Vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography imaging method and system

Abstract

The application provides a kind of vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography imaging method and system, it is related to electromagnetic imaging field, method includes: sailing acquisition and pose synchronization;Time series sampling data is mapped to unified geodetic coordinate according to pose and is rastered and merged, and equivalent space aperture is constructed;Noise suppression and shutdown correction are carried out to observation data, and late channel signal is extracted;The actual receiving and transmitting position and attitude of each measuring point are substituted into forward operator, and the sensitivity matrix for compensating motion geometric distortion is constructed;Three-dimensional resistivity grid is established, and the objective function is constructed by data fitting term plus spatial smoothing and depth weighting regularization term, and three-dimensional resistivity distribution is obtained by inversion;Abnormal body is extracted and bound to road mileage stake and depth, and three-dimensional imaging map and hidden danger list are generated.The application upgrades point measurement to continuous sailing, does not need to seal road, and operation efficiency is improved by about one order of magnitude, and the output is organized according to stake and depth, which can be directly included in highway maintenance account book.

Description

Technical Field

[0001] This application relates to the field of electromagnetic imaging, and in particular to a vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method and system. Background Technology

[0002] Transient Electromagnetic Method (TEM), a time-domain electromagnetic exploration method, excites a primary field in the subsurface medium by passing a pulsed current through a transmitting coil, and observes the secondary field generated by induced eddy currents in the subsurface geological body during the current-off period. Because it is unaffected by the primary field and sensitive to low-resistivity bodies, it is widely used in engineering exploration, groundwater resource exploration, and other fields. However, traditional transient electromagnetic detection mainly relies on single-point, fixed-point measurements, resulting in low operational efficiency (approximately 1 km / 8 h), often requiring road closures, and producing mostly one-dimensional sounding curves or two-dimensional profiles, making it difficult to provide a three-dimensional spatial morphology of potential hazards. If the TEM transceiver is placed on a vehicle for continuous mobile data acquisition, problems arise such as the vehicle's movement causing changes in the transceiver geometry over time, sparse and irregular distribution of measurement points along the survey line, and geometric errors introduced by attitude fluctuations. Direct three-dimensional inversion then exhibits severe ill-formed results, with distorted or even divergent imaging.

[0003] In existing technologies, although there have been attempts to introduce the concept of synthetic aperture into transient electromagnetic detection, these have mainly focused on aerial or ground-based fixed-point observation modes and have not yet solved the problem of motion distortion compensation and three-dimensional joint inversion under vehicle-mounted mobile conditions. Although efficient algorithms based on finite element and finite difference have been developed for three-dimensional inversion methods, they are mostly designed for static observation data and are not suitable for motion geometry time-varying scenarios of continuous mobile acquisition.

[0004] Therefore, existing technologies lack an imaging method that organizes continuous vehicle-mounted mobile sampling into an equivalent spatial aperture and performs motion-compensated three-dimensional joint inversion of all mobile data to meet the needs of rapid survey of hidden defects in highway subgrades under non-road closure conditions. Summary of the Invention

[0005] The purpose of this invention is to address the problem that existing transient electromagnetic detection technologies cannot achieve three-dimensional tomographic imaging for motion distortion compensation of hidden roadbed defects under continuous vehicle-mounted navigation conditions without road closures, and to provide a vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomographic imaging method and system.

