A method for constructing a water film thickness prediction model of a mobile pavement in a low-lying area

Abstract

The application provides a low-lying area mobile pavement water film thickness prediction model construction method, which is applied to the water film thickness prediction technical field, and through real-time acquisition of a pavement plate load structure state and pavement fluid physical properties, the pavement dynamics is divided into micro regions and the structure characteristics of each region are determined. According to the matching initial behavior parameters of the structure characteristics and the fluid properties, the changes of the two are continuously monitored, when the changes exceed the threshold value, the parameters are adaptively adjusted according to the structure deformation increment or the fluid evolution trend. The adjusted parameters and the fluid properties are substituted into a fluid dynamics model for solving, and finally the prediction value of the mixed fluid covering layer thickness of each micro region is obtained. The method realizes the adaptive response to the pavement dynamic deformation and the fluid property change, so as to accurately calculate the covering layer thickness under complex working conditions.

Description

A method for constructing a water film thickness prediction model for mobile pavements in low-lying areas Technical Field

[0001] This application relates to the field of water film thickness prediction technology, and in particular to a method for constructing a water film thickness prediction model for mobile pavement in low-lying areas. Background Technology

[0002] In mobile pavement systems that utilize high-strength composite modular pavement panels for rapid installation in low-lying areas, accurate prediction of the thickness of the slippery layer formed by accumulated water is crucial for ensuring the safety of heavy vehicle traffic. However, existing conventional prediction methods fail to adapt to the complex dynamic changes in actual operating environments, facing the following technical challenges:

[0003] First, repeated passage of heavy vehicles will compact the natural foundation below. Due to the uneven soil texture, uneven settlement will occur, causing the originally flat road slab to tilt and shift. The preset drainage slope will completely fail, and the water accumulation area and thickness will deviate completely from the initial design.

[0004] Secondly, the gaps between modules are squeezed or stretched under the displacement of the road panel, forming irregular micro-drainage channels. The rainwater runoff path changes dynamically, and some areas accumulate water due to the tightness of the joints. Traditional models cannot simulate the real-time evolution of this geometry.

[0005] Even more challenging is that, under the combined effects of rainfall and wheel loads, fine silt particles in the foundation are squeezed upwards from deformed joints, mixing with accumulated water to form highly viscous mud. Its fluidity is significantly different from that of clear water, making it more difficult to drain and more prone to accumulating into thick layers. Prediction methods based on the characteristics of clear water completely fail in the face of this type of non-Newtonian fluid.

[0006] In summary, how to construct a predictive method that can adapt to the dynamic deformation of the pavement caused by foundation settlement, the real-time changes in the joint network, and the transformation of mud fluid properties, and accurately calculate the thickness of the mixed overburden layer composed of mud and water, is a technical problem that urgently needs to be solved. Summary of the Invention

[0007] In view of the shortcomings of the prior art, this application provides a method for constructing a water film thickness prediction model for mobile pavement in low-lying areas, which has the advantage of being able to adaptively adjust and thus accurately calculate the thickness of the cover layer formed by the mixed fluid of mud and water on the constantly changing pavement.

[0008] A first aspect is a method for constructing a prediction model for the water film thickness of a movable pavement in a low-lying area, the method comprising the following steps:

[0009] S1: Obtain real-time structural state information of the moving pavement during the loading process and real-time physical property information of the pavement fluid;

[0010] S2: Based on the real-time structural state information, the road surface is divided into multiple micro-regions, and the structural feature description of each micro-region is determined according to the degree of deformation of each micro-region;

[0011] S3: Based on the structural feature description and the instantaneous physical property information, match the corresponding initial fluid behavior parameters for each micro-region;

[0012] S4: Continuously monitor the dynamic changes of the real-time structural state information and the real-time physical property information. When the change exceeds a preset threshold, adaptively adjust the initial fluid behavior parameters of the micro-region according to the increment of structural deformation or the evolution trend of fluid properties.

[0013] S5: Substitute the adaptively adjusted fluid behavior parameters and the instantaneous physical property information into the fluid dynamics model for numerical solution to obtain the predicted value of the overburden thickness of each micro-region under the current structural state.

[0014] Furthermore, step S1 includes:

[0015] S11: Obtain the local tilt angle of the road panel through a preset angle sensing unit;

[0016] S12: Obtain wavelength drift data at the road panel joint through a preset deformation sensing unit, and convert the wavelength drift data into the joint displacement; wherein, the real-time structural state information includes at least the tilt angle of the road panel and the joint displacement.

[0017] S13: Obtain the light intensity attenuation value when the fluid flows through the monitoring point, and calculate the turbidity value of the fluid based on the ratio of the light intensity attenuation value to the initial light intensity;

[0018] S14: Obtain the vibration decay time constant of the preset detection unit in the fluid, and invert the viscosity of the fluid based on the vibration decay time constant; the instantaneous physical property information includes at least the turbidity value and viscosity of the fluid.

[0019] Furthermore, the instantaneous physical property information also includes the fluid density, and after step S14, the following is also included:

[0020] S15: Determine the category of the fluid based on the turbidity value and the viscosity. When the turbidity value exceeds the turbidity threshold and the viscosity exceeds the viscosity threshold, determine that the fluid is mud; otherwise, determine that the fluid is clear water.

[0021] S16: When the fluid is mud, estimate the fluid density of the mud type based on the turbidity value and the preset mud density; when the fluid is clear water, use the density of water as the fluid density.

[0022] Furthermore, step S2 includes:

[0023] S21: Divide the movable road surface into virtual grid units;

[0024] S22: For each virtual grid cell, calculate the average tilt, joint opening and closing degree, and settlement relative to the initial laying state based on the local tilt angle and the joint displacement.

[0025] S23: When the average tilt or the settlement exceeds the first preset threshold, the structural feature description of the corresponding virtual grid unit is marked as a local depression;

[0026] S24: When the opening degree of the joint exceeds the second preset threshold, the structural feature description of the corresponding virtual mesh unit is marked as a wear joint.

