A method for calculating fatigue reliability of cement concrete pavement with data and physical fusion

Abstract

The application relates to a cement concrete pavement fatigue reliability calculation method based on data and physics fusion. The method comprises the following steps: predicting each temperature load data in a prediction time period according to axle load data and environment data of a target road in a target time period; for each group of structure parameters of the target road, a pre-trained target network model is used to process a target data set to generate a stress mapping table corresponding to the structure parameters, the target data set is obtained by combining the structure parameters and various working condition data, and the stress mapping table is used to represent stress values corresponding to the structure parameters and the working condition data; according to the stress mapping table, target stress data corresponding to each temperature load data is determined, according to each target stress data and a fatigue life prediction model, fatigue failure prediction results corresponding to the structure parameters are determined; and according to the fatigue failure prediction results corresponding to each structure parameter, a reliability evaluation result of the target road is determined.

Description

Technical Field

[0001] This application relates to the field of road engineering technology, and in particular to a method for calculating the fatigue reliability of cement concrete pavement by integrating data and physics. Background Technology

[0002] With the rapid development of road transportation, cement concrete pavement has been widely used in heavy traffic scenarios such as highways, airport runways, and ports due to its advantages of high load-bearing capacity, good durability, and long service life. However, cement concrete pavement is subjected to the combined effects of vehicle loads and environmental factors over a long period of time, making it highly susceptible to structural problems such as fatigue cracking and slab breakage, which seriously affect driving safety and service life.

[0003] In related technologies, when faced with various road conditions, the reliability performance of the road surface can only be evaluated based on empirical formulas or deterministic mechanical models, using material parameters, structural dimensions and traffic loads as fixed values ​​and introducing empirical safety factors. This results in low accuracy of the determined road surface reliability. Summary of the Invention

[0004] Therefore, it is necessary to provide a data- and physics-integrated method for calculating the fatigue reliability of cement concrete pavements to address the aforementioned technical problems.

[0005] Firstly, this application provides a data- and physics-integrated method for calculating the fatigue reliability of cement concrete pavements, including:

[0006] Based on the axle load data and environmental data of the target road within the target time period, predictions are made to obtain the temperature load data for each time period within the prediction time period.

[0007] For each set of structural parameters of the target road, a pre-trained target network model is used to process the target dataset to generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with various working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data.

[0008] Based on the stress mapping table, the target stress data corresponding to each of the temperature load data is determined, and based on each of the target stress data and the fatigue life prediction model, the fatigue failure prediction result corresponding to the structural parameters is determined.

[0009] Based on the fatigue failure prediction results corresponding to each of the structural parameters, the reliability assessment results of the target road are determined.

[0010] In one embodiment, the target time period includes multiple historical time units, and the prediction time period includes multiple prediction time units. The step of predicting based on axle load data and environmental data of the target road within the target time period to obtain temperature load data for each prediction time period includes:

[0011] Statistical analysis is performed on the axle load data to determine the historical axle loads within each axle load interval in each historical time unit;

[0012] Based on the historical axle loads within each axle load interval and the traffic growth data within the predicted time period, the target axle loads for each predicted time unit are determined.

[0013] Within each of the prediction time units, the target axle and each of the historical axles are processed to determine the corresponding target axle load data, and the target axle and temperature gradient distribution function are processed to determine the corresponding first predicted temperature gradient and second predicted temperature gradient.

[0014] The target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient of each predicted time unit are fused to construct temperature load data.

[0015] In one embodiment, processing the target axle number and each of the historical axle numbers to determine the corresponding target axle load data includes:

[0016] The ratio of the historical axle count within each axle load interval to the total historical axle count within its respective historical time unit is determined as the probability data corresponding to each axle load interval within the historical time unit.

[0017] The probability data and the target axle are fused together to determine the target axle load data for each prediction time unit.

[0018] In one embodiment, processing the target axis and temperature gradient distribution function to determine the corresponding first predicted temperature gradient and second predicted temperature gradient includes:

[0019] Using the target axis as the sampling number, sampling is performed according to the temperature gradient distribution function to determine the first predicted temperature gradient.

[0020] Sampling is performed on the historical temperature gradients corresponding to each of the first predicted temperature gradients to determine the corresponding second predicted temperature gradients.

[0021] In one embodiment, the environmental data includes historical meteorological data and historical solar radiation data. Before the step of processing the target axis and temperature gradient distribution function to determine the corresponding first predicted temperature gradient and second predicted temperature gradient, the method further includes:

[0022] By using a temperature transfer model, the historical meteorological data and the historical solar radiation data are processed to determine the temperature distribution data of the target road at each historical moment;

[0023] The temperature distribution data at each historical moment are processed to determine the first historical temperature gradient and the second historical temperature gradient for each historical time unit.

[0024] Statistical analysis is performed on each of the first historical temperature gradients to determine the temperature gradient distribution function corresponding to each historical time unit.

[0025] In one embodiment, the stress mapping table includes a first stress mapping table and a second stress mapping table, and the target stress data includes first target stress data and second target stress data. The step of determining the target stress data corresponding to each of the temperature load data based on the stress mapping table includes:

[0026] Based on the temperature load data, a search is performed in the first stress mapping table to determine the corresponding first target stress data;

[0027] Based on the temperature gradient data in each of the temperature load data, a search is performed in the second stress mapping table to determine the corresponding second target stress data.

[0028] In one embodiment, determining the fatigue failure prediction result corresponding to the structural parameters based on the target stress data and fatigue life prediction model includes:

[0029] Using the fatigue life prediction model, the first target stress data and the second target stress data are processed to determine the theoretical fatigue life of the temperature load data;

[0030] Based on the theoretical fatigue life and the number of times the corresponding temperature load data appears within the predicted time period, the fatigue damage data of the temperature load data is determined.

[0031] The fatigue damage data corresponding to each temperature load data in the structural parameters are superimposed to determine the cumulative fatigue damage data.

[0032] Based on the relationship between the cumulative fatigue damage data and the preset threshold, the fatigue failure prediction result corresponding to the structural parameters is determined.

[0033] In one embodiment, determining the reliability assessment result of the target road based on the fatigue failure prediction results corresponding to each of the structural parameters includes:

[0034] Among the structural parameters, the fatigue failure prediction result is the target number of fatigue failures.

[0035] The ratio of the target quantity to the total number of structural parameters is determined as the fatigue failure probability of the target road, and the reliability assessment result is determined based on each fatigue failure probability.

[0036] In one embodiment, prior to the step of processing the target dataset using a pre-trained target network model, the method further includes:

[0037] Obtain a sample dataset, which includes multiple sample data, each of which is obtained by combining sample structure parameters with multiple sample operating condition data;

[0038] The sample dataset is processed using the finite element model of the target road to obtain the actual stress data;

[0039] The sample data are processed using the network model to be trained to obtain the predicted stress data;

[0040] Based on the difference between the actual stress data and the actual stress data, the network model to be trained is trained to obtain a pre-trained target network model.

[0041] Secondly, this application also provides a data-physical fusion device for calculating the fatigue reliability of cement concrete pavement, comprising:

[0042] The prediction module is used to predict the axle load data and environmental data of the target road within the target time period, and obtain the temperature load data for each time period within the prediction time period.

[0043] The generation module is used to process the target dataset using a pre-trained target network model for each set of structural parameters of the target road, and generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with multiple working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data.

[0044] The first determining module is used to determine the target stress data corresponding to each of the temperature load data according to the stress mapping table, and to determine the fatigue failure prediction result corresponding to the structural parameters according to each of the target stress data and the fatigue life prediction model.

[0045] The second determining module is used to determine the reliability assessment result of the target road based on the fatigue failure prediction results corresponding to each of the structural parameters.

[0046] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0047] Based on the axle load data and environmental data of the target road within the target time period, predictions are made to obtain the temperature load data for each time period within the prediction time period.

[0048] For each set of structural parameters of the target road, a pre-trained target network model is used to process the target dataset to generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with various working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data.

[0049] Based on the stress mapping table, the target stress data corresponding to each of the temperature load data is determined, and based on each of the target stress data and the fatigue life prediction model, the fatigue failure prediction result corresponding to the structural parameters is determined.

[0050] Based on the fatigue failure prediction results corresponding to each of the structural parameters, the reliability assessment results of the target road are determined.

[0051] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0052] Based on the axle load data and environmental data of the target road within the target time period, predictions are made to obtain the temperature load data for each time period within the prediction time period.

