A method for detecting moisture content of subgrade based on Topp model array radar

Abstract

This application belongs to the field of electronic digital data processing and relates to a method for detecting subgrade moisture content using array radar based on the Topp model. The method includes: step S1, system preparation and field setup; step S2, synchronous data acquisition; step S3, dynamic CMP reconstruction and signal processing; step S4, establishing a Topp model considering compaction correction, including establishing a Topp model considering compaction correction to retrieve moisture content; and step S5, real-time moisture content retrieval and intelligent early warning imaging. The subgrade moisture content non-destructive testing system and method involved in this application integrates "high-efficiency array acquisition, accurate ray tracing retrieval, and multi-field coupling model correction," significantly improving the quantitative accuracy and physical authenticity of moisture content retrieval.

Description

Technical Field

[0001] This invention belongs to the field of electronic digital data processing, specifically relating to a method for detecting roadbed moisture content using array radar based on the Topp model. Background Technology

[0002] As the main load-bearing component of transportation infrastructure, changes in the moisture content of the roadbed can alter mechanical properties such as the resilient modulus and cumulative plastic strain, leading to roadbed instability, subsidence, deformation, and other damages, as well as pavement cracking and uneven settlement. Therefore, efficient and accurate detection of roadbed moisture content is crucial for ensuring project quality.

[0003] Methods for detecting the moisture content of roadbeds mainly include the drying method, sand cone method, time domain reflectance (TDR), and conventional ground-penetrating radar (GPR) technology. Firstly, traditional contact-based detection methods (such as the drying method, sand cone method, and TDR) are mostly destructive or invasive point measurements, only reflecting the state of discrete points. This not only damages the roadbed structure but also fails to provide continuous detection profiles, making it difficult to meet the timeliness requirements of large-scale surveys. Secondly, although GPR technology can achieve non-destructive detection, it still faces three major technical bottlenecks in the actual accurate quantitative detection of roadbeds: acquisition efficiency, inversion algorithms, and physical models. Firstly, while the traditional common center point (CMP) method can obtain velocity information, it requires fixed-point operation, resulting in extremely low efficiency. Secondly, while the continuous scanning method is highly efficient, it lacks multi-offset data, making it difficult to guarantee inversion accuracy. However, the traditional common center point (CMP) method requires manual adjustment of antenna spacing for point-by-point measurements, resulting in extremely low efficiency and unsuitability for rapid scanning of long-distance roadbeds. While continuous scanning based on a fixed offset is efficient, it lacks multi-angle travel time information, making it difficult to directly invert layer velocities and leading to significant deviations in dielectric constant estimation. Secondly, existing layer velocity inversion algorithms have theoretical flaws in thin-layer roadbed detection. Current GPR data processing often uses the Dix formula from petroleum seismic exploration, which is based on the assumptions of horizontally layered media and small offsets. However, roadbed structures are characterized by thin layers and large differences in wave impedance, causing significant refraction of electromagnetic waves at the interface between the surface and base layers. In such thin media as roadbeds, the Dix formula, which ignores the refraction path, amplifies even small travel time errors, leading to severe distortion of the inverted layer velocities and failing to meet the requirements for refined detection. Finally, the traditional Topp formula is based only on natural soil and does not consider the high compaction characteristics of artificial roadbeds. It ignores the coupling effect of changes in pore structure on dielectric properties, and its direct application will produce large systematic errors. In summary, the existing technology still lacks a non-destructive testing system and method for roadbed moisture content that integrates "efficient array acquisition, ray tracing accurate inversion, and multi-field coupling model correction". Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this invention provides a method for detecting roadbed moisture content using an array radar based on the Topp model. The improvement lies in that the method includes: Step S1, system preparation and on-site setup, including: Step S11: Install an antenna array on the detection vehicle; the antenna array includes one transmitting antenna and four receiving antennas; Step S12: Construct a multi-channel observation system with different offsets, measure and verify the array geometric parameters; input the geometric parameters into the observation system of the acquisition software; Step S2: Synchronous data acquisition. Control all channels of the array antenna to acquire data synchronously once, and write the latitude, longitude and elevation information output by GNSS into the data header file; Step S3, dynamic CMP reconstruction and signal processing, includes: Step S31: Standardize and preprocess the collected multi-channel raw data; Step S32: Perform dynamic CMP reconstruction to construct dynamic CMP gathers distributed along the survey line; Step S33: Calculate the weighted similarity coefficient for each CMP gather. scanning; Step S34: Construct a layer velocity inversion model based on ray tracing; Step S35: Perform iterative layer stripping inversion; Step S36: Calculate the relative permittivity to obtain the accurate relative permittivity of the target roadbed. ; Step S4: Establish a Topp model considering compaction correction, including establishing a Topp model considering compaction correction to invert moisture content, as shown in the following formula:

[0005] Using a numerical inversion algorithm, the accurate relative permittivity obtained in step S3 is... Substituting the known subgrade design compaction degree K into the Topp model with the compaction degree correction, the corrected subgrade volumetric moisture content is obtained by inverse solving. ; Step S5: Real-time moisture content inversion and intelligent early warning imaging.

[0006] Preferably, step S12, constructing a multi-channel observation system with different offset distances, includes: setting the center distance between the transmitting antenna Tx and the first receiving antenna Rx1 to x0=0.6m, the spacing between adjacent receiving antennas Δx=0.2m, the offset distance between the transmitting antenna Tx and the first receiving antenna Rx1 to x1=0.6m; the offset distance between the transmitting antenna Tx and the second receiving antenna Rx2 to x2=0.8m; the offset distance between the transmitting antenna Tx and the third receiving antenna Rx3 to x3=1.0m; and the offset distance between the transmitting antenna Tx and the fourth receiving antenna Rx4 to x4=1.2m.