[0027] Furthermore, step S3 includes:

[0028] S31: Based on the fluid category determined by the instantaneous physical property information, select the corresponding basic parameter set from the preset behavioral parameter set, wherein the basic parameter set includes the basic roughness coefficient and the basic retention coefficient;

[0029] S32: Based on the structural feature description, call the corresponding parameter adjustment function, and use the parameter adjustment function to correct the basic roughness coefficient and basic retention coefficient to generate initial fluid behavior parameters that match the current situation.

[0030] Furthermore, step S32 includes:

[0031] S321: When the structural feature description includes local depressions, call the depression retention adjustment function to increase the basic retention coefficient according to the settlement amount to obtain the corrected retention coefficient;

[0032] S322: When the structural feature description includes wear joints, call the joint seepage adjustment function, and increase the basic roughness coefficient according to the joint opening degree to obtain the corrected roughness coefficient; the initial fluid behavior parameters include the retention coefficient and the roughness coefficient.

[0033] Furthermore, in step S4, adaptively adjusting the initial fluid behavior parameters of the micro-region based on the increment of structural deformation or the evolution trend of fluid properties includes the following steps:

[0034] S41: Establish a first corrected mapping relationship between the increment of the structural deformation and the roughness coefficient in the fluid behavior parameters;

[0035] S42: Establish a second corrected mapping relationship between the evolution trend of the fluid properties and the retention coefficient in the fluid behavior parameters;

[0036] S43: Based on the first correction mapping relationship and the second correction mapping relationship, the initial fluid behavior parameters are compensated in real time to obtain adaptively adjusted fluid behavior parameters.

[0037] Furthermore, step S5 includes:

[0038] S51: Construct a fluid dynamics model based on two-dimensional shallow water equations;

[0039] S52: Calculate the bottom shear force in the fluid dynamics model based on the viscosity and fluid density in the instantaneous physical property information and the roughness coefficient in the fluid behavior parameters;

[0040] S53: Correct the source and sink terms in the fluid dynamics model based on the seepage correction term in the fluid behavior parameters;

[0041] S54: The modified fluid dynamics model is solved using a numerical solution method to obtain the predicted values ​​of fluid flow velocity, spreading range, and cover layer thickness of each micro-region at the current moment.

[0042] Furthermore, step S54 includes:

[0043] S541: Within each solution time step, identify the micro-regions whose predicted overburden thickness is greater than zero and which are adjacent to the region whose predicted overburden thickness is zero, and mark them as leading edge boundary regions.

[0044] S542: Obtain surface wetting state information of the leading edge boundary region;

[0045] S543: Adjust the bottom shear force corresponding to the leading edge boundary region according to the surface wetting state information to correct the spreading resistance of the fluid dynamics model in the leading edge boundary region, thereby obtaining the predicted values ​​of fluid flow velocity, spreading range and capping layer thickness of each micro-region at the current moment.

[0046] Furthermore, step S5 includes the following:

[0047] S6: Obtain the ultrasonic measured thickness data of the movable road surface;

[0048] S7: Calculate the deviation between the predicted thickness of the cover layer and the ultrasonically measured thickness data;

[0049] S8: Feedback calibration is performed on the adaptive adjustment logic based on the deviation value to correct the adjustment weights of the fluid behavior parameters.

[0050] Beneficial Effects: The proposed method for constructing a water film thickness prediction model for mobile pavements in low-lying areas obtains real-time structural state information and real-time physical property information of the pavement fluid during loading. Based on the real-time structural state information, the pavement is divided into multiple micro-regions, and its structural feature description is determined according to the deformation degree of each micro-region. Based on the structural feature description and real-time physical property information, corresponding initial fluid behavior parameters are matched for each micro-region. The dynamic changes of real-time structural state information and real-time physical property information are continuously monitored. When the change exceeds a preset threshold, the initial fluid behavior parameters of the micro-region are adaptively adjusted according to the increment of structural deformation or the evolution trend of fluid properties. The adaptively adjusted fluid behavior parameters and real-time physical property information are substituted into the fluid dynamics model for numerical solution to obtain the predicted value of the cover layer thickness of each micro-region under the current structural state. This method has the advantage of being able to adaptively adjust and thus accurately calculate the cover layer thickness formed by the mixed fluid of silt and water on the constantly changing pavement. Attached Figure Description

[0051] Figure 1 is a flowchart of a method for constructing a water film thickness prediction model for a movable pavement in a low-lying area proposed in this application.

[0052] Figure 2 is a schematic diagram of a method for constructing a water film thickness prediction model for a movable pavement in a low-lying area proposed in this application. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0054] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0055] In low-lying mobile pavements, a series of complex and dynamically evolving factors arise, including uneven ground settlement caused by heavy vehicle traffic, dynamic tilting and displacement of pavement slabs, real-time changes in the joint morphology between slabs, and the formation of mud slurry with continuously changing physical properties due to the upwelling of ground sediment caused by vehicle rolling and its mixing with accumulated water. Therefore, constructing an adaptive predictive model to accurately calculate the thickness of the overburden layer formed by the mixture of sediment and water on the constantly changing pavement has become a pressing technical challenge.

[0056] Therefore, referring to Figures 1 and 2, this application provides a method for constructing a water film thickness prediction model for mobile pavements in low-lying areas. The method includes the following steps:

[0057] S1: Obtain real-time structural state information of the moving pavement during the loading process and real-time physical property information of the pavement fluid;

[0058] S2: Based on real-time structural state information, the road surface is divided into multiple micro-regions, and the structural feature description of each micro-region is determined according to the degree of deformation of each micro-region;

[0059] S3: Based on the structural feature description and real-time physical property information, match the corresponding initial fluid behavior parameters for each micro-region;

[0060] S4: Continuously monitor the dynamic changes of real-time structural state information and real-time physical property information. When the change exceeds the preset threshold, adaptively adjust the initial fluid behavior parameters of the micro-region according to the increment of structural deformation or the evolution trend of fluid properties.

[0061] S5: Substitute the adaptively adjusted fluid behavior parameters and real-time physical property information into the fluid dynamics model for numerical solution to obtain the predicted value of the overburden thickness of each micro-region under the current structural state.

