A large data driving-based pavement structure state intelligent inversion method and system

Abstract

The application discloses a road surface structure state intelligent inversion method and system based on big data driving. The method comprises the following steps: acquiring deflection basin data of a road surface, detecting environment parameters and road surface structure reference data; correcting the deflection basin data of the road surface according to the environment parameters to obtain corrected deflection basin data; determining a reasonable distribution interval of a road surface structure inversion result according to the corrected deflection basin data; performing data amplification on the corrected deflection basin data to obtain amplified deflection basin data; inverting the road surface structure according to the amplified deflection basin data to obtain a plurality of groups of initial inversion results of the road surface structure; and selecting a group of results that is in the reasonable distribution interval of the road surface structure inversion result and most similar to the road surface structure reference data from the plurality of groups of initial inversion results of the road surface structure as a final road surface structure inversion result. The method realizes accurate inversion of the road surface structure.

Description

Technical Field

[0001] This invention relates to the field of pavement structure inversion technology, and in particular to a method and system for intelligent inversion of pavement structure state based on big data. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] To achieve precise highway maintenance, accurate detection of the road surface's structural condition is necessary.

[0004] Currently, when detecting the pavement structure condition, deflection basin data from pavement detection points is obtained, and then the pavement structure condition is inverted based on the deflection basin data. The pavement structure condition inversion results of the current method are highly sensitive to small fluctuations in the deflection basin data. However, in the process of deflection basin data detection, there are inevitably problems such as deflection measurement errors and structural layer thickness measurement errors, which ultimately lead to inaccurate pavement structure condition inversion results.

[0005] Furthermore, the current method mainly provides modulus parameters for pavement structure state inversion, which cannot reflect the interlayer bonding state and lacks comprehensiveness and refinement in reflecting the pavement structure state. Summary of the Invention

[0006] To address the aforementioned problems, this invention proposes a big data-driven intelligent inversion method and system for pavement structure state, which enables accurate inversion of pavement structure parameters.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, a big data-driven intelligent inversion method for pavement structure state is proposed, including: Acquire road surface deflection basin data, detect environmental parameters, and road surface structure reference data; The deflection basin data of the road surface is corrected based on the environmental parameters of the test to obtain the corrected deflection basin data. Based on the corrected deflection basin data, determine the reasonable distribution range of the pavement structure inversion results; Data augmentation was performed on the corrected bend basin data to obtain augmented bend basin data. The pavement structure was inverted based on the amplified sinkhole data to obtain multiple sets of initial inversion results for the pavement structure. The final pavement structure inversion result is selected from multiple initial pavement structure inversion results that fall within a reasonable distribution range and are most similar to the pavement structure reference data.

[0008] Furthermore, environmental parameters to be detected include road surface humidity and road surface temperature; Determine the humidity correction factor and temperature correction factor based on the road surface humidity and road surface temperature; The humidity correction factor and temperature correction factor are multiplied by the deflection basin data of the road surface to obtain the corrected deflection basin data.

[0009] Furthermore, based on the corrected deflection basin data, the deflection basin inflection point and the far-end deflection value index are calculated and determined. Based on the deflection basin inflection point, the far-end deflection value index, and the deflection index-modulus relationship model, the set confidence band of the pavement structure data is calculated and determined as the reasonable distribution range of the pavement structure inversion results.

[0010] Furthermore, based on the corrected deflection basin data, within a set range, multiple sets of deflection data similar to the corrected deflection basin data are randomly generated; the multiple sets of randomly generated deflection data and the corrected deflection basin data together form the amplified deflection basin data.

[0011] Furthermore, a deflection inversion model is adopted to invert the pavement structure based on the amplified deflection basin data, obtaining multiple sets of initial inversion results for the pavement structure. The deflection inversion model takes the deflection basin data as input and the pavement structure parameters as output. It is constructed using a neural network and trained with training data. The training data includes multiple deflection basin data obtained through pavement structure parameter working condition mechanics calculations, and the pavement structure parameters corresponding to each deflection basin data.