[0007] Preferably, step S2, dynamic acquisition of vehicle-mounted continuous data, includes: Step S21: Drive the testing vehicle into the starting point of the road section to be tested and keep the vehicle centered. Step S22: Drive the testing vehicle at a constant speed of 20 km / h along the roadbed survey line; Step S23: The DMI encoder outputs a TTL trigger signal at fixed intervals of dx=0.05m based on the vehicle's travel distance. In step S24, after receiving the trigger signal, the radar host controls all channels of the array antenna to synchronously collect data once, and at the same time writes the latitude, longitude and elevation information output by GNSS into the data header file to form a raw radar data stream with accurate mileage and coordinate information.

[0008] Preferably, in step S33, a weighted similarity coefficient is calculated for each CMP gather. The scanning range is set to 0.03~0.17 m / ns, as shown in the following formula:

[0009] In the formula, Zero offset time t 0 and probing speed v The following is the similarity coefficient; M The number of channels in the CMP channel set; A i For the first i The amplitude value of the Dao; W The width of the time window; j A sliding index for sampling points within a time window; The sampling time interval for radar data; t i (v) For the hyperbolic trajectory time; Step S34: Construct a layer velocity inversion model based on ray tracing, as shown in the following equation:

[0010] In the formula, Let be the incident angle of the i-th layer; v i Let be the layer velocity of the i-th medium layer, and there are a total of n medium layers; p is the ray parameter; Obtain the total horizontal offset of electromagnetic waves x and round trip t As shown in the following formula:

[0011]

[0012] In the formula, h i For the first i Layer thickness; Step S35, perform iterative layer stripping inversion, including: Construct the objective function The theoretical travel time and the measured CMP gather picking travel time t were obtained. obs The residuals between:

[0013] In the formula, M This represents the number of channels in the array radar. x j For the first j The offset between the transmitting and receiving antennas corresponding to each channel; t obs (x j ) The first one picked up from the measured CMP channel set j Two-way travel time corresponding to each offset; The first one obtained based on ray tracing forward modeling j The offset probes the speed at the nth layer. v n Theoretical two-way travel time is as follows;

[0014] Iterative optimization of the objective function: when the objective function If the velocity is less than the preset convergence threshold, then this layer is the optimal layer, and the corresponding velocity is the optimal layer velocity v. int,n ; Step S36, Calculation of relative permittivity Based on the optimal layer velocity v int,n Using the formula Obtain the precise relative permittivity of the target road base layer. .

[0015] Preferably, step S4 establishes the Topp model considering compaction correction, including the following steps: Step S41: Select the subgrade fill soil of the road section to be tested, and obtain the optimum moisture content (OMC) and maximum dry density of the fill soil. ; Step S42: Design four different compaction gradient specimens; and design five different moisture content gradient specimens. Step S43, calculate the required dry soil mass and water addition: Dry soil mass m d,i Based on the target compaction degree K i The calculation is shown in the following formula: ； V is the known volume of the standard specimen mold; Based on the target quality moisture content Calculate the required amount of water to add m w,i As shown in the following formula: ; Step S44: Weigh the dry soil. Water After mixing, the mixture is sealed and pressed into shape under static pressure, then compacted in layers and placed into a mold. Step S45: Determine the relative permittivity. The arithmetic mean of multiple measurements is taken as the effective relative permittivity of the specimen under the set compaction degree and moisture content conditions. ; Drying to obtain the true mass and moisture content of the specimen And convert it to the actual volumetric moisture content. ; Step S46, organize the obtained multiple sets of data pairs ( K , , The data is fitted to a three-dimensional surface using the nonlinear least squares method to establish a modified Topp forward model that includes moisture content and compaction degree, as shown in the following equation:

[0016] In the formula, The effective relative permittivity; This refers to the volumetric water content. K For compaction degree, This is the compaction sensitivity coefficient. Based on the fundamental dielectric constant parameter (i=0,1,2,3), the... and These are fixed constants obtained through indoor three-dimensional calibration experiments.

[0017] Preferably, step S5, real-time water level inversion and intelligent early warning imaging, includes: Step S51: Determine the precise relative permittivity of the target roadbed layer obtained in step S3. Combining the target compaction degree, or the compaction degree information obtained on-site, and substituting it into the corrected model established in step S4, the continuous volumetric moisture content inside the subgrade is calculated. As shown in the following formula: ; Obtain high-precision volumetric moisture content after compaction correction. ( ); Step S52: Perform median filtering or moving average processing on the original moisture content data stream; Step S53: Construct a two-dimensional spatiotemporal coordinate system with the vehicle mileage as the horizontal axis and the roadbed depth as the vertical axis; use pseudo-color mapping technology to draw a cloud map of roadbed moisture content distribution, with different colors representing different moisture contents; Step S54, Hierarchical intelligent early warning mechanism: Set moisture content early warning threshold When the moisture content is calculated by inversion > When this occurs, the area is automatically identified as an area with abnormal humidity. Step S55: After the inspection is completed, a quality inspection report is generated. Compared with the prior art, the beneficial effects of the present invention are as follows: First, this invention significantly improves the quantitative accuracy and physical authenticity of moisture content inversion. By constructing a ternary correction model of "compaction degree-relative permittivity-moisture content" based on a large number of indoor coupled experiments, it quantifies for the first time the nonlinear coupling effect of soil compaction structure on dielectric properties under high-pressure compaction conditions. Combined with known subgrade design compaction degree or compaction degree information obtained in the field through other rapid methods, it performs constrained inversion, effectively correcting the dielectric constant drift caused by changes in soil skeleton density, significantly reducing systematic errors under different compaction conditions, and achieving a leap from "qualitative trend judgment" to "high-precision quantitative evaluation". Secondly, considering the structural characteristics of roadbeds with "thin layers and large differences in wave impedance," this invention eliminates the inversion divergence problem caused by neglecting refraction effects in the traditional Dix formula. By constructing a dynamic common center point (CMP) gather using an array antenna, an iterative layer-stripping inversion algorithm based on ray tracing theory is employed. This algorithm follows Snell's law, fully considering the refraction path of electromagnetic waves in layered media, and accurately solves for the true layer velocity of each structural layer by minimizing the travel time residual, thereby obtaining an accurate relative permittivity. This improvement eliminates the theoretical blind spot in shallow medium inversion, obtaining high-confidence model input parameters without relying on on-site core drilling calibration, truly achieving non-destructive, independent, and accurate inversion; furthermore, this system balances detection efficiency and data quality. This invention utilizes a multi-channel array antenna to simulate common center point (CMP) data acquisition during movement, replacing the traditional "walk-stop-walk" operation, thereby improving detection efficiency by more than 70%. At the same time, the high-resolution velocity spectrum and superimposed profile constructed based on multi-channel data effectively suppress random environmental noise and interlayer multiple wave interference, significantly improving the signal-to-noise ratio of weak signals in deep roadbeds, and ensuring that extremely high data quality can still be maintained under high-speed acquisition conditions. Finally, the technical solution integrates a fully automated processing module, supporting real-time calculation from raw data to moisture content results. The continuous moisture content color raster profile generated by the system can intuitively reflect the overall uniformity of the roadbed working surface, and combined with GNSS high-precision positioning information, it realizes automatic alarm and station locking of abnormal areas (overly wet / overly dry), constructing a "test-and-evaluate, traceable" digital archive for roadbed construction quality, effectively avoiding the omission of local hidden dangers caused by traditional point measurement methods. Attached Figure Description