[0062] The core working principle of this method revolves around the real-time perception of road surface structure and fluid properties, and a closed-loop cycle of continuously adjusting and optimizing fluid behavior rules based on this perceived information.

[0063] First, a large number of miniature sensors distributed throughout the mobile road surface enable comprehensive, real-time monitoring of the road surface condition. These sensors act like nerve endings on the road surface, continuously capturing subtle deformations that occur under the pressure of heavy vehicles, such as local tilting of the road slab, opening and closing of joints between slabs, and cumulative settlement of the foundation. At the same time, miniature detection units deployed at specific locations on the road surface act like taste buds, instantly analyzing whether the water on the road surface is pure rainwater or mud mixed with silt, and quantifying its key physical properties such as viscosity and density.

[0064] All this real-time structural and fluid property information is transmitted to a central processing unit. Upon receiving the data, this unit matches a set of fluid behavior parameters best suited to the current real-world conditions for each pre-defined small area, or micro-region, on the road surface. For example, if a micro-region forms a depression due to foundation settlement and the water on the road surface has turned into viscous mud, the processing unit will assign a specific set of parameters to this depression. These parameters will accurately reflect the physical characteristics of the viscous mud—slow flow and easy stagnation—within this depression.

[0065] Furthermore, when new changes occur in the road surface structure—for example, a small depression becomes deeper due to subsequent vehicle traffic, or the surrounding joints wear more severely due to repeated stretching—the processing unit does not simply apply the previous parameters. Instead, it performs an evolutionary adjustment or optimization of the fluid behavior parameters previously matched to that micro-region based on these incremental structural changes. This means that if the depression depth increases, the processing unit will immediately and proportionally increase the retention coefficient of the mud in that area, thus predicting a greater water accumulation thickness; if joint wear intensifies, the processing unit will adjust the parameters at the joints accordingly to more accurately simulate the complex situation where water flow may accelerate leakage or be more significantly impeded at the joints. This adjustment is continuous, ensuring that the fluid behavior parameters remain highly synchronized with the real-time physical conditions of the road surface.

[0066] Finally, the processing unit utilizes these dynamically adjusted fluid behavior parameters, combined with established fluid dynamics equations, to accurately calculate the thickness of the mud cover layer in various micro-regions of the road surface. Simultaneously, to ensure the long-term accuracy of the model, actual thickness measurement devices are deployed at key locations on the road surface. By comparing the predicted results with measured data, a feedback calibration mechanism is established to continuously optimize the logic and weighting of parameter adjustments.

[0067] This working principle differs fundamentally from conventional prediction methods. Conventional methods typically assume that pavement geometry and fluid properties are relatively fixed or pre-defined; for example, they calculate based on a single measurement of pavement slope and the standard physical properties of clear water. This approach is only suitable for structurally stable rigid pavements or scenarios with minimal environmental changes. However, in the special scenario of mobile pavements in low-lying areas, the passage of heavy vehicles is the core driving force causing uneven ground settlement and continuous changes in pavement geometry. Simultaneously, the upwelling of ground sediment causes accumulated water to dynamically evolve into mud with complex physical properties. Conventional methods cannot perceive these continuously occurring changes in real time, nor can they adjust their fixed calculation parameters, thus their predictions inevitably deviate significantly from reality. The method proposed in this application, by establishing an instantaneous correlation and continuous adjustment mechanism between structural and fluid behavior, can perfectly adapt to such complex and unexpected dynamic changes, thereby providing highly accurate overburden thickness predictions.

[0068] Furthermore, step S1 includes:

[0069] S11: Obtain the local tilt angle of the road panel through a preset angle sensing unit;

[0070] S12: Obtain wavelength drift data at the road panel joint through a preset deformation sensing unit, and convert the wavelength drift data into joint displacement; wherein, the real-time structural state information includes at least the tilt angle of the road panel and the joint displacement.

[0071] S13: Obtain the light intensity attenuation value when the fluid flows through the monitoring point, and calculate the turbidity value of the fluid based on the ratio of the light intensity attenuation value to the initial light intensity;

[0072] S14: Obtain the vibration decay time constant of the preset detection unit in the fluid, and invert the viscosity of the fluid based on the vibration decay time constant; the real-time physical property information includes at least the turbidity value and viscosity of the fluid.

[0073] To achieve accurate sensing of the road surface structure, various miniature sensors are strategically embedded in the modular pavement panels of the movable pavement. Angle sensing units can specifically be microelectromechanical tilt sensors, also known as MEMS tilt sensors. These sensors are placed at the four corners or center of each pavement panel and can accurately measure the local tilt angle of the pavement panel in two orthogonal directions at extremely high frequencies, such as hundreds of times per second. Their output is typically a digital signal, directly representing the tilt angle value in degrees or radians, and can instantly reflect the attitude changes of the pavement panel caused by foundation settlement.

[0074] The deformation sensing unit can be specifically a fiber Bragg grating strain sensor, also known as an FBG strain sensor. This sensor consists of a specially treated optical fiber cleverly embedded or adhered to the connection structure between pavement slabs, such as dovetail joints or near pin connection holes. When the pavement slabs undergo relative displacement due to foundation settlement or vehicle loads, the joint is pulled open or squeezed tight, causing the optical fiber to be stretched or compressed. This minute deformation causes a change in the period of the Bragg grating in the optical fiber, thus altering the center wavelength of its reflected light. Assuming a dedicated fiber grating demodulator is responsible for transmitting a broadband light source to the fiber optic network and monitoring the wavelength of the light signals returned from each FBG sensor in real time, by analyzing the wavelength drift relative to the initial laying state, the relative displacement at the joint can be accurately deduced, with an accuracy reaching the millimeter level or even higher.

[0075] For example, a wavelength shift towards longer wavelengths corresponds to the joint being stretched open, while a shift towards shorter wavelengths corresponds to the joint being squeezed tight. In the application of fiber Bragg grating sensors, stretching or compression at the joint causes changes in the grating period and effective refractive index, resulting in a shift in the center wavelength of the reflected light. This shift has a linear proportional relationship with strain. By dividing the detected wavelength change value by a pre-calibrated strain sensitivity coefficient, the micro-strain value at the joint can be obtained. Since the joint displacement is the relative motion between road slabs, the micro-strain value can be integrated over the length of the grating sensing section, or converted according to the geometric proportional relationship between the grating length and deformation, to calculate the joint displacement in millimeters in real time.