[0053] For each set of structural parameters of the target road, a pre-trained target network model is used to process the target dataset to generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with various working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data.

[0054] Based on the stress mapping table, the target stress data corresponding to each of the temperature load data is determined, and based on each of the target stress data and the fatigue life prediction model, the fatigue failure prediction result corresponding to the structural parameters is determined.

[0055] Based on the fatigue failure prediction results corresponding to each of the structural parameters, the reliability assessment results of the target road are determined.

[0056] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0057] Based on the axle load data and environmental data of the target road within the target time period, predictions are made to obtain the temperature load data for each time period within the prediction time period.

[0058] For each set of structural parameters of the target road, a pre-trained target network model is used to process the target dataset to generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with various working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data.

[0059] Based on the stress mapping table, the target stress data corresponding to each of the temperature load data is determined, and based on each of the target stress data and the fatigue life prediction model, the fatigue failure prediction result corresponding to the structural parameters is determined.

[0060] Based on the fatigue failure prediction results corresponding to each of the structural parameters, the reliability assessment results of the target road are determined.

[0061] The above-mentioned method for calculating the fatigue reliability of cement concrete pavement by integrating data and physics can first predict the axle load data and environmental data of the target road within a target time period to obtain the temperature load data within the prediction time period. Then, for each set of structural parameters of the target road, a pre-trained target network model is used to process the target dataset to generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining structural parameters with multiple working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data. Then, according to the stress mapping table, the target stress data corresponding to each temperature load data is determined. According to each target stress data and the fatigue life prediction model, the fatigue failure prediction result corresponding to the structural parameters is determined. Finally, the reliability assessment result of the target road can be determined according to the fatigue failure prediction result corresponding to each structural parameter. Therefore, based on the statistical patterns within the target time period, load-temperature coupled temperature data for the predicted time period can be generated. Then, by using multiple randomly generated sets of structural parameters and combining them with the target network model, a stress mapping table covering all possible working conditions can be constructed. Subsequently, fatigue life prediction is performed based on the stress mapping table, and the fatigue failure prediction results of each set of structural parameters are statistically analyzed to determine the reliability assessment results of the target road. In the above process, the influence of the randomness of structural parameters and the synergistic effect of load and temperature on pavement fatigue performance is fully considered. By using the stress mapping table to determine the target stress data, the limitation of large-scale random simulation on computing resources is effectively avoided, thereby effectively improving the computational efficiency and accuracy of pavement fatigue reliability assessment results. Attached Figure Description

[0062] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0063] Figure 1 This is a flowchart illustrating a data-physical fusion method for calculating the fatigue reliability of cement concrete pavement in one embodiment.

[0064] Figure 2 This is a flowchart illustrating the process of determining temperature load data in one embodiment;

[0065] Figure 3 This is a flowchart illustrating a data-physical fusion method for calculating the fatigue reliability of cement concrete pavement in another embodiment.

[0066] Figure 4 This is a flowchart illustrating a data-physical fusion method for calculating the fatigue reliability of cement concrete pavement in yet another embodiment.

[0067] Figure 5 This is a schematic diagram of the structure of a finite element model in one embodiment;

[0068] Figure 6 This is a schematic diagram of the architecture of the network model to be trained in one embodiment;

[0069] Figure 7 This is a flowchart illustrating a data-physical fusion method for calculating the fatigue reliability of cement concrete pavement in another embodiment.

[0070] Figure 8 This is a schematic diagram illustrating the determination of shared time-history sampling in one embodiment;

[0071] Figure 9 This is a structural block diagram of a data and physics fusion calculation device for fatigue reliability of cement concrete pavement in one embodiment.

[0072] Figure 10 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0074] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0075] In one embodiment, such as Figure 1 As shown, a method for calculating the fatigue reliability of cement concrete pavement by fusing data and physics is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and to a system including both a terminal and a server, and can be implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0076] Step 102: Based on the axle load data and environmental data of the target road within the target time period, make predictions to obtain the temperature load data for each time period within the prediction time period.

[0077] The target road can be a road whose pavement performance is to be evaluated. It can be a cement concrete road or other types of roads, etc. This application does not limit it.

[0078] In addition, the target time period can be the number of years of historical monitoring or data collection, such as 1 year, 2 years, 5 years, etc., and the predicted time period can be the future design reference period or service operation period, such as 20 years, 30 years, etc. This application does not limit this.

[0079] In addition, axle load data can be the axle load of vehicles passing through the target road during the target time period; environmental data can include road surface temperature data, solar radiation data, etc., of the target road during the target time period, and this application does not limit these.

[0080] Furthermore, temperature load data can be understood as a dataset representing the coupling effect of vehicle load and road surface temperature environment at the same point in time within the prediction period. In this embodiment, temperature load data may include each prediction time unit within the prediction period, such as the target axle load data corresponding to each month or each hour, the first predicted temperature gradient, and the second predicted temperature gradient.

[0081] The first predicted temperature gradient can be a linearly varying temperature gradient component along the thickness direction of the pavement structure layer; the second predicted temperature gradient is a non-linearly varying temperature gradient component along the thickness direction of the pavement structure layer, excluding the linear component. Therefore, in this embodiment, the complex temperature field distribution inside the target road pavement can be accurately characterized using the first and second predicted temperature gradients.

[0082] It is understandable that by statistically analyzing the axle load data and environmental data obtained from the target road within the target time period, historical data characteristics can be obtained, and predictions can be made based on these historical data characteristics to obtain the target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient within the prediction time period.

[0083] Step 104: For each set of structural parameters of the target road, the target dataset is processed using a pre-trained target network model to generate a stress mapping table corresponding to the structural parameters.

[0084] The target dataset is obtained by combining structural parameters with various working condition data, and the stress mapping table can be used to characterize the stress values ​​corresponding to the structural parameters and working condition data.

[0085] In addition, structural parameters can be parameters that characterize the structural geometry and material mechanical properties of the target road. To comprehensively consider the variability of actual road construction, multiple sets of structural parameters can be obtained by randomly sampling within a preset range. Each set of structural parameters represents a possible cement concrete pavement structure, and each set of structural parameters may include the surface layer thickness, surface layer elastic modulus, and the comprehensive resilient modulus of the base course top. The comprehensive resilient modulus of the base course top can be jointly determined by the base course structure and the subgrade.

[0086] Optionally, each set of structural parameters may also include the concrete flexural strength f. r Alternatively, other parameters may be added as needed, but this application does not limit this.

[0087] In addition, the operating condition data can include reference axle load data, a first reference temperature gradient, and a second reference temperature gradient. Each set of operating condition data can be presented as a (reference axle load data, first reference temperature gradient, second reference temperature gradient) operating condition triplet. For example, the input spaces of the reference axle load data, the first reference temperature gradient, and the second reference temperature gradient can be divided into grids according to appropriate preset step sizes. For example, the grid can be divided according to a temperature gradient step size of 10℃ / m and an axle load step size of 10kN, or it can be divided according to a temperature gradient step size of 15℃ / m and an axle load step size of 20kN, or it can be divided into grids according to other preset step sizes as needed. The midpoint value of each grid interval in each dimension is taken as the node value, thereby obtaining multiple operating condition data. Then, the structural parameters of the current group can be combined with all the above operating condition data to form a target dataset that combines structural parameters with multiple operating condition data.

[0088] In addition, the target network model can be a pre-trained regression model, which can be a multi-layer perceptron (MLP), a back propagation (BP) neural network, or any other model that can perform regression processing, etc. This application does not limit this.

[0089] Understandably, given that the current structural parameters are fixed, all data in the target dataset—namely, the current surface layer thickness, surface layer elastic modulus, and the comprehensive resilient modulus at the top of the base layer—can be combined with each load case data and used as input data for the target network model. This target network model then processes the data to obtain the reference stress data corresponding to each load case. Since the load case data represents the midpoint value of a grid interval, the reference stress data corresponding to that load case can be used as the representative stress of that grid interval. Subsequently, the current structural parameters, the load case data, and the corresponding reference stress data can be stored in the form of a matrix or lookup table to generate a stress mapping table corresponding to that set of structural parameters.

[0090] Therefore, in this embodiment of the application, for each set of structural parameters, the set of structural parameters can be combined with multiple working condition data to obtain the corresponding target dataset. Then, the target network model is used to generate the stress mapping table corresponding to the set of structural parameters, providing a solid data foundation for subsequent processing.