[0018] Figure 1 This is a flowchart of the roadbed moisture content array radar detection method involved in the present invention. Detailed Implementation

[0019] To better understand this invention, the following description, in conjunction with the accompanying drawings and examples, will further illustrate the invention.

[0020] To address the following technical deficiencies in the existing technology: (1) The traditional Topp formula and existing modified models are mostly based on naturally loose soil, ignoring the nonlinear coupling effect of the compaction characteristics of the subgrade on the dielectric properties, resulting in a large deviation in the water content inversion of the single variable model under the subgrade working conditions with significant changes in compaction. (2) The conventional fixed offset (single channel) mode cannot directly obtain wave velocity, and can only rely on empirical estimation or core drilling calibration, which leads to doubts about the calculation of dielectric constant; while the traditional common midpoint (CMP) method requires fixed-point operation, which is inefficient and cannot meet the problem of continuous and real-time acquisition of velocity spectrum for long-distance subgrades. (3) Conventional inversion algorithms (such as the Dix formula) are mainly applicable to deep thick media, ignoring the significant refraction effect of electromagnetic waves in the thin subgrade structure, resulting in serious distortion of the layer velocity inversion results; and the single-channel scanning data has low signal-to-noise ratio and weak anti-interference ability in deep areas. (4) To address the problem that existing technologies lack a system that integrates efficient data acquisition and multi-physical quantity collaborative inversion, making it impossible to construct a three-element coupled inversion mechanism of "compaction degree-relative permittivity-moisture content", and thus difficult to achieve high-precision, non-destructive real-time evaluation of roadbed construction quality, this application provides a roadbed moisture content array radar detection method based on the Topp model. The detection system involved in this application includes: a multi-channel array radar host connected to the data processing unit, a DMI ranging encoder, and an RTK-GNSS positioning module; The data processing unit includes a ruggedized industrial computer with multi-threaded parallel processing capabilities; Antenna: Includes a multi-degree-of-freedom rigid telescopic bracket, mounted at the rear of the vehicle; A high-precision electronic level is used to calibrate the antenna array, ensuring that the bottom surfaces of all antenna elements are on the same horizontal plane and that the polarization direction is strictly orthogonal to the driving direction, so as to ensure that electromagnetic waves enter the ground in the best coupling mode.

[0021] like Figure 1 As shown, the detection methods include: Step S1 System preparation and on-site setup includes: Step S11: Install an antenna array on the detection vehicle; the antenna array includes one transmitting antenna and four receiving antennas.

[0022] Specifically, a suspended air-coupled antenna array is mounted on the rear of the inspection vehicle using a rigid bracket. The antenna array consists of one transmitting antenna (Tx) and four receiving antennas (Rx1-Rx4), with the center frequency set at 500MHz. The receiving antennas are arranged at equal intervals in a straight line along the direction of travel. The bracket height is adjusted to ensure that the clearance between the bottom of the antenna and the roadbed surface remains within the range of 30cm to 50cm. The antenna is kept horizontal, and the bracket joints are locked to balance signal coupling efficiency and traffic passability.

[0023] The detection system involved in this application also includes: a sensor for antenna geometry calibration: after system startup, a thermal equilibrium preheating period of no less than 10 minutes is performed. Then, geometry calibration is performed: the antenna height is finely adjusted via the servo motor of the bracket or a manual knob, controlling the antenna bottom surface to be 40 mm from the roadbed surface. 2cm.

[0024] Step S12: Construct a multi-channel observation system with different offset distances, measure and verify the array geometric parameters, and input the geometric parameters into the observation system of the acquisition software. Specifically, set the center distance between the transmitting antenna Tx and the first receiving antenna Rx1 to x0 = 0.6m, and the spacing between adjacent receiving antennas Δx = 0.2m, thereby constructing a multi-channel observation system with offset distances of x1 = 0.6m, x2 = 0.8m, x3 = 1.0m, and x4 = 1.2m between the transmitting antenna Tx and each receiving antenna, respectively, and measure and verify the array geometric parameters. Accurately input these geometric parameters into the "Observation System" settings of the acquisition software.

[0025] Specifically, when constructing the observation system configuration file in the software, input the precisely measured spatial coordinate offsets (Offset: 0.6m, 0.8m, 1.0m, 1.2m) between the transmitting antenna Tx and each receiving antenna Rx1~Rx4. Set the acquisition parameters: set the time window depth to 60ns (adjust according to the roadbed thickness), the number of sampling points to 512 or 1024, and the waveform stacking to 4~8 times to improve the signal-to-noise ratio.