[0006] The above-mentioned objective of this application is achieved through the following technical solution: S1: Mobile data acquisition and attitude synchronization: In the continuous driving state of the vehicle, transient electromagnetic transmission is triggered at fixed time intervals to synchronously acquire the secondary field response data of the multi-channel receiving coil, and the position and attitude information of the vehicle at each transmission moment is recorded by the positioning and attitude unit. S2: Dynamic Aperture Construction: The secondary field response data of each transmission cycle is mapped to a unified geodetic coordinate system based on the position and attitude information. Combined with the installation offset of the transmitting coil and receiving coil relative to the vehicle body, the geodetic coordinates and normal attitude of the center of the transmitting coil and each receiving coil in each transmission cycle are restored. The grid is then merged along the survey line at a preset interval. The data of multiple transmission cycles merged into the same grid are superimposed to form the observation data set of the measuring point. The transmit and receive geometric set of all grid measurement points within a frame segment constitutes the equivalent spatial sampling aperture of that frame segment; S3: Data preprocessing: Noise suppression, current normalization and shutdown correction are performed on the observation data set. Late channel signals are extracted at logarithmic equal time intervals to form the attenuation vector of each measurement point and convert it into an apparent resistivity curve. S4: Sensitivity kernel construction: Establish a three-dimensional inversion mesh. For each measurement point, substitute its actual transmit / receive position and attitude into the forward modeling operator to calculate the sensitivity of the measurement point to each mesh cell and construct a sensitivity matrix that compensates for motion geometric distortion. S5: Three-dimensional regularized joint inversion: The initial three-dimensional resistivity model is generated by interpolating the apparent resistivity curves of each measurement point. An objective function containing data fitting terms and model constraint terms is constructed. Combined with the sensitivity matrix, the Gauss-Newton iteration method is used to solve the problem and obtain the three-dimensional resistivity distribution. S6: Anomaly Extraction and Station Number Binding Output: Based on the three-dimensional resistivity distribution, the background resistivity is statistically analyzed, and anomalies below or above the set threshold of the background resistivity are extracted. The spatial attributes of the anomalies are calculated and bound to the road mileage station number and depth, generating a three-dimensional imaging map and a hazard list.

[0007] Optionally, step S1 includes: The positioning and attitude determination unit includes RTK / GNSS, wheel speed odometer and IMU attitude unit, and outputs the vehicle's position and attitude angle at each launch time through combined navigation filtering; The acquisition trigger and transmission shutdown are hardware aligned, and each transmission cycle is time-stamped by a unified clock.

[0008] Optionally, step S2 includes: The preset spacing for grid merging is no more than 0.5m. Data from multiple transmission cycles within the same grid are aligned by flipping positive and negative polarities and then weighted and superimposed according to noise variance.

[0009] Optionally, step S4 includes: The three-dimensional inversion mesh is a three-dimensional hexahedral mesh along the station, horizontal and depth directions, with a horizontal size of 0.5 to 2 m and a logarithmic densification in the depth direction. The shallow mesh size is set to 0.5 m and the deep mesh size is set to 5 m. The forward modeling operator adopts a three-dimensional sensitivity calculation method combining a one-dimensional hierarchical analytical kernel with the Born approximation, or adopts a time-domain finite-difference forward modeling method.

[0010] Optionally, step S5 includes: The objective function is:

[0011] in, This is a three-dimensional resistivity model vector. This is the vector of all underway observation data. Forward response vector Weight the data into a diagonal matrix. For the gradient term in the model space, For depth-weighted terms, and The regularization coefficient is used. The Gauss-Newton iterative update formula is as follows:

[0012] in, For the first The model vector for the next iteration; Indicates transpose; The sensitivity matrix is ​​represented as follows:

[0013] in, For the first Forward response at each measurement point For the first The conductivity or resistivity model parameters of the nth grid cell, the partial derivative of which is expressed as the nth... Calculation of the actual transmit and receive positions and attitudes of each measuring point.

[0014] Optionally, step S6 includes: The anomaly extraction includes: those below 40% to 60% of the background median are identified as low-resistivity candidate anomalies, and those above 60% to 100% of the background median are identified as high-resistivity candidate anomalies; these are then merged into anomalies through three-dimensional connected component analysis, and isolated anomalies smaller than a preset volume are removed. The list of potential hazards includes the center three-dimensional coordinates, outer envelope size, volume, average and extreme resistivity, relative deviation from the background, and corresponding station number and lateral offset of the anomaly.

[0015] A vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography system, the system comprising: The towing vehicle's speed is linked to the transmission repetition frequency to ensure the spatial density of the measurement points. The rear-mounted platform is made of non-metallic materials and is kept at a set distance from the rear of the vehicle. A multi-turn rectangular transmitting coil and a multi-channel receiving coil array are installed on the platform. A high-power transmitter that outputs bipolar square wave current, with adjustable turn-off time and a real-time monitoring channel for the transmission current. Multi-channel synchronous acquisition unit, multi-channel synchronous acquisition, unified clock and trigger, hardware alignment of sampling and transmission cycles; The positioning and attitude determination unit, including RTK / GNSS, wheel speed odometer and IMU attitude unit, is used to output the vehicle's position and attitude information; The vehicle-mounted industrial control computer and imaging software are used to complete data acquisition and monitoring, data preprocessing, dynamic aperture construction, sensitivity kernel calculation, three-dimensional regularization inversion, and image output.