[0076] To achieve real-time identification of the physical properties of road surface fluids, miniature detection units are installed at specific monitoring points on the road surface, such as the center or lowest point of each pre-defined analysis grid cell. To obtain the turbidity value of the fluid, an optical turbidity sensor based on the principle of transmitted light can be used. This sensor consists of a light source, such as an 850 nm infrared light-emitting diode, and a photodetector, such as a photodiode array, arranged opposite each other. When the road surface fluid flows through the tiny channel between the light source and the detector, the light beam emitted by the light source penetrates the fluid. If the fluid is clear water, the light attenuation is minimal; however, if the fluid is slurry containing sediment particles, these suspended particles scatter and absorb the light, resulting in a significant attenuation of the light intensity reaching the detector. The central processing unit calculates the turbidity value of the fluid in real time by comparing the currently received light intensity with the reference light intensity measured under initial pure water conditions and using pre-calibrated mathematical relationships, such as the turbidity value being equal to a calibration coefficient multiplied by the logarithm of the ratio of the initial light intensity to the current light intensity. A higher turbidity value indicates a higher sediment content in the fluid.

[0077] To obtain the viscosity of a fluid, a miniature vibratory viscometer can be used. The core of this device is a miniature piezoelectric ceramic oscillator that vibrates at its inherent high frequency when an electrical signal is applied. When this oscillator is immersed in the fluid on the road surface, the viscous resistance of the fluid dampens its vibration. The higher the viscosity of the fluid, the stronger the damping effect, and the faster the vibration of the oscillator decays after the excitation stops. The central processing unit accurately measures the time required for the oscillator's vibration amplitude to decay to a specific proportion, i.e., the vibration decay time constant. Then, by referring to a pre-calibrated viscosity-decay time constant relationship curve or lookup table using a standard fluid, the current viscosity of the fluid can be calculated in real time.

[0078] The data acquired by these sensors is periodically transmitted to the central processing unit via low-power wireless communication modules, such as those based on long-range radio technology, providing the most original and immediate field data for subsequent analysis and modeling.

[0079] Furthermore, the instantaneous physical property information also includes the fluid density, and after step S14, the following is also included:

[0080] S15: Determine the fluid category based on turbidity and viscosity. When the turbidity exceeds the turbidity threshold and the viscosity exceeds the viscosity threshold, the fluid is classified as mud; otherwise, the fluid is classified as clear water.

[0081] S16: When the fluid is mud, estimate the fluid density of the mud category based on the turbidity value and the preset mud density; when the fluid is clear water, use the density of water as the fluid density.

[0082] The purpose of this step is to accurately classify road surface fluids and assign them precise density values, as density is a key parameter in fluid dynamics calculations, directly affecting the fluid's inertial forces and gravitational effects. The judgment logic employs a dual-condition judgment: the fluid is only classified as mud when both turbidity and viscosity values ​​simultaneously exceed their respective preset thresholds. For example, the turbidity threshold can be set to 50 NTU, and the viscosity threshold to 1.5 mPa·s. This dual judgment standard effectively avoids misclassification; for instance, some water containing dissolved chemicals may have slightly higher viscosity but very low turbidity, and thus will not be misclassified as mud.

[0083] Once a fluid is classified as clean water, its density is directly adopted using the standard density of water, typically 1000 kg / m³. If the fluid is classified as mud, its density needs to be estimated based on the sediment content. Since turbidity values ​​are strongly correlated with suspended sediment concentration, an approximate conversion relationship from turbidity value to sediment volume percentage can be established. For example, a pre-defined lookup table or polynomial fitting function can be used to establish this approximate conversion relationship. This lookup table is pre-constructed under laboratory conditions through turbidity measurements and sediment volume sedimentation experiments on mud samples with different sediment concentrations. For example, when the turbidity sensor reports a turbidity value of 500 NTU, the processing unit maps it to a sediment volume percentage of 10% using the lookup table. When the turbidity value is 1000 NTU, it is mapped to a sediment volume percentage of 25%. For turbidity values ​​between discrete points in the lookup table, the processing unit uses a linear interpolation method to obtain the corresponding sediment volume percentage.

[0084] Then, using the formula for calculating the density of a mixture—that is, mud density is approximately equal to the density of water multiplied by the volume percentage of water—plus a preset density of sediment particles multiplied by the volume percentage of sediment, the current fluid density of the mud is estimated. The preset sediment density can be determined in advance based on the local soil type, for example, 2650 kg per cubic meter. In this way, dynamically changing, more realistic density values ​​can be assigned to mud of different concentrations.

[0085] Furthermore, step S2 includes:

[0086] S21: Divide the movable road surface into virtual grid cells;

[0087] S22: For each virtual grid cell, calculate the average tilt, joint opening and closing, and settlement relative to the initial laying state based on the local tilt angle and joint displacement.

[0088] S23: When the average tilt or settlement exceeds the first preset threshold, the structural feature description of the corresponding virtual grid unit is marked as a local depression;

[0089] S24: When the opening degree of the joint exceeds the second preset threshold, the structural feature description of the corresponding virtual mesh unit is marked as a wear joint.

[0090] First, to facilitate spatial analysis and calculation, the entire movable road surface is logically divided into a two-dimensional grid composed of multiple virtual grid units. The size of each grid unit can be set according to the size of the road surface slab and the required calculation accuracy; for example, it can be set as a 2m × 2m square unit.

[0091] Then, for each virtual grid cell, the structural state of its interior and boundaries is quantified. Using data from all tilt sensors falling within the grid cell, an average tilt angle representing the overall tilt state of the cell is calculated. Similarly, using data from all FBG strain sensors at the cell boundary, the average joint opening is calculated. Settlement can be obtained by spatial integration of the tilt data or directly from dedicated settlement sensors, and the average settlement value of the cell relative to the initial laying height is calculated.