[0026] Step S13: Perform elevation correction. Specifically, a high-precision photoelectric encoder (DMI) with a resolution of 1000 pulses / cycle is installed on the rear axle of the detection vehicle and connected to the "TriggerIn" port of the radar host via a dedicated shielded cable. An RTK-GNSS module is installed directly above the center of the antenna array. The height from the GNSS antenna phase center to the ground is measured and input into the software for elevation correction, eliminating geometric distortions caused by ground undulations and changes in antenna height. The photoelectric encoder (DMI) outputs TTL pulse signals based on wheel rotation.

[0027] Step S14: Perform air wave correction. Specifically, start the radar host and industrial control computer. Set the radar host pulse repetition frequency (PRF) to ≥500kHz and the time window range to 100ns. Check the background noise of each channel waveform. If the background noise is too high, check the grounding and cable shielding. Perform "air wave correction": Park the vehicle on a flat surface, place a metal plate horizontally on the road surface, record the arrival time of the direct wave and the wave reflected from the metal plate, adjust the system's time zero point t0, and ensure that the time zero point drift error is ≤0.5ns.

[0028] Step S2 involves synchronous data acquisition, where all channels of the control array antenna are synchronously acquired once, and the latitude, longitude, and elevation information output by GNSS is written into the data header file.

[0029] Specifically, the implementation process of dynamic acquisition of vehicle-mounted continuous data in step S2 is as follows: S21. Drive the testing vehicle into the starting point of the section to be tested and keep the vehicle centered.

[0030] S22. Drive the inspection vehicle at a constant speed of 20 km / h along the roadbed survey line. The S23 and DMI encoders output TTL trigger signals at fixed intervals of dx=0.05m based on the vehicle's travel distance. Specifically, as the test vehicle travels at a constant speed along the test line, the DMI encoder outputs TTL pulse signals based on the rotation of the wheels.

[0031] S24. After receiving the trigger signal, the radar host controls all channels of the array antenna to synchronously acquire data once. Simultaneously, it writes the latitude, longitude, and elevation information output by GNSS into the data header file, forming a raw radar data stream with precise mileage and coordinate information. Specifically, at the instant the radar host receives the pulse trigger, it synchronously controls the transmitting antenna to emit an electromagnetic pulse and controls the four receiving antennas to synchronously record the echo signal. At the same time, the GNSS module outputs an NMEA format positioning data stream at a frequency of 1Hz or higher, and the software spatially matches the positioning coordinates with each radar data channel through timestamp interpolation.

[0032] During the data acquisition process, the software automatically performs data stream processing in the background, executing steps S3 and S4: First, the raw radar data undergoes DC removal and bandpass filtering; then, the common offset profile is reconstructed into a dynamic CMP gather using a geometric rearrangement algorithm; the roadbed reflection layer is automatically tracked through energy scanning to extract the two-way travel time at different offsets; finally, the layer velocity V is accurately calculated by using the built-in ray tracing iterative inversion algorithm, based on Snell's law, to minimize the travel time residual. int The actual volumetric moisture content was calculated using the Topp model, which takes compaction correction into account. θ v .

[0033] Specifically, step S3, dynamic CMP reconstruction and signal processing, includes: S31. Perform standardization preprocessing on the acquired multi-channel raw data. Specifically, perform standardization preprocessing on the acquired multi-channel raw data: perform DC removal to eliminate zero drift; perform Dewow processing to eliminate low-frequency induced noise; and perform background removal to suppress interference from ground direct waves and antenna internal coupled waves.

[0034] S32. Perform dynamic CMP reconstruction to construct dynamic CMP gathers distributed along the survey line. The geometric rearrangement algorithm uses the relative geometric positions of each channel of the array antenna and the precise movement distance recorded by DMI to calculate the spatial coordinates of the transmission and reception points of each data point. Through spatial coordinate matching, common offset (CO) data collected at different times but with the same spatial midpoint position are extracted and recombined to form dynamic CMP gathers that reflect the same underground location but have different offset information.

[0035] S33. Calculate the weighted similarity coefficient for each CMP gather. Scanning. Specifically, a weighted similarity coefficient is applied to each CMP gather. The scan range is set to 0.03~0.17 m / ns, as shown in the following formula:

[0036] In the formula, Zero offset time t 0 and probing speed v The following is the similarity coefficient; M The number of channels in the CMP channel set (M=4 in this embodiment); A i For the first i The amplitude value of the Dao; The time window width (in this embodiment, it is taken as 1 / 2 of the radar wavelet width, i.e., 2ns) is used to perform energy statistical smoothing on the time axis and eliminate instantaneous noise interference; t i (v) For the hyperbolic trajectory time; j A sliding index for sampling points within a time window; This represents the sampling time interval for radar data.

[0037] S34. Construct a ray-tracing-based velocity inversion model for the roadbed. Specifically, the roadbed is considered as a horizontally layered medium, assuming that electromagnetic waves propagate according to geometric optics principles. For the i-th layer of the roadbed (out of a total of n layers), when the offset between the transmitting and receiving antennas is x, the propagation path of the electromagnetic wave is determined by the refraction points at each layer interface, and Snell's law is satisfied at each layer interface, as shown in the following equation:

[0038] In the formula, Let be the incident angle of the i-th layer. v i Let be the layer velocity of the i-th layer, and p be the ray parameter (Snell constant).

[0039] At this time, the total horizontal offset of the electromagnetic wave x and round trip t It can be represented as a function of the parameters of each layer:

[0040]

[0041] In the formula, h i For the first i The thickness of the layer.

[0042] S35. Perform iterative layer stripping inversion. Specifically, perform iterative layer stripping inversion: solve for the parameters layer by layer in a top-down order, with the following steps: (1) First layer inversion: For the first layer of the subgrade (usually the surface layer), there is no interference from the overlying layer. The velocity of the first layer is directly obtained by hyperbolic fitting. v 1 and thickness h 1 .

[0043] (2) Layer-by-layer recursion: Assume the velocity of the first to n-1 layers v i and thickness h i Given the given information, we now need to determine the parameters of the nth layer (target roadbed).