[0091] Step 106: Based on the stress mapping table, determine the target stress data corresponding to each temperature load data, and based on each target stress data and the fatigue life prediction model, determine the fatigue failure prediction results of the structural parameters.

[0092] This method allows for the combination of current structural parameters and various temperature load data, and rapid interpolation or retrieval in the stress mapping table to determine the target stress data corresponding to each temperature load data. This avoids the need for online finite element mechanical simulation calculations for massive amounts of temperature load data, saving a significant amount of time and greatly improving computational efficiency.

[0093] In addition, the fatigue life prediction model can be used to process the determined target stress data to obtain the cumulative fatigue damage data under the set of structural parameters. If the cumulative fatigue damage data is greater than the preset threshold, it can be determined that the set of structural parameters has failed due to fatigue. If the cumulative fatigue damage data is less than or equal to the preset threshold, it can be considered that the set of structural parameters has not failed due to fatigue.

[0094] It is understandable that different structural parameters, due to their varying thicknesses, exhibit different stress response characteristics, thus requiring different stress mapping tables. Therefore, in this embodiment, a corresponding stress mapping table can be constructed for each set of structural parameters, followed by fatigue life prediction to determine whether failure occurs under each set of structural parameters, thereby fully considering the impact of road structure parameter variability on fatigue life.

[0095] Optionally, the stress mapping table may include a first stress mapping table and a second stress mapping table, and the target stress data may include first target stress data and second target stress data. The first target stress data can be determined by searching the first stress table based on the temperature load data, and the second target stress data can be determined by searching the second stress table based on the temperature gradient data in the various operating conditions.

[0096] The first stress mapping table is generated by the combined effect of axial load data and temperature data under this set of structural parameters. The corresponding first target stress data is the coupled stress of load and temperature. The second stress mapping table is generated by the effect of temperature only under this set of structural parameters. At this time, the axial load data is zero, and the corresponding second target stress data is the stress under the effect of temperature.

[0097] The temperature load data can include target axle load data, a first predicted temperature gradient, and a second predicted temperature gradient. Based on the current structural parameters, target axle load data, first predicted temperature gradient, and second predicted temperature gradient, the data can be searched in a first stress mapping table corresponding to the structural parameters to determine the corresponding first target stress data. For example, the axle load interval to which the target axle load data belongs, the first temperature gradient interval to which the first predicted temperature gradient belongs, and the second temperature gradient interval to which the second predicted temperature gradient belongs can be determined separately. Then, the stress values ​​corresponding to these intervals in the first stress mapping table can be used as the first target stress data.

[0098] Then, the corresponding second target stress data can be determined by searching the second stress mapping table corresponding to the structural parameters based on the first and second predicted temperature gradients.

[0099] Therefore, in this embodiment, for different structural parameters, a first stress mapping table and a second stress mapping table corresponding to the structural parameter can be determined firstly. Then, based on the temperature load data, the first and second stress mapping tables can be searched to determine the corresponding first and second target stress data. Since the axial load range, first temperature gradient range, and second temperature gradient range in the first and second stress mapping tables cover all possible value ranges of the axial load data, the first temperature gradient, and the second temperature gradient, the corresponding target stress data can be determined in the stress mapping table for any temperature load data.

[0100] Optionally, determining the fatigue failure prediction result corresponding to the structural parameter based on the target stress data and fatigue life prediction model may include: first, using the fatigue life prediction model to process the first target stress data and the second target stress data to determine the theoretical fatigue life of the temperature load data; then, based on the theoretical fatigue life and the number of times the corresponding temperature load data appears within the prediction time period, determining the fatigue damage data corresponding to the temperature load data; then, superimposing the fatigue damage data corresponding to the structural parameter to determine the cumulative fatigue damage data; and finally, determining the fatigue failure prediction result corresponding to the structural parameter based on the relationship between the cumulative fatigue damage data and a preset threshold.

[0101] The fatigue life prediction model can be expressed as follows:

[0102] ;

[0103] Where, N i The theoretical fatigue life is given by S, where S is the stress level and represents the first target stress data. With the flexural strength f of concrete r The ratio of can be expressed as R represents the stress ratio, which is the second target stress data. Compared with the first target stress data The ratio of can be expressed as a and b are fatigue coefficients, such as a can be 1.0 and b can be 0.057, or they can be adjusted as needed. This application does not limit them.

[0104] Optionally, each of the above sets of structural parameters may include the concrete flexural strength f. rThe concrete flexural strength f corresponding to the first target stress data and the second target stress data can be used. r This allows for the determination of theoretical fatigue life, thereby effectively ensuring the accuracy and reliability of the theoretical fatigue life determination.

[0105] In addition, the first target stress data It can also be called the maximum fatigue stress, which characterizes the coupled flexural tensile stress generated at the critical load position of the road surface when vehicle load and temperature gradient act simultaneously; the second target stress data It can also be called minimum fatigue stress, which characterizes the temperature-induced bending tensile stress at the critical load position of the road surface when there is no load and only a temperature gradient acts alone. Therefore, temperature stress .

[0106] Based on the above physical correspondence, and considering that the temperature stress at the critical load site and the second target stress are the same during a single loading process, the above fatigue life prediction model can be transformed into:

[0107] ;

[0108] in, For the first target stress data, For temperature stress, f r Let denoted as , and 'a' be the fatigue coefficient. Based on the mechanical assumptions of a thin plate on an elastic foundation, the structural contact conditions and mechanical constitutive model are linear, thus satisfying the principle of linear superposition, i.e. .

[0109] in, The stress at the critical load position is the stress generated when a vehicle load acts solely on a cement pavement. Therefore, the above fatigue life prediction model can also be expressed as:

[0110] ;

[0111] After determining the theoretical fatigue life of each temperature load using the aforementioned fatigue life prediction model, we can count the number of times the temperature load data corresponding to each theoretical fatigue life appears in all temperature load data within the prediction time period, and denote this as n. i Then, the ratio of the frequency of occurrence of the temperature load data to the corresponding theoretical fatigue life can be determined as the fatigue damage data of that temperature load data. The fatigue damage data corresponding to all temperature load data under this set of structural parameters are then superimposed, and the superposition result is taken as the cumulative fatigue damage data. The above process can be expressed as:

[0112] ;

[0113] Where, n iN represents the number of times the current temperature load data occurs within the prediction time period. i This represents the theoretical fatigue life corresponding to the current temperature load data.

[0114] In addition, the preset threshold can be a pre-set value, such as 1, or it can be adjusted as needed, etc. This application does not limit this.

[0115] Understandably, after determining the fatigue cumulative damage data corresponding to each set of structural parameters, the relationship between each fatigue cumulative damage data and a preset threshold can be compared. If the fatigue cumulative damage data is greater than the preset threshold, fatigue failure can be considered to have occurred; if the fatigue cumulative damage data is less than or equal to the preset threshold, fatigue failure can be considered not to have occurred. Thus, the fatigue failure prediction results corresponding to each set of structural parameters can be determined.

[0116] Step 108: Determine the reliability assessment result of the target road based on the fatigue failure prediction results corresponding to each structural parameter.

[0117] The reliability assessment results can be presented in various forms, such as fatigue failure probability, fatigue failure reliability, reliability index, etc. This application does not limit the presentation of these results.

[0118] Understandably, after obtaining the fatigue failure prediction results corresponding to each structural parameter, these results can be integrated to determine the reliability assessment result of the target road. For example, if the number of fatigue failures in the fatigue failure prediction results exceeds a certain threshold, it can be determined that the target road has experienced fatigue failure and has low reliability.

[0119] Optionally, one can first count the number of targets whose fatigue failure prediction results are fatigue failures among all structural parameters, then determine the fatigue failure probability of the target road by the ratio of the target number to the total number of structural parameters, and then determine the reliability assessment result based on each fatigue failure probability.

[0120] The fatigue failure probability P of the target road can be calculated in the following way. f :

[0121] ;

[0122] Where, N MC X represents the total number of structural parameter sets; （k） The kth group of structural parameters obtained from the kth sampling; For structural function, if This indicates that the pavement structure corresponding to this set of structural parameters has experienced fatigue failure. This is an indicator function; it takes a value of 1 when the condition is met, i.e., fatigue failure occurs, and a value of 0 when fatigue failure does not occur.R For reliability.

[0123] Among them, the reliability P of the target road R It can be represented as:

[0124] ;

[0125] Reliability index reflecting the mechanical safety margin of the pavement It can be represented as:

[0126] ;

[0127] in, It is the inverse function of the cumulative function of the standard normal distribution.

[0128] Therefore, in the embodiments of this application, the reliability P can be... R Reliability Indicators and the corresponding fatigue failure probability P f As the reliability assessment result of the target road, a performance evaluation report is generated, providing a reliable data foundation for subsequent structural design optimization, maintenance strategy formulation, or life cycle cost analysis of the target road.

[0129] In the above-mentioned method for calculating the fatigue reliability of cement concrete pavement by integrating data and physics, predictions can be made first based on the axle load data and environmental data of the target road within the target time period to obtain the temperature load data within the prediction time period. Then, for each set of structural parameters of the target road, the target dataset is processed using a pre-trained target network model to generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining structural parameters with multiple working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data. Then, based on the stress mapping table, the target stress data corresponding to each temperature load data is determined. Based on each target stress data and the fatigue life prediction model, the fatigue failure prediction results corresponding to the structural parameters are determined. Finally, based on the fatigue failure prediction results corresponding to each structural parameter, the reliability assessment results of the target road are determined. Therefore, based on the statistical patterns within the target time period, load-temperature coupled temperature data for the predicted time period can be generated. Then, by using multiple randomly generated sets of structural parameters and combining them with the target network model, a stress mapping table covering all possible working conditions can be constructed. Subsequently, fatigue life prediction is performed based on the stress mapping table, and the fatigue failure prediction results of each set of structural parameters are statistically analyzed to determine the reliability assessment results of the target road. In the above process, the influence of the randomness of structural parameters and the synergistic effect of load and temperature on pavement fatigue performance is fully considered. By using the stress mapping table to determine the target stress data, the limitation of large-scale random simulation on computing resources is effectively avoided, thereby effectively improving the computational efficiency and accuracy of pavement fatigue reliability assessment results.

[0130] It is understood that the target time period may include multiple historical time units, and the prediction time period may include multiple prediction time units. In an exemplary embodiment, such as... Figure 2 As shown, step 102 includes steps 202 to 208. Wherein:

[0131] Step 202: Perform statistical analysis on the axle load data to determine the historical axle loads within each axle load interval in each historical time unit.

[0132] The target time period may include multiple historical time units, such as dividing historical time into regions by month, day or hour. In this embodiment, (month, hour) is used as the time unit.

[0133] In addition, the axle load interval can be a set of intervals obtained by discretizing the original vehicle axle load measured by the dynamic weighing system or toll station according to a preset load interval. For example, it can be divided with an interval step of 10kN. Etc., multiple consecutive axle load intervals.

[0134] In addition, historical axle loads can be understood as the number of times a historical vehicle axle loads fall within each axle load range within a certain historical time unit.

[0135] Understandably, by analyzing and processing the axle load data of all vehicles passing through the monitored road segment within the target time period, and categorizing them into corresponding historical time units (month, hour) based on their recorded timestamps, the axle load interval can be determined according to the magnitude of the axle load value. For the axle load data within each historical time unit, statistical counting can be performed according to preset axle load intervals to obtain the historical axle counts within each interval. Therefore, in this embodiment, by statistically analyzing data by time unit and axle load interval, the temporal distribution characteristics and axle load distribution characteristics of traffic load are preserved, providing an accurate and reliable statistical basis for subsequent predictions.

[0136] Step 204: Determine the target axle number in each prediction time unit based on the historical axle number in each axle load interval and the traffic growth data in the prediction time period.

[0137] The forecast period may include multiple forecast time units, such as each forecast time unit corresponding to a future (month, hour). Traffic growth data may be parameters of future traffic development trends obtained based on historical traffic flow, such as the annual average growth rate, or other traffic growth parameters; this application does not limit this.

[0138] In addition, the target axle load refers to the total number of cumulative traffic axle loads expected to act on the road structure within each predicted time unit in the future.

[0139] It is understandable that, for the same historical time unit, the historical axle counts of each axle load interval within that historical time unit can be summed to obtain the historical axle counts of that historical time unit.

[0140] Optionally, the target axis number for each prediction time unit within the prediction period can be represented as follows:

[0141] ;

[0142] in, Y represents the historical axis of each historical time unit (m, h) within the target time period; o Y represents the number of years within the target time period. d Let r be the number of years in the predicted time period, and r be the annual traffic volume growth rate.

[0143] Therefore, in this embodiment of the application, the historical axes of each historical time unit can be scaled based on traffic growth data to obtain the target axes of each prediction time unit. Only the sample size is changed without changing the probability structure, thus reflecting the long-term growth trend of traffic volume with economic development and avoiding excessive assumptions about the distribution pattern of future years.

[0144] Step 206: Within each prediction time unit, process the target axle load data according to the target axle load and each historical axle load, and process the target axle load data according to the temperature gradient distribution function to determine the corresponding first predicted temperature gradient and second predicted temperature gradient.

[0145] Optionally, the ratio of the historical axle load intervals to the total historical axle loads within the corresponding historical time unit can be used as the probability data for each axle load interval within the historical time unit. The probability data can then be fused with the target axle loads to determine the target axle load data for each prediction time unit.

[0146] The probability data corresponding to each axle load interval within the aforementioned historical time unit can be expressed as follows:

[0147] ;

[0148] Where m and h are the month and hour indices, respectively, and b is the axis load range index. For historical time units (m, h) within the axle load interval b, For all axle load intervals within the historical time unit (m, h) Total historical timeline.

[0149] Optionally, the midpoint of the axle load interval can also be called the bin midpoint. The midpoint of the axle load interval can be used as the representative axle load data of the axle load interval, thus preserving all the characteristics of the original distribution and avoiding the distortion that may exist in the parameter fitting in extreme areas.

[0150] In addition, in the process of determining the target axle load data for each prediction time unit, the probability data of each axle load interval can be multiplied by the target axle number in the corresponding prediction time unit to obtain the allocated axle number for each axle load interval. Then, the representative value of each axle load interval, such as the midpoint value, is used as the axle load data to obtain the target axle load data for each prediction time unit.

[0151] Alternatively, the total number of samplings can be based on the target axle load within each prediction time unit, and the above probability data can be used as the probability of occurrence for each axle load interval to generate target axle load data within each prediction time unit.

[0152] It should be noted that the above-described method for determining the target axle load data for each prediction time unit is merely an illustrative example and should not be construed as a limitation on the method for determining the target axle load data for each prediction time unit in the embodiments of this application.

[0153] Optionally, the sampling process can be performed based on the temperature gradient distribution function, with the target axis as the sampling number, to determine the first predicted temperature gradient. Then, sampling can be performed on the historical temperature gradients corresponding to each first predicted temperature gradient to determine the corresponding second predicted temperature gradient.

[0154] Due to the randomness of the temperature gradient, the target axle number can be used as the total number of samplings. Independent random sampling can be performed according to the temperature gradient distribution function to obtain the first predicted temperature gradient corresponding to each axle load.

[0155] It is understandable that, since there are related physical constraints between the actual nonlinear temperature distribution of the road surface, i.e. the second temperature gradient, and the linear temperature distribution, i.e. the first temperature gradient, after obtaining the first predicted temperature gradient, secondary sampling can be performed on the historical temperature gradient corresponding to the first predicted temperature gradient to obtain the corresponding second predicted temperature gradient.

[0156] For example, for a certain first predicted temperature gradient obtained from the current sampling, a first historical temperature gradient that is the same as the first predicted temperature gradient can be found in the corresponding historical time unit, and the second historical temperature gradient that appears most frequently among the multiple second historical temperature gradients corresponding to the first predicted temperature gradient can be used as the second predicted temperature gradient, etc. This application does not limit this.

[0157] Optionally, the maximum and minimum values ​​of all first historical temperature gradients within the current historical time unit can be counted first, and the data can be evenly divided into multiple sub-intervals. Then, for each historical data sample within the historical time unit, based on the sub-interval in which the first historical temperature gradient falls, the corresponding second historical temperature gradient can be added to the data list of that sub-interval, thus establishing the correlation between the first and second historical temperature gradients for each sub-interval. Therefore, after determining the first predicted temperature gradient, the second predicted temperature gradient corresponding to the first predicted temperature gradient can be determined based on the correlation between the first and second historical temperature gradients.

