value]. If the percentage is less than 10%, retain the offspring; then query again. If the mapping structure parameters and adjacent gradients all meet the requirements, then... Moving on to the next generation.

[0140] Furthermore, the retained child encoding vectors are re-examined for the continuity constraints of the bearing capacity guidance index, particle size distribution, and structural parameter gradient constraints of adjacent permeable concrete pavement sections. If any constraint is violated, local linear interpolation is first performed on the transition pavement section; if local linear interpolation still cannot satisfy the constraints, the corresponding segment is restored to the parameter segment from the optimal source before the crossover.

[0141] 7) Based on a comprehensive evaluation of stress residuals, sparsity of abnormal pavement sections, and load-bearing guiding performance during the construction phase, the fitness of the offspring coding vectors is evaluated and selected. Specifically, the offspring coding vectors... Input the finite element model for the construction phase The stress residual, number of abnormal paving sections, mean load-bearing guidance index, and structural parameter gradient violations during the construction phase were calculated and compared with those of individual frogs in the parent swarm.

[0142] In practical implementation, priority is given to retaining child encoding vectors that reduce stress residuals during construction, decrease the number of abnormal pavement sections, and whose structural parameter gradients in adjacent pavement sections do not violate constraints. If a child encoding vector only reduces stress residuals during construction but increases the number of abnormal pavement sections, it is penalized using an anomaly sparsity penalty function to prevent the optimization result from spreading local anomalies to more pavement sections. If a child encoding vector simultaneously increases the mean of the load-bearing guidance index and reduces the maximum stress residual, it is selected as a priority individual in the next generation of the swarm. If a child encoding vector violates the rule base matching conditions or causes abrupt changes in parameters of adjacent pavement sections, it is not included in the next generation of the swarm.

[0143] In one embodiment, as an example, the first If anomalies are detected in the 4th and 5th permeable concrete pavement sections, and the optimal complementary candidate individual exhibits low stress residuals and meets the particle size distribution continuity requirement in both sections, then the template cross-mask vector... In the fourth and fifth permeable concrete pavement sections, the corresponding dimension is set to 0, prioritizing the introduction of aggregate morphology parameter fragments from the optimal complementary candidate individuals; in the remaining permeable concrete pavement sections, the corresponding dimension is set to 1, prioritizing the retention of the first... Parameter fragments of each frog in the swarm. After crossover, the structural parameter prediction rule base is used. The structural parameters are remapped, and the transition area between adjacent paving sections is corrected by the feasible region projection operator, so that the offspring coding vector can maintain the continuity of aggregate gradation and load-bearing orientation while reducing local stress residuals.

[0144] In one embodiment, the distribution of structural parameters of each pavement section after optimization is analyzed using a line graph, such as... Figure 7As shown, the distribution of concrete stiffness reduction factor, interlayer shear correction factor, compaction sensitivity coefficient, and temporary construction load amplification factor along the pavement section in the optimal scheme is illustrated. The horizontal axis represents the permeable concrete pavement section index (dimensionless), and the vertical axis represents the structural parameter values ​​(dimensionless). Experiments show that the gradient changes of the four optimized structural parameters in adjacent pavement sections are all continuously varied within the corresponding maximum allowable gradient range, without any local abrupt changes. This proves that the structural parameter prediction rule base mapping and gradient constraint mechanism can ensure the physical continuity of the input parameters of the finite element model during the construction stage.

[0145] It should be noted that, through global information reorganization, frog swarm search can utilize the complementarity of aggregate morphology and the differences in abnormal patterns to carry out directional parameter exchange, so that the transition of aggregate gradation, the coordination of load-bearing guidance and the stress control during construction can be achieved in a unified manner within the same optimization framework, thereby obtaining a permeable concrete aggregate morphology combination that meets physical constraints, has a lower stress residual during construction and is more sparse in abnormal parts.