[0016] The transmitting coil is a rectangular multi-turn coil with a side length of 1-2m; the receiving coil array is an 8-channel receiving coil; the transmitting shutdown time is 0.1-10µs; and the multi-channel synchronous acquisition unit is an 8-channel 24-bit synchronous acquisition unit.

[0017] Optionally, the system also includes a map inversion module, which is used to divide the survey line into overlapping sub-aperture maps according to the mileage station number and perform independent inversion. The overlap between adjacent maps is not less than 20%, and the overlapping area is subject to model consistency constraints and weighted fusion.

[0018] The combined navigation filtering of the positioning and attitude determination unit is a loose combination or a tight combination extended Kalman filter; when RTK loses lock, positioning is maintained by wheel speed odometer and IMU dead reckoning, and overall adjustment is performed after recovery.

[0019] The beneficial effects of the technical solution provided in this application are: Dynamic aperture construction: The time-series sampling data of continuous vehicle-mounted navigation is mapped to a unified geodetic coordinate system based on RTK / IMU pose and organized into an equivalent spatial aperture, breaking through the traditional single-point fixed-point observation mode.

[0020] Motion-compensated sensitivity kernel: The actual transmit and receive positions and attitudes of each measurement point are substituted into the forward modeling sensitivity kernel to construct a point-by-point model, which explicitly compensates for the geometric distortion caused by vehicle motion and eliminates the imaging distortion caused by traditional static data processing methods.

[0021] Multi-line 3D regularized joint inversion: A unified 3D resistivity grid is established for all mobile data, and the data fitting term plus spatial smoothing and depth-weighted regularization term are used for joint inversion to upgrade the 2D profile to a 3D CT volume.

[0022] Station number sliding slab incremental imaging: The slab is inverted by overlapping station numbers, and the map is generated while the data is being collected. The results are organized and output according to "station number + depth". Attached Figure Description

[0023] The present application will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a block diagram of the vehicle-mounted dynamic synthetic aperture electromagnetic CT system in the embodiments of this application; Figure 2 This is a schematic diagram of the equivalent spatial aperture formed by continuous navigation in the embodiments of this application; Figure 3 This is a step diagram of an embodiment of this application; Figure 4 This is a schematic diagram of the construction of a sensitivity kernel with motion compensation in an embodiment of this application; Figure 5 This is an example diagram of the three-dimensional resistivity inversion and anomaly delineation results in the embodiments of this application; Figure 6 This is a schematic diagram of abnormal body station number binding and segmented output in the embodiments of this application; Figure 7 This is a schematic diagram of the output results in the embodiments of this application; Figure 8 This is the first three-dimensional tomographic image in the embodiments of this application; Figure 9 This is the second three-dimensional tomographic image in the embodiments of this application. Detailed Implementation

[0024] To provide a clearer understanding of the technical features, objectives, and effects of this application, the specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0025] The embodiments of this application provide a vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method.

[0026] Please refer to Figure 3 , Figure 3 This is a flowchart illustrating the steps of a vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method according to an embodiment of this application, including: Mobile data acquisition is synchronized with attitude measurement. (a) Survey line planning: Parallel survey lines are laid out along the driving lane / shoulder (example survey line spacing 1.5-3.5m, 3 survey lines covering a single road surface); (b) Operating parameters: vehicle speed example 5-30km / h, linked with the transmission repetition frequency (example 25-75Hz bipolar), ensuring that the spacing between survey points after merging is no more than 0.5m; (c) Time stamping mechanism: each transmission cycle is time-stamped by a unified clock, and attitude data is recorded synchronously at 10-100Hz. The acquisition trigger and transmission shutdown hardware are aligned; (d) On-site quality control: late-stage amplitude, superposition number and RTK status are displayed in real time. When the signal-to-noise ratio or positioning quality exceeds the limit, an alarm prompts to reduce speed or retest.