[0158] Therefore, in this embodiment of the application, by determining the second predicted temperature gradient based on the first predicted temperature gradient, the situation that may result in a combination of temperature fields that do not exist physically could be avoided by generating linear and nonlinear temperature differences independently and randomly. This ensures that the obtained first and second predicted temperature gradients conform to the basic characteristics of atmospheric physical conduction, improves the authenticity and reliability of the data, and provides a reliable data foundation for subsequent data processing.

[0159] Step 208: The target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient of each predicted time unit are fused to construct temperature load data.

[0160] For triplet combinations where the target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient are all identical, the frequency of occurrence of this combination within the prediction time period can be counted first, and only one identical combination can be retained. Therefore, in the temperature load data obtained in this embodiment, there are no two completely identical combinations of target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient.

[0161] For example, target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient within the same prediction time unit can be associated and combined to form a triplet. Then, the values ​​in the triplet can be preprocessed, such as rounding the target axle load data to obtain an integer value and retaining the same number of decimal places for each temperature gradient data. Next, an association table can be constructed, using the preprocessed triplet as the key and the frequency of occurrence as the value. All data can be traversed; if the current triplet already exists in the association table, its frequency is incremented by 1; if it does not exist, it is treated as a new entry with its frequency recorded as 1. Finally, the deduplicated, non-repeating triplets and their corresponding frequencies are obtained, thus constituting the final temperature load data.

[0162] In this embodiment, by statistically analyzing historical axle load data by time unit and axle load interval, and combining it with traffic growth data to predict the target axle load for each future time unit, and then performing axle load sampling and temperature gradient sampling based on the target axle load, the prediction of the load-environment coupling effect within the predicted time period is achieved through the deduplicated temperature load data, providing an accurate data foundation for subsequent pavement fatigue reliability assessment.

[0163] It is understood that environmental data may include historical meteorological data and historical solar radiation data. In an exemplary embodiment, such as... Figure 3 As shown, a method for calculating the fatigue reliability of cement concrete pavement by fusing data and physics is provided, including steps 302 to 306. Wherein:

[0164] Step 302: Using a temperature transfer model, process historical meteorological data and historical solar radiation data to determine the temperature distribution data of the target road at each historical moment.

[0165] Historical meteorological data can include temperature data at various historical moments within the target time period; historical solar radiation data can be the total amount of shortwave solar radiation acting on the road surface at various historical moments.

[0166] In addition, the temperature transfer model can be a model based on the laws of thermodynamics that describes the heat transfer from the external meteorological environment to the interior of the road surface. For example, it can be a partial differential equation model constructed by using a one-dimensional transient heat conduction equation and combining the boundary conditions of heat convection and heat radiation on the road surface. This application does not limit this.

[0167] In addition, the temperature distribution data can be a numerical vector of the temperature field along the thickness, or depth, of the road surface structure at a certain historical moment. For example, it can be the corresponding temperature value at every 1cm depth position.

[0168] It is understandable that historical meteorological data and historical solar radiation data can be measured data of the target road within the target time period, or they can be determined by processing publicly available datasets based on the location information of the target road.

[0169] In one possible implementation, to ensure that the temperature boundary conditions closely resemble real climate records, when using a publicly available dataset, spatial interpolation algorithms can be employed to interpolate the measured sequences of nearby meteorological stations, combining the specific latitude and longitude coordinates of the target road (i.e., the target point), thereby reconstructing the instantaneous surface temperature and solar radiation data at the corresponding location of the target point. For example, nearby meteorological stations can be multiple stations within a preset radius of the target point in a global meteorological station network.

[0170] If the distance from the target point to the i-th nearest weather station is d i The measured temperature at the nearby meteorological station at time t is T. i (t). According to the spatial decay law of the first law of geography, the temperature T of the target road at time t can be obtained by combining inverse distance weighting (IDW). air (t) can be represented as follows:

[0171] ;

[0172] Where n is the number of neighboring stations participating in the interpolation, such as the 3 or 4 nearest stations around the target point; T represents the distance weighting coefficient for the i-th neighboring meteorological station; i (t) represents the measured temperature at time t.

[0173] Additionally, if the target road is relatively close to a specific weather station, numerical singularities may occur due to the denominator approaching zero; therefore, a distance threshold can be set. If the distance between the two is less than the preset distance threshold, the weather station can be considered to spatially overlap with the target road, and the observed temperature value T of the weather station can be used. i (t) is the temperature data for the target road.

[0174] Optionally, after calculating the target road temperature sequence using the spatial inverse distance weighted algorithm, in order to unify the time step, cubic spline interpolation can be used to perform time-series interpolation on the temperature sequence, so as to unify and align the time resolution to the hourly resolution.

[0175] In addition, the horizontal solar radiation flux data of the target road at each time point, i.e., the total solar shortwave radiation, can be obtained by retrieving satellite reanalysis product datasets. The solar shortwave radiation data of the grid node closest to the latitude and longitude location of the target road can be identified as the corresponding historical solar radiation data.

[0176] In this embodiment, the temperature transfer model can be represented as follows:

[0177] ;

[0178] in, Let be the density of the pavement structural layer material, c be the specific heat capacity of the pavement structural layer material, and k be the thermal conductivity of the pavement structural layer material. This represents the temperature at time t and depth z within the road panel; z is the coordinate along the depth direction of the road panel.

[0179] The surface of the road, i.e., at z=0, can comprehensively reflect three heat exchange mechanisms: energy input from solar shortwave radiation, convective heat transfer between the surface and the atmosphere, and longwave radiation heat dissipation from the surface to the sky. Its boundary conditions can be expressed as:

[0180] ;

[0181] Where k is the thermal conductivity of the pavement structure layer material. h represents the surface solar radiation absorptivity. c Indicates the convective heat transfer coefficient. Indicates the long-wavelength emissivity of the surface. This is the Stefan-Boltzmann constant; This represents historical solar radiation data. T represents the target road temperature value at time t obtained through interpolation. s This represents the solution value at the road surface temperature, i.e., T(0, t). sky The equivalent temperature of the sky.

[0182] The lower surface of the road surface, i.e. the ground boundary, is placed in a sufficiently deep subgrade to satisfy the deep isothermal assumption, and the temperature at its boundary remains constant over time, or is set as an adiabatic boundary condition with zero heat flux density, etc. This application does not limit this.

[0183] Therefore, in this embodiment, the Crank-Nicolson implicit finite difference scheme can be used to solve the temperature transfer model. To accurately capture the drastic temperature fluctuations and nonlinear mechanical response characteristics of the top surface of the slab under the strong coupling boundary conditions of radiation and convection, a non-uniform meshing method can be used along the slab thickness direction (i.e., the z-axis) during mesh generation, with mesh refinement in the region near the top surface. By solving the partial differential equations, the nonlinear temperature distribution data of the concrete panel along the depth direction at various times can be calculated. .

[0184] Step 304: Process the temperature distribution data at each historical moment to determine the first historical temperature gradient and the second historical temperature gradient for each historical time unit.

[0185] Understandably, based on the design requirements and historical research of cement concrete pavements, the nonlinear temperature distribution of cement pavements has a significant impact on the stress of the pavement panels. To quickly obtain nonlinear distribution data within the prediction timeframe, a quadratic function can be used to fit the nonlinear temperature distribution of the cement pavement.

[0186] Since the temperature warping stress of the concrete panel is independent of the set zero temperature point, only two temperatures at a defined depth are needed to derive a quadratic function to express the nonlinear temperature. In this embodiment, two characteristic temperature gradients can be extracted from the temperature profile at each historical moment to fit the nonlinear distribution, namely the full-thickness temperature gradient and the half-thickness temperature gradient.

[0187] The full-thickness temperature gradient, which is the overall temperature difference gradient between the top and bottom of the plate, is used to characterize the main component of linear change. The half-thickness temperature gradient, which is the temperature difference gradient between the middle and bottom of the plate, can be expressed as follows:

[0188] ;

[0189] ;

[0190] Where h is the thickness of the surface layer. Let be the temperature at the top of the slab at time t, which is the temperature value at a depth z=0 from the road surface. Let be the temperature at the bottom of the slab at time t, which is the temperature value at a depth z = h from the road surface. Let be the temperature in the slab at time t, which is the temperature value at a depth z = 1 / 2h from the road surface. It can be obtained by linear interpolation of the temperature profile at the midpoint of the plate thickness.