[0146] S4. Optimization Results Output and Determination of Construction Parameters for Permeable Concrete Pavement Sections After the hybrid frog-jump optimization, the optimal frog swarm individuals need to be converted into aggregate configuration schemes and structural parameter schemes for the permeable concrete pavement section and the construction stage. This step, after optimization, determines the optimal frog swarm individuals based on fitness values, constraint satisfaction, and finite element verification results during the construction stage, and outputs the aggregate morphology parameters, particle size distribution curves, load-bearing guidance index, and structural parameter vectors for each permeable concrete pavement section. The specific steps are as follows: 1) Determine whether the hybrid frog-jump optimization process meets the termination conditions. Termination conditions include the maximum number of iterations reaching a preset value, the decrease in the optimal fitness value for several consecutive generations being less than a preset threshold, all permeable concrete pavement sections meeting the structural parameter gradient constraints, or the maximum stress residual during the construction stage not exceeding the design allowable range.

[0147] In practical implementation, the maximum number of iterations The range can be from 100 to 300; continuous stable algebra A value of 10 to 20 is acceptable; the relative decrease threshold for the optimal fitness value can be set to... When any termination condition is met, the frog with the lowest fitness value and satisfying all constraints is selected from the current frog population as the optimal frog population individual. , It refers to the final candidate solution obtained after mixed frog leap search, which is used to output the aggregate morphology parameters and structural parameters of the permeable concrete pavement section.

[0148] 2) Decoding the optimal frog population individual The optimal aggregate morphology parameter vector, optimal particle size distribution curve, optimal load-bearing guidance index, and optimal structural parameter vector for each permeable concrete pavement section were obtained.

[0149] Specifically, for the first For a permeable concrete pavement section, the optimal aggregate morphology parameter vector is denoted as... The optimal particle size distribution curve is denoted as The optimal load-bearing guidance index is denoted as The optimal structure parameter vector is denoted as .in, It refers to the first The average edge angle, needle-like index, and aggregate volume fraction were ultimately used for each permeable concrete pavement section. It refers to the first The final aggregate particle size distribution used in each permeable concrete pavement section; Used to characterize the The ability of the final aggregate contact force chain network of a permeable concrete pavement section to form a load-bearing path along the target force transmission direction; Used to verify the first in the finite element model during the construction phase. Equivalent mechanical response of a permeable concrete pavement section.

[0150] In one embodiment, the distribution of aggregate morphology parameters in the optimized permeable concrete pavement section is analyzed using a line graph, such as... Figure 8 As shown, the average edge angle, needle-like and flaky index, and aggregate volume fraction of the optimal frog swarm individual obtained by the hybrid frog jumping algorithm vary along the paving direction in each paving segment. The horizontal axis represents the permeable concrete paving segment number (dimensionless), the left vertical axis represents the average edge angle and needle-like and flaky index (dimensionless), and the right vertical axis represents the aggregate volume fraction (dimensionless). The blue solid circle represents the average edge angle, the green square dashed line represents the needle-like and flaky index, and the red diamond-shaped dotted line represents the aggregate volume fraction. The experiment shows that the three aggregate morphology parameters after optimization generally show a slow decreasing trend, and the variation between adjacent paving segments is small.

[0151] 3) Generate the construction mix control range based on the optimal aggregate morphology parameter vector and the optimal particle size distribution curve. The construction mix control range refers to converting the optimized continuous aggregate morphology parameters into material control indicators that can be executed on site, including the mass ratio of sieve aperture intervals, the control value of the proportion of needle-shaped and flaky particles, the aggregate edge angle grade, and the amount of aggregate per unit volume.

[0152] In practical implementation, the optimal particle size distribution curve is first... The aggregate mass ratio is converted to the range of each sieve size; then the aggregate content per unit volume of permeable concrete is determined based on the optimal aggregate volume fraction; then the upper limit of allowable needle-shaped and flaky particles is determined based on the optimal needle-shaped and flaky index; finally, the aggregate source screening standard or crushing and shaping requirements are determined based on the optimal average edge angle.

[0153] In one embodiment, for example, if the optimal particle size distribution curve of the 6th permeable concrete pavement section is [0.25, 0.50, 0.25], then the aggregate mass ratio of the three sieve aperture ranges of 4.75mm to 9.5mm, 9.5mm to 13.2mm, and 13.2mm to 16.0mm can be controlled at around 25%, 50%, and 25%, respectively, and minor adjustments can be made based on the on-site moisture content and density conversion.

[0154] 4) Input the optimal structural parameter vector into the finite element model during the construction phase. A review is conducted. During the review, the maximum tensile stress, maximum compressive stress, maximum deflection, and bottom tensile stress of all construction stages are recalculated according to the actual construction sequence, and it is checked whether they meet the design allowable values ​​and monitoring control values. This can prevent schemes that only perform well in the optimization fitness function but do not meet the safety control requirements in the key construction stages from being put into actual construction.

[0155] In practice, if the maximum tensile stress of any permeable concrete pavement segment exceeds the design allowable tensile stress, or the deflection exceeds the construction control limit, the corresponding permeable concrete pavement segment will be marked as an abnormal pavement segment for review, and the process will return to S301 to increase the abnormal sparsity perception coefficient or force chain guiding strength of the pavement segment again; if all permeable concrete pavement segments meet the design allowable value and construction control limit, the final aggregate morphology optimization scheme will be output.

[0156] 5) During construction, update the aggregate morphology optimization scheme for the permeable concrete pavement section based on monitoring data. Online construction monitoring data includes pavement section stress, pavement deflection, permeability coefficient, crack records, concrete age, temperature, and changes in temporary construction loads. If the monitoring data and the finite element model during the construction phase... If the deviation between the verification results exceeds a preset threshold, the structural parameter prediction rule base will be recalibrated using monitoring data from the already poured paved section. The structural parameter range in the text is determined, and S202 to S302 are re-executed for the unpoured permeable concrete pavement section.

[0157] In practical implementation, if the measured deflection of the already poured pavement section exceeds 10% of the finite element prediction value for two consecutive construction stages, the priority of the concrete stiffness reduction factor correction for subsequent unpoured permeable concrete pavement sections can be increased, and this can be added to the structural parameter prediction rule base. In this approach, high-confidence rules with relatively stable stiffness reduction factors are prioritized. If the temporary construction load is significantly higher than the planned value, the upper limit of the temporary construction load amplification factor can be increased, and the gradient constraints of structural parameters in adjacent pavement sections can be re-verified. Based on this, the optimization scheme can be modified according to changes in actual construction conditions, avoiding the continuous accumulation of deviations.

[0158] Based on this, the final output of this invention is a chain of aggregate morphology parameters, a chain of particle size distribution curves, a chain of load-bearing guidance indexes, and a chain of structural parameters for permeable concrete pavement sections arranged along the continuous paving and compaction construction sequence. This output simultaneously demonstrates the feasibility of material formulation, the finite element calculation capability during the construction phase, and the physical continuity of adjacent pavement sections. It can be used to guide the selection of aggregates for permeable concrete pavement sections, verify stress during the construction phase, and provide feedback and corrections during subsequent monitoring.

[0159] In one embodiment, for example, a continuous pavement is divided into 12 permeable concrete pavement sections using continuous paving and compaction construction. First, geometric dimensions, construction loads, and 300 sets of aggregate morphology sample data for each pavement section were collected to establish a finite element model and a discrete element specimen model. Second, the load-bearing guidance index of each combination was calculated through the S2 step to mine the structural parameter prediction rule base. Then, 60 initial frog population individuals that meet the constraints were generated, with a subpopulation number of 6. After 150 generations of mixed frog jumping optimization, the optimal frog population individuals were obtained. Finally, the optimal frog population individuals were decoded to obtain the optimal parameter values. For pavement sections 1 to 4, it is recommended to use aggregate combinations with an average edge angle of [0.58, 0.65], a needle-like index of [0.10, 0.15], and an aggregate volume fraction of [0.62, 0.66]. The particle size distribution is mainly in the range of 9.5mm to 13.2mm. The load-bearing guidance index of pavement sections 5 to 8 is all higher than 0.78, and the structural parameter gradient is stable. Finite element verification during the construction stage shows that the stress of all pavement sections meets the design requirements, and the aggregate gradation transitions smoothly along the pavement sections.