[0027] Dynamic aperture construction. (a) Pose calculation: Perform combined navigation filtering (extended Kalman filter) on RTK, IMU and wheel speed odometer to output the vehicle's position and attitude angle at each transmission cycle; (b) Geometric reconstruction: Superimpose the installation offset of the transmitting / receiving coils relative to the vehicle body to obtain the geodetic coordinates and normal attitude of the transmitting coil and the center of each receiving coil in each cycle; (c) Grid merging: Establish a grid of measuring points along the survey line at intervals of 0.5m (example), merge the data of each cycle according to the nearest grid, and multiple cycles of the same grid constitute the superimposed set of the measuring point; (d) Aperture definition: The transmit and receive geometric set of all grid measuring points within a frame window constitutes the equivalent spatial sampling aperture of that frame (see Figure 2 ).

[0028] Data preprocessing. (a) Multi-period superposition: Periodic data within the same grid are first aligned by flipping positive and negative polarities, and then superimposed by noise variance to suppress random noise; (b) Interference suppression: Notch filtering is applied to power frequency and its harmonics, and wavelet threshold denoising is applied to low-frequency disturbances introduced by motion; (c) Current normalization and turn-off correction: The echo is normalized using the measured transmit current waveform, and early channel distortion is corrected using the equivalent turn-off time method; (d) Late channel extraction: Channels are extracted at logarithmic equal time intervals (30-64 gates in the example) to form the attenuation vector of each measurement point and convert it into an apparent resistivity curve; (e) Bad channel rejection: Channels with signal-to-noise ratio below the threshold or extremely abnormal amplitude are rejected and recorded, and measurement points with a rejection rate exceeding 30% (example) are marked for retesting.

[0029] Sensitivity kernel construction. (a) Mesh generation: Establish a three-dimensional hexahedral mesh along station number × horizontal × depth, with horizontal dimensions ranging from 0.5 to 2 m in the example, and logarithmically refined in the depth direction (approximately 0.5 m for shallow parts and approximately 5 m for deep parts); (b) Forward modeling operator: Use a one-dimensional hierarchical analytical kernel combined with the Born approximation for three-dimensional sensitivity calculation. For deep or complex models, finite-difference time-domain forward modeling can be selected; (c) Motion compensation: For each measurement point, substitute the actual transmit / receive position and attitude obtained from S2 into the Jacobian matrix (Equation 2) to ensure that the inversion operator is strictly consistent with the motion sampling geometry; (d) Matrix assembly: Assemble the sparse sensitivity matrix according to the sizing window and cache it for reuse in iterative inversion.

[0030] Three-dimensional regularized joint inversion. (a) Initial model: A pseudo-three-dimensional initial resistivity volume is generated by interpolation of the apparent resistivity curves at each measurement point; (b) Objective function: Constructed according to Equation 1, including data fitting term, spatial smoothing term, and depth weighting term; (c) Iterative solution: The model is updated according to Equation 3 using the Gauss-Newton method, and the inner layer uses conjugate gradients to solve the linear equations. A line search is performed in each iteration to determine the step size; the convergence criterion is that the data fitting difference is less than a set threshold or the maximum number of iterations is reached (example: 10-20 times); (d) Regularization parameters: λ1 and λ2 are determined by the L-curve method or empirical values ​​(example: order of magnitude 10). - ²~10 - ¹, data calibration is required; (e) Segment overlap: the overlap between adjacent segments is not less than 20%, and model consistency constraints are applied to the overlap area and weighted fusion is performed to ensure continuous and seamless long-mileage imaging.

[0031] (Equation 1)

[0032] in, This is a three-dimensional resistivity model vector. This is the vector of all underway observation data. Forward response vector Weight the data into a diagonal matrix. For the gradient term in the model space, For depth-weighted terms, and The regularization coefficient is used. In Equation 1, the first term requires the forward modeling response to fit the observed data, the second term suppresses spatial abrupt changes in the model (smoothing), and the third term weights the deep units to balance the shallow-depth resolution; λ1 and λ2 are regularization coefficients.