[0012] Furthermore, results within a reasonable distribution range of pavement structure inversion results are selected from multiple initial pavement structure inversion results as candidate pavement structure inversion data; Calculate the similarity between candidate inversion data of pavement structure and reference data of pavement structure; From the candidate inversion data of pavement structure, the set of data with the highest similarity is selected as the final pavement structure inversion result.

[0013] Secondly, a big data-driven intelligent inversion system for pavement structure state is proposed, including: The data acquisition unit is used to acquire pavement deflection basin data, detection environmental parameters, and pavement structure reference data; The temperature and humidity correction unit is used to correct the deflection basin data of the road surface according to the detected environmental parameters, and obtain the corrected deflection basin data. The distribution interval determination unit is used to determine the reasonable distribution interval of the pavement structure inversion results based on the corrected deflection basin data. The data augmentation unit is used to augment the corrected sinkhole data to obtain augmented sinkhole data. The pavement structure inversion unit is used to invert the pavement structure based on the amplified sinkhole data to obtain multiple sets of initial pavement structure inversion results. The inversion result determination unit is used to select the set of results that is within the reasonable distribution range of the pavement structure inversion results and is most similar to the pavement structure reference data from multiple initial inversion results of pavement structure, and to take it as the final pavement structure inversion result.

[0014] Thirdly, a computer device is proposed, the device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the big data-driven intelligent inversion method for road structure state proposed in the first aspect.

[0015] Fourthly, a computer-readable storage medium is proposed, which stores a computer program adapted to be loaded and executed by a processor, namely, the intelligent inversion method for road structure state based on big data driven proposed in the first aspect.

[0016] Fifthly, a computer program product is proposed, which includes a computer program. When the computer program is executed by a processor, it implements the big data-driven intelligent inversion method for road structure state proposed in the first aspect.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes a big data-driven intelligent inversion method and system for pavement structure state. The method first corrects the pavement deflection basin data by detecting environmental parameters, obtaining corrected deflection basin data. Then, based on the corrected deflection basin data, a reasonable distribution range for the pavement structure inversion results is determined. Data amplification is performed on the corrected deflection basin data, and the pavement structure is inverted based on the amplified data, obtaining multiple sets of initial pavement structure inversion results. Finally, the set of results that falls within the reasonable distribution range and is most similar to the pavement structure reference data is selected as the final pavement structure inversion result, ensuring the accuracy of the final pavement structure inversion result. This overcomes the problem of inaccurate pavement structure inversion results caused by measurement errors in the pavement deflection basin data.

[0018] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0019] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute an undue limitation of this application.

[0020] Figure 1 This is a flowchart of a big data-driven intelligent inversion method for road structure state proposed in an embodiment of the present invention. Figure 2 This is the overall technical route of a big data-driven intelligent inversion method for road structure state proposed in an embodiment of the present invention; Figure 3 This is the deflection index-modulus relationship model proposed in this embodiment of the invention; Figure 4 This is a schematic diagram of the deflection inversion model structure proposed in an embodiment of the present invention; Figure 5 This is a schematic diagram of the road surface structure data proposed in an embodiment of the present invention; Figure 6 This is an example diagram of a front-end page proposed in an embodiment of the present invention; Figure 7 This is a schematic diagram of the back-end calculation logic for deflection inversion proposed in an embodiment of the present invention; Figure 8 This is a schematic diagram of the layout of deflection data acquisition points proposed in an embodiment of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0023] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0024] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0025] First, the application scenarios of the intelligent inversion method for road structure state based on big data driven proposed in the embodiments of the present invention will be described.

[0026] The present invention proposes a big data-driven intelligent inversion method for pavement structure state, which is applied to the application scenario of inverting pavement structure state based on pavement deflection basin data.

[0027] Currently, the stock of highway infrastructure is enormous, with most highways having exceeded their designed service life (15 years), and transportation infrastructure is entering a phase of large-scale maintenance. Therefore, how to efficiently, accurately, and non-destructively assess the condition of pavement structures and materials has become an urgent problem that needs to be solved in the field of highway maintenance.

[0028] Falling weight deflectometer (FWD) is a testing method for evaluating the bearing capacity of pavement structures and inverting material modulus. It measures the instantaneous deflection basin of the road surface under impact load by using a falling weight deflectometer, thereby inverting information such as the modulus of the roadbed and pavement. It is currently the most widely used non-destructive testing method in highway maintenance practice.