[0044] Set the initial trial value for the velocity of the nth layer. v n (Root mean square velocity can be taken) V rms As initial values). Using the ray tracing equation established in step S34, the known parameters of the first n-1 layers ( v 1 ... v n-1 , h 1 ... h n-1 ) and the trial speed of the nth layer v n Substitute. According to Snell's law, at a given horizontal offset distance... x j The optimal ray parameters are then searched iteratively. p Solve for the electromagnetic wave propagation path that satisfies the total offset constraint, and thus obtain the corresponding theoretical two-way travel time. .

[0045] Construct the objective function: Set the initial trial value v for the velocity of the nth layer. n,0(Vrms can be taken as the initial value), and the theoretical travel time t at different offsets can be calculated using the forward modeling of the ray tracing equation. cal Construct the objective function Characterizing the theoretical travel time and the measured CMP gather picking travel time t obs The residuals between them are shown in the following formula:

[0046] In the formula, M This refers to the number of channels (offset range) of the array radar. x j For the first j The offset between the transmitting and receiving antennas corresponding to each channel; t obs (x j ) The first one picked up from the measured CMP channel set j Two-way travel time corresponding to each offset; The first one obtained based on ray tracing forward modeling j The offset probes the speed at the nth layer. v n Theoretical two-way travel time The objective function is iteratively optimized using the Gauss-Newton method or damped least squares method. The trial speed is continuously adjusted. v n This aims to make the theoretical travel time curve fit the measured travel time point as closely as possible. When the objective function... F ( v n When the velocity is less than the preset convergence threshold, the corresponding velocity is the optimal layer velocity v for that layer. int,n .

[0047] S36. Calculate the relative permittivity: Based on the optimal layer velocity v obtained from the inversion... int,n Using the formula The precise relative permittivity of the target road base layer was calculated. .

[0048] Step S4: Establish a Topp model considering compaction correction, including establishing a Topp model considering compaction correction to invert moisture content. Based on the characteristics of subgrade engineering, a model incorporating compaction correction is constructed. K An improved Topp model with a correction factor. This invention utilizes a pre-built indoor physical model test library to prepare models covering different compaction degrees ( K ) and volumetric moisture content ( Multiple sets of standard specimens with gradients were used to determine the true relative permittivity under various working conditions using a TDR dielectric constant meter. A positive evolution model of relative permittivity driven by "moisture content-compaction degree" was established using a three-dimensional surface fitting method.

[0049]

[0050] During on-site testing, a numerical inversion algorithm was used to obtain the precise relative permittivity from step three. With known subgrade design compaction degree K Substituting into the model, the corrected subgrade volumetric moisture content is obtained by inverse kinematics. .

[0051] The specific implementation process of step S4, establishing the Topp model considering compaction correction, is as follows: S41. Select the subgrade fill soil for the road section to be tested, and determine the optimum moisture content (OMC) and maximum dry density of the soil through standard compaction tests. Air dry, crush and sieve, then dry for later use.

[0052] S42. Design four compaction gradients, with K values ​​of 96%, 93%, 90%, and 85%, respectively. Design five moisture content gradients, using the optimum moisture content OMC as the baseline, and prepare five different moisture conditions at 0.8×OMC, 0.9×OMC, 1.0×OMC, 1.1×OMC, and 1.2×OMC, respectively. A total of 4 [preparations were made]. 5 = 20 sets of standard test specimens.

[0053] S43. To accurately control the compaction degree of each specimen, the required dry soil mass and water volume must be strictly calculated. The known volume of the standard specimen mold is V = 269.39 cm³. 3 dry soil quality m d,i Based on the target compaction degree K i calculate:

[0054] Based on the target quality moisture content Calculate the required water volume (m) d,i

[0055] S44. Weigh the calculated dry soil mass. Water After thorough mixing, seal and let it sit for 24 hours to ensure even distribution of moisture among the soil particles. Use a hydraulic jack for static compaction, compacting the weighed wet soil into three layers before filling the mold. Roughen the surface after each layer is compacted.

[0056] S45. Determine the relative permittivity using a TDR dielectric constant meter. To eliminate random measurement errors caused by local soil inhomogeneity and probe contact gap, three independent insertion measurements were performed on each statically pressed specimen. Three different representative positions were selected at the center and around the specimen surface, and the probe was inserted vertically to read the three relative permittivity values. , , The arithmetic mean of the three measurements is calculated as the effective relative permittivity of the specimen under specific compaction and moisture content conditions. Immediately after the test, samples were taken from the center of the specimens, and the true mass moisture content of each specimen was determined using the drying method (constant temperature drying at 105℃). And convert it into actual volumetric moisture content. .

[0057] S46. Obtain the 20 sets of data pairs ( K , , Based on the mixed-medium theory, the relative permittivity is considered a function of compaction degree and moisture content. A modified Topp forward model is established by performing surface fitting on the data using the nonlinear least squares method, as follows:

[0058] In the formula, The effective relative permittivity; This refers to the volumetric water content. K For compaction degree, This is the compaction sensitivity coefficient. The basic dielectric constant parameter (i=0,1,2,3) is given by where and These are fixed constants obtained through indoor three-dimensional calibration experiments.

[0059] This model quantitatively characterizes the specific influence of moisture content variation on relative permittivity under different compaction degrees, serving as the core algorithm for subsequent field inversion.

[0060] Step S5, real-time moisture content inversion and intelligent early warning imaging, is implemented as follows: S51. Obtain the precise layer relative permittivity from the array radar velocity spectrum analysis in step S3. The target compaction degree specified in the construction design documents, or the compaction degree information obtained on-site using rapid non-destructive equipment such as a nuclear density meter, is substituted into the corrected model established in step S4. The continuous volumetric moisture content within the subgrade is then calculated. Because the corrected model is A cubic equation, within the scope of physical meaning ( Solve the equation using Newton's iteration method or a lookup table interpolation method with a preset step size:

[0061] This yields a high-precision volumetric moisture content after compaction correction. .

[0062] S52. To eliminate local abrupt changes in values ​​caused by underground inhomogeneities (such as boulders and metallic debris) and improve the robustness of the inversion results, median filtering or moving average processing is applied to the original water content data stream. A spatial filtering window is set (e.g., a range of 0.5m along the survey line direction), and data with deviations exceeding a certain threshold are removed. The outliers are identified to ensure the continuity and smoothness of the final output curve.