[0033] (Equation 2)

[0034] in, For the first Forward response at each measurement point For the first The conductivity or resistivity model parameters of the nth grid cell, the partial derivative of which is expressed as the nth... Calculation of the actual transmit and receive positions and attitudes of each measuring point.

[0035] Equation 2 represents the sensitivity of the i-th observation to the resistivity of the k-th grid cell; since the transmit and receive geometry of each measuring point takes the true value in motion, the geometric changes caused by motion are explicitly brought into the inversion operator.

[0036] (Equation 3)

[0037] in, For the first The model vector for the next iteration; Indicates transpose; Equation 3 is the Gaussian-Newton iterative update formula: driven by the fitting residuals of the current model, the model increment is solved and updated under regular constraints until convergence. In the formula, m is the three-dimensional resistivity model vector, d is all the underway observation data, F(·) is the forward modeling operator, and W_d is the data weighting matrix.

[0038] Anomaly extraction and station-binding output: (a) Anomaly extraction: Statistically analyze the background resistivity distribution of the inverted volume. Volumes below a certain percentage (e.g., 40%–60%) are identified as low-resistivity (water-rich) candidates, and volumes above a certain percentage (e.g., 60%–100%) are identified as high-resistivity (void / loose) candidates. These are then merged into anomalies through 3D connected component analysis, and isolated volumes smaller than the minimum volume (e.g., 1 m³) are removed. (b) Attribute calculation: Output the center 3D coordinates, outer envelope size, volume, average and extreme resistivity, and their deviation from the background for each anomaly. (c) Station binding: Calculate the planar coordinates back to the road mileage station and lateral offset, with the depth taken as the depth below the road surface. (d) Output results (see...) Figure 7 The following are sample hazard lists: ① Three-dimensional resistivity volume and anomaly isosurface map; ② Two-dimensional apparent resistivity profile along the survey line; ③ Anomaly plane projection map; ④ Hazard list linked to station number. An example of the hazard list format is as follows:

[0039] The list can also be expanded to record fields such as confidence probability P, uncertainty σ (when used with intelligent interpretation methods), verification method and disposal status, which can be directly imported into the maintenance management ledger and re-inspected and compared according to the same station number after disposal.

[0040] Key parameters (examples, adjustable according to engineering calibration). Typical detection depth 50–80m; A / D sampling bit depth 24bit; shutdown time 0.1–10µs; transmission repetition frequency 25–75Hz; vehicle speed 5–30km / h; survey line spacing 1.5–3.5m; survey point merging spacing ≤0.5m; time gates 30–64 (logarithmically equal intervals); grid horizontal size 0.5–2m, depth logarithmic refinement; swathe length 50–200m, overlap ≥20%; inversion iterations 10–20 times.

[0041] Anomaly Handling. For RTK lost-lock sections, positioning is maintained using wheel speed odometer and IMU dead reckoning, and overall adjustment is performed after recovery; sections with excessive speed or sudden changes are automatically downweighted or removed; sections with strong interference (high-voltage lines, dense metal guardrails) are marked as low-confidence areas and prompted for retesting; when the bad road rate of a certain section exceeds the limit or the inversion fails to converge, the entire section is marked for retesting, without affecting the mapping of adjacent sections.

[0042] As one example, a field measurement on a closed road is taken. The inspection vehicle continuously travels along three parallel survey lines to collect data (vehicle speed and repetition frequency are set as in the example above), completes pose synchronization and 0.5m grid merging, superimposes and denoises, and extracts late-stage roads. A three-dimensional regularized joint inversion is performed using a sensitivity kernel with motion compensation to obtain a two-dimensional apparent resistivity profile and a three-dimensional resistivity volume. The delineated low-resistivity, water-rich anomaly area is consistent with the high-density electrical resistivity cross-validation results in terms of location and morphology. The anomaly body is used to generate a three-dimensional imaging map and a hazard list (format as in the table above) according to the station number and delivered to the owner, which includes the disposal recommendations.