[0092] Next, the structural characteristics of each grid cell are qualitatively described by setting thresholds. For example, the first preset threshold is set to 2 degrees and 5 millimeters. If the average tilt of a grid cell exceeds 2 degrees, or its settlement exceeds 5 millimeters, the cell is considered to have formed a significant water accumulation potential, and its structural characteristic description is marked as a local depression. As another example, the second preset threshold is set to 3 millimeters. If the average joint opening of a grid cell exceeds 3 millimeters, the joint at that location is considered to be severely worn, potentially becoming a major seepage channel or flow obstruction, and its structural characteristic description is marked as a worn joint. A grid cell can simultaneously have multiple structural characteristic descriptions, such as being both a local depression and a worn joint. In this way, complex and continuous pavement deformation is discretized into a series of micro-regions with clear structural labels, providing a clear physical context for subsequent parameter matching. As a specific implementation method, the monitoring accuracy is dynamically improved when dealing with extreme conditions such as sudden high-intensity rainfall and continuous passage of heavy convoys. The entire mobile pavement is divided into finely refined grid cells with a side length of 1 meter. As the convoy passed, sensors collected data at a higher frequency, and the processing unit captured continuous fluctuations in deformation caused by ground softening. For one unit located on the path of a heavily loaded wheel, its settlement was detected to surge from 2 mm to 15 mm in a short period, far exceeding the first preset threshold. At this point, the structural feature description of this unit was instantly updated to a deep depression.

[0093] Furthermore, step S3 includes:

[0094] S31: Based on the fluid category determined by real-time physical property information, select the corresponding basic parameter set from the preset behavioral parameter set. The basic parameter set includes the basic roughness coefficient and the basic retention coefficient.

[0095] S32: Based on the structural feature description, call the corresponding parameter adjustment function, use the parameter adjustment function to correct the basic roughness coefficient and basic retention coefficient, and generate initial fluid behavior parameters that match the current situation.

[0096] The process of matching initial fluid behavior parameters to each micro-region consists of two steps. The first step is to select basic parameters based on the fluid type. The central processing unit pre-stores sets of behavior parameters for different fluid types. For example, there is a set of basic parameters for clean water and a set for mud. The basic roughness coefficient in the clean water set might be a small value, such as a Manning coefficient of 0.013, reflecting the smoothness of clean water flow on the composite material surface; its basic retention coefficient is also small. In contrast, the basic roughness coefficient in the mud set would be significantly larger, such as a Manning coefficient of 0.035, to reflect the greater flow resistance caused by the high viscosity of mud; its basic retention coefficient is also correspondingly larger. When the fluid in a micro-region is identified as mud, the mud basic parameter set is retrieved from the set as the starting point for subsequent adjustments.

[0097] The second step is parameter correction based on structural characteristics. After selecting the basic parameter set, the corresponding parameter adjustment functions are invoked based on the structural characteristics of the micro-region to refine the basic parameters. These adjustment functions are pre-established mathematical models that quantify the impact of specific structural features on fluid behavior. Through this step, the local geometric changes of the road surface are transformed into parameter changes that the fluid dynamics model can understand, thereby generating a set of initial fluid behavior parameters that perfectly match the specific context of the current micro-region.

[0098] Furthermore, step S32 includes:

[0099] S321: When the structural feature description includes local depressions, call the depression retention adjustment function to increase the foundation retention coefficient according to the settlement amount to obtain the corrected retention coefficient;

[0100] S322: When the structural feature description includes wear joints, call the joint seepage adjustment function to increase the basic roughness coefficient according to the joint opening degree to obtain the corrected roughness coefficient; the initial fluid behavior parameters include the retention coefficient and the roughness coefficient.

[0101] For example, when the structural feature description of a micro-region includes local depressions, the processing unit invokes a depression retention adjustment function. This function takes the settlement of the region as an input variable. A simple function form could be: the corrected retention coefficient equals the base retention coefficient plus an increment proportional to the settlement. For example, the increment could be equal to an adjustment coefficient multiplied by the settlement in millimeters. The deeper the settlement, the more pronounced the depression morphology, and the larger the calculated retention coefficient. Physically, this corresponds to a region where fluid is more likely to accumulate and water is more difficult to drain.

[0102] Similarly, when the structural feature description of a micro-region includes wear joints, the processing unit invokes a joint seepage adjustment function. This function takes the joint opening degree of the region as an input variable. The function's effect can be to increase the base roughness coefficient. For example, the corrected roughness coefficient equals the base roughness coefficient plus an increment related to the joint opening degree. The larger the joint opening degree, the stronger the disturbance and obstruction effect on water flow, and therefore the corresponding roughness coefficient should also be larger.

[0103] The retention coefficient and roughness coefficient obtained after correction by these adjustment functions together constitute the initial fluid behavior parameters of this micro-region. When dealing with local depression characteristics, the depression retention adjustment function is defined as follows: the corrected retention coefficient is equal to the basic retention coefficient multiplied by an exponential enhancement factor determined by the settlement amount. This factor is in the form of a power function of the natural constant, and the power is determined by the ratio of the settlement amount to the standard settlement threshold.

[0104] For wear joint characteristics, the joint seepage adjustment function incorporates the joint opening degree into the correction of the Manning roughness coefficient. The corrected roughness coefficient equals the basic roughness coefficient plus a linear compensation term, which is obtained by multiplying the joint opening degree by a preset friction increment constant. Through these specific mathematical functions, the qualitative structural description is transformed into precise numerical values ​​for use in fluid dynamics calculations.

[0105] Furthermore, in step S4, the initial fluid behavior parameters of the micro-region are adaptively adjusted according to the increment of structural deformation or the evolution trend of fluid properties, including the following steps:

[0106] S41: Establish the first corrected mapping relationship between the increment of structural deformation and the roughness coefficient in the fluid behavior parameters;

[0107] S42: Establish a second corrected mapping relationship between the evolution trend of fluid properties and the retention coefficient in fluid behavior parameters;

[0108] S43: Based on the first and second correction mapping relationships, the initial fluid behavior parameters are compensated in real time to obtain the adaptively adjusted fluid behavior parameters.