[0063] S53. Construct a two-dimensional spatiotemporal coordinate system with the mileage (station number) of the inspection vehicle as the horizontal axis and the roadbed depth as the vertical axis. Use pseudo-color mapping technology to draw a cloud map of the roadbed moisture content distribution, setting blue to represent low moisture content (dry), green to represent the optimal moisture content range (moderate), and red to represent high moisture content (overly wet).

[0064] S54. Hierarchical Intelligent Early Warning Mechanism: The system has built-in quality monitoring logic and sets a moisture content early warning threshold. (For example: optimum moisture content + 2%). When the moisture content is calculated through inversion... > When the area is identified as an "abnormal humidity zone," the algorithm automatically identifies it. A red warning box immediately pops up on the software interface, accompanied by a buzzer alarm, alerting the inspection personnel. The system automatically locks the start and end chainages of the abnormal road section (e.g., K10+200 to K10+250) and the corresponding 3D GNSS coordinates, and marks the abnormal information in the database.

[0065] S55. After the test is completed, the system supports one-click generation of a standardized quality inspection report. The report automatically calculates the average moisture content, extreme moisture content, and standard deviation of the measured line, and automatically lists the coordinates and extent of all out-of-limit anomalies, providing a digital basis for subsequent precise compaction, drying, or replacement treatment.

[0066] Specifically, the quality warning and report generation process is as follows: The software draws a color profile cloud map of the roadbed moisture content on the screen in real time, with the horizontal axis representing mileage and the vertical axis representing depth, and displays the moisture content value at the current measuring point location. When the detected moisture content exceeds a preset threshold (such as optimum moisture content + 2%), the system automatically marks the abnormal road section station number on the interface and issues an audible and visual alarm. After the inspection is completed, all raw data and inversion results are automatically stored in the database, and a standard inspection report containing inspection conclusions, abnormal point coordinates, and statistical analysis can be generated with one click.

[0067] The following is a preferred embodiment 1 of this application, which aims to illustrate the specific construction process of the "Topp model considering compaction degree correction" in this invention. Through indoor geotechnical tests, the measured relative permittivity under different combinations of compaction degree and moisture content is obtained, thereby quantifying the influence of compaction degree parameters on dielectric properties.

[0068] 1. Experimental Design: Roadbed soil from the Chongqing construction site was selected as the experimental subject. Standard compaction tests determined the optimum moisture content (OMC) of the soil sample to be 9.0%, and the maximum dry density (…). The concentration is 2.09 g / cm³. The part was manufactured using a modified hydrostatic jack device in conjunction with a standard cylindrical steel mold. The mold diameter is 70 mm, the height is 70 mm, and the mold volume V = 269.39 cm³. 3 Based on the roadbed construction specifications and actual site conditions, the design compaction degree ( K ) and moisture content ( The system employs a bivariate gradient, with the compaction gradient set at four levels: 96%, 93%, 90%, and 85%. The moisture content gradient is based on the optimum moisture content (OMC=9%), with five humidity states configured at multiples of 0.8, 0.9, 1.0, 1.1, and 1.2, corresponding to design mass moisture contents of 7.2%, 9.0%, 10.8%, 12.6%, and 14.4%, respectively.

[0069] 2. Experimental steps: (1) Soil sample preparation and batching calculation: Based on the target compaction degree K With moisture content Accurately calculate the required dry soil mass m d With water volume m w Taking a compaction degree of 96% and a design moisture content of 9.0% (i.e., optimum moisture content) as an example: target dry density According to the calculation, weigh 540.50g of dry soil and add 48.65g of water. After thoroughly mixing the soil sample, put it into a sealed bag and let it stand for 24 hours. (2) Put the soaked soil (total mass about 589.15g) into the mold in three layers. Use the modified jack device to perform layered static pressing, and roughen the layers to ensure the overall homogeneity of the specimen until it is pressed to the specified volume to obtain the standard specimen. (3) For high compaction specimens with a compaction degree of 93% and 96%, due to the extremely dense soil structure, directly forcibly inserting the TDR probe is very likely to cause the probe to bend or the soil surface to crack brittlely, which seriously affects the contact effect. This experiment adopts the process of "minimally invasive pre-hole and interference fit": wrap the prepared specimen with two layers of plastic wrap to give it vertical constraint, use a 3mm diameter drill bit with positioning bracket to pre-drill holes vertically on the surface of the specimen, and then insert a 4.5mm diameter TDR probe vertically into the pre-drilled hole. This measure effectively eliminated the interference of air layering around the probe, ensuring the accuracy of relative permittivity measurement under high-pressure actual working conditions. (4) Data acquisition and model establishment: The relative permittivity under various working conditions was collected using a dedicated TDR dielectric constant measuring instrument. Three points (center and edge) were selected for testing on each specimen, and the average value was taken. Immediately after testing, the soil sample was dried to constant weight using the drying method (105℃±2℃, dried to constant weight) according to the "Specifications for Geotechnical Testing of Highways" (JTG3430-2020). According to the formula, the actual mass moisture content Converted to volumetric moisture content .

[0070]

[0071] According to the compaction test, the maximum dry density of this soil sample is:

[0072] The actual dry density of the specimen can be calculated from the degree of compaction:

[0073] Therefore, the conversion formula for volumetric moisture content is:

[0074] Finally, based on 20 sets of measured data, the results were analyzed. The modified Topp model parameters were obtained by fitting using the nonlinear least squares method, and an inversion equation for the water content of this type of subgrade soil in Chongqing was established. The modified model obtained in this embodiment is as follows (example form):

[0075]

[0076] Where A(K) = -9.8065 + 42.8281K, B(K) = -203.2456 + -230.0967K, C(K) = 3098.1736 + -909.3295K, and D(K) = -7366.8615 + 4690.0564K. Table 1. Basic experimental data of the Topp model considering compaction degree correction.