[0043] Existing TEM equipment performs point-based / fixed-point observations, or simply inverts mobile data point-by-point into one-dimensional / two-dimensional results without constructing spatial aperture, performing motion compensation, or performing multi-line three-dimensional joint inversion. This invention achieves three-dimensional tomographic imaging under unclosed mobile conditions using "dynamic aperture scaling + motion compensation sensitivity kernel + three-dimensional regularized joint inversion + station-based segmented output," which differs from existing technologies and is not readily apparent.

[0044] The beneficial effects of this application are as follows: First, the point measurement has been upgraded to continuous mobile surveying, which does not require road closures. The survey efficiency is improved by about an order of magnitude compared to traditional electrical resistivity point measurement methods (for example, from about 1km / 8h to about 1km / 0.5h).

[0045] Secondly, the two-dimensional profile has been upgraded to a three-dimensional CT volume, which can directly provide the three-dimensional location and morphology of cavities, loose structures, and water-rich abnormalities, supporting precise treatment.

[0046] Third, the results are organized into imaging maps and hazard lists according to "station number + depth", which can be directly written into the highway maintenance ledger and support re-inspection and comparison, making the project highly usable.

[0047] Please refer to Figure 1 , Figure 1 This is a system composition diagram of a vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography system according to an embodiment of this application, including: Towing vehicle: Light-duty van-type testing vehicle, with driving speed and emission repetition frequency linked for control to ensure spatial density of measuring points; Rear-mounted platform: Non-metallic (fiberglass, etc.) towing platform, maintaining a certain distance from the rear of the vehicle (2-3m in example) to avoid eddy current interference from the vehicle body; a multi-turn rectangular transmitting coil (1-2m in example side length) and an 8-channel receiving coil array are installed on the platform; High-power transmitter: outputs bipolar square wave current, peak current in the tens of amperes (example), turn-off time adjustable from 0.1 to 10µs, and has a real-time monitoring channel for transmit current; Multi-channel synchronous acquisition unit: 8-channel 24-bit synchronous acquisition, unified clock and trigger, hardware alignment of sampling and transmission cycles; Positioning and attitude determination unit: RTK / GNSS (centimeter level), wheel speed odometer and IMU attitude unit, output frequency 10~100Hz (example); Vehicle-mounted industrial control computer and 3D imaging software: complete data acquisition and monitoring, real-time preprocessing, sectional inversion and map output.

[0048] The data flow is as follows: master control triggers → transmits pulse and quickly shuts down → 8 channels synchronously acquire secondary fields → uniform time stamp is applied in each cycle and stored together with pose → preprocessing and inversion according to slatted windows → output imaging map and hazard list.

[0049] The core inventive points of this invention are as follows: 1. Dynamic Aperture Optimization for Mobile Sampling – Traditional TEM relies on single-point fixed-point observations, with spatial coverage achieved through repeated station relocation. This invention utilizes continuous vehicle movement to map time-series samples acquired over time to a unified geodetic coordinate system based on RTK / wheel speed odometer and IMU pose data. This transforms the continuous mobile operation of a small transceiver system into an equivalent large-aperture array deployed along the survey line (see...). Figure 2 The point measurement has been upgraded to continuous mobile measurement, and the spatial sampling density is controlled by vehicle speed and pulse repetition rate.

[0050] 2. Sensitivity Kernel for Motion Distortion Compensation—Vehicle motion causes changes in the relative geometry and attitude of the transmitter and receiver within each pulse cycle. This invention substitutes the actual transmit and receive positions and attitudes of each measuring point into the forward modeling sensitivity kernel to construct it point by point (see...). Figure 4 The geometric changes introduced by motion are explicitly compensated by the inversion operator, eliminating the imaging distortion caused by "treating motion data as static data".

[0051] 3. Multi-line Joint 3D Regularized Inversion—A unified 3D resistivity grid is established for all mobile data from multiple survey lines on the same road segment. A joint inversion is performed using data fitting terms, spatial smoothing, and depth-weighted regularization terms, upgrading the 2D profile to a 3D CT volume. The 3D location and morphology of anomalies can be directly output (see...). Figure 5 ).