[0191] Optionally, by processing the temperature distribution data at each historical moment to obtain the full-thickness temperature gradient and half-thickness temperature gradient corresponding to each historical moment, the full-thickness temperature gradient and half-thickness temperature gradient corresponding to each historical moment can be used as the first historical temperature gradient and the second historical temperature gradient of the corresponding historical time unit according to the month and hour to which each historical moment belongs.

[0192] Therefore, in this embodiment of the application, by using the fitting method of full-thickness temperature gradient and half-thickness temperature gradient, the infinite-dimensional temperature curve is decoupled into a calculable standard parameter, which preserves the nonlinear physical characteristics of the temperature gradient and provides good data conditions for subsequent fatigue reliability assessment.

[0193] Step 306: Perform statistical analysis on each first historical temperature gradient to determine the temperature gradient distribution function corresponding to each historical time unit.

[0194] Among them, the temperature gradient distribution function can be the probability density function of the probability of the temperature gradient occurring within a certain historical time unit. The historical time unit can be a discrete time interval formed by grouping according to the periodicity of meteorological and physical characteristics in two layers: month and hour. The month dimension can reflect seasonal changes, and the hour dimension can reflect the diurnal pattern.

[0195] Since the changes in date and year are relatively weak compared to the seasonal changes in the atmosphere (monthly dimension) and the periodic changes in the alternation of day and night (hourly dimension), this embodiment treats the temperature gradient fluctuations caused by different years and different days in the same month as random disturbances, and regards the differences in months and hours as deterministic spatiotemporal structures.

[0196] For any historical time unit (m, h), the first set of historical temperature gradients contained in that historical time unit can be statistically analyzed. Since the road surface temperature gradient exhibits a symmetrical distribution at night, but shows a significant right-skewed, long-tailed characteristic during the day due to the influence of strong solar radiation, this embodiment uses a parametric skewed normal distribution (Skew-Normal Distribution) to perform maximum likelihood estimation (MLE) fitting on the first set of historical temperature gradients to construct the temperature gradient distribution function corresponding to that historical time unit.

[0197] Specifically, a partially normal distribution can be used to fit the first historical temperature gradient of each historical time unit using maximum likelihood estimation, which can be expressed as follows:

[0198] ;

[0199] in, The first historical temperature gradient to be fitted; It is the standard normal density function. With cumulative distribution function, For unknown parameters, For scale parameters, For shape parameters.

[0200] By solving the problem using maximum likelihood estimation, we can obtain the partially normal distribution parameter set of the first historical temperature gradient corresponding to each historical time unit. Ultimately, a set of parameters storing the aforementioned partially normal distribution can be constructed over all 288 historical time units. A two-dimensional lookup table, where each set of partially normal distribution parameters corresponds to a temperature gradient distribution function.

[0201] Optionally, the maximum and minimum values ​​of all first historical temperature gradients within the current historical time unit can be counted first, and the gradients can be evenly divided into multiple sub-intervals. Then, each set of historical temperature gradients within the historical time unit can be traversed. Based on the sub-interval in which the first historical temperature gradient falls, the second historical temperature gradient in the set of historical temperature gradients can be placed into the data list corresponding to the sub-interval, thereby establishing the correlation between the sub-interval in which each first historical temperature gradient is located and the second historical temperature gradient.

[0202] After determining the first predicted temperature gradient within the prediction time period, a two-dimensional lookup table can be used, with the month and hour corresponding to the current prediction time unit as indices, to determine the corresponding skewed normal distribution parameter set. Random sampling is then performed based on this temperature gradient distribution function to obtain the first predicted temperature gradient. Next, the sub-interval containing the first predicted temperature gradient data is determined, and a second historical temperature gradient is randomly selected from the data list corresponding to this sub-interval as the second predicted temperature gradient to match the first predicted temperature gradient obtained above.

[0203] In this embodiment, by performing conditional sampling on the second temperature gradient, the physical consistency between the generated first and second predicted temperature gradients is ensured, effectively avoiding the problem of temperature distribution curve distortion caused by independent random sampling. By decoupling the infinite-dimensional nonlinear temperature curve into two computable standard gradient parameters, the computational dimensionality is greatly reduced while preserving the nonlinear physical characteristics of temperature. By transforming continuous temperature time-series prediction into a discrete process of double-layer sampling by month and hour, not only is the seasonality and diurnal evolution of meteorology reproduced, but also, because the subsequent traffic axle load sequence is divided into the same time period, it covers the spatiotemporal coupling law between the traffic load borne by the pavement structure and the temperature gradient, significantly improving the efficiency and reliability of pavement service life performance prediction.

[0204] In one exemplary embodiment, such as Figure 4 As shown, the method for calculating the fatigue reliability of cement concrete pavement by integrating data and physics may include steps 402 to 408. Wherein:

[0205] Step 402: Obtain the sample dataset, which includes multiple sample data, each of which is obtained by combining sample structure parameters with multiple sample working condition data.

[0206] Among them, the sample structure parameters are the geometric and material characteristic parameters corresponding to the target road, which may include the sample surface layer thickness h and the sample surface layer elastic modulus E. c The comprehensive resilience modulus E of the top of the sample base layer t Sample operating condition data may include sample axle load data, sample full-thickness temperature gradient, and sample half-thickness temperature gradient.

[0207] Optionally, within the allowable range of values ​​for the above parameters, Latin Hypercube Sampling (LHS) can be used to perform multidimensional spatial random sampling to generate N independent parameter combinations. Each parameter combination constitutes a sample data point, which corresponds to a completely determined set of working conditions, thereby obtaining a sample dataset containing N sample data points.

[0208] Step 404: Process the sample dataset using the finite element model of the target road to obtain the real stress data.

[0209] The finite element model can be used to simulate the three-dimensional stress state of a target road under the combined effects of vehicle load and ambient temperature. The finite element model can have various design parameters, such as... Figure 5 As shown, the structure includes three concrete slabs joined along the driving direction, which can be denoted as Slab-1, Slab-2, and Slab-3, representing the base, subbase, and foundation. Spring elements (spring2) can be used at the joints between adjacent slabs to simulate the shear and bending force transmission of the load-bearing rods. The surface slabs can be discretized using 3D 8-node solid elements (C3D8), while the base and subsequent structural layers can use 3D 8-node linear reduced integral elements (C3D8R). Local mesh refinement is applied to the slab areas subjected to wheel loads, with a mesh size of 0.05m or other sizes.

[0210] The subgrade surface is subject to triaxial displacement constraints, while the remaining soil outer boundary surfaces are subject to displacement constraints in the normal direction perpendicular to each boundary. Interlayer contact can be vertically continuous and horizontally smooth.

[0211] Understandably, the sample axle load data from the sample dataset can be applied to the longitudinal joint edge of the surface layer slab in the form of a rectangular uniformly distributed load with a single axle and two wheels. Based on the full-thickness and half-thickness temperature gradients in this sample data, a nonlinear continuous temperature field with a quadratic polynomial distribution along the thickness direction of the surface layer slab can be reconstructed and applied to the aforementioned finite element model. By analyzing this data, for example by running a mechanical solver, the maximum flexural tensile stress at the bottom of the cement concrete surface layer slab can be extracted as the true stress data corresponding to the sample data.

[0212] Step 406: Use the network model to be trained to process the sample data to obtain the predicted stress data.

[0213] The network model to be trained can be any type of neural network regression model, and its model architecture can be designed according to actual needs. For example, it can be like... Figure 6As shown, it can include an input layer, four hidden layers and an output layer. The input layer can contain 6 feature vectors. Hidden layer 1 has 256 neuron units, hidden layer 2 has 128 neuron units, hidden layer 3 has 64 neuron units, and hidden layer 4 has 32 neuron units.

[0214] The input features of the input layer may include the sample surface layer thickness h and the sample surface layer elastic modulus E. c The comprehensive resilience modulus E of the top of the sample base layer t Sample axle load data P, sample full thickness temperature gradient G full Temperature gradient G at half thickness of sample half The activation functions of each hidden layer can be rectified linear functions (ReLU) to improve the model's ability to fit nonlinear physical and mechanical features; the output layer can use linear activation units (linear) to output predicted stress data.

[0215] Understandably, each sample data can be used as input to the network model to be trained. After matrix multiplication, biasing, and activation function mapping of the multi-layer neurons within the network model, the corresponding estimated mechanical response, i.e., the predicted stress data, can be obtained.

[0216] Step 408: Based on the difference between the real stress data and the actual stress data, train the network model to be trained to obtain the pre-trained target network model.