[0077] The following is a preferred embodiment 2 of this application, which aims to quantitatively evaluate the accuracy of the method of the present invention in the detection of subgrade moisture content by comparing field measurements with standard true values, and to verify the necessity of the compaction degree correction model.

[0078] 1. Experimental Design: A section of roadbed under construction in Shuangfu New District, Chongqing (soil quality consistent with Experiment Example 1) was selected. A 200m long test section (from chainage K10+100 to K10+300) that had just been compacted was chosen. The design compaction degree requirements were as follows: K =96%, compacted layer thickness approximately 30cm. The vehicle-mounted array ground-penetrating radar system described in Example 1 was used, with an antenna center frequency of 500MHz and a ground clearance of 40cm. The actual dry density was determined using the sand cone method according to the "Specifications for Geotechnical Testing of Highways" (JTG3430-2020). And calculate the actual compaction degree K real Simultaneously, samples were taken at the corresponding measuring points to determine the actual mass moisture content using the drying method. The mass moisture content is converted into the true volume moisture content using a formula, which serves as the reference true value for model verification.

[0079]

[0080] in The density of water is usually taken as... .

[0081] The standard Topp formula (without considering compaction correction) was used as a comparison to evaluate its applicability under high compaction conditions of the subgrade. Five test points (numbered P1 to P5) were evenly selected within the test section, covering areas with different dry and wet conditions.

[0082] 2. Experimental Procedure (1) System setup and parameter settings: Park the test vehicle at the test start position and allow it to heat up for no less than 10 minutes. Calibrate the antenna height to 40cm using the servo bracket, set the sampling window to 60ns, the channel spacing to 0.05m, and the number of stacking operations to 8.

[0083] (2) Testing of the method of this invention (non-contact scanning): The test vehicle is driven at a constant speed of 20 km / h along the survey line, and the system synchronously collects multi-channel radar data. Data processing: The software automatically performs dynamic CMP reconstruction in the background, and uses an iterative inversion algorithm based on ray tracing theory to minimize the travel time residual using Snell's law, and calculates the precise layer velocity V of each measuring point. int The measured relative permittivity was then calculated. The measured dielectric constant will be used. and on-site design compaction K Substituting the ternary coupled correction model constructed in Experiment Example 1, the moisture content of the present invention is calculated. (3) Reference True Value Acquisition (Contact Verification): After the radar scan is completed, a sand cone test is immediately conducted at the measuring points P1 to P5. The mass of the excavated wet soil and the mass of the filled sand are recorded, and the wet density and dry density are calculated to obtain the actual compaction degree. Simultaneously, the soil samples were sealed and brought back to the laboratory, dried at 105°C to constant weight, and the true moisture content was determined. Subsequently, based on the maximum dry density parameter of the soil samples, it was uniformly converted into the true volumetric water content. , which serves as the reference truth value for the final comparison.

[0084] (4) Data recording and error analysis: Simultaneously, the standard Topp formula (excluding the compaction correction term) is used to... Convert to volumetric moisture content. Calculate the absolute and relative errors of the volumetric moisture content obtained by the method of this invention and the standard Topp method relative to the actual volumetric moisture content obtained by the drying method.

[0085] 3. Experimental Results and Analysis The test data from the five measuring points were summarized, and the results are shown in Table 2. Experimental data show that the average relative error of the standard Topp model in inverting volumetric moisture content is as high as 12.5%, while the average relative error of the method of this invention is only 2.8%. This proves that the present invention has extremely high reliability in detecting subgrade moisture content and can meet the requirements of refined engineering testing. Addressing the defect of the standard Topp model, which generally leads to overestimation of the inversion results due to neglecting high compaction, the "compaction degree-relative permittivity-moisture content" ternary coupled model introduced in this invention effectively corrects the system bias, making the detection results more consistent with actual working conditions. The ray tracing iterative algorithm used in this invention overcomes the refraction distortion problem of the traditional Dix formula in the detection of thin subgrade layers. By accurately simulating the electromagnetic wave path, the obtained layer velocity and permittivity are closer to the physical true values, solving the problem of low accuracy in shallow medium inversion. Compared with existing technologies, the method of this invention, while maintaining non-contact and high-efficiency acquisition, achieves a leap from "qualitative trend analysis" to "high-precision quantitative evaluation" of moisture content, with significantly better results than the standard Topp model.

[0086] Table 2 Comparison of accuracy verification results between the method of the present invention and the standard Top model.

[0087] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0088] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0089] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0090] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0091] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A method for detecting roadbed moisture content using array radar based on the Topp model, characterized in that, The method includes: Step S1, system preparation and on-site setup, including: Step S11: Install an antenna array on the detection vehicle; the antenna array includes one transmitting antenna and four receiving antennas; Step S12: Construct a multi-channel observation system with different offsets, measure and verify the array geometric parameters; input the geometric parameters into the observation system of the acquisition software; Step S2: Synchronous data acquisition. Control all channels of the array antenna to acquire data synchronously once, and write the latitude, longitude and elevation information output by GNSS into the data header file; Step S3, dynamic CMP reconstruction and signal processing, includes: Step S31: Standardize and preprocess the collected multi-channel raw data; Step S32: Perform dynamic CMP reconstruction to construct dynamic CMP gathers distributed along the survey line; Step S33: Calculate the weighted similarity coefficient for each CMP gather. scanning; Step S34: Construct a layer velocity inversion model based on ray tracing; Step S35: Perform iterative layer stripping inversion; Step S36: Calculate the relative permittivity to obtain the accurate relative permittivity of the target roadbed. ; Step S4: Establish a Topp model considering compaction correction, including establishing a Topp model considering compaction correction to invert moisture content, as shown in the following formula: ； Using a numerical inversion algorithm, the accurate relative permittivity obtained in step S3 is... Substituting the known subgrade design compaction degree K into the Topp model with the compaction degree correction, the corrected subgrade volumetric moisture content is obtained by inverse solving. ; Step S5: Real-time moisture content inversion and intelligent early warning imaging.