[0109] This step is the core of the entire method, demonstrating the model's adaptive evolution capability. It is not a one-time parameter matching, but a continuous and dynamic parameter optimization.

[0110] Specifically, the first correction mapping relationship aims to directly link the continuous degradation process of the pavement structure with changes in fluid flow resistance. For example, a relationship can be established whereby the increment of the roughness coefficient is a function of the increment of the joint opening. When the joint opening in a micro-region increases from 3 mm to 4 mm (an increment of 1 mm) over a period of time, the processing unit calculates a corresponding roughness coefficient compensation value using this mapping relationship and adds it to the current roughness coefficient. This allows the model to reflect the negative impact of accelerated pavement wear on fluid flow in a timely manner.

[0111] The second correction mapping aims to link the evolution trend of fluid properties with the fluid's retention characteristics on the road surface. For example, a relationship can be established whereby the increment of the retention coefficient is a function of the rate of change of fluid viscosity. If the processing unit detects a continuous upward trend in the fluid viscosity of a micro-region over the past five minutes, this mapping will output a positive retention coefficient compensation value, even if the current viscosity value has not yet reached the next level. This is a proactive adjustment; the model anticipates that the mud is becoming increasingly viscous and therefore increases its retention parameters in advance to more accurately predict impending, more severe water accumulation.

[0112] Through these two mapping relationships, the processing unit performs real-time, minute compensation and correction on the initial fluid behavior parameters of each micro-region, thereby obtaining a set of continuously evolving and adaptively adjusted fluid behavior parameters. This ensures that the input parameters used by the fluid dynamics model can always most accurately reflect the latest dynamics of the road surface and the fluid. Both mapping relationships can be implemented by establishing a dynamic incremental compensation matrix.

[0113] The first modified mapping relationship targets the increment of structural deformation, taking the rate of change of joint displacement as the independent variable, and maps it to the compensation value of the roughness coefficient through a second-order polynomial fitting function, thereby capturing the hindering effect of road surface micro-topological degradation on flow velocity.

[0114] The second modified mapping relationship targets the evolution trend of fluid properties and monitors the derivative of viscosity on the time axis. When the viscosity increases at a faster rate, the mapping relationship outputs a nonlinear retention compensation weight, which is proportional to the square root of the viscosity change rate, thus pre-reflecting the precipitous drop in fluidity caused by mud thickening in the model.

[0115] Immediately following, the model automatically triggered a rapid evolution logic. Based on the first correction mapping relationship, due to the extremely high rate of change of joint displacement, the roughness coefficient was given an instantaneous jump increment, compensated from the basic 0.035 to 0.052. Simultaneously, considering the massive influx of sediment caused by rainfall erosion, the second correction mapping relationship significantly increased the weighting of the retention coefficient based on the dramatic increase in viscosity. This adjustment enabled the numerical solution process to instantly reflect the state of fluid rapidly converging from the surroundings to the center and being difficult to expel when calculating the water depth in this area. The final calculated overburden thickness was corrected from 0.8 cm to 4.2 cm within three seconds, and the warning terminal immediately issued an emergency obstacle avoidance signal to subsequent drivers.

[0116] Furthermore, step S5 includes:

[0117] S51: Construct a fluid dynamics model based on two-dimensional shallow water equations;

[0118] S52: Calculate the bottom shear force in the fluid dynamics model based on the viscosity and fluid density in the instantaneous physical property information, and the roughness coefficient in the fluid behavior parameters;

[0119] S53: Correct the source and sink terms in the fluid dynamics model based on the seepage correction term in the fluid behavior parameters;

[0120] S54: The modified fluid dynamics model is solved using a numerical solution method to obtain the predicted values ​​of fluid flow velocity, spreading range, and overlay thickness of each micro-region at the current moment.

[0121] First, a mathematical model is constructed to describe the flow of fluid in a two-dimensional plane. The two-dimensional shallow water equations, also known as the Saint-Venant equations, are a classic and effective tool for describing such problems. These equations describe the changes in water depth and flow velocity in time and space through the laws of conservation of mass and momentum.

[0122] Solving this system of equations requires calculating several key terms. The bottom shear force represents the frictional resistance of the pavement to the water flow, and its magnitude directly affects the fluid velocity. According to fluid mechanics theory, the bottom shear force is proportional to the fluid density, the square of the velocity, and a friction coefficient. Here, the fluid density is an instantaneously estimated density of mud or water, while the friction coefficient is directly derived from an adaptively adjusted roughness coefficient. In this way, instantaneous fluid properties and pavement structural conditions are directly coupled into the model's dynamic calculations.

[0123] Source and sink terms represent the increase or decrease of fluid. Besides rainfall as the primary source term, adjustments need to be made based on the condition of worn joints. If the joint opening degree of a micro-region is large, a seepage velocity can be calculated based on its opening degree and water depth, and this velocity can be added as a negative sink term to the equation, representing fluid loss from the joint. In two-dimensional shallow water equations, source and sink terms typically represent the rate of change of fluid volume per unit area. For the specific characteristics of movable pavements, the correction process involves introducing a negative sink term to simulate leakage at the joints. This leakage is calculated based on the joint opening degree and the current mesh water depth, combined with the orifice outflow formula: the leakage flow rate equals the product of the flow coefficient, the joint opening area, and the square root of the water depth. The calculated fluid volume lost through the joint per unit time is subtracted from the total water volume of the micro-region, ensuring the model accurately reflects water loss due to loose pavement slabs.

[0124] Finally, the two-dimensional shallow water equations, which include all these terms corrected for real-time information, are solved in a computational domain composed of virtual grid cells using established numerical methods, such as the finite volume method or the finite difference method. By iterating step-by-step over time, the average fluid flow velocity, the fluid spread range, and the final predicted overburden thickness for each micro-region at the current moment can be calculated.

[0125] To further improve the accuracy of numerical solutions, especially when simulating details of fluid spreading processes, the specific steps of the numerical solution method include:

[0126] S541: Within each solution time step, identify micro-regions whose predicted overburden thickness is greater than zero and which are adjacent to regions whose predicted overburden thickness is zero, and mark them as leading edge boundary regions.