[0052] 4. Incremental Imaging with Sliding Maps Along Station Numbers—The survey line is divided into overlapping sub-aperture map inversions based on station numbers, with mapping performed simultaneously. This approach balances the real-time performance of long-mileage surveys with the quality of 3D imaging. Results are output organized by "station number + depth" (see...). Figure 6 , Figure 7 ), can be directly included in the maintenance ledger.

[0053] In one embodiment, such as Figure 8 and Figure 9 As shown, the shallow layer of the survey area exhibits a relatively continuous electrical distribution characteristic, and the anomalous responses of each survey line correspond and connect well spatially, indicating that the data collected jointly by multiple survey lines have good spatial consistency. From the perspective of the overall electrical structure, the upper part of the survey area is dominated by relatively high resistivity characteristics, with obvious low-resistivity anomaly areas in some areas, and there is a certain degree of continuity between different survey lines.

[0054] Within the survey area, two distinct low-resistivity anomalies are visible along survey line 1, appearing as blue or cyan areas in the 3D imaging results. This type of low-resistivity response is typically associated with underground aquifers or increased local soil moisture content. Considering the experimental area's location within an urban expressway environment, this anomaly may be related to underground drainage pipes, stormwater and sewage networks, or other drainage facilities. These underground pipelines and their surrounding backfill soil often have high moisture content and high electrical conductivity, thus appearing as low-resistivity anomalies in apparent resistivity imaging. Furthermore, the localized low-resistivity anomalies exhibit a certain tendency to extend in the depth direction, indicating the possible presence of aquifer channels or locally loose backfill structures distributed along the pipeline direction. Long-term leakage from underground drainage facilities or relatively loose surrounding soil can create locally water-rich areas, which is also a possible cause of the low-resistivity anomaly. Considering the distribution of underground municipal facilities in the road area, this anomaly characteristic has a certain degree of engineering rationale.

[0055] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure.

[0056] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method, characterized in that, The method includes the following steps: S1: Mobile data acquisition and attitude synchronization: In the continuous driving state of the vehicle, transient electromagnetic transmission is triggered at fixed time intervals to synchronously acquire the secondary field response data of the multi-channel receiving coil, and the position and attitude information of the vehicle at each transmission moment is recorded by the positioning and attitude unit. S2: Dynamic Aperture Construction: The secondary field response data of each transmission cycle is mapped to a unified geodetic coordinate system based on the position and attitude information. Combined with the installation offset of the transmitting coil and receiving coil relative to the vehicle body, the geodetic coordinates and normal attitude of the center of the transmitting coil and each receiving coil in each transmission cycle are restored. The grid is then merged along the survey line at a preset interval. The data of multiple transmission cycles merged into the same grid are superimposed to form the observation data set of the measuring point. The transmit and receive geometric set of all grid measurement points within a frame segment constitutes the equivalent spatial sampling aperture of that frame segment; S3: Data preprocessing: Noise suppression, current normalization and shutdown correction are performed on the observation data set. Late channel signals are extracted at logarithmic equal time intervals to form the attenuation vector of each measurement point and convert it into an apparent resistivity curve. S4: Sensitivity kernel construction: Establish a three-dimensional inversion mesh. For each measurement point, substitute its actual transmit / receive position and attitude into the forward modeling operator to calculate the sensitivity of the measurement point to each mesh cell and construct a sensitivity matrix that compensates for motion geometric distortion. S5: Three-dimensional regularized joint inversion: The initial three-dimensional resistivity model is generated by interpolating the apparent resistivity curves of each measurement point. An objective function containing data fitting terms and model constraint terms is constructed. Combined with the sensitivity matrix, the Gauss-Newton iteration method is used to solve the problem and obtain the three-dimensional resistivity distribution. S6: Anomaly Extraction and Station Number Binding Output: Based on the three-dimensional resistivity distribution, the background resistivity is statistically analyzed, and anomalies below or above the set threshold of the background resistivity are extracted. The spatial attributes of the anomalies are calculated and bound to the road mileage station number and depth, generating a three-dimensional imaging map and a hazard list.