[0217] In this process, the actual stress data corresponding to each sample data point, i.e., the finite element calculation value, can be used as the supervised ground truth value, and the corresponding actual stress data, i.e., the neural network output value, can be used as the model prediction value. The loss function is used to calculate the loss value between the two, and backpropagation of the error is performed based on the obtained loss value to adjust the connection weights and bias parameters in the neural network regression model until the model meets the preset iteration stopping conditions, such as reaching the set maximum number of iterations, the loss value converging to the preset minimum threshold, or triggering the early stopping mechanism, etc. The training is then completed and the model parameters are saved to obtain the target network model.

[0218] Optionally, the weight update of the network model can use the Adam optimizer, and the mean square error (MSE) can be used as the loss function to quantify the difference between the true stress data and the true stress data. It can be combined with L2 weight regularization and early stopping training mechanism based on the validation set to suppress overfitting during the model training process.

[0219] Understandably, the aforementioned target network model has established a nonlinear mapping relationship between pavement structural parameters, environmental temperature field conditions, traffic axle load conditions, and the mechanical response of the concrete surface layer. In actual pavement performance evaluation, the structural parameters to be evaluated, target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient can be input into this target network model to quickly obtain the corresponding target stress data, thus providing reliable conditions for the subsequent generation of stress mapping tables.

[0220] In this embodiment, a sample dataset is first acquired. Then, the sample dataset is processed using a finite element model and a network model to be trained, respectively, to obtain true stress data and true stress data. Based on the difference between the true stress data and the true stress data, the network model to be trained is trained to obtain a pre-trained target network model. Thus, this target network model, by coupling structural parameters, axle load data, and the first and second predicted temperature gradients characterizing the nonlinear temperature field, enables the model to predict stress under multi-physics coupling conditions of temperature and load. This provides a reliable data foundation for subsequent cumulative fatigue damage prediction within the prediction period, effectively improving the efficiency and accuracy of long-term service performance prediction for cement concrete pavements.

[0221] The data-physical integration method for calculating the fatigue reliability of cement concrete pavement provided in this application can be applied to pavements of any geographical location and any specification. The following is a brief explanation of the calculation process of the data-physical integration method for cement concrete pavement provided in this application, using cement concrete pavement as an example.

[0222] Among them, it can be like Figure 7 As shown, it is divided into three stages: Stage 1 is the shared time history sampling (month, hour), Stage 2 is the Monte Carlo cycle stage, and Stage 3 is the reliability calculation.

[0223] In Stage 1, temperature gradient sequences are obtained through partial normal sampling and conditional sampling within historical time units, and axle load sequences are obtained through frequency sampling. Then, according to historical time units, the temperature gradient sequences and axle load sequences are combined and deduplicated to obtain temperature load data, including multiple triplets composed of target axle load data, a first predicted temperature gradient, and a second predicted temperature gradient.

[0224] Specifically, it can be like Figure 8As shown, historical meteorological data and historical solar radiation data within the target time period are first processed to obtain the temperature gradient distribution function, the first historical temperature gradient, and the second historical temperature gradient for each historical time unit. Statistical analysis is then performed on historical axle load data to obtain axle load data, historical axle numbers, and probability data corresponding to axle load intervals within each historical time unit. Subsequently, based on the axle load data and traffic growth rate within the historical time unit, the target axle number within each prediction time unit can be calculated.

[0225] For each prediction time unit, sampling can be performed based on the temperature gradient distribution function according to the corresponding target axle number to obtain the first predicted temperature gradient for each prediction time unit. Then, conditional sampling can be performed based on the first predicted temperature gradient and the second historical temperature gradient to obtain the associated second predicted temperature gradient. For each prediction time unit, sampling can be performed based on the probability data of each axle load interval according to the corresponding target axle number to obtain the corresponding target axle load data.

[0226] The first predicted temperature gradient, the second predicted temperature gradient, and the target axis are combined for each predicted time unit to obtain a unit triplet sequence. Then, the frequency of each unit triplet sequence is counted, and the 288 unit triplet sequences are deduplicated to obtain the total time history data for the design period, i.e., the temperature load data.

[0227] In the Monte Carlo loop of stage 2, in each Monte Carlo loop, the structural parameters of the current loop can be randomly sampled first to obtain the corresponding structural parameters. Then, this set of parameters can be combined with various working condition data to obtain the target dataset. Specifically, the target network model can be used to process the complete target dataset, that is, to process the target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient, generating a first stress mapping table corresponding to this set of structural parameters. The axle load data in the target dataset is set to 0. The target network model is then used to process the structural parameters, the first predicted temperature gradient, and the second predicted temperature gradient to generate a second stress mapping table corresponding to this set of structural parameters. .

[0228] Then, based on the temperature load data, the corresponding first target stress data and second target stress data can be obtained by searching in the first stress mapping table and the second stress mapping table, and the fatigue failure prediction results corresponding to this set of structural parameters can be calculated using the fatigue life prediction model.

[0229] Then we can determine whether the current loop count k has reached the maximum loop count N. MC If k is less than N MCThen, the loop count can be incremented by 1 to calculate the next loop count, regenerate new structural parameters, and generate the corresponding first and second stress mapping tables. This allows for the calculation of fatigue failure prediction results for the reconfigured structural parameters. This process continues until the maximum number of loops is obtained.

[0230] After reaching the maximum Monte Carlo cycle, we can proceed to stage 3, where we superimpose the fatigue failure prediction results corresponding to each set of structural parameters and calculate the reliability.

[0231] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0232] Based on the same inventive concept, this application also provides a data-and-physics fusion-based cement concrete pavement fatigue reliability calculation device for implementing the data-and-physics fusion-based cement concrete pavement fatigue reliability calculation method described above. The solution provided by this device is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more data-and-physics fusion-based cement concrete pavement fatigue reliability calculation device embodiments provided below can be found in the limitations of the data-and-physics fusion-based cement concrete pavement fatigue reliability calculation method described above, and will not be repeated here.

[0233] In one exemplary embodiment, such as Figure 9 As shown, a data-physical fusion fatigue reliability calculation device 900 for cement concrete pavement is provided, comprising: a prediction module 910, a generation module 920, a first determination module 930, and a second determination module 940, wherein:

[0234] The prediction module 910 is used to make predictions based on the axle load data and environmental data of the target road within the target time period, and to obtain the temperature load data within the prediction time period.

[0235] The generation module 920 is used to process the target dataset using a pre-trained target network model for each set of structural parameters of the target road, and generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with multiple working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data.

[0236] The first determining module 930 is used to determine the target stress data corresponding to each of the temperature load data according to the stress mapping table, and to determine the fatigue failure prediction result corresponding to the structural parameters according to each of the target stress data and the fatigue life prediction model.

[0237] The second determining module 940 is used to determine the reliability assessment result of the target road based on the fatigue failure prediction results corresponding to each of the structural parameters.

[0238] In one embodiment, the target time period includes multiple historical time units, the prediction time period includes multiple prediction time units, and the prediction module 910 includes:

[0239] The first determining unit is used to perform statistical analysis on the axle load data and determine the historical axle number in each axle load interval within each historical time unit.

[0240] The second determining unit is used to determine the target axle number in each of the predicted time units based on the historical axle number in each of the axle load intervals and the traffic growth data in the predicted time period.

[0241] The third determining unit is used to process the target axle and each of the historical axles within each of the prediction time units to determine the corresponding target axle load data, and to process the target axle and the temperature gradient distribution function to determine the corresponding first predicted temperature gradient and second predicted temperature gradient.

[0242] The fusion unit is used to fuse the target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient of each of the predicted time units to construct temperature load data.

[0243] In one embodiment, the third determining unit is specifically used for:

[0244] The ratio of the historical axle count within each axle load interval to the total historical axle count within its respective historical time unit is determined as the probability data corresponding to each axle load interval within the historical time unit.

[0245] The probability data and the target axle are fused together to determine the target axle load data for each prediction time unit.

[0246] In one embodiment, the third determining unit is specifically used for:

[0247] Using the target axis as the sampling number, sampling is performed according to the temperature gradient distribution function to determine the first predicted temperature gradient.

[0248] Sampling is performed on the historical temperature gradients corresponding to each of the first predicted temperature gradients to determine the corresponding second predicted temperature gradients.