[0029] In the research on using deflection basins to invert pavement structural layer modulus, in the 1970s, foreign scholar Scrivner proposed to invert pavement modulus based on deflection basins measured by FWD and compiled inversion nomographs. Hoffman of the University of Illinois analyzed the deflection basin parameters of different pavement structures and regressed the relationship between structural layer modulus and deflection basin parameters, and drew nomographs for inverting pavement structural mechanical parameters.

[0030] Ni Fujin et al. discovered a good linear relationship between the logarithm of the modulus ratio and the logarithm of the deflection value, and performed modulus back-calculation using the POWELL method. Sun Lijun et al. calculated the deflection basins of different pavement structures, discovered the existence of "inertia points" in the pavement structure deflection basins, demonstrated the uniqueness of these "inertia points," and established the relationship between the location of the "inertia points," the deflection value of the "inertia points," and the mechanical parameters of the pavement structure layers. Subsequently, Zhang Xiaoning, Zang Guoshuai et al. used the "inertia points" of the deflection basins to inversely calculate the modulus of the pavement structure layers. Zhu Jie et al. proposed the optimal inversion point for asphalt pavements, through which the modulus of the surface layer and base layer can be inversely calculated. They compared this method with conventional search methods, genetic algorithms, homotopy methods, and other inversion methods, finding that this method is more accurate and faster.

[0031] While the aforementioned research has yielded a series of results, its application in engineering testing within the industry is limited. The basic principles of commonly used modulus inverse calculation software are mainly divided into database methods and iterative methods.

[0032] Database-based methods focus on data fitting, with software like MODULUS being a typical example. MODULUS employs techniques such as Lagrange interpolation to achieve fitting, resulting in a relatively stable solution process. However, this method relies on a large and highly specific pavement deflection basin database—a separate database needs to be constructed for each pavement structure type and combination of structural layer thicknesses. Due to a lack of extraction and summarization of the inherent patterns in the data, this type of method has poor versatility and is difficult to adapt to diverse engineering scenarios.

[0033] In contrast, iterative methods are more widely used in engineering practice and are the commonly adopted inversion method by current testing units. Representative software includes MODCOMP, WESDEF, and SIDMOD. However, this type of method still has several limitations, such as the inversion results being quite sensitive to the selection of initial parameters, resulting in large variability; the output is limited to modulus parameters and cannot effectively identify interlayer bonding states; and the comprehensiveness, refinement, and intelligence level of the inversion results still need to be improved.

[0034] The traditional iterative method for inverting pavement structure currently faces the following technical problems: (1) Traditional iterative methods have stringent requirements for the selection of initial values. Improper selection of initial values ​​may lead to non-convergence of the inversion process or non-uniqueness of the inversion results, which can easily result in high variability of the inversion results. In the FWD detection process, problems such as deflection measurement error and structural layer thickness measurement error are unavoidable. These may cause disturbances and changes in the input initial values, and these small changes will lead to large deviations in the inversion results. Related studies have shown that the coefficient of variation of the back-calculated subgrade modulus at different points in the same road segment is generally around 10%, and the coefficient of variation of other layers is even larger, sometimes reaching 40% to 50%, indicating insufficient reliability of the back-calculated results.

[0035] (2) The inversion results of traditional inversion methods are mainly modulus parameters, which cannot reflect the interlayer bonding state. Furthermore, the influence of temperature and humidity on the deflection inversion results is not considered in the inversion software. The inversion software does not reflect the pavement structure state in a comprehensive and detailed manner.