2. The vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method as described in claim 1, characterized in that, Step S1 includes: The positioning and attitude determination unit includes RTK / GNSS, wheel speed odometer and IMU attitude unit, and outputs the vehicle's position and attitude angle at each launch time through combined navigation filtering; The acquisition trigger and transmission shutdown are hardware aligned, and each transmission cycle is time-stamped by a unified clock.

3. The vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method as described in claim 1, characterized in that, Step S2 includes: The preset spacing for grid merging is no more than 0.5m. Data from multiple transmission cycles within the same grid are aligned by flipping positive and negative polarities and then weighted and superimposed according to noise variance.

4. The vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method as described in claim 1, characterized in that, Step S4 includes: The three-dimensional inversion mesh is a three-dimensional hexahedral mesh along the station, horizontal and depth directions, with a horizontal size of 0.5 to 2 m and a logarithmic densification in the depth direction. The shallow mesh size is set to 0.5 m and the deep mesh size is set to 5 m. The forward modeling operator adopts a three-dimensional sensitivity calculation method combining a one-dimensional hierarchical analytical kernel with the Born approximation, or adopts a time-domain finite-difference forward modeling method.

5. The vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method as described in claim 1, characterized in that, Step S5 includes: The objective function is: in, This is a three-dimensional resistivity model vector. This is the vector of all underway observation data. Forward response vector Weight the data into a diagonal matrix. For the gradient term in the model space, For depth-weighted terms, and The regularization coefficient is used. The Gauss-Newton iterative update formula is as follows: in, For the first The model vector for the next iteration; Indicates transpose; The sensitivity matrix is ​​represented as follows: in, For the first Forward response at each measurement point For the first The conductivity or resistivity model parameters of the nth grid cell, the partial derivative of which is expressed as the nth... Calculation of the actual transmit and receive positions and attitudes of each measuring point.

6. The vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method as described in claim 1, characterized in that, Step S6 includes: The anomaly extraction includes: those below 40% to 60% of the background median are identified as low-resistivity candidate anomalies, and those above 60% to 100% of the background median are identified as high-resistivity candidate anomalies; these are then merged into anomalies through three-dimensional connected component analysis, and isolated anomalies smaller than a preset volume are removed. The list of potential hazards includes the center three-dimensional coordinates, outer envelope size, volume, average and extreme resistivity, relative deviation from the background, and corresponding station number and lateral offset of the anomaly.

7. A vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography system, used to implement the vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography method as described in any one of claims 1-6, characterized in that, The system includes: The towing vehicle's speed is linked to the transmission repetition frequency to ensure the spatial density of the measurement points. The rear-mounted platform is made of non-metallic materials and is kept at a set distance from the rear of the vehicle. A multi-turn rectangular transmitting coil and a multi-channel receiving coil array are installed on the platform. A high-power transmitter that outputs bipolar square wave current, with adjustable turn-off time and a real-time monitoring channel for the transmission current. Multi-channel synchronous acquisition unit, multi-channel synchronous acquisition, unified clock and trigger, hardware alignment of sampling and transmission cycles; The positioning and attitude determination unit, including RTK / GNSS, wheel speed odometer and IMU attitude unit, is used to output the vehicle's position and attitude information; The vehicle-mounted industrial control computer and imaging software are used to complete data acquisition and monitoring, data preprocessing, dynamic aperture construction, sensitivity kernel calculation, three-dimensional regularization inversion and image output; The transmitting coil is a rectangular multi-turn coil with a side length of 1-2m; the receiving coil array is an 8-channel receiving coil; the transmitting shutdown time is 0.1-10µs; and the multi-channel synchronous acquisition unit is an 8-channel 24-bit synchronous acquisition unit.

8. The vehicle-mounted dynamic synthetic aperture transient electromagnetic three-dimensional tomography system as described in claim 7, characterized in that, The system also includes a map inversion module, which is used to divide the survey line into overlapping sub-aperture maps according to the mileage station number and perform independent inversion. The overlap between adjacent maps is not less than 20%, and the overlapping area is subject to model consistency constraints and weighted fusion. The combined navigation filtering of the positioning and attitude determination unit is a loose combination or a tight combination extended Kalman filter; when RTK loses lock, positioning is maintained by wheel speed odometer and IMU dead reckoning, and overall adjustment is performed after recovery.

Images