Rock-soil body internal defect identification method based on wave field inversion

Abstract

The invention discloses a rock-soil body internal defect identification method based on wave field inversion, and the method comprises the following steps: S1, carrying out the wave field numerical simulation of a target region through a coarse grid discretization model, and generating distorted wave field data containing a numerical frequency dispersion effect; s2, frequency dispersion characteristic fragments related to defects are extracted from the distorted wave field data; s3, matching the frequency dispersion characteristic fragments with a pre-stored or dynamically generated defect characteristic template library, and determining a defect candidate area; s4, when matching fails, decomposing a scattering unit based on distortion wave field residual data, generating a new defect template through parameterized folding recombination, and updating the new defect template to a template library; s5, replacing the coarse grid numerical solution in the candidate region with an analytic wave field solution or a mixed solution of a newly generated template, and inverting defect parameters; through core technologies of coarse grid efficient calculation, dynamic template adaptive generation, physical constraint strong verification and the like, leap-forward progress is realized in the aspects of defect identification efficiency, complex form adaptability, environment robustness and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of geotechnical engineering safety monitoring, and specifically relates to a method for identifying internal defects of geotechnical bodies based on wave field inversion. Background Art

[0002] Accurately identifying internal defects in rock and soil (such as cracks, holes, and loose areas) is a core issue in underground engineering safety assessment and geological disaster prevention. Current mainstream technologies rely primarily on elastic wavefield inversion methods, inferring defect parameters by matching numerical simulations with measured wavefield data. However, these methods still suffer from the following significant drawbacks: 1. It is difficult to strike a balance between computational efficiency and accuracy: To ensure inversion accuracy, existing methods (such as the finite element method and the spectral element method) require the use of ultra-fine grids (grid size Δx ≤ λ_min / 20, where λ_min is the minimum wavelength), which results in an exponential increase in the computational complexity of the three-dimensional model. For example, for a 10 m×10 m×10 m area, when the highest frequency is 10 kHz, the number of grid nodes exceeds 10. 9 , a single simulation takes several hours, which cannot meet the real-time requirements of the project. If a coarse grid is used to simplify the calculation, the numerical dispersion effect will seriously distort the wave field characteristics, resulting in a defect location error of more than 20%.

[0003] 2. Poor adaptability to complex defect morphology: Traditional template matching methods require a pre-existing database of wavefield responses for regular defects (such as circular holes and linear cracks), making them ineffective in identifying complex defects such as bifurcated cracks and multi-hole complexes. Studies have shown that the misjudgment rate for irregular defects can be as high as 40%, and the reconstructed size error often exceeds 25%. Existing parametric inversion methods (such as elliptical approximation and polynomial expansion) make overly simplistic geometric assumptions and therefore struggle to characterize the topological characteristics of real defects.

[0004] 3. Insufficient robustness in noisy environments: In actual engineering, interference such as loose sensor coupling and environmental vibration can cause the signal-to-noise ratio (SNR) to drop below 0 dB. Existing methods rely on fixed-threshold filtering and rigid template matching, causing defect detection rates to plummet below 50% in such scenarios. Furthermore, wavefield splicing of numerical simulations and measured data often involves direct superposition, which causes energy jumps (>20%) at the boundaries, further reducing inversion reliability.

[0005] 4. High hardware cost and maintenance complexity: High-precision inversion requires supercomputing clusters or GPU servers, with hardware costs exceeding millions of yuan. Furthermore, the predefined template library requires regular manual updates to cover new defