Long-life pavement structure optimization method and system based on artificial intelligence

Abstract

The invention relates to the technical field of intelligent road engineering, in particular to a long-life pavement structure optimization method and system based on artificial intelligence. The method comprises the following steps: obtaining pavement structure data, and constructing a structural mechanical model according to the pavement structure data to obtain the structural mechanical model; performing fatigue loss simulation according to the structural mechanical model to obtain fatigue loss data; carrying out rapid physical simulation according to the fatigue loss data to obtain loss simulation data; and performing physical prior life prediction according to the loss simulation data to obtain life prediction data. According to the method, through fracture entropy field modeling and crack density analysis, accurate quantification of spatial distribution and evolution trend of fatigue damage is realized. In combination with artificial intelligence-driven multi-time frame evolution simulation, the precision of pavement structure life prediction and the pertinence of design optimization can be significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent road engineering, and in particular to a long-life pavement structure optimization method and system based on artificial intelligence. BACKGROUND

[0002] Long-life pavement refers to a high-durability pavement structure form that its main structural layers (such as base layer, subbase layer and soil base) will not be structurally damaged within a design service period of more than 30 years under normal traffic load and appropriate maintenance conditions, and only needs to maintain the service performance (such as flatness, skid resistance, etc.) through periodic maintenance of the surface layer. The existing pavement structure design method mainly relies on traditional mechanical theory and empirical formula to select and evaluate the service life of the structure parameters, but in the long-term service process, the nonlinear factors of the pavement material at the microscopic level will significantly affect the structural performance. In addition, factors such as traffic load and environmental changes in actual operation make it difficult for the traditional method to accurately predict the service life of the structure, thereby affecting the scientificity and engineering economy of the pavement maintenance strategy. SUMMARY

[0003] The present application relates to the technical field of intelligent road engineering, and in particular to a long-life pavement structure optimization method and system based on artificial intelligence.

[0004] The present application provides a long-life pavement structure optimization method based on artificial intelligence, which comprises the following steps: S1, obtaining pavement structure data, constructing a structural mechanics model according to the pavement structure data, and obtaining the structural mechanics model; S2, simulating fatigue loss according to the structural mechanics model, and obtaining fatigue loss data; S3, performing rapid physical simulation according to the fatigue loss data, and obtaining loss simulation data; S4, performing physical a priori life prediction according to the loss simulation data, and obtaining life prediction data.

[0005] In the present application, the structural mechanics modeling and the fatigue loss simulation driven by artificial intelligence are fused to efficiently depict the multi-dimensional performance of the pavement structure. Compared with the traditional design method relying on empirical formula or single mechanical calculation, the present application can dynamically reflect the micro-damage evolution process caused by material performance degradation, load accumulation effect and environmental factors. The rapid physical simulation technology reduces the numerical calculation cost, improves the structural evaluation efficiency while ensuring the accuracy. Combined with the physical a priori life prediction model, the durability performance of different design schemes within the service period can be quantitatively estimated, which has superior engineering adaptability and promotion value, and is helpful to promote the intelligent optimization and green and sustainable development of long-life pavement structure.

[0006] Optionally, S1 comprises: Obtaining pavement structure data, performing hierarchical division according to the pavement structure data to obtain structure layer data, wherein the structure layer data comprises surface layer data, middle surface layer data, base layer data and soil base data; Performing interlayer condition setting on the structure layer data to obtain an interlayer model; Performing physical condition setting on the interlayer model to obtain a structure physical model; Performing material performance degradation condition setting on the structure physical model to obtain a structure mechanical model.

[0007] In the present application, the structure mechanical model is constructed with high fidelity by hierarchical modeling and condition setting on the pavement structure data, and has a significant physical expression advantage. The hierarchical division of the structure makes each functional layer (such as the surface layer, the middle surface layer, the base layer and the soil base) have a clear mechanical boundary in the model, which is beneficial to simulate the stress action and deformation response that each layer bears; the introduction of the interlayer condition enhances the reflection ability of the model to local failure behaviors such as interlayer shear slip and bond failure. By setting the physical boundary and the material degradation function, not only can the influence of environmental and use factors such as temperature and humidity coupling and load accumulation on the material performance be dynamically simulated, but also the time series evolution basis for life assessment can be provided. The modeling process can effectively improve the accuracy and adaptability of pavement structure simulation, and provides a solid data and physical basis for fatigue analysis and life prediction.

[0008] Optionally, S2 comprises: Generating a fracture entropy field according to the structure mechanical model to obtain fracture entropy field data; Calculating crack density according to the structure mechanical model to obtain crack density data; Performing model labeling on the structure mechanical model according to the fracture entropy field data and the crack density data to obtain fatigue loss data.

[0009] In the present application, the fatigue loss simulation is effectively improved in physical interpretation and spatial resolution by the dual characteristics of fracture entropy field and crack density. The fracture entropy field reflects the local energy dissipation and damage accumulation trend of the material, and has a highly sensitive damage precursor identification capability; the crack density characterizes the actual evolution trend of micro-damage from two dimensions of quantity and spatial distribution. The fusion of the two for labeling the structure mechanical model not only strengthens the expression ability of the model to the damage degree of each heterogeneous region, but also realizes effective mapping from micro-scale features to macro fatigue performance, significantly improving the accuracy and reliability of fatigue simulation.

[0010] Optionally, the fracture entropy field generation comprises: Performing local scanning on the stress distribution of the structure layer data under the action of external load disturbance according to the structure mechanical model to obtain local scanning data; According to the local scanning data, equivalent stress fluctuation amplitude calculation, shear stress gradient calculation and bonding interface instability index calculation are carried out, and equivalent stress fluctuation amplitude data, shear stress gradient data and bonding interface instability data are obtained respectively; According to the equivalent stress fluctuation amplitude data, shear stress gradient data and bonding interface instability data, discrete grid division is carried out, and grid microstate data are obtained. According to the grid microstate data, local fracture entropy is calculated, and fracture entropy field data are obtained.

[0011] In the application, through extraction of multi-dimensional stress response characteristics and local energy dissipation analysis, high-resolution modeling of structure micro-damage potential is realized. By local scanning of stress distribution in the structure layer, local stress concentration area under external load disturbance can be accurately captured, and the deficiency of easily ignoring weak instability area in traditional overall analysis is made up. Through equivalent stress fluctuation amplitude, shear stress gradient and bonding interface instability index, the potential cracking risk of the structure can be revealed from different mechanical angles. Combined with the discrete grid division strategy, the mechanical response data are converted into grid elements with position semantics and physical state, and the calculability and expression ability of fracture entropy are improved.

[0012] Optionally, the crack density calculation comprises: According to the structure mechanics model, crack criterion is carried out, and crack data are obtained. According to the crack data, number density calculation and length density calculation are carried out, and number density data and length density data are obtained. According to the number density data, spatial crack density map mapping is carried out, and spatial crack density map data are obtained. According to the spatial crack density map data, local density mutation extraction is carried out, and local density mutation data are obtained. The number density data, length density data and local density mutation data are integrated, and crack density data are obtained.

[0013] In the application, the crack criterion identifies the key damage position based on the structure mechanics model, ensuring that the input data has physical credibility. Through the double-index modeling of number density and length density, not only the crack occurrence frequency is captured, but also the crack development scale is quantified, which helps to evaluate the rate and degree of structure integrity degradation. The mapping of spatial crack density map projects the original crack data into a distribution atlas with spatial semantics, combined with the extraction of local density mutation, effectively revealing the local damage aggregation area and risk hotspots in the structure.

[0014] Optionally, the model labeling comprises: According to the crack data, crack direction feature extraction is carried out, and crack direction feature data are obtained. According to the fracture entropy field data and the crack direction feature data, anisotropy feature extraction is performed to obtain anisotropy feature data; The fracture entropy field data, the crack density data and the anisotropy feature data are marked on a structural mechanics model to obtain fatigue loss data.

[0015] In the present application, the crack direction feature extraction can accurately depict the dominant path of crack propagation, avoiding the neglect of the influence of crack orientation in traditional fatigue evaluation. The crack direction feature and the fracture entropy field data are coupled to perform anisotropy feature extraction, effectively reflecting the difference in energy dissipation and crack evolution in different directions of the material, and enhancing the adaptability of the model to the response of heterogeneous and asymmetric structures. By integrating the fracture entropy field, crack density and anisotropy feature into the structural mechanics model, a fatigue loss label system based on physical mechanism driving is formed, so that the model has clear damage semantic identification and traceability in the prediction and optimization process.

[0016] Optionally, S3 comprises: According to the fatigue loss data, multi-time frame data extraction is performed to obtain multi-time frame data; According to the multi-time frame data, crack evolution slope calculation is performed to obtain crack evolution slope data; According to the multi-time frame data corresponding to the fracture entropy field data, entropy increase rate extraction is performed to obtain entropy increase rate data; According to the multi-time frame data, risk area hot spot evolution is performed to obtain risk area hot spot data; According to the crack evolution slope data, the entropy increase rate data and the risk area hot spot data, a reduced-order modeling space expansion simulation is performed to obtain loss simulation data.

[0017] In the present application, through multi-time frame sequence analysis and multi-physical field fusion calculation, the analysis accuracy and simulation efficiency of the damage evolution process are significantly improved. The crack evolution slope reflects the time sequence intensity characteristics of crack growth, and can reveal the main trend of fatigue expansion. The entropy increase rate index combines the change of the fracture entropy field to capture the energy dissipation dynamics caused by microstructure instability. The risk area hot spot evolution describes the local damage aggregation path from the spatial distribution angle. By integrating the above multi-source dynamic characteristics and using the reduced-order modeling strategy for space expansion simulation, not only the calculation complexity is effectively reduced, but also the high-dimensional structure state can be quickly simulated while maintaining physical reliability, enhancing the applicability and response speed of life evaluation and structure optimization in actual engineering.

[0018] Optionally, the entropy increase rate extraction comprises: According to the multi-time frame data corresponding to the fracture entropy field data, global entropy increase rate calculation and neighborhood entropy increase rate calculation are performed to obtain global entropy increase rate data and neighborhood entropy increase rate data; According to the global entropy increase rate data and the neighborhood entropy increase rate data, entropy increase time-varying variance calculation is performed to obtain entropy increase time-varying variance data; According to the entropy increase time-varying variance data, a local entropy flow graph is constructed to obtain local entropy flow graph data; According to the local entropy flow graph data, cross-resolution entropy fusion is performed to obtain the entropy increase rate data.

[0019] In the present application, by introducing the multi-scale and multi-time frame fracture entropy evolution characteristics, the dynamic perception ability of the local instability trend of the material is improved, and the physical explanation of the structure fatigue evolution prediction is also enhanced. The global entropy increase rate calculation can macroscopically reflect the overall energy dissipation trend, and the neighborhood entropy increase rate calculation can accurately capture the micro disturbance response of the local area; the entropy increase time-varying variance analysis makes the system sensitive to the instability behavior fluctuation degree in different periods, which is beneficial to identify the potential critical evolution inflection point. The construction of the local entropy flow graph realizes the flow modeling of information in the time and space micro interval, and the cross-resolution entropy fusion mechanism effectively coordinates the feature consistency under high and low resolution, and improves the adaptability and accuracy of the model to damage propagation in the heterogeneous scale scene.

[0020] Optionally, the risk area hot spot evolution includes: According to the multi-time frame data, a fatigue concentrated area is selected to obtain hot spot data; The contour of the hot spot data is extracted to obtain hot spot contour data; According to the hot spot contour data, the center of gravity trajectory is extracted to obtain hot spot center of gravity data; According to the hot spot center of gravity data, dynamics evolution is performed to obtain risk area hot spot data.

[0021] In the present application, by gradually extracting and tracking the spatial features and dynamic response behavior of the fatigue concentrated area under multi-time frame, the identification accuracy and time sequence evolution analysis ability of the local high-risk area are effectively improved. The preliminary screening of the fatigue concentrated area can focus on the potential failure core area, which significantly reduces the redundant analysis overhead; the contour extraction clearly defines the hot spot boundary form, providing accurate geometric basis for subsequent structure change modeling; the center of gravity trajectory extraction introduces spatial-time bidirectional information fusion, which strengthens the continuity modeling ability of the regional displacement and evolution trend; the dynamics evolution analysis based on the center of gravity data not only captures the stress concentration, expansion or migration behavior of the area, but also excavates the dominant mechanism of the hot spot development.

[0022] Optionally, the present application also provides a long-life pavement structure optimization system based on artificial intelligence, which is used for executing the long-life pavement structure optimization method based on artificial intelligence as described above, and the long-life pavement structure optimization system based on artificial intelligence comprises: A structural mechanics model construction module is configured to acquire pavement structure data, construct a structural mechanics model according to the pavement structure data, and obtain a structural mechanics model. a fatigue loss simulation module for simulating fatigue loss according to a structural mechanics model to obtain fatigue loss data; a fast physical simulation module for performing fast physical simulation according to the fatigue loss data to obtain loss simulation data; a physical prior life prediction module for performing physical prior life prediction according to the loss simulation data to obtain life prediction data.

[0023] The present application aims to improve the physical authenticity and adaptability of modeling by introducing real pavement structure data to construct a structural mechanics model containing interlayer contact stress, bonding conditions and environment coupling boundaries in step S1; ensure the fatigue loss evaluation to have material degradation and anisotropy characteristics by simulating the micro-damage evolution process under real working conditions by fusing fatigue indexes such as crack density and fracture entropy on the basis of the model in step S2; realize fast order reduction modeling and extended simulation at the physical level by combining multi-time frame sequences and hot spot evolution paths based on the fatigue loss data in step S3, which greatly reduces the traditional simulation calculation overhead; realize the forward-looking evaluation of pavement service period and the suggestion of structure life extension strategy by fusing the loss simulation results to construct a prior life prediction model in step S4. BRIEF DESCRIPTION OF DRAWINGS

[0024] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-restrictive embodiments made with reference to the accompanying drawings: Fig. 1 a step flow chart of a long-life pavement structure optimization method based on artificial intelligence is shown; Fig. 2 a step flow chart of a structural mechanics model construction method is shown; Fig. 3 a step flow chart of a fatigue loss simulation method is shown; Fig. 4 a step flow chart of a fast physical simulation method is shown; The implementation of the present application, functional features and advantages will be further described with reference to the accompanying drawings. DETAILED DESCRIPTION

[0025] The technical method of the present application will be described clearly and completely below in combination with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0026] Furthermore, the drawings are merely schematic and may not be drawn to scale. A software form can be employed to implement the functional entities, or the functional entities can be implemented in one or more hardware modules or integrated circuits, or the functional entities can be implemented in different network and / or processor methods and / or microcontroller methods.

[0027] It should be understood that, although terms such as "first" and "second" can be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to differentiate one element from another. For example, without departing from the scope of the example embodiments, a first element can be referred to as a second element, and similarly a second element can be referred to as a first element. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0028] Reference is made to Figs. 1 to 4 The application provides an artificial intelligence-based long-life pavement structure optimization method, which comprises the following steps: S1, acquiring pavement structure data, constructing a structural mechanics model according to the pavement structure data, and obtaining a structural mechanics model; Specifically, the key material parameters of each layer of the pavement structure (including the surface layer, the base layer, the bottom base layer, and the soil base) are collected, including the thickness, the elastic modulus, the Poisson's ratio, and other basic mechanical properties of each layer, which are used to represent the structural stiffness and deformation capacity. Traffic load condition data are obtained, including typical axle load values, tire contact pressure distribution, vehicle speed, and load repetition frequency, etc.; and combined with external environmental condition information, including regional temperature and humidity range, diurnal temperature difference cycle, and other factors. Based on the above structural layer data and condition information, a multi-layer structural mechanics analysis model is constructed. In the model, each structural layer is regarded as a sub-structural unit with independent mechanical properties, and a contact surface is set between each structural layer to simulate the stress transfer and deformation coordination between layers. In the layer boundary condition setting, the interface bonding model is considered, and the coupling combination form including the Coulomb friction model and the shear slip model is preferably selected to simulate the friction bonding and relative slip behavior between each structural layer. Temperature gradient, humidity penetration depth, and other parameters are combined with the stress-strain relationship of the structural unit in the model to simulate the physical behavior caused by thermal expansion and contraction and moisture content change. A time decay function is set in the model. The time decay function is represented in an exponential form, i.e., the modulus of the i-th layer at time t is represented as The modulus degradation behavior of the i-th layer is described by the function wherein represents the stiffness decay factor of the i-th layer at time t, ​​a material degradation rate coefficient, an evolution time.

[0029] S2, fatigue loss simulation is performed according to the structural mechanics model to obtain fatigue loss data; Specifically, a time-varying traffic load is applied to the completed structural mechanics model. The traffic load simulation considers a typical traffic pattern and is constructed according to parameters such as axle load spectrum, tire contact area, repeated passing frequency, and vehicle speed, is applied to the surface layer surface node, and is loaded in time sequence distribution to form a dynamic load input field.

[0030] After the load is applied, the system solves the stress-strain response of each structural unit in the model, extracts the axial stress, shear stress, principal strain and other mechanical variables at each time, and provides basic data for subsequent damage calculation.

[0031] Subsequently, based on the disturbance response results, a fracture entropy field generation model is used to analyze the local energy fluctuation of each unit. The model combines information such as stress fluctuation amplitude, shear stress gradient, and interface stability index to measure the energy imbalance of each structural unit under external load disturbance, and outputs fracture entropy field data representing the local thermal imbalance trend of the system, which is used to identify potential damage active areas. The system identifies potential crack locations according to the maximum principal stress criterion. That is, for a certain structural unit, when its principal stress exceeds the material set fracture threshold, it is considered as a potential crack risk area. In the identified area, the number of cracks per unit area (i.e. crack number density) and the total length of cracks per unit area (i.e. crack length density) are calculated. Based on the fracture entropy field data and crack density data, combined with the local stress principal direction and crack propagation direction information, the anisotropic damage characteristics of each structural unit are extracted. The fracture entropy data, crack density data and anisotropic characteristics are fused / annotated, and the spatial annotation of the damage factor in the structural mechanics model is completed, and the fatigue loss data of each structural unit is output. The fatigue loss data includes but is not limited to fatigue damage index (indicating the cumulative damage degree), energy dissipation rate (reflecting the energy dissipation ability per unit time), and critical proximity rate (used to represent the relative distance from the current state to the material failure critical value).

[0032] S3, fatigue loss data is used to perform rapid physical simulation to obtain loss simulation data; Specifically, based on the fatigue loss data obtained above, a multi-time-frame sequence is constructed. Each time frame corresponds to the fatigue state of the pavement structure after continuous loading under a certain working condition, including time-varying indices such as crack density, fracture entropy value, and hot spot characteristics. Based on this multi-time-frame sequence, the system calculates the crack evolution slope, representing the rate of change of the total crack length or crack density per unit area per unit time. The local fracture entropy growth rate is calculated by the change in fracture entropy values ​​of adjacent structural units in consecutive time frames, i.e., the energy imbalance growth per unit time. The hot spot trajectory evolution characteristics are obtained by extracting the spatial contour of the hot spot region based on the high-value clustered areas in the aforementioned entropy field, and by fitting the trajectory of the hot spot centroid coordinates in consecutive time frames to obtain the spatial migration path and evolution direction of the hot spot. The system uses principal component analysis or an autoencoder neural network model to perform feature compression processing on the above high-dimensional time-series fields, generating a low-dimensional potential expression space. Based on the compressed representation space, a physical simulation network for spatial expansion simulation is constructed. This network combines time frame indexes, loading conditions (such as wheel and axle loads, temperature and humidity environment), structural location, and low-dimensional state variables to predict the evolution of structural fatigue state in the spatial domain at future time steps. This simulation network can provide rapid simulation support for different loading conditions (such as high temperature and high frequency, humid erosion, etc.).

[0033] S4. Based on the loss simulation data, perform physical prior lifetime prediction to obtain lifetime prediction data.

[0034] Specifically, based on the aforementioned rapid physical simulation results, the system extracts high-risk regions within the structure and tracks the cumulative damage data for these regions. The cumulative damage data is a continuous sequence of damage factors across multiple time frames, denoted as... ,in Indicates the first The system inputs local damage intensity indices (such as fracture entropy and crack density) at each frame. This time-series data is then fed into a deep learning model built on a Long Short-Term Memory (LSTM) network for lifetime trend prediction. The model structure is as follows: Input is a multidimensional damage sequence. ,in This represents the damage value. This is the spatial location information of the corresponding structural unit (which can be embedded as a vector). For working condition characteristic vectors (such as temperature and humidity, load frequency, etc.); input to the intermediate layer, which includes an LSTM encoding layer containing one to two layers of hidden units, with 64 to 128 nodes per layer; a fully connected output layer that outputs the damage prediction sequence for the next N frames; and mean square error (MSE) as the training indicator. The system is based on empirical data or expert knowledge to pre-set physical prior data, i.e., to set fatigue limit conditions as the life assessment threshold, including but not limited to the fatigue life threshold, which represents the maximum allowable number of cycles for a material under a specific load working condition; the critical crack density threshold, which represents the maximum allowable crack density before structural failure; and the maximum fracture entropy threshold, which represents the energy dissipation limit. The system determines the intersection point between the above threshold values and the predicted curve output by the LSTM, i.e., identifies the position of the first threshold exceeding in the predicted sequence as the remaining life time of the structure unit. This life is defined as the time difference between the current time and the intersection time, with units of days or load cycles. Based on the unit life prediction, the system performs fusion and weighting processing according to the structure level (surface layer, middle surface layer, base layer, soil base). Specifically, the life of each layer is calculated, and the weight is set according to the layer thickness, modulus, and structure importance factor to obtain the overall remaining life index: wherein is the remaining life data of the overall structure, is the index of the structure layer, is the number of structure levels (such as surface layer, middle surface layer, base layer, and soil base, totaling 4 layers), is the weighting coefficient of the th layer, which is set based on empirical data or expert knowledge, is the predicted life value of the th structure layer, and an analysis graph of the life contribution of each structure layer is output to identify the layer level with the shortest life. Based on the life prediction results, the system generates structure reinforcement suggestions and material replacement scheme recommendations through a pre-set suggestion parameter library, including but not limited to setting a paving thickness suggestion at the life critical layer; recommending new structural materials with higher fatigue strength or bonding performance; and outputting visual life extension simulation graphs for decision-making.

[0035] Optionally, S1 includes: S11, acquiring pavement structure data, performing structure level division based on the pavement structure data to obtain structure layer data, wherein the structure layer data includes surface layer data, middle surface layer data, base layer data, and soil base data; Specifically, the system collects and aggregates raw structural parameter data related to the target road section, including the thickness of each structural layer (in centimeters), material type identification, construction batch information, and compaction degree indicators. This data can be sourced from design drawings, on-site testing records, and historical construction archives. The system deconstructs the raw data into hierarchical structural layer datasets based on the standard hierarchical division logic commonly used in road engineering. Surface layer data: This layer is constructed using asphalt mixtures such as AC-13 or AC-20, with a conventional laying thickness ranging from 4 to 8 centimeters. The system determines the surface layer type based on the material identification field, extracts its thickness and construction compaction degree as surface structure elements. Middle surface layer data: This layer is mainly used for fatigue buffering and stress dispersion, including stress absorbing layers (SAMI) or chip seal layers. The system detects the presence or absence of the middle surface layer based on the interlayer material identification field, and labels its fatigue-resistant material type and structural level position. Base layer data: The base layer is the main bearing structural layer, commonly using cement stabilized gravel layers or lime-fly ash stabilized gravel layers, with a thickness set between 20 and 35 centimeters. The system automatically extracts the thickness, material composition, mix proportion, and construction strength grade of this layer. Subsoil layer data: This layer is the natural foundation or artificial fill layer, and the system obtains its California bearing ratio (CBR value), compaction degree, and water content as foundation bearing capacity indicators based on in-situ testing or geological exploration data, serving as the input basis for the mechanical properties of the structural bottom. Through the above standardized structural layer division process, the system can output structural layer datasets with hierarchical identification, physical parameters, and construction attributes.

[0036] S12, interlayer condition setting is performed on the structural layer data to obtain an interlayer model; Specifically, the system identifies the contact interface between each adjacent layer in the structural layer data, and sets the corresponding interlayer boundary condition for each layer pair (such as the surface layer and the middle surface layer, the middle surface layer and the base layer, and the base layer and the soil base). According to the road construction process record, the design parameters or the sensor feedback information, the system determines the contact type between each pair of structural layers and classifies it into one of the following three contact states: full bonding state (ideal bonding), indicating that the upper and lower structural layers form a completely firm connection without relative slip, the system sets the friction coefficient of the contact interface to approach infinity and does not consider the effects of sliding or interface separation; partial bonding state (frictional bonding), indicating that there is a certain degree of frictional bonding effect at the contact interface, allowing limited slip but still having a certain shear capacity, the system sets the value range of the friction coefficient μ to be 0.3 to 0.6 according to the on-site material conditions; unbonding state (slip contact), indicating that the upper and lower structural layers do not form effective bonding, allowing complete slip, the system sets the friction coefficient to zero, the interface contact only bears the normal pressure stress and does not have shear transmission capacity. The system performs mechanical modeling processing on the above contact states, adopts a nonlinear contact element modeling method to construct the interlayer interaction mechanism, and specifically can adopt the Coulomb friction model, in which the shear slip force is determined by the friction coefficient μ and the normal contact force, and has a shear limit determination function to realize dynamic simulation of physical behaviors such as interlayer bonding failure and slip expansion. The system outputs the numbering of each structural layer pair, the contact boundary type (including the contact state, the friction parameter) and the contact model form as a standard interlayer model data structure.

[0037] S13, physical condition setting for the interlayer model, to obtain a structural physical model; Specifically, the system applies stress loading conditions to the pavement structure. Based on the preset parameters, a tire-structure coupling contact model is set on the top surface of the structural model, considering the contact pressure distribution and its dynamic transmission characteristics between the tire and the pavement. The load form can be selected as a moving load (simulating vehicle driving) or a periodic load (simulating repeated traffic), and the axle load peak P is set according to different vehicle types, with a value range of 40 to 80 kilonewtons (kN). The frequency parameter of the traffic load is set based on historical experience or preset parameters, between 1 to 5 hertz (Hz). This frequency parameter, together with the Poisson's ratio of the structural material, affects the lateral stress distribution and dynamic response of the structure, and the system performs coupling term modeling processing in the calculation. The system sets the boundary conditions of the structural model, including symmetric boundary setting, for the geometric center line or mirror image area of the structural model, an offset symmetric boundary condition is set, i.e. zero displacement constraints in the horizontal and vertical directions are applied at the boundary, to avoid non-physical deformation caused by artificial boundaries. The bottom boundary setting, according to the foundation processing method, sets the foundation model, including setting it as a completely fixed boundary, i.e. the bottom nodes have no displacement in three directions; or setting it as an elastic support boundary, introducing the foundation elastic coefficient k (unit: kN / m) and the foundation damping coefficient c (unit: kN / m / s) to simulate the dynamic response of the foundation to the load. This is to reflect the elastic response characteristics of the soil foundation.

[0038] S14. Set material property degradation conditions for the structural physical model to obtain the structural mechanical model.

[0039] Specifically, a time-dependent material degradation function, such as modulus degradation, is set for each structural layer. ,in for The elastic modulus of the material at any given time (unit: MPa). The initial elastic modulus, This is the degradation factor, which is the rate at which the stiffness of a material decreases with time or the number of load cycles during the loading process. Its value is determined based on material fatigue test results or empirical estimation, for example, 0.005-0.05 / year. It is an exponential function. To load the duration or usage time; or to use a degradation model based on cumulative strain, such as... ,in The modulus after N cycles. To accumulate traffic load cycles (e.g., equivalent standard axle load cycles). This is the logarithmic degradation coefficient of the modulus, such as 0.02-0.1. The system is configured with load cycle count; fatigue performance parameters are set, including the fatigue limit of the surface layer material. ,in This is the fatigue strength proportionality coefficient (usually an empirical value, depending on the material type). The current elastic modulus; the critical strain for base layer cracking. The system embeds the aforementioned degradation model and fatigue performance parameters into the finite element material definition module, enabling the material parameters to be dynamically updated over time or load cycles, forming a spatiotemporally coupled structural mechanics model. This model can adjust material stiffness, load-bearing capacity, and fatigue tolerance in real time during physical simulation.

[0040] Optionally, S2 includes: S21. Generate the fracture entropy field based on the structural mechanics model to obtain fracture entropy field data; Specifically, for each element mesh or voxel in the structural mechanics model, the following mechanical responses are extracted based on external disturbances (such as vehicle loads and temperature stress): equivalent stress fluctuation amplitude (equivalent stress fluctuation amplitude = maximum principal stress − minimum principal stress), shear stress gradient (shear stress gradient = square root of the sum of the squares of the rates of change of shear stress in the horizontal and vertical directions), and bond interface instability indices (such as shear slip or normal crack opening). The system performs local fracture entropy calculation, using the following formula: ,in For the first The fracture entropy value of each unit. is the Boltzmann constant, which can be set as a normalization constant or an empirical weight, is the micro-response category index, is the total number of micro-response categories, including equivalent stress fluctuation amplitude, shear stress gradient, and bond interface instability indicators, is the probability of the unit to exhibit the micro-response under multiple perturbation loads (which can be obtained based on finite element sub-step statistics). The system obtains the corresponding fracture entropy value for each grid unit; forms a three-dimensional tensor as the fracture entropy field data.

[0041] S22, crack density calculation according to the structural mechanics model, to obtain crack density data; Specifically, the system determines which unit regions have crack initiation risks according to the preset crack criterion. When the maximum principal strain of a unit exceeds the critical principal strain threshold of the material, it is considered that there is a crack initiation risk; or if the stress intensity factor in the fixed spatial window range of the crack tip is greater than the fracture toughness threshold of the material (i.e. the fracture strength factor), the location is marked as a potential crack source region. The crack number density calculation is to count the number of crack center points that meet the above crack criterion in each preset spatial window (for example, 10 cm 10 cm 2 cm three-dimensional region). By dividing the total number of cracks by the volume of the spatial window, the crack number density of the region is obtained. The system calculates the total length of all cracks in the above region and normalizes it to unit volume to obtain the crack length density. The above two types of density values (crack number density and length density) are mapped back to the grid units in the original structure model according to their corresponding positions to construct a three-dimensional spatial distribution map, denoted as the crack density map.

[0042] S23, model annotation of the structural mechanics model according to the fracture entropy field data and the crack density data, to obtain fatigue loss data.

[0043] Specifically, based on the crack density map, if crack geometry information can be obtained, the spatial orientation of each crack can be modeled: the system fits the crack morphology to line segments and extracts the principal direction vector of the crack; it statistically analyzes the direction distribution of cracks within the region and constructs a direction histogram; the system constructs a principal direction tensor (i.e., a structural tensor) based on a preset region range (e.g., 3x3x3). The fracture entropy field data and crack direction information are jointly analyzed; the corresponding local fracture entropy value is extracted for each directional dimension (e.g., along 0°, 45°, and 90°); the statistical variance of the fracture entropy value in different directions is calculated to obtain the degree of fatigue anisotropy in that region. Based on existing fracture entropy field data, crack density map, and anisotropy index, the system performs fatigue annotation operations on each element in the structural mechanics model. The following three types of data are spatially mapped and indexed, unified into the structural model mesh element: the fracture entropy value of each element (representing its energy imbalance under multiple perturbation conditions); the crack density value (number density or length density) of each element; and the fatigue anisotropy index of each element.

[0044] Optionally, the generation of the fracture entropy field includes: Based on the structural mechanics model, the stress distribution of the structural layer data under external load disturbance is locally scanned to obtain local scan data. Specifically, the system applies representative external disturbance boundary conditions to the target structure, including but not limited to simulating typical traffic load conditions, such as periodic axle loads (modeled as sinusoidal loads). ,in For a moment Traffic disturbance load. This represents the peak value of the traffic disturbance load. It is a sine function. The load disturbance angular frequency, (Time-based); localized impact load or short-term overload simulation. Based on historical celestial data, the temperature difference distribution range is set to simulate structural thermal stress caused by diurnal or seasonal temperature variations, specifically using a periodic sinusoidal temperature function. Establish a thermo-mechanical coupling field, in which For a moment Temperature value, The average temperature reference value, The amplitude of the temperature disturbance. It is a sine function. The angular frequency of the temperature difference change. is a time variable. Uneven settlement displacement boundary is applied to the bottom or boundary of the structural model to simulate the dynamic changes of the foundation soil under the environmental background of rainfall, compaction, etc. After completing the disturbance boundary setting, the system scans each finite element unit in the structural layer in the time domain, collects its stress indicators, obtains the stress tensor data of each unit under the disturbance, records the complete stress change trajectory at different disturbance times, and represents the stress response data as a four-dimensional tensor sequence, i.e. constructing a space-time stress response surface, and obtaining local scanning data.

[0045] According to the local scanning data, the equivalent stress fluctuation amplitude calculation, shear stress gradient calculation and bond interface instability index calculation are carried out, and the equivalent stress fluctuation amplitude data, shear stress gradient data and bond interface instability data are obtained respectively; Specifically, for any time point, the stress is calculated as: , wherein is the equivalent stress value, is the normal stress in the direction, is the normal stress in the direction, is the normal stress in the direction, is the plane shear stress, is the plane shear stress, is the plane shear stress. For all time points in the scanning period, the maximum and minimum values are calculated, and the stress fluctuation amplitude is calculated as: wherein is the equivalent stress fluctuation amplitude, is the maximum time point stress operation, is the minimum time point stress operation. The system calculates the three-dimensional shear stress gradient field: wherein is the shear stress spatial gradient modulus, is the plane shear stress, is the plane shear stress, is the plane shear stress, is the spatial coordinate axis, is the spatial coordinate axis, is the spatial coordinate axis. The normal cracking displacement and the tangential slip displacement are extracted for the interface between the structural layers (such as the interface between the base layer and the surface layer), and the interface damage index is calculated as: wherein is the instability index of the bonding interface, is the tangential slip displacement, is the normal cracking displacement, is the critical tangential slip displacement, is the critical normal cracking displacement. If , it is considered that the instability trend is obvious.

[0046] Discrete grid division is performed according to equivalent stress fluctuation amplitude data, shear stress gradient data and bonding interface instability data to obtain grid microstate data; Specifically, the structure model space is discretized into small grid units (also referred to as voxels) according to a fixed voxel size (such as 5 cm3); for each small grid unit, if the unit contains multiple finite element units or scanning point data, mean value or maximum value is used for dimension reduction; or if the unit contains multi-layer material data, it can be synthesized in a weighted manner (such as according to the volume proportion, layer position, etc.).

[0047] Local fracture entropy is calculated according to the grid microstate data to obtain fracture entropy field data.

[0048] Specifically, the occurrence frequency of each state in the voxel is counted within a perturbation period (such as time or load cycle), and is converted into a probability value. For the i-th voxel (i.e. small grid unit), the occurrence probability of the j-th state is denoted as Pij, which satisfies the normalization condition that the sum of all state probabilities is equal to 1. The entropy is calculated as follows: wherein Pij is the static probability of the j-th state in the i-th voxel, and the sum of all state probabilities is equal to 1. The entropy is calculated as follows: wherein H is the fracture entropy value of the i-th voxel, N is the state category index, M is the total number of state categories, Pij is the static probability of the j-th state in the i-th voxel, and the sum of all state probabilities is equal to 1. The entropy is calculated as follows: The fracture entropy value of each voxel is combined to form the overall fracture entropy field.

[0049] Optionally, the crack density calculation comprises: Crack criterion is performed according to the structure mechanics model to obtain crack data; ​​​​​​​​​​​​​​​​​Specifically, the three-dimensional stress tensor output by the structural mechanics model of the previous step is used as the criterion input. For each discrete node or mesh element in the structure, the first principal stress value (i.e., the component with the maximum stress in the principal axis direction) is extracted. If this value exceeds the preset critical stress threshold for fracture, it is determined that there is a tendency for cracks to initiate at this location. Alternatively, if the shear stress experienced by the structure exceeds the shear limit of the material, it is determined that there is a possibility of shear crack formation. Alternatively, the system uses the J-integral method to construct a local path integration region for each suspected crack tip location, calculates the energy release rate at the crack tip of the structure, and if it is greater than the material fracture toughness, it is considered that the crack has a tendency to expand. For any spatial element that meets any of the crack criteria, the following structural crack information is extracted: for each mesh element that meets the crack formation / propagation criteria, its coordinates, crack direction (principal stress direction), crack length (if there is extension in the simulation), and time frame are extracted.

[0050] According to the crack data, the number density calculation and the length density calculation are performed to obtain number density data and length density data; Specifically, in each voxel element, the number of crack entries in the crack data set located in the space is counted. The crack number density of the voxel region is defined as the crack number density equal to the number of cracks in the region divided by the volume of the voxel. For the crack data in each voxel, the length of each crack is extracted and summed in the region. The crack length density of the voxel region is defined as the crack length density equal to the sum of the lengths of all cracks in the region divided by the volume of the voxel.

[0051] According to the number density data, a spatial crack density map is mapped to obtain spatial crack density map data; Specifically, the crack number density value or length density value at the discrete voxel position point is input, and spatial interpolation is performed on the entire structure region to generate a continuous density distribution field. Cross sections are cut at preset heights, depths, or key stress regions to display the crack density distribution on the corresponding cross sections; different colors or gray levels are generated according to the density value interval to highlight the dense crack regions or potential fatigue hot spot regions; the output spatial crack density map data is a continuous distribution mapping of the crack density in the structure region, i.e., the spatial crack density map data.

[0052] According to the spatial crack density map data, local density mutations are extracted to obtain local density mutation data; Specifically, spatial gradient analysis is performed on the crack density map generated in the previous stage, and first-order partial derivatives of the crack density function in X, Y, and Z directions are calculated in a three-dimensional coordinate space; a density gradient vector field is constructed based on the first-order partial derivatives, and the density gradient value reflects the rate and direction of crack density change in the neighborhood of the point; the gradient calculation can be realized by using finite difference method, central difference method, or convolution filter operator (such as Sobel operator). After obtaining the density gradient field, local mutation regions are identified by setting a mutation judgment threshold, specifically including setting a mutation judgment threshold, for example, using the average value of the overall density change rate plus two times the standard deviation as the critical value; for any spatial point, when the length of its density gradient exceeds the threshold, it is determined that the corresponding region is a local density mutation region; optionally, an edge enhancement algorithm (such as a high-pass filter based on the Laplacian operator) is used to enhance the density gradient change. The spatial profile boundary of the local mutation region is defined based on Sobel, Canny or region growing method, and the output local density mutation data is a set of mutation points or mutation regions, including the positions of the centers of the mutation points or regions in the three-dimensional structure coordinates; the corresponding density gradient length value, i.e. the density change rate strength; the output result can be organized as a point set, a region layer or a density mutation vector tensor, and is uniformly represented as a local density mutation data set.

[0053] The number density data, the length density data, and the local density mutation data are integrated to obtain the crack density data.

[0054] Specifically, for each grid cell, a density description vector is constructed, and the number density data, the length density data, and the local density mutation data are represented in vector form to obtain the crack density data.

[0055] Optionally, the model labeling includes: Crack direction feature extraction is performed according to the crack data to obtain crack direction feature data; Specifically, for each effective crack, the system obtains the components of a three-dimensional direction vector by subtracting the start point coordinates from the end point coordinates of the crack; the direction vector is normalized in length to convert it into a unit direction vector, which represents the propagation direction of the crack. The crack direction distribution in the local region is clustered and induced by K-means clustering, spherical Gaussian mixture model, or principal component analysis, etc. The dominant direction or direction distribution pattern is extracted; the crack direction vectors in each grid cell are averaged or modeled to generate a regional crack main direction density map. The crack direction feature data is obtained by integrating the foregoing data.

[0056] Anisotropy feature extraction is performed according to the fracture entropy field data and the crack direction feature data to obtain anisotropy feature data; Specifically, the system performs spatial gradient calculation on the fracture entropy field in the entire three-dimensional structure grid to obtain an entropy gradient vector at each voxel unit. The vector is composed of three directional components, respectively representing the change rate of the entropy value in the x, y, and z spatial dimensions, and collectively indicating the direction in which the fracture entropy changes most drastically. For each grid unit that has an effective crack direction vector, the system performs cosine similarity calculation on the unit direction vector of the crack and the entropy gradient vector of the unit. When the cosine value is close to 1, it indicates that the crack direction is highly consistent with the entropy change direction, which means that there is a clear dominant directionality in the region and the region has significant anisotropic characteristics. When the cosine value is close to 0 or negative, it indicates that the crack direction is random or inconsistent with the entropy change direction, i.e., the structure tends to be isotropic or there is no dominant path for energy diffusion. In each structure grid unit or local analysis window, the system aggregates and averages the similarity values between all crack directions and entropy gradient directions to obtain an anisotropy index for the region, which is used to quantitatively represent the degree of directionality. The output anisotropy feature data includes an anisotropy feature field, which assigns a scalar value to each spatial location (i.e., grid unit) to represent the directionality strength at that location, and a three-dimensional anisotropy atlas organized in the form of a tensor or a voxel map.

[0057] The fracture entropy field data, crack density data, and anisotropy feature data are labeled to the structure mechanics model to obtain fatigue loss data.

[0058] Specifically, the system performs three-dimensional grid registration on the fracture entropy field data, crack density data, and anisotropy feature data to make them share a unified coordinate domain with the structure finite element mechanics model. If the structure model uses hexahedral elements, nearest neighbor interpolation or trilinear interpolation algorithms are performed based on node coordinates. After registration is completed, the system constructs a fatigue representation joint feature vector for each structure unit or node. The vector is composed of the following four items: fracture entropy value (reflecting the degree of local evolution chaos of the material), number density and length density (corresponding to the number and total length of cracks per unit volume, respectively), and anisotropy feature value (indicating the consistency degree of the crack and the energy dissipation direction). The system constructs a fatigue loss index function to map the multi-dimensional representation vector to a single fatigue score indicator. The scoring function can adopt a linear weighting form, assigning adjustable weight coefficients to each type of feature value to form the following expression: the fatigue loss score is equal to the weighted sum of the fracture entropy value and the number density, length density, and anisotropy feature. For composite materials or nonlinear behavior cases, a lightweight neural network or regression model can also be used alternatively to establish and optimize the mapping function through training and learning. The system maps the calculated fatigue loss score to the corresponding unit position in the structure mechanics model to form a three-dimensional fatigue loss field of the structure.

[0059] Optionally, S3 includes: S31. Extract multi-time frame data based on fatigue loss data to obtain multi-time frame data; Specifically, the system selects fatigue loss data of the structure under different loading cycle stages, and each specific fatigue loss data point is represented at a specific time point. The system measures the fatigue sensitivity of each element at different locations. The total simulation duration is set to T, divided into N equally spaced time frame sequences, for example, every 100,000 loading cycles as the time frame interval. For each time frame, the system records the distribution of fatigue loss in the three-dimensional space of the structure, forming a tensor field. The system combines the fatigue loss tensors from all time frames in a time-series format to form a complete multi-time-frame fatigue loss dataset.

[0060] S32. Calculate the crack evolution slope based on multi-time frame data to obtain crack evolution slope data; Specifically, the system first selects index fields to describe the crack state, including crack length density and crack number density, which represent the crack state over time. The system calculates the volume-normalized total crack length and number of cracks at a given time and spatial location. Based on this, the system uses a temporal difference method to estimate the crack evolution rate across adjacent time frames. Taking crack length density as an example, the crack growth slope is defined as follows: ,in This represents the rate of increase in crack length density at that point per unit time. For the first +1 frame time Below, spatial point Crack length density, For the first +1 time frame corresponding to the loading time for Spatial coordinates for Spatial coordinates for Spatial coordinates For the first Frame time Below, spatial point Crack length density, For the first The loading time corresponding to each time frame The system assigns a time frame index number. The system performs spatial mean filtering on the obtained slope field. The system calculates the standard deviation of the slope variation. The output is a crack evolution slope map.

[0061] S33. Extract the entropy growth rate from the fracture entropy field data corresponding to the multi-time frame data to obtain the entropy growth rate data; Specifically, the system uses fracture entropy field data corresponding to multiple loading time frames as input, including loading cycle times and spatial coordinates. For two consecutive time frames, the system calculates the global and local entropy growth rates respectively. On the global scale, the system integrates the difference in the entropy field across the entire spatial domain and divides it by the time interval to obtain the average entropy growth rate index; the global entropy growth rate... ,in For the global entropy growth rate, For the first +1 time frame position The fracture entropy value, In three-dimensional space Spatial coordinates In three-dimensional space Spatial coordinates In three-dimensional space Spatial coordinates For the first Location in each time frame The fracture entropy value, For the first +1 time frame corresponding to the loading time For the first Each time frame corresponds to a loading moment. At a local scale, the system calculates the rate of entropy change per unit time point by point, forming a local entropy acceleration field. ,in The rate of increase of local entropy indicates the position. The local unit-time entropy growth is calculated, and other parameters can be referenced from previous data. The system calculates the temporal variance of the entropy growth rate at each point in multiple time frames to obtain a local entropy growth fluctuation index. Integrating the above, the entropy growth rate data is obtained.

[0062] S34. Based on the multi-time frame data, the hot spot evolution in the risk area is performed to obtain the hot spot data in the risk area; Specifically, the system identifies regions where fatigue loss values ​​are significantly higher than the overall average level at each time frame. A threshold for the fatigue loss field is set (e.g., the 90th percentile is selected as the global quantile threshold). The set of spatial locations where the local fatigue loss exceeds the threshold is defined as the hot spot region, i.e., the extraction time point. Corresponding hot spot set: , For time frames Hot spot clusters / hot spot regions These are three-dimensional spatial coordinates, representing the position of a voxel, node, or discrete element in the structural model. For time frames The fatigue loss value, The threshold value for the fatigue loss field represents the stress concentration region or potential crack accumulation region of the structure under the current fatigue loading state. The system performs boundary extraction operations on the spatial contour of each hot spot region. Edge detection algorithms (such as 3D Sobel filtering) or voxel envelope construction (such as...) are employed. The system employs shape-based methods and convex hull algorithms to extract the external geometric boundaries of the hotspot. It tracks the position of the geometric center (centroid) of the hotspot region in each time frame, recording the sequence of changes in this center across all time frames to form a continuous hotspot trajectory line. After acquiring the trajectory, the system performs dynamic characteristic analysis on the hotspot motion based on the trajectory line, including calculating the velocity vector (displacement per unit time), acceleration vector (rate of change of velocity), and trajectory perturbation amplitude (measuring the extent of deviation from the main trend line) of the hotspot's geometric center. The system outputs a hotspot dynamic evolution dataset including the hotspot's spatial extent, boundary contour information, the hotspot's geometric center trajectory line, and corresponding dynamic characteristic indices for each time frame.

[0063] S35. Based on the crack evolution slope data, entropy growth rate data, and hot spot data in the risk area, a reduced-order modeling spatial expansion simulation is performed to obtain loss simulation data.

[0064] Specifically, the system constructs a state feature vector for each structural mesh element. This vector includes the crack evolution slope (representing the local crack growth rate), the fracture entropy growth rate (reflecting the intensity of material energy state changes), and a hotspot path association identifier (used to indicate whether the element is within the hotspot trajectory range, taking a Boolean value or a normalized value of the distance function). Specifically, the state vector of a mesh element can be represented as: ,in This indicates the crack evolution slope of the unit. This represents the local growth rate of fracture entropy. The unit is represented as a dynamic indicator of structural evolution, indicating whether it falls into the vicinity of the hot spot dynamic evolution path. The system performs reduced-order modeling processing on the above state vector data. This process can be implemented in various ways, including principal component analysis (PCA) for linear feature compression, autoencoder for nonlinear representation learning, or multi-channel attention mechanism based on graph structure to enhance the structural perception ability between features. The reduced-order model outputs compressed low-dimensional latent variables. In the modeling stage, the system uses time series neural networks (such as long short-term memory networks) or graph neural networks (such as time series graph convolutional networks) to dynamically learn the above reduced-order features and simulate the state changes in the future time frames of the structure. If combined with physical mechanism constraints, an energy release rate threshold can also be set to consider cracks exceeding the threshold as continuing to expand, thereby embedding into the subsequent processing module to realize an evolution simulation process combining physical driving and data driving. The system maps the latent loss representation data corresponding to the predicted time frames back to the structure space grid, restoring it to a three-dimensional loss field form. This mapping process supports trilinear interpolation, nearest neighbor interpolation, or a trained back-projection decoder to realize spatial restoration of high-dimensional prediction results. The output is the structural fatigue loss simulation data at the predicted time point.

[0065] Optionally, the entropy increase rate extraction includes: According to the fracture entropy field data corresponding to the multi-time frame data, global entropy increase rate calculation and neighborhood entropy increase rate calculation are performed to obtain global entropy increase rate data and neighborhood entropy increase rate data; Specifically, the system input is a series of fracture entropy field data sequences arranged in chronological order, denoted as wherein represents the local fracture entropy value of the structure at the spatial position in the time frame. This data set reflects the change process of local microscopic energy state of the structure during the loading cycle. On a global scale, the system calculates the voxel mean value of each frame entropy field and further estimates the growth rate of the entropy average value over time. The specific calculation formula is: wherein is the global entropy increase rate value, is the total number of time frames, is the current time frame number, is the voxel mean value of the fracture entropy field in the frame, is the spatial mean value of the fracture entropy field in the frame, is the local fracture entropy value at the spatial position in the +1 time frame, local break-up entropy value at spatial position at time frame is the time interval between adjacent time frames. On the local scale, the system extracts the local entropy increment between adjacent time frames for each grid point , and calculates the neighborhood entropy increment rate based on the local entropy increment. The definition of the local entropy increment rate is as follows: , is the local entropy increment rate value, is the local break-up entropy value at spatial position at time frame is the local break-up entropy value at spatial position at time frame is the local break-up entropy value at spatial position at time frame is the time interval between adjacent frames. The neighborhood entropy increment rate calculation also includes calculating the local entropy increment of the adjacent space to obtain the local spatial entropy increment of the current frame; based on the local spatial entropy increment between adjacent time frames, the neighborhood entropy increment rate calculation is performed to obtain the neighborhood entropy increment rate data. The output results include two parts: (1) the global entropy increment rate value, which is a single scalar index; (2) the neighborhood entropy increment rate atlas, which is a three-dimensional spatial distribution data representing the spatial distribution of the growth trend of the structure internal position unit or neighborhood difference energy state.

[0066] According to the global entropy increment rate data and the neighborhood entropy increment rate data, entropy increment time-varying variance calculation is performed to obtain entropy increment time-varying variance data; Specifically, based on the generated neighborhood entropy increment rate atlas sequence , the system constructs a sliding time window structure to extract local time fluctuation features. The time window size can be set to positive and negative w time frames (such as w=3), then each sliding time corresponds to a window , which is 2w+1 local entropy increment rate data of time frames. The system performs local time series variance statistics for each spatial position , and defines the entropy increment time-varying variance index of the position within the window. Specifically, assuming that the entropy increment rate sequence of the point within the current window is , then its time-varying variance is defined as: , wherein represents the variance of the entropy increment rate change of the spatial position within the current window. The system can optionally use the exponential weighted sliding variance method for processing, that is, a higher weight is given to the time points close to the current frame. The output is the entropy increment time-varying variance map data, which identifies the strength of the position energy state fluctuation on the three-dimensional space.

[0067] According to the entropy increment time-varying variance data, a local entropy flow map is constructed to obtain local entropy flow map data; Specifically, the system extracts the spatial gradient of the entropy increase variance value at each position in the three-dimensional space based on the generated entropy increase time-varying variance atlas. Specifically, for each spatial grid point, its gradient vector is composed of partial derivative arrays in three directions, that is, the entropy increase variance change rate in the x direction, the y direction and the z direction is calculated respectively, so as to obtain a local gradient vector field. The gradient vector can be regarded as a local directionality index representing the diffusion trend of the entropy disturbance, that is, the entropy flow direction. The system constructs a local streamline structure based on the above gradient vector field. At each initial point, the current gradient direction is taken as the initial speed, the next position point is calculated through a fourth-order integration process, and the direction vector is updated until the preset termination condition (such as reaching the boundary, the direction gradient amplitude tends to zero or the trajectory length exceeds the threshold value) is reached. The path is the local entropy flow path reflecting the propagation trend of the entropy disturbance. As an alternative, the system can also construct a graph structure based on the grid nodes: taking each three-dimensional grid point as a node in the graph, a directed edge is established between adjacent nodes, the direction of the edge is indicated by the gradient vector, and the weight of the edge is set to the gradient modulus or its normalized value. The system forms a complete local entropy flow graph structure, denoted as wherein represents a node set, corresponding to the spatial discrete grid points; represents a directed edge set, representing the propagation direction of the local entropy disturbance; represents an edge weight set.

[0068] According to the local entropy flow graph data, cross-resolution entropy fusion is performed to obtain the entropy increase rate data.

[0069] Specifically, the system constructs a multi-resolution representation based on the entropy increase time-varying variance atlas obtained in the previous step. A Gaussian pyramid construction method is used to perform down-sampling processing of the entropy variance atlas at multiple scales, generating a sequence of entropy variance atlas layers at different spatial scales. Each layer of the atlas represents the observation of local energy state fluctuations at different spatial resolutions. The system introduces a structure guiding factor as a weighted reference in the fusion process, forming a spatially adaptive entropy variance fusion strategy. The structure guiding factor can be selected as an index closely related to the structure stress state, such as a local shear stress gradient map, a historical stress fluctuation atlas, or a weight map based on the node centrality of the graph structure. The system normalizes the above candidate indicators and uses a weighted combination strategy to form a spatial guiding weight map. The weighted sum of all scale layers at the corresponding points is performed to generate the fused entropy increase rate map. The system maps the fused entropy increase rate atlas value to the graph nodes of the local entropy flow graph structure as the updated attribute information of each node. The output is a fusion result map, which is represented as three-dimensional spatial distribution of entropy increase rate fusion data.

[0070] Optionally, the risk area hot spot evolution includes: According to the multi-time frame data, a fatigue concentrated area is selected to obtain hot spot data; Specifically, the system input is a sequence of multi-time frame fatigue loss images, each frame being a fatigue intensity spatial distribution map of the structure at a specific loading cycle stage. Each map corresponds to a two-dimensional spatial field representing the local fatigue accumulation degree of each position on the structure surface, denoted as a fatigue loss image sequence. The system sets a fatigue intensity threshold in each time frame. This threshold can be adaptively set in the following ways: one is to use the upper percentile of the fatigue intensity value of the whole map, for example, the upper 15% quantile value; the second is to use the mean value of the fatigue intensity of each frame image plus twice the standard deviation as the dynamic threshold. Based on the threshold set above, the system marks the pixel points in each frame image that satisfy the fatigue intensity greater than or equal to the threshold. The system uses the 8-neighbor connectivity judgment criterion to merge adjacent high fatigue intensity pixels into a hot spot region and performs numbering and boundary filling processing to extract continuous fatigue concentration regions with certain area and shape characteristics. The system outputs the hot spot mask image of each frame as a binary image consistent with the size of the original fatigue map, where the hot spot region pixel value is 1 and the rest is 0.

[0071] Specifically, further, the hot spot region is selected according to the multi-time frame data through the preset screening parameter data to obtain preliminary hot spot data; the preliminary hot spot data is subjected to elastic hot spot migration to obtain elastic hot spot data; and the elastic hot spot data is subjected to multi-frame cooperative potential field processing to obtain hot spot data. The system obtains stress response data (such as equivalent stress, fracture entropy value, crack density, etc.) of each grid element in the structure model at consecutive time frames, and constructs a mechanical response tensor of the three-dimensional structure in time sequence. For each time frame, the system sets the following hot spot screening parameters: when the equivalent stress value of a certain grid element is higher than 1.2 times the average stress of the whole structure, it is determined as an abnormal region; when the fracture entropy value of a certain element is higher than 1.5 times the average fracture entropy of the structure, it is considered as a locally active region of damage evolution; when the crack density exceeds 0.8, it is considered as a crack aggregation area. The grid elements that meet any of the conditions are determined as potential hot spots, and the preliminary hot spot region set is output. For each preliminary hot spot element, a spatial neighborhood is constructed, which can be set as a local cubic region with the element as the center and a radius of 1 to 2 grid elements; the frequency of hot spot element appearance in the neighborhood of the element is counted within a certain number of frames (such as k frames before and after), forming a time sequence activity frequency index of the element; if a certain element does not meet the hot spot judgment standard in the current time frame, but the hot spot active signal frequently appears in its neighborhood, the system will compensate it to be included in the hot spot region; if the hot spot state of a certain element appears only in less than 3 time frames, it is considered to be caused by incidental disturbance, and the system excludes it from the hot spot judgment. Constructing the hot spot intensity map. Assign 1 to all hot spot cells in the three-dimensional space of the structure and 0 to non-hot spot cells to form the hot spot state tensor at the time frame; for each time frame, the system performs spatial weighting on the hot spot region based on the Gaussian kernel function to calculate the local hot spot density distribution function; the system accumulates the hot spot density data of all time frames to form the cooperative potential function. The time weighting factor is obtained in an exponential decay manner, giving higher weight to data closer to the current frame. The weight of a certain time frame can be set as wherein represents the current time frame number, represents the historical time frame number, is an adjustment parameter for controlling the decay rate; the system performs local maximum value detection and threshold filtering on the cooperative potential function. Only the spatial regions that meet the following two conditions are retained: they are local extreme points in the cooperative potential function, and their potential values exceed the preset threshold. The system determines these regions as the hot spot region set.

[0072] Contour extraction is performed on the hot spot data to obtain hot spot contour data; Specifically, the system performs edge extraction operation on the hot spot mask map corresponding to each time frame. Preferably, an edge detection algorithm (such as Canny operator) is used to identify the boundary line between the hot spot region and the background; a contour extraction method (such as the findContours function in OpenCV) is called to analyze the edge map and extract the boundary point set of the closed contour. Based on the fact that there are multiple spatially discontinuous hot spot regions in the actual structure, the system performs contour extraction on each independent connected hot spot region and establishes an independent contour point set for each region to form a contour set. For contours with a large number of boundary points, the system takes the first and last two points in the contour point set as the initial line segment. Calculate the perpendicular distance of the remaining points to the line segment to find the point with the maximum distance. If the maximum distance is greater than a preset threshold, the point is considered as a reserved point, and the current point set is divided into two segments, and recursive compression processing is performed on each segment; if the maximum distance is less than or equal to the threshold, the intermediate points are considered redundant and can be approximated as a straight line, and the intermediate points are removed, leaving only the first and last two points; all reserved points will form a boundary point sequence that is more concise than the original contour, achieving feature simplification of the original boundary point set. The system outputs the hot spot contour data set of each frame. The data set contains multiple contours, each of which consists of a set of ordered boundary point pairs.

[0073] According to the hot spot contour data, the barycenter trajectory is extracted to obtain the hot spot barycenter data; Specifically, the system performs center of gravity calculation on the hot spot contours extracted in each frame. For any hot spot contour, the boundary of which is composed of a number of pixel points, the system respectively averages the horizontal and vertical coordinates of all boundary points to obtain the geometric center position of the hot spot contour in the frame, denoted as the center of gravity coordinates of the hot spot. The center of gravity can be represented as a pair of numerical values, i.e. the center coordinates in the horizontal and vertical directions. In order to track the evolution trajectory of the same hot spot in different time frames, the system adopts a cross-frame matching strategy. Preferably, the maximum Intersection over Union (IoU) matching method or the minimum Euclidean distance matching method is used to correspondingly match the center of gravity positions between the current frame and the next frame, thereby establishing the hot spot identity association relationship between consecutive frames. The system establishes the complete center of gravity trajectory sequence for each initial hot spot in the entire time sequence. The trajectory is composed of center of gravity coordinates arranged in time sequence. When the value is missing, the system can choose to complete the missing position in the center of gravity trajectory using linear interpolation, or mark it as a "trajectory interruption" state.

[0074] According to the dynamics evolution of the hot spot center of gravity data, the risk area hot spot data is obtained.

[0075] Specifically, the system performs derivative calculation on each hot spot trajectory to extract its motion features. The system performs difference operation on the center of gravity coordinates between adjacent time frames to calculate the motion velocity vector at each frame time; the system calculates the acceleration vector based on the change trend of the velocity vector. Based on the above velocity and acceleration data, the system constructs a dynamics evolution model of the hot spot area. If a certain hot spot area shows a significant speed growth trend in consecutive time frames, or its center of gravity continuously accelerates in a certain direction, then the area is in a state of fatigue evolution aggravation. The system sets a threshold for speed or acceleration, or directly performs clustering calculation for classification, for example, when the hot spot speed exceeds the preset upper limit, the risk label of the area is triggered. The system classifies all hot spot areas according to risk levels. The classification indicators include but are not limited to the total length of the trajectory, the average or maximum speed, the acceleration change rate, and the historical entropy growth rate. The system can perform weighted calculation on the hot spot area according to the above multi-dimensional indicators (weights generated based on empirical data for regression calculation), and divide it into high-risk, medium-risk or low-risk areas according to the risk assessment information. The system outputs the hot spot data structure containing the risk assessment information. Each risk hot spot object includes its geometric center coordinates, complete trajectory sequence, corresponding risk level, and other related attributes.

[0076] Optionally, the present application also provides an artificial intelligence-based long-life pavement structure optimization system for executing the artificial intelligence-based long-life pavement structure optimization method as described above, the artificial intelligence-based long-life pavement structure optimization system comprising: The structural mechanics model construction module is configured to acquire pavement structure data, construct a structural mechanics model according to the pavement structure data, and obtain the structural mechanics model. The fatigue loss simulation module is configured to simulate fatigue loss according to the structural mechanics model, and obtain fatigue loss data. The rapid physical simulation module is configured to perform rapid physical simulation according to the fatigue loss data, and obtain loss simulation data. The physical prior life prediction module is configured to perform physical prior life prediction according to the loss simulation data, and obtain life prediction data.

[0077] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, the scope of the application being defined by the appended application file and not by the above description, and it is intended to encompass all variations falling within the meaning and scope of the equivalent requirements of the application file.

[0078] The above description is merely a specific implementation of the present application, enabling those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for optimizing long-life pavement structures based on artificial intelligence, characterized in that, The method includes: S1. Obtain pavement structure data, construct a structural mechanics model based on the pavement structure data, and obtain the structural mechanics model; S2. Perform fatigue loss simulation based on the structural mechanics model to obtain fatigue loss data; S3. Perform rapid physical simulation based on fatigue loss data to obtain loss simulation data; S4. Based on the loss simulation data, perform physical prior lifetime prediction to obtain lifetime prediction data.

2. The method according to claim 1, characterized in that, S1 includes: Obtain pavement structure data, divide the pavement structure into structural layers based on the pavement structure data, and obtain structural layer data, which includes surface layer data, intermediate layer data, base layer data, and subgrade data; By setting inter-layer conditions for the structural layer data, an inter-layer model is obtained. Physical conditions are set for the interlayer model to obtain the structural physical model; By setting material property degradation conditions for the structural physical model, a structural mechanical model is obtained.

3. The method according to claim 1, characterized in that, S2 include: Fracture entropy field data is obtained by generating the fracture entropy field based on the structural mechanics model. Crack density was calculated based on the structural mechanics model to obtain crack density data; The structural mechanics model was annotated based on fracture entropy field data and crack density data to obtain fatigue loss data.

4. The method according to claim 3, characterized in that, The generation of the fracture entropy field includes: Based on the structural mechanics model, the stress distribution of the structural layer data under external load disturbance is locally scanned to obtain local scan data. Based on the local scan data, the equivalent stress fluctuation amplitude, shear stress gradient, and bond interface instability index were calculated, and the equivalent stress fluctuation amplitude data, shear stress gradient data, and bond interface instability data were obtained respectively. Discrete meshes are generated based on equivalent stress fluctuation amplitude data, shear stress gradient data, and bond interface instability data to obtain mesh microstate data. The fracture entropy field data is obtained by calculating the local fracture entropy based on the grid microstate data.

5. The method according to claim 3, characterized in that, The crack density calculation includes: Crack data were obtained by using a structural mechanics model to determine crack criterion. Based on the crack data, the number density and length density are calculated to obtain the number density data and length density data. Spatial crack density map data is obtained by mapping spatial crack density map based on quantity density data; Local density mutation data are obtained by extracting local density mutation data from the spatial crack density map data. The crack density data is obtained by integrating the number density data, length density data, and local density abrupt change data.

6. The method according to claim 5, characterized in that, The model annotation includes: Crack direction features are extracted from crack data to obtain crack direction feature data; Anisotropic features are extracted based on fracture entropy field data and crack direction feature data to obtain anisotropic feature data. The fracture entropy field data, crack density data, and anisotropic characteristic data are used to annotate the structural mechanics model to obtain fatigue loss data.

7. The method according to claim 1, characterized in that, S3 include: Multi-time frame data is extracted from fatigue loss data to obtain multi-time frame data; Crack evolution slope data is obtained by calculating the crack evolution slope based on multi-time frame data. Entropy growth rate data is obtained by extracting the entropy growth rate data from the fracture entropy field data corresponding to multiple time frames. The evolution of hot spots in the risk area is analyzed based on multi-time frame data to obtain hot spot data of the risk area; Based on crack evolution slope data, entropy growth rate data, and hot spot data in the risk area, a reduced-order modeling spatial expansion simulation was performed to obtain loss simulation data.

8. The method according to claim 7, characterized in that, The entropy growth rate extraction includes: The global entropy growth rate and the neighborhood entropy growth rate are calculated based on the fracture entropy field data corresponding to the multi-time frame data, and the global entropy growth rate data and the neighborhood entropy growth rate data are distributed. The entropy increase time-varying variance is calculated based on the global entropy growth rate data and the neighborhood entropy growth rate data to obtain the entropy increase time-varying variance data; Local entropy flow graphs are constructed based on entropy-increasing time-varying variance data to obtain local entropy flow graph data; Entropy growth rate data is obtained by performing cross-resolution entropy fusion based on local entropy flow graph data.

9. The method according to claim 7, characterized in that, The evolution of hotspots in the risk area includes: Fatigue concentration areas are selected based on multi-time frame data to obtain hot spot data; Contour extraction is performed on the hot spot data to obtain hot spot contour data; The centroid trajectory is extracted based on the hot spot contour data to obtain the hot spot centroid data; Dynamic evolution was performed based on hotspot centroid data to obtain hotspot data for the risk area.

10. A long-life pavement structure optimization system based on artificial intelligence, characterized in that, For executing the AI-based long-life pavement structure optimization method as described in claim 1, the AI-based long-life pavement structure optimization system comprises: The structural mechanics model building module is used to acquire pavement structure data and build a structural mechanics model based on the pavement structure data to obtain the structural mechanics model. The fatigue loss simulation module is used to simulate fatigue loss based on the structural mechanics model and obtain fatigue loss data. The rapid physical simulation module is used to perform rapid physical simulations based on fatigue loss data to obtain loss simulation data. The physical prior lifetime prediction module is used to predict physical prior lifetime based on loss simulation data, and obtain lifetime prediction data.

Images