[0249] In one embodiment, the environmental data includes historical meteorological data and historical solar radiation data, and the third determining unit is further configured to:

[0250] Using a temperature transfer model, the historical meteorological data and the historical solar radiation data are processed to determine the temperature distribution data of the target road at each historical moment;

[0251] The temperature distribution data at each historical moment are processed to determine the first historical temperature gradient and the second historical temperature gradient for each historical time unit.

[0252] Statistical analysis is performed on each of the first historical temperature gradients to determine the temperature gradient distribution function corresponding to each historical time unit.

[0253] In one embodiment, the stress mapping table includes a first stress mapping table and a second stress mapping table, the target stress data includes first target stress data and second target stress data, and the first determining module 930 is specifically used for:

[0254] Based on the temperature load data, a search is performed in the first stress mapping table to determine the corresponding first target stress data;

[0255] Based on the temperature gradient data in each of the temperature load data, a search is performed in the second stress mapping table to determine the corresponding second target stress data.

[0256] In one embodiment, the first determining module 930 is specifically used for:

[0257] Using the fatigue life prediction model, the first target stress data and the second target stress data are processed to determine the theoretical fatigue life of the temperature load data;

[0258] Based on the theoretical fatigue life and the number of times the corresponding temperature load data appears within the predicted time period, the fatigue damage data of the temperature load data is determined.

[0259] The fatigue damage data corresponding to each temperature load data in the structural parameters are superimposed to determine the cumulative fatigue damage data.

[0260] Based on the relationship between the cumulative fatigue damage data and the preset threshold, the fatigue failure prediction result corresponding to the structural parameters is determined.

[0261] In one embodiment, the second determining module 940 is specifically used for:

[0262] Among the structural parameters, the fatigue failure prediction result is the target number of fatigue failures.

[0263] The ratio of the target quantity to the total number of structural parameters is determined as the fatigue failure probability of the target road, and the reliability assessment result is determined based on each fatigue failure probability.

[0264] In one embodiment, the generation module 920 is further configured to:

[0265] Obtain a sample dataset, which includes multiple sample data, each of which is obtained by combining sample structure parameters with multiple sample operating condition data;

[0266] The sample dataset is processed using the finite element model of the target road to obtain the actual stress data;

[0267] The sample data are processed using the network model to be trained to obtain the predicted stress data;

[0268] Based on the difference between the actual stress data and the actual stress data, the network model to be trained is trained to obtain a pre-trained target network model.

[0269] The various modules in the aforementioned data and physical fusion-based cement concrete pavement fatigue reliability calculation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0270] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 10As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a data- and physical fusion method for calculating the fatigue reliability of cement concrete pavement. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0271] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0272] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0273] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0274] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0275] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0276] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0277] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0278] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for calculating the fatigue reliability of cement concrete pavement by integrating data and physics, characterized in that, The method includes: Based on the axle load data and environmental data of the target road within the target time period, predictions are made to obtain the temperature load data for each time period within the prediction time period. For each set of structural parameters of the target road, a pre-trained target network model is used to process the target dataset to generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with various working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data. Based on the stress mapping table, the target stress data corresponding to each of the temperature load data is determined, and based on each of the target stress data and the fatigue life prediction model, the fatigue failure prediction result corresponding to the structural parameters is determined. Based on the fatigue failure prediction results corresponding to each of the structural parameters, the reliability assessment results of the target road are determined.

2. The method according to claim 1, characterized in that, The target time period includes multiple historical time units, and the prediction time period includes multiple prediction time units. The prediction is performed based on the axle load data and environmental data of the target road within the target time period to obtain various temperature load data within the prediction time period, including: Statistical analysis is performed on the axle load data to determine the historical axle loads within each axle load interval in each historical time unit; Based on the historical axle loads within each axle load interval and the traffic growth data within the predicted time period, the target axle loads for each predicted time unit are determined. Within each of the prediction time units, the target axle and each of the historical axles are processed to determine the corresponding target axle load data, and the target axle and temperature gradient distribution function are processed to determine the corresponding first predicted temperature gradient and second predicted temperature gradient. The target axle load data, the first predicted temperature gradient, and the second predicted temperature gradient of each predicted time unit are fused to construct temperature load data.

3. The method according to claim 2, characterized in that, The process of processing the target axle number and each of the historical axles to determine the corresponding target axle load data includes: The ratio of the historical axle count within each axle load interval to the total historical axle count within its respective historical time unit is determined as the probability data corresponding to each axle load interval within the historical time unit. The probability data and the target axle are fused together to determine the target axle load data for each prediction time unit.

4. The method according to claim 2, characterized in that, The process of processing the target axis and temperature gradient distribution function to determine the corresponding first predicted temperature gradient and second predicted temperature gradient includes: Using the target axis as the sampling number, sampling is performed according to the temperature gradient distribution function to determine the first predicted temperature gradient. Sampling is performed on the historical temperature gradients corresponding to each of the first predicted temperature gradients to determine the corresponding second predicted temperature gradients.

5. The method according to claim 2, characterized in that, The environmental data includes historical meteorological data and historical solar radiation data. Before the step of processing the target axis and temperature gradient distribution function to determine the corresponding first predicted temperature gradient and second predicted temperature gradient, the method further includes: Using a temperature transfer model, the historical meteorological data and the historical solar radiation data are processed to determine the temperature distribution data of the target road at each historical moment; The temperature distribution data at each historical moment are processed to determine the first historical temperature gradient and the second historical temperature gradient for each historical time unit. Statistical analysis is performed on each of the first historical temperature gradients to determine the temperature gradient distribution function corresponding to each historical time unit.

6. The method according to claim 1, characterized in that, The stress mapping table includes a first stress mapping table and a second stress mapping table, and the target stress data includes first target stress data and second target stress data. Determining the target stress data corresponding to each of the temperature load data based on the stress mapping table includes: Based on the temperature load data, a search is performed in the first stress mapping table to determine the corresponding first target stress data; Based on the temperature gradient data in each of the temperature load data, a search is performed in the second stress mapping table to determine the corresponding second target stress data.

7. The method according to claim 6, characterized in that, The step of determining the fatigue failure prediction result corresponding to the structural parameters based on the target stress data and fatigue life prediction model includes: Using the fatigue life prediction model, the first target stress data and the second target stress data are processed to determine the theoretical fatigue life of the temperature load data; Based on the theoretical fatigue life and the number of times the corresponding temperature load data appears within the predicted time period, the fatigue damage data of the temperature load data is determined. The fatigue damage data corresponding to each temperature load data in the structural parameters are superimposed to determine the cumulative fatigue damage data. Based on the relationship between the cumulative fatigue damage data and the preset threshold, the fatigue failure prediction result corresponding to the structural parameters is determined.

8. The method according to claim 1, characterized in that, The step of determining the reliability assessment result of the target road based on the fatigue failure prediction results corresponding to each of the structural parameters includes: Among the structural parameters, the fatigue failure prediction result is the target number of fatigue failures. The ratio of the target quantity to the total number of structural parameters is determined as the fatigue failure probability of the target road, and the reliability assessment result is determined based on each fatigue failure probability.

9. The method according to claim 1, characterized in that, Prior to the step of processing the target dataset using a pre-trained target network model, the method further includes: Obtain a sample dataset, which includes multiple sample data, each of which is obtained by combining sample structure parameters with multiple sample operating condition data; The sample dataset is processed using the finite element model of the target road to obtain the actual stress data; The sample data are processed using the network model to be trained to obtain the predicted stress data; Based on the difference between the actual stress data and the actual stress data, the network model to be trained is trained to obtain a pre-trained target network model.

10. A data and physics-integrated fatigue reliability calculation device for cement concrete pavement, characterized in that, The device includes: The prediction module is used to predict the axle load data and environmental data of the target road within the target time period, and obtain the temperature load data for each time period within the prediction time period. The generation module is used to process the target dataset using a pre-trained target network model for each set of structural parameters of the target road, and generate a stress mapping table corresponding to the structural parameters. The target dataset is obtained by combining the structural parameters with multiple working condition data. The stress mapping table is used to characterize the stress values ​​corresponding to the structural parameters and working condition data. The first determining module is used to determine the target stress data corresponding to each of the temperature load data according to the stress mapping table, and to determine the fatigue failure prediction result corresponding to the structural parameters according to each of the target stress data and the fatigue life prediction model. The second determining module is used to determine the reliability assessment result of the target road based on the fatigue failure prediction results corresponding to each of the structural parameters.

Images