[0036] To address the problems of high initial value sensitivity, large result variability, and inaccurate and incomplete inversion results in traditional asphalt pavement deflection inversion methods, this invention proposes a big data-driven intelligent inversion method for pavement structure state. Based on a pavement elastic layered system mechanics calculation program, a large dataset of "pavement structure parameters—deflection basin" is constructed through large-scale parallel computation. Using this large dataset, nonlinear regression analyses are performed on the asphalt pavement surface layer modulus, subgrade modulus, and deflection basin parameters. The "deflection basin inflection point" and "far-end deflection value" are selected as characterization indicators of the surface layer and subgrade modulus state, respectively, serving as constraints on the reasonable distribution range of the inversion results. This significantly improves the robustness of the pavement structure parameter inversion results and reduces the variability of the results. A multi-task learning architecture is used to train a neural network model, and combined with Monte Carlo data augmentation and methods for selecting reasonable distribution ranges of inversion results, a supporting program is developed to achieve intelligent inversion of pavement structure layer thickness, material modulus, and interlayer contact state. The research findings will provide intelligent solutions for the refined inversion of pavement structure and material parameters, which is of great significance for improving the accuracy of asphalt pavement health status assessment and maintenance decisions.

[0037] like Figures 1-8As shown in the figure, an embodiment of the present invention proposes a big data-driven intelligent inversion method for pavement structure state, comprising: Acquire road surface deflection basin data, detect environmental parameters, and road surface structure reference data; The deflection basin data of the road surface is corrected based on the environmental parameters of the test to obtain the corrected deflection basin data. Based on the corrected deflection basin data, determine the reasonable distribution range of the pavement structure inversion results; Data augmentation was performed on the corrected bend basin data to obtain augmented bend basin data. The pavement structure was inverted based on the amplified sinkhole data to obtain multiple sets of initial inversion results for the pavement structure. The final pavement structure inversion result is selected from multiple initial pavement structure inversion results that fall within a reasonable distribution range and are most similar to the pavement structure reference data.

[0038] To avoid deviations of the deflection basin data from the standard state due to different detection conditions, reduce the impact of the environment on the pavement structure inversion results, and improve the accuracy of the pavement structure inversion results, this embodiment of the invention obtains detection environment parameters; based on the detection environment parameters, the pavement deflection basin data is corrected to obtain corrected deflection basin data.

[0039] Among them, the environmental parameters to be detected include road surface humidity and road surface temperature; Determine the humidity correction factor based on road surface moisture and road surface temperature. K 1 and temperature correction factor K 3; Humidity correction factor K 1. Temperature correction factor K 3. Deflection basin data of the road surface l Multiply by 0 to obtain the corrected deflection basin data.

[0040] Specifically:

[0041] l - Corrected road surface deflection value (0.01mm); -Measured road surface deflection value (0.01mm); - The humidity correction factor can be determined according to B.7.2 of JTG D50-2017, or based on local experience; users can also customize it. -Temperature correction factor.

[0042] In some embodiments, the inflection point of the deflection basin and the far-end deflection value index are calculated and determined based on the corrected deflection basin data. The set confidence band of the pavement structure data is calculated and determined based on the inflection point of the deflection basin, the far-end deflection value index and the deflection index-modulus relationship model, which serves as the reasonable distribution range of the pavement structure inversion results.

[0043] like Figure 5 As shown, the pavement structure mentioned in this invention includes a surface layer, a base layer, and a subgrade. The reasonable distribution range of the determined pavement structure inversion results includes the reasonable distribution range of the surface layer modulus and the reasonable distribution range of the subgrade modulus.

[0044] Each set of deflection basin data represents a deflection detection point. Based on the corrected deflection basin data, the deflection basin inflection point and the far-end deflection value index are calculated and determined. Based on the deflection index-modulus relationship model, the deflection basin inflection point and the far-end deflection value index, the surface layer modulus curve and the subgrade modulus curve are determined. The upper and lower limits of the 95% confidence band are selected from the surface layer modulus curve and the subgrade modulus curve. The surface layer modulus data and subgrade modulus data that are between the upper and lower limits are the reasonable distribution range of the surface layer modulus and the reasonable distribution range of the subgrade modulus.

[0045] like Figure 3 As shown, the deflection index-modulus relationship model includes the deflection index-surface layer modulus relationship model and the deflection index-subgrade modulus relationship model. The two models are obtained by performing regression analysis on a large number of deflection indices and their corresponding surface layer moduli, and on the deflection indices and their corresponding subgrade models. The regression analysis model can adopt a negative exponential model.