[0160] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete, characterized in that, Includes the following steps: S1. Collect basic data of permeable concrete pavement sections and establish finite element models and discrete element specimen generation rules for the construction stage. S2. By extracting the discrete element force chain guidance index, generating structural parameter association rules, and initializing and mapping the frog swarm constraints based on the similarity propagation of the bearing guidance index, the micro-aggregate morphology is bound to the macro-structural parameter constraints, so that the initial frog swarm individuals meet the requirements of aggregate morphology value range, bearing guidance continuity, structural parameter gradient stability and finite element calculation feasibility in the construction stage during the generation stage, and finally the initial frog swarm that meets the physical constraints is obtained. S3. Divide the initial frog population into multiple subpopulations to perform local evolution; after all subpopulations have completed local evolution, perform global information recombination, merge the frog individuals of each subpopulation into the global candidate pool, screen complementary candidate individuals based on aggregate carrying capacity guidance sequence, particle size distribution continuity and abnormal part differences, generate offspring coding vectors through templated crossover, evaluate and select the fitness of offspring coding vectors to obtain the optimal frog population individuals; S4. Convert the optimal frog swarm individuals into aggregate configuration schemes for permeable concrete pavement sections and structural parameter schemes for the construction stage.

2. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 1, characterized in that, Step S1 specifically includes: The pavement structure to be optimized is divided into paving sections according to the continuous paving and compaction construction sequence. One permeable concrete pavement section; Collect aggregate morphology data for permeable concrete, including average angularity, needle-like and flaky index, aggregate volume fraction, and particle size distribution curve; Establishing a finite element model for the construction phase Finite element model during construction phase The inputs include pavement section geometry data, material reference parameters, construction sequence, roller self-weight, vibration load, paver force, temperature gradient load during curing, and interlayer constraint load; finite element model for the construction stage. The output includes the maximum tensile stress, maximum compressive stress, maximum deflection, and bottom tensile stress of each permeable concrete pavement section at each construction stage; Finite element model from the construction phase Extract the target force transmission direction vector , It refers to the first The average tangent direction of the principal compressive stress trace of a permeable concrete pavement section; Establish discrete element specimen generation rules, including specimen size and target porosity, particle generation method, particle contact parameters, loading boundary conditions, and loading stop conditions.

3. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 1, characterized in that, In step S2, the discrete element force chain guidance index extraction and structural parameter association rule generation specifically include: Construct a sample set of aggregate morphology combinations, the first The aggregate morphology parameter vector is ,in, Indicates an index of aggregate morphology combinations. Indicates the first The average edge angle of each aggregate shape combination Indicates the first The flaky and needle-like index of aggregate morphology combinations Indicates the first Aggregate volume fraction for each aggregate morphology combination; particle size distribution curve for each aggregate morphology combination. ; Discrete element specimen models are established based on a set of aggregate morphology combination samples. The input of the discrete element specimen model is the aggregate morphology combination and the bound particle size distribution curve, and the output is the mechanical response, aggregate contact force chain network and peak strength of the specimen under the corresponding aggregate morphology combination. From the Effective force chains are extracted from the aggregate contact force chain network corresponding to each aggregate morphology combination to obtain the first effective force chain. The set of effective force chains corresponding to each aggregate morphology combination; the unit direction vector of each effective force chain is calculated based on the three-dimensional coordinate sequence of the particle contact center. ; Based on the target force transmission direction vector Calculate the effective force chain length weight The load-bearing guidance index is calculated based on the degree of consistency between the effective force chain direction and the target force transmission direction. The load-bearing guidance level is classified according to the load-bearing guidance index, including high guidance level, medium guidance level and low guidance level. Structural parameter vectors corresponding to aggregate morphology combinations and permeable concrete pavement sections were obtained based on finite element calibration. , For the first The aggregate morphology combination is mapped to the first The equivalent pavement mechanical parameter vector obtained when a permeable concrete pavement section is used; The aggregate morphology range, load-bearing guidance level, and structural parameter range are combined into transaction items, and a rule base for structural parameter prediction is mined. , It is a set of queryable mappings composed of association rules.

4. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 1, characterized in that, In step S2, the frog swarm constraint initialization and structural parameter mapping specifically include: Definition of the first Complete encoding vector of each frog swarm individual , Encode the candidate solution that simultaneously includes the aggregate morphology parameter vector, particle size distribution curve, load-bearing orientation index, and structural parameter vector of all permeable concrete pavement sections; define the first... Among the individuals in the frog pod, the first Structural parameter vector of a permeable concrete pavement section And unify the range of structural parameter values; based on the structural parameter prediction rule base The high-confidence rules initialize the aggregate morphology parameters of the first permeable concrete pavement section; based on the bearing-guiding index similarity propagation, the second to third sections are generated according to the continuous paving and compaction construction sequence. Aggregate morphology parameters of a permeable concrete pavement section; prediction rule base using structural parameters. The aggregate morphology parameters are mapped to structural parameter vectors; the gradient of structural parameter changes between adjacent permeable concrete pavement sections is constrained. The gradient constraint of structural parameter changes is used to prevent unreasonable jumps in the stiffness, compaction response or construction load parameters of local pavement sections; the above process of pavement section generation, structural parameter mapping and gradient constraint correction is repeated to form an initial frog swarm that satisfies physical constraints.

5. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 1, characterized in that, In step S3, the local evolution of the subpopulation specifically includes: Before performing local evolution, the fitness value of each frog swarm individual was calculated based on the finite element model calculation results during the construction phase, the load-bearing guidance index of each paving section, and the distribution of abnormal paving sections. The initial frog population was sorted and grouped according to its fitness value: After sorting all frogs in the population from best to worst fitness value, they were divided into groups using an alternating sequential allocation method. A subpopulation, in the first From the subpopulation, the frog individual with the best fitness is selected as the optimal individual, and the complete encoding vector of the optimal individual is defined as follows: The frog individual with the worst fitness is selected as the worst individual, and the complete encoding vector of the worst individual is defined as follows: Based on the difference between the complete encoding vectors of the best and worst individuals, and through amplitude truncation and constraint checks using the local constraint maintenance operator, the local update direction from which the worst individual learns to the best individual is obtained. Anomaly indicator vectors are generated based on the median of stress residuals and deflection residuals of each pavement section calculated using the finite element model during the construction phase, as well as the degree of deviation from the structural parameters of adjacent pavement sections. An abnormally sparse penalty function is constructed based on the smoothed absolute deviation between the structural parameters of each pavement segment of the worst individual and the mean of adjacent pavement segments, and the correction direction of the abnormally sparse penalty function is calculated; the gradient direction vector is obtained according to the gradient components of each parameter dimension of the abnormally sparse penalty function. ; The force chain space guidance vector is calculated based on the load-bearing guidance sensitivity of aggregate morphology parameters to the load-bearing guidance index. ; Based on the local update direction, anomaly indication vector, gradient direction vector, and force chain space guidance vector learned from the worst individual to the best individual, a local evolutionary update is performed on the worst individual to obtain the worst individual update encoding vector; the worst individual update encoding vector is then corrected by the feasible region projection operator.

6. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 5, characterized in that, Fitness calculation specifically includes: The first The complete encoding vector for each individual frog in the swarm Input the finite element model of the construction stage, calculate the maximum tensile stress, maximum compressive stress, maximum deflection, and bottom tensile stress of each permeable concrete pavement section in each construction stage according to the continuous paving and compaction construction sequence, and compare the calculation results with the design allowable value, monitoring value, or target control value to obtain the stress residual and deflection residual of the construction stage; combine the mean of the bearing guidance index, the degree of violation of the gradient of structural parameters of adjacent pavement sections, and the number of abnormal pavement sections to form the fitness value.

7. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 1, characterized in that, In step S3, the global information reorganization specifically includes: A global candidate pool was constructed based on all frog individuals in each subpopulation after local evolution. For the global candidate pool Each individual frog in the swarm was re-performed with the finite element calculation of the construction stage to obtain stress residuals, abnormal location indicator vectors, load-bearing guidance index sequences, and particle size distribution curve sequences. The complementary index of aggregate morphology is calculated based on the similarity of bearing guidance sequence, the overlapping area of ​​particle size distribution, and the difference ratio of abnormal parts. Based on aggregate morphology complementary index Determine the optimal complementary candidate individual: For the first... Complete encoding vector of each frog swarm In the global candidate pool The search engine identifies candidate individuals that satisfy the requirements of cosine similarity of the bearing-guiding index sequence, overlap area of ​​particle size distribution curves, and feasible region constraints. The candidate individual with the highest complementary aggregate morphology index is selected and defined as the optimal complementary candidate individual. The complete encoding vector of the optimal complementary candidate individual is defined as follows: ; Generate a template cross-mask vector based on the anomaly indicator vectors of the current individual and the optimal complementary candidate individual. Based on template cross mask vector Templated crossover is performed to obtain the offspring encoding vector. ; Based on the structural parameter prediction rule base and feasible region projection operator, structural parameter mapping and feasibility correction are re-executed on the child encoding vector; The fitness evaluation and selection of the offspring coding vectors are based on a comprehensive evaluation of stress residuals, sparsity of abnormal pavement sections, and load-bearing guidance performance during the construction phase.

8. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 7, characterized in that, The specific calculation process for the similarity of the bearing guidance sequence, the overlap area of ​​particle size distribution, and the proportion of differences in abnormal parts is as follows: No. The carrying orientation index sequence of each frog swarm is represented as follows: For the first The first frog group and the first The calculation method for the cosine similarity of the bearing guidance index sequence for each candidate individual is expressed as follows: ,in, Indicates the first The first frog group and the first Cosine similarity of the bearing guidance index sequence among candidate individuals; Indicates the first The carrying guidance index sequence of each candidate individual; express Transpose of; This represents the stability constant used in similarity calculations. Represents the Euclidean norm; Area of ​​overlap of particle size distribution curves The calculation method is as follows: It is obtained by summing the minimum volume fractions of two adjacent paving sections in each sieve aperture interval. ,in, Indicates the first Among the individuals in the frog pod, the first The first permeable concrete pavement section and the first The area of ​​overlap of particle size distribution curves between permeable concrete pavement sections; Indicates the first Among the individuals in the frog pod, the first The permeable concrete pavement section is in the first The volume fraction of aggregate particles within each sieve aperture interval; Indicates the first Among the individuals in the frog pod, the first The permeable concrete pavement section is in the first The volume fraction of aggregate particles within each sieve aperture interval; This represents a function that takes the minimum value. The difference ratio of abnormal parts is calculated by the difference ratio between two abnormal part indicator vectors. First, count the number of different values ​​of corresponding elements in the abnormal part indicator vectors of individuals in the two frog groups, and then divide by the total number of structural parameter dimensions to obtain the difference ratio of abnormal parts.

9. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 7, characterized in that, Re-performing structural parameter mapping and feasibility correction on the child encoding vector specifically includes: Extracting the child encoding vector The aggregate morphology parameter vector and load-bearing guidance index of each permeable concrete pavement section are obtained; the structural parameter prediction rule base is queried. The corresponding structural parameter range is obtained; if a high-confidence rule is hit, the structural parameter vector is regenerated; if a high-confidence rule is not hit, the aggregate morphology parameters are corrected to the nearest rule coverage range using the feasible region projection operator; if the correction exceeds 10% of the aggregate morphology parameter value range, it indicates that the sub-encoding vector deviates too much from the rule base coverage area, and the sub-encoding vector is deemed insufficiently feasible and discarded.

10. The method for optimizing the aggregate morphology for load-bearing guidance in permeable concrete according to claim 1, characterized in that, Step S4 specifically includes: Decoding the optimal frog swarm individual The optimal aggregate morphology parameter vector, optimal particle size distribution curve, optimal load-bearing guidance index, and optimal structural parameter vector for each permeable concrete pavement section are obtained. The construction mix control range is generated based on the optimal aggregate morphology parameter vector and optimal particle size distribution curve. The optimal structural parameter vector is then input into the finite element model for the construction stage. The verification was carried out; during the construction process, the aggregate morphology optimization plan for the permeable concrete pavement section was updated based on the monitoring data.

Images