[0127] S542: Obtain surface wetting status information of the leading edge boundary region;

[0128] S543: Adjust the bottom shear force corresponding to the leading edge boundary region based on the surface wetting state information to correct the spreading resistance of the fluid dynamics model in the leading edge boundary region, thereby obtaining the predicted values ​​of fluid flow velocity, spreading range and capping layer thickness of each micro-region at the current moment.

[0129] Within each tiny time step of the numerical solution, it is necessary to address the boundary of fluid expansion from a water-rich region to a dry region; this boundary is called the leading edge boundary or wet-dry boundary. The speed at which this boundary moves is crucial for predicting the extent of fluid spread. Therefore, it is first necessary to accurately identify these leading edge boundary regions, which are micro-regions where the water depth is greater than zero on their own, but the water depth of an adjacent grid cell is zero.

[0130] After identifying the leading edge boundary region, it is necessary to obtain information on the surface wetting status of the road surface in these areas. Whether the road surface is completely dry or already covered with a thin film of water will result in different resistances to the spread of new fluid. Dry surfaces will generate greater resistance to the fluid leading edge due to effects such as surface tension. This wetting status information can be obtained through a dedicated surface humidity sensor or inferred from calculations over previous time steps.

[0131] Finally, based on the obtained surface wetting information, the formula for calculating the bottom shear force in the leading edge boundary region is locally adjusted. For example, if the surface is dry, a correction factor greater than 1 can be multiplied in the standard bottom shear force calculation formula to simulate increased paving resistance. This refined treatment of the leading edge boundary region can effectively avoid instability problems in numerical calculations and more realistically simulate the paving process of fluid on the road surface, thereby improving the prediction accuracy of paving range and overburden thickness.

[0132] As a specific implementation, in the scenario of the first rainfall after a long period of drought, the road surface is extremely dry. At this time, the initial spread of fluid on the composite panel is significantly affected by surface tension and capillary resistance. Within each time step of the numerical solution, the computational program scans all mesh cells, precisely identifying those cells whose current predicted overburden thickness is greater than zero and which are adjacent to dry regions with zero thickness, marking them as active leading-edge boundary regions.

[0133] To quantify the impact of surface wetting state on the fluid front, a wetting delay coefficient was introduced. Reflectivity data of the leading edge boundary region was acquired through an optical monitoring unit to determine whether a continuous water film had formed on the road surface. When the surface was determined to be dry, the wetting delay coefficient was set to 1.8. This coefficient directly affects the calculation of the bottom shear force, causing the flow resistance term in that region to be locally amplified. In the numerical solver, this resistance correction limits the unrestricted diffusion of mud at the leading edge, simulating the head accumulation effect of mud on dry roads due to high resistance. As the calculation progresses, once the cumulative water depth of the leading edge unit exceeds the wetting critical value, the surface wetting state is updated to wet, and the delay coefficient drops back to 1.0. This refined boundary control ensures that the model's predicted coverage area is highly consistent with the actual spread boundary transmitted back by the ultrasonic rangefinder when simulating the slow spread of water accumulation in the early stages of light rain.

[0134] After completing the prediction, the method also includes:

[0135] S6: Obtain ultrasonic measurement data of the thickness of the moving road surface;

[0136] S7: Calculate the deviation between the predicted thickness of the overburden layer and the ultrasonically measured thickness data;

[0137] S8: Feedback calibration is performed on the adaptive adjustment logic based on the deviation value to correct the adjustment weights of the fluid behavior parameters.

[0138] This is a closed-loop feedback calibration mechanism designed to continuously optimize the model's predictive capabilities using real-world observational data. First, non-contact ultrasonic ranging sensors are deployed at several key locations on the road surface, such as historical waterlogging hotspots or areas of significant structural deformation. These waterproof sensors emit ultrasonic waves downwards and receive signals reflected from the fluid surface. By calculating the time-of-flight of the sound waves, the actual thickness of the fluid cover at the current location is accurately measured.

[0139] After obtaining these measured thickness data, they are compared with the predicted thickness of the overburden layer output by the model at the same location and time, and the deviation between the two is calculated.

[0140] If the deviation consistently exceeds a preset tolerance threshold, such as 0.5 mm, a feedback calibration procedure is triggered. This procedure adjusts the correction mapping relationship in the aforementioned adaptive adjustment step in reverse, based on the sign and magnitude of the deviation. For example, if the model's predicted value for a certain depression is consistently and systematically lower than the measured value, the calibration procedure will automatically and slightly increase the weight coefficient in the depression retention adjustment function. This calibration can be implemented using optimization algorithms such as gradient descent or Kalman filtering. Through this continuous feedback calibration, the model can continuously learn from errors and improve its internal parameter adjustment logic, thereby maintaining high-accuracy predictive performance over long-term operation.

[0141] As a specific implementation method, in the mid-to-late stages of road surface service, due to microscopic wear on the composite material surface, the preset parameter adjustment function may experience systematic drift. At this time, ultrasonic ranging sensors deployed at the center of depressions and the main drainage channel play a crucial calibration role. During a three-hour rainfall monitoring session, the ultrasonic sensors transmitted a set of precise measured thickness data back to the processing unit every five minutes. The processing unit synchronously compared these measured values ​​with the predicted values ​​output by the hydrodynamic model at the corresponding coordinate points.

[0142] When analysis revealed that the deviation values ​​at ten consecutive sampling points all showed a positive shift, meaning the measured values ​​were generally about 15% higher than the predicted values, the feedback calibration mechanism was activated. The calibration logic no longer simply modifies a single parameter, but rather performs a global correction to the adaptively adjusted weight matrix. Specifically, the gradient descent algorithm is used to find the optimal weights in the parameter space that minimize the sum of squared deviations. The calibration process fine-tuned the coefficients in the depression retention adjustment function, increasing them from 0.12 to 0.14, and reduced the decay rate of the joint seepage correction term. After two rounds of iterative calibration, the prediction deviation value rapidly contracted to within 0.2 mm in the following half hour. This closed-loop optimization process not only corrected the current prediction error but also, by adjusting the weights, enabled the model to better adapt to changes in flow characteristics caused by pavement material aging, ensuring prediction reliability throughout the entire lifecycle.