[0046] In some embodiments, based on the corrected deflection basin data, within a set range, multiple sets of deflection data similar to the corrected deflection basin data are randomly generated; the multiple sets of randomly generated deflection data and the corrected deflection basin data together constitute the amplified deflection basin data.

[0047] In this embodiment of the invention, for each deflection detection point, using the Monte Carlo method, multiple additional sets of deflection data are randomly generated within a set range (e.g., ±2%) around each deflection data point in the corrected deflection basin data. This data amplification, combined with the original corrected deflection basin data, forms a data matrix. This data matrix, representing the amplified deflection basin data, serves as the input for that deflection detection point. According to the "Calibration Specification for Laser-Type High-Speed ​​Deflection Measuring Instrument," the maximum relative indication error of deflection is ≤ ±5%. This step, by amplifying the corrected deflection basin data within the possible error range, is used to offset the deviation in the inversion results caused by detection errors.

[0048] In some embodiments, a deflection inversion model is used to invert the pavement structure based on the amplified deflection basin data to obtain multiple sets of initial inversion results of the pavement structure. The deflection inversion model takes the deflection basin data as input and the pavement structure parameters as output. It is constructed using a neural network and trained with training data. The training data includes multiple deflection basin data obtained through pavement structure parameter working condition mechanics calculations, and the pavement structure parameters corresponding to each deflection basin data.

[0049] After constructing the initial model of the deflection inversion model using a neural network model, the initial model of the deflection inversion model is trained using training data. Once the training is complete, the deflection inversion model is obtained. The training data includes pavement deflection basin data with multiple known pavement structure parameters.

[0050] The process of acquiring training data includes: Step 1: Based on the theory of elastic layered systems, develop a mechanical calculation program that can meet the requirements of the "Specifications for Design of Asphalt Pavement of Highway" (JTGD50-2017) for calculating the mechanical response index of pavement structure. Its single calculation accuracy should be comparable to that of pavement mechanical analysis software, and it should also meet the requirements of efficient parallel calculation for large-scale working conditions to meet the calculation needs of a large number of working conditions with varying structural thickness / material parameters.

[0051] Step 2: Based on a typical semi-rigid base asphalt pavement structure, set up a combination of calculation parameters (pavement structure parameters), mainly including the thickness range, modulus range and variation gradient of each structural layer, Poisson's ratio, interlayer contact state, and other parameters, thus forming a batch of mechanical calculation conditions. A random number generator can be used to automatically generate a specified number of random numbers within the target range, thereby achieving the random generation of a large batch of mechanical calculation conditions. An example of setting up calculation conditions for a semi-rigid base asphalt pavement structure is shown in the table below: Table 1. Examples of working conditions for mechanical calculation of semi-rigid base pavement structures Layer Thickness range / cm Thickness variation gradient / cm Modulus range / MPa Modulus change gradient / MPa Poisson's ratio Interlayer contact coefficient surface layer H1=10~25 1.5 E1=1000~20000 100 0.25 Continuous / Semi-continuous / Discontinuous (α1=0 / 0.7 / 1) grassroots H2=42~64 2.0 E2=1000~30000 200 0.25 Continuous / Semi-continuous / Discontinuous (α1=0 / 0.7 / 1) Roadbed — E3=40~400 20 0.4 — Step 3: Construction of the deflection basin dataset. Using the developed mechanical calculation program, batch calculations were performed for various working conditions to obtain deflection basin data (D0, D20, D30, D45, D60, D90, D120, D150, D180, etc.). The calculation results of the deflection basin data were then mapped one-to-one with the input data (pavement structure parameters) to construct a pavement structure parameter-deflection basin database, as shown in Table 2. This provides a data foundation for subsequent data mining analysis and deflection inversion model training.