2. The method for detecting roadbed moisture content using array radar based on the Topp model as described in claim 1, characterized in that, Step S12, constructing a multi-channel observation system with different offset distances, includes: setting the center distance between the transmitting antenna Tx and the first receiving antenna Rx1 to x0=0.6m, the spacing between adjacent receiving antennas Δx=0.2m, the offset distance between the transmitting antenna Tx and the first receiving antenna Rx1 to x1=0.6m; the offset distance between the transmitting antenna Tx and the second receiving antenna Rx2 to x2=0.8m; the offset distance between the transmitting antenna Tx and the third receiving antenna Rx3 to x3=1.0m; and the offset distance between the transmitting antenna Tx and the fourth receiving antenna Rx4 to x4=1.2m.

3. The method for detecting roadbed moisture content using array radar based on the Topp model as described in claim 1, characterized in that, The dynamic acquisition of continuous vehicle data in step S2 includes: Step S21: Drive the testing vehicle into the starting point of the road section to be tested and keep the vehicle centered. Step S22: Drive the testing vehicle at a constant speed of 20 km / h along the roadbed survey line; Step S23: The DMI encoder outputs a TTL trigger signal at fixed intervals of dx=0.05m based on the vehicle's travel distance. In step S24, after receiving the trigger signal, the radar host controls all channels of the array antenna to synchronously collect data once, and at the same time writes the latitude, longitude and elevation information output by GNSS into the data header file to form a raw radar data stream with accurate mileage and coordinate information.

4. The method for detecting roadbed moisture content using array radar based on the Topp model as described in claim 1, characterized in that, Step S33: Calculate the weighted similarity coefficient for each CMP gather. The scanning range is set to 0.03~0.17 m / ns, as shown in the following formula: ； In the formula, Zero offset time t 0 and probing speed v The following is the similarity coefficient; M The number of channels in the CMP channel set; A i For the first i The amplitude value of the Dao; W The width of the time window; j A sliding index for sampling points within a time window; The sampling time interval for radar data; t i (v) For the hyperbolic trajectory time; Step S34: Construct a layer velocity inversion model based on ray tracing, as shown in the following equation: ； In the formula, Let be the incident angle of the i-th layer; v i Let be the layer velocity of the i-th medium layer, and there are a total of n medium layers; p is the ray parameter; Obtain the total horizontal offset of electromagnetic waves x and round trip t As shown in the following formula: ； ； In the formula, h i For the first i Layer thickness; Step S35, perform iterative layer stripping inversion, including: Construct the objective function The theoretical travel time and the measured CMP gather picking travel time t were obtained. obs The residuals between: ； In the formula, M This represents the number of channels in the array radar. x j For the first j The offset between the transmitting and receiving antennas corresponding to each channel; t obs (x j ) The first one picked up from the measured CMP channel set j Two-way travel time corresponding to each offset; The first one is obtained based on ray tracing forward modeling. j The offset probes the speed at the nth layer. v n Theoretical two-way travel time is as follows; Iterative optimization of the objective function: when the objective function If the velocity is less than the preset convergence threshold, then this layer is the optimal layer, and the corresponding velocity is the optimal layer velocity v. int,n ; Step S36, Calculation of relative permittivity Based on the optimal layer velocity v int,n Using the formula Obtain the precise relative permittivity of the target road base layer. .

5. The method for detecting roadbed moisture content using array radar based on the Topp model as described in claim 1, characterized in that, Step S4 establishes the Topp model considering compaction correction, including the following steps: Step S41: Select the subgrade fill soil of the road section to be tested, and obtain the optimum moisture content (OMC) and maximum dry density of the fill soil. ; Step S42: Design four different compaction gradient specimens; and design five different moisture content gradient specimens. Step S43, calculate the required dry soil mass and water addition: Dry soil mass m d,i Based on the target compaction degree K i The calculation is shown in the following formula: ； V is the known volume of the standard specimen mold; Based on the target quality moisture content Calculate the required amount of water to add m w,i As shown in the following formula: ； Step S44: Weigh the dry soil. Water After mixing, the mixture is sealed and pressed into shape under static pressure, then compacted in layers and placed into a mold. Step S45: Determine the relative permittivity. The arithmetic mean of multiple measurements is taken as the effective relative permittivity of the specimen under the set compaction degree and moisture content conditions. ; Drying to obtain the true mass and moisture content of the specimen And convert it to the actual volumetric moisture content. ; Step S46, organize the obtained multiple sets of data pairs ( K , , The data is fitted to a three-dimensional surface using the nonlinear least squares method to establish a modified Topp forward model that includes moisture content and compaction degree, as shown in the following equation: ； In the formula, The effective relative permittivity; This refers to the volumetric water content. K For compaction degree, This is the compaction sensitivity coefficient. Based on the fundamental dielectric constant parameter (i=0,1,2,3), the... and These are fixed constants obtained through indoor three-dimensional calibration experiments.

6. The method for detecting roadbed moisture content using array radar based on the Topp model as described in claim 1, characterized in that, Step S5, real-time water level inversion and intelligent early warning imaging, includes: Step S51: Determine the precise relative permittivity of the target roadbed layer obtained in step S3. Combining the target compaction degree, or the compaction degree information obtained on-site, and substituting it into the corrected model established in step S4, the continuous volumetric moisture content inside the subgrade is calculated. As shown in the following formula: ; Obtain high-precision volumetric moisture content after compaction correction. ( ); Step S52: Perform median filtering or moving average processing on the original moisture content data stream; Step S53: Construct a two-dimensional spatiotemporal coordinate system with the vehicle mileage as the horizontal axis and the roadbed depth as the vertical axis; use pseudo-color mapping technology to draw a cloud map of roadbed moisture content distribution, with different colors representing different moisture contents; Step S54, Hierarchical intelligent early warning mechanism: Set moisture content early warning threshold When the moisture content is calculated by inversion > When this occurs, the area is automatically identified as an area with abnormal humidity. Step S55: After the test is completed, a quality inspection report is generated.

Images