[0143] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for constructing a water film thickness prediction model for movable pavements in low-lying areas, characterized in that, The method includes the following steps: S1: Obtaining real-time structural state information of the mobile pavement during the loading process and real-time physical property information of the pavement fluid; S2: Based on the real-time structural state information, the road surface is divided into multiple micro-regions, and the structural feature description of each micro-region is determined according to the degree of deformation; S3: Based on the structural feature description and the real-time physical property information, corresponding initial fluid behavior parameters are matched for each micro-region; S4: Continuously monitor the dynamic changes of the real-time structural state information and the real-time physical property information. When the change exceeds a preset threshold, adaptively adjust the initial fluid behavior parameters of the micro-region according to the increment of structural deformation or the evolution trend of fluid properties. S5: Substitute the adaptively adjusted fluid behavior parameters and the instantaneous physical property information into the fluid dynamics model for numerical solution to obtain the predicted value of the overburden thickness of each micro-region under the current structural state.

2. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 1, characterized in that, Step S1 includes: S11: Obtaining the local tilt angle of the road panel through a preset angle sensing unit; S12: Obtaining wavelength drift data at the joint of the road panel through a preset deformation sensing unit, and converting the wavelength drift data into the joint displacement; wherein, the real-time structural state information includes at least the tilt angle of the road panel and the joint displacement; S13: Obtaining the light intensity attenuation value when the fluid flows through the monitoring point, and calculating the turbidity value of the fluid based on the ratio of the light intensity attenuation value to the initial light intensity; S14: Obtaining the vibration decay time constant of the preset detection unit in the fluid, and inverting the viscosity of the fluid based on the vibration decay time constant; the real-time physical property information includes at least the turbidity value and viscosity of the fluid.

3. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 2, characterized in that, The instantaneous physical property information also includes the fluid density. After step S14, the following steps are added: S15: Determine the category of the fluid based on the turbidity value and the viscosity. When the turbidity value exceeds the turbidity threshold and the viscosity exceeds the viscosity threshold, the fluid is determined to be mud; otherwise, the fluid is determined to be clear water. S16: When the fluid is mud, estimate the fluid density of the mud category based on the turbidity value and the preset mud density. When the fluid is clear water, use the density of water as the fluid density.

4. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 2, characterized in that, Step S2 includes: S21: Dividing the movable pavement into virtual grid units; S22: For each virtual grid unit, calculating the average inclination, joint opening and closing degree, and settlement relative to the initial paving state based on the local tilt angle and the joint displacement; S23: When the average inclination or the settlement exceeds a first preset threshold, marking the structural feature description of the corresponding virtual grid unit as a local depression; S24: When the joint opening and closing degree exceeds a second preset threshold, marking the structural feature description of the corresponding virtual grid unit as a wear joint.

5. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 4, characterized in that, Step S3 The process includes: S31: Selecting a corresponding set of basic parameters from a preset set of behavioral parameters based on the fluid category determined by the instantaneous physical property information, wherein the set of basic parameters includes a basic roughness coefficient and a basic retention coefficient; S32: Calling a corresponding parameter adjustment function based on the structural feature description, and using the parameter adjustment function to correct the basic roughness coefficient and the basic retention coefficient to generate initial fluid behavior parameters that match the current situation.

6. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 5, characterized in that, Step S32 includes: S321: When the structural feature description includes local depressions, the depression retention adjustment function is invoked to increase the foundation retention coefficient according to the settlement amount, thereby obtaining a corrected retention coefficient; S322: When the structural feature description includes wear joints, the joint seepage adjustment function is invoked to increase the foundation roughness coefficient according to the joint opening degree, thereby obtaining a corrected roughness coefficient; The initial fluid behavior parameters include the retention coefficient and the roughness coefficient.

7. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 1, characterized in that, In step S4, the initial fluid behavior parameters of the micro-region are adaptively adjusted according to the increment of structural deformation or the evolution trend of fluid properties, including the following steps: S41: Establishing a first correction mapping relationship between the increment of structural deformation and the roughness coefficient in the fluid behavior parameters; S42: Establish a second corrected mapping relationship between the evolution trend of the fluid properties and the retention coefficient in the fluid behavior parameters; S43: Based on the first correction mapping relationship and the second correction mapping relationship, the initial fluid behavior parameters are compensated in real time to obtain the adaptively adjusted fluid behavior parameters.

8. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 1, characterized in that, Step S5 includes: S51: Constructing a fluid dynamics model based on two-dimensional shallow water equations; S52: Calculating the bottom shear force in the fluid dynamics model based on the viscosity and fluid density in the instantaneous physical property information and the roughness coefficient in the fluid behavior parameters; S53: Correcting the source and sink terms in the fluid dynamics model based on the seepage correction term in the fluid behavior parameters; S54: Solving the corrected fluid dynamics model using a numerical solution method to obtain the predicted values ​​of fluid flow velocity, spreading range, and overburden thickness of each micro-region at the current moment.

9. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 8, characterized in that, Step S54 includes: S541: In each solution time step, identify the micro-regions whose predicted capping thickness is greater than zero and which are adjacent to the region whose predicted capping thickness is zero, and mark them as leading edge boundary regions; S542: Obtain the surface wetting state information of the leading edge boundary regions; S543: Adjust the bottom shear force corresponding to the leading edge boundary regions according to the surface wetting state information to correct the spreading resistance of the fluid dynamics model in the leading edge boundary regions, thereby obtaining the fluid flow velocity, spreading range and predicted capping thickness of each micro-region at the current moment.

10. The method for constructing a water film thickness prediction model for a movable pavement in a low-lying area according to claim 1, characterized in that, After step S5, Includes: S6: Acquiring ultrasonically measured thickness data of the movable road surface; S7: Calculate the deviation between the predicted thickness of the cover layer and the ultrasonically measured thickness data; S8: Feedback calibration is performed on the adaptive adjustment logic based on the deviation value to correct the adjustment weights of the fluid behavior parameters.

Images