[0052] Table 2. Examples of Pavement Structure Parameters—Deflection Basin Database

[0053] Step 4: Deflection-Modulus Correlation Analysis. For semi-rigid base asphalt pavement structures, the surface layer modulus is correlated with the deflection basin inflection point (D0-2). The correlation between D20+D30 and the subgrade modulus is good; the correlation between the subgrade modulus and the distal deflection value (D120+D150+D180) / 3 is also good. Based on the database constructed in step 3, two types of deflection value indicators, deflection basin inflection point and distal deflection value, and their corresponding moduli were selected. Regression analysis was performed on the selected data to obtain the following results: Figure 3 The deflection index-surface layer modulus relationship model shown in (a) and as follows Figure 3 The deflection index-subgrade modulus relationship model shown in (b) is generally a negative exponential model. The 95% confidence band of the corresponding model is determined as the reasonable distribution range of the surface layer modulus inversion results and the reasonable distribution range of the subgrade modulus inversion results.

[0054] Wherein, the inflection point of the deflection basin = D0-2 D20+D30, such as Figure 8 As shown, D0 is the maximum deflection value at the center of the FWD bearing plate, D20 is the maximum deflection value 20cm away from the center of the bearing plate, and D30 is the maximum deflection value 30cm away from the center of the bearing plate.

[0055] The deflection value at the far end is calculated as (D120 + D150 + D180) / 3, where D120 is the maximum deflection value at 120cm from the center of the FWD bearing plate, D150 is the maximum deflection value at 150cm from the center of the bearing plate, and D180 is the maximum deflection value at 180cm from the center of the bearing plate.

[0056] Step 5: Temperature and Humidity Correction Method for Deflection Data. To avoid deviations in deflection data from standard conditions due to varying temperature and humidity during testing, a humidity correction factor is adopted according to the relevant provisions of the "Specifications for Field Testing of Highway Subgrade and Pavement" (JTG 3450-2019) and the "Specifications for Design of Highway Asphalt Pavement" (JTG D50-2017). K 1 and temperature correction factor K 3 pairs of measured deflection data l A correction of 0 is applied, and the corrected deflection value is... l The specific formula is as follows:

[0057] l - Corrected road surface deflection value (0.01mm); -Measured road surface deflection value (0.01mm); - The humidity correction factor can be determined according to B.7.2 of JTG D50-2017, or based on local experience; users can also customize it. -Temperature correction factor, determined by the following formula.

[0058]

[0059] In the formula: - The road surface temperature (°C) at the road surface detection point; -Thickness (mm) of asphalt binder materials can be customized by the user; - Resilient modulus of the top surface of the roadbed under equilibrium humidity conditions (MPa).

[0060] and The design modulus of the roadbed is derived from the design data of the pavement.

[0061] Step 6: Construction of a deflection inversion model based on AI and big data. To achieve regression prediction of continuous variables such as the thickness and modulus of each pavement structural layer, and classification prediction of the contact state (discrete variable) between layers, it is recommended to use a multi-task learning architecture to build a neural network model. A schematic diagram of the model architecture is attached. Figure 4 As shown, by adjusting key hyperparameters such as the depth of the shared layer and task-specific layer, activation function, and loss function, deflection data is used as input, and the pavement structure inversion results, including the thickness, modulus, and interlayer contact state of each structural layer, are used as output. The performance is evaluated using a test set, and the model structure is continuously optimized to improve inversion accuracy. Finally, the trained deflection inversion model is saved and exported.

[0062] In some embodiments, the pavement structure reference data includes the surface layer reference thickness and the subgrade reference thickness, which can be obtained by core sampling of the pavement.

[0063] The process of selecting the final pavement structure inversion result from multiple sets of initial pavement structure inversion results in this embodiment of the invention includes: Results that fall within a reasonable distribution range of pavement structure inversion results are selected from multiple initial pavement structure inversion results and used as candidate pavement structure inversion data. Calculate the similarity between candidate inversion data of pavement structure and reference data of pavement structure; From the candidate inversion data of pavement structure, the set of data with the highest similarity is selected as the final pavement structure inversion result.

[0064] This embodiment of the invention also includes the design of the front-end page for a pavement structure state parameter inversion tool. The front-end page of the tool should have the function of inputting the following data: (1) Initial setting parameters: For semi-rigid base pavement structures, input the subgrade design modulus (from design data) and the reference data of each pavement structure (from core sampling).

[0065] (2) Detect environmental parameters: obtain road surface humidity and road surface temperature, and determine humidity correction factor and temperature correction factor based on road surface humidity and road surface temperature.

[0066] (3) Deflection input: Obtain one or more sets of deflection basin test values, including D0, D20, D30, D45, D60, D90, D120, D150, D180, etc., and support users to copy and paste in batches from Excel.

[0067] (4) Results display window: When the user clicks to start the analysis, the final pavement structure inversion results will appear in the display window below. The pavement structure inversion results displayed in the display window are, in order, surface layer thickness H1, base layer thickness H2, surface layer modulus E1, base layer modulus E2, subgrade modulus E3, surface layer / base layer contact coefficient α1 and base layer / subgrade contact coefficient α2.

[0068] This invention proposes a big data-driven intelligent inversion method for pavement structure state. It fully leverages the powerful nonlinear modeling capabilities of artificial neural networks to deeply analyze the complex nonlinear relationship between multi-modal parameters and deflection in multi-layer pavement structures. Furthermore, by incorporating temperature and humidity correction for deflection values ​​and a reasonable distribution range for inversion results, it overcomes the technical challenges of traditional iterative inversion algorithms, such as sensitivity to initial values, high variability, and incomplete inversion results. This method achieves comprehensive inversion of pavement structure layer thickness, material modulus, and base-surface contact state. Its specific advantages are as follows: (1) The inversion results have richer parameters. By combining a multi-task learning architecture to train a neural network model, the complex relationship between multi-modal parameters such as the thickness, modulus and interlayer contact state of the pavement structure and deflection is deeply analyzed. This enables accurate inversion of various parameters such as the thickness of the pavement structure layer, material modulus and interlayer contact state, and the health status parameters are richer.

[0069] (2) Reduced variability of inversion results. The original deflection data is augmented using the Monte Carlo method to offset the deviations in the inversion results that may be caused by detection errors, thereby reducing the sensitivity of the inversion process to fluctuations in initial values. Furthermore, nonlinear regression analyses of the asphalt pavement surface modulus, subgrade modulus, and deflection basin parameters are conducted using a large dataset to provide reasonable distribution range constraints for the inversion results, significantly improving the robustness of the pavement structure parameter inversion results and reducing the variability of the inversion results.

[0070] The embodiments of this invention will overcome a series of bottlenecks in the theory and application of existing deflection inversion methods, provide intelligent solutions for the refined inversion of pavement structure and material parameters, better support and serve the pavement health assessment and scientific maintenance decision-making in the new era, and have important engineering application value and socio-economic benefits.

[0071] This invention also proposes a big data-driven intelligent inversion system for pavement structure state, comprising: The data acquisition unit is used to acquire pavement deflection basin data, detection environmental parameters, and pavement structure reference data; The temperature and humidity correction unit is used to correct the deflection basin data of the road surface according to the detected environmental parameters, and obtain the corrected deflection basin data. The distribution interval determination unit is used to determine the reasonable distribution interval of the pavement structure inversion results based on the corrected deflection basin data. The data augmentation unit is used to augment the corrected sinkhole data to obtain augmented sinkhole data. The pavement structure inversion unit is used to invert the pavement structure based on the amplified sinkhole data to obtain multiple sets of initial pavement structure inversion results. The inversion result determination unit is used to select the set of results that is within the reasonable distribution range of the pavement structure inversion results and is most similar to the pavement structure reference data from multiple initial inversion results of pavement structure, and to take it as the final pavement structure inversion result.

[0072] It should be noted that the above-described intelligent inversion system for pavement structure state based on big data is only illustrated by the division of the functional modules described above. In practical applications, the functions can be assigned to different functional modules as needed, that is, the internal structure of the equipment can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the above-described intelligent inversion system for pavement structure state based on big data and the embodiment of an intelligent inversion method for pavement structure state based on big data belong to the same concept, and their specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0073] The present invention also discloses a computer device, the device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements a big data-driven intelligent inversion method for road structure state disclosed in the embodiments of the present invention.

[0074] The present invention also discloses a computer-readable storage medium storing a computer program adapted for loading and execution by a processor of a big data-driven intelligent inversion method for road structure state disclosed in the embodiments of the present invention.

[0075] The present invention also discloses a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements a big data-driven intelligent inversion method for road structure state disclosed in the embodiments of the present invention.

[0076] The method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The software modules can reside in readily available storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0077] Those skilled in the art will recognize that the units and algorithm steps described in conjunction with the embodiments herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0078] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for intelligent inversion of pavement structure state based on big data, characterized in that, include: Acquire road surface deflection basin data, detect environmental parameters, and road surface structure reference data; The deflection basin data of the road surface is corrected based on the environmental parameters of the test to obtain the corrected deflection basin data. Based on the corrected deflection basin data, determine the reasonable distribution range of the pavement structure inversion results; Data augmentation was performed on the corrected bend basin data to obtain augmented bend basin data. The pavement structure was inverted based on the amplified sinkhole data to obtain multiple sets of initial inversion results for the pavement structure. The final pavement structure inversion result is selected from multiple initial pavement structure inversion results that fall within a reasonable distribution range and are most similar to the pavement structure reference data.

2. The method for intelligent inversion of pavement structure state based on big data as described in claim 1, characterized in that, The environmental parameters to be tested include road surface humidity and road surface temperature; Determine the humidity correction factor and temperature correction factor based on the road surface humidity and road surface temperature; The humidity correction factor and temperature correction factor are multiplied by the deflection basin data of the road surface to obtain the corrected deflection basin data.

3. The method for intelligent inversion of pavement structure state based on big data as described in claim 1, characterized in that, Based on the corrected deflection basin data, the deflection basin inflection point and the far-end deflection value index are calculated and determined. Based on the deflection basin inflection point, the far-end deflection value index, and the deflection index-modulus relationship model, the set confidence band of the pavement structure data is calculated and determined as the reasonable distribution range of the pavement structure inversion results.

4. The intelligent inversion method for road structure state based on big data as described in claim 1, characterized in that, Based on the corrected deflection basin data, within a set range, multiple sets of deflection data that are similar to the corrected deflection basin data are randomly generated; the multiple sets of randomly generated deflection data and the corrected deflection basin data together form the amplified deflection basin data.

5. The method for intelligent inversion of pavement structure state based on big data as described in claim 1, characterized in that, A deflection inversion model was adopted to invert the pavement structure based on the amplified deflection basin data, and multiple sets of initial inversion results of the pavement structure were obtained. The deflection inversion model takes the deflection basin data as input and the pavement structure parameters as output. It is constructed using a neural network and trained with training data. The training data includes multiple deflection basin data obtained by pavement structure parameter working condition mechanics calculations, and the pavement structure parameters corresponding to each deflection basin data.

6. The intelligent inversion method for pavement structure state based on big data as described in claim 1, characterized in that, Results that fall within a reasonable distribution range of pavement structure inversion results are selected from multiple initial pavement structure inversion results and used as candidate pavement structure inversion data. Calculate the similarity between candidate inversion data of pavement structure and reference data of pavement structure; From the candidate inversion data of pavement structure, the set of data with the highest similarity is selected as the final pavement structure inversion result.

7. A big data-driven intelligent inversion system for pavement structure state, characterized in that, include: The data acquisition unit is used to acquire pavement deflection basin data, detection environmental parameters, and pavement structure reference data; The temperature and humidity correction unit is used to correct the deflection basin data of the road surface according to the detected environmental parameters, and obtain the corrected deflection basin data. The distribution interval determination unit is used to determine the reasonable distribution interval of the pavement structure inversion results based on the corrected deflection basin data. The data augmentation unit is used to augment the corrected sinkhole data to obtain augmented sinkhole data. The pavement structure inversion unit is used to invert the pavement structure based on the amplified sinkhole data to obtain multiple sets of initial pavement structure inversion results. The inversion result determination unit is used to select the set of results that is within the reasonable distribution range of the pavement structure inversion results and is most similar to the pavement structure reference data from multiple initial inversion results of pavement structure, and to take it as the final pavement structure inversion result.

8. An electronic device, characterized in that, The device includes: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the intelligent inversion method for road structure state based on big data as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed by a processor to provide a big data-driven intelligent inversion method for road structure state as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the intelligent inversion method for road structure state based on big data as described in any one of claims 1-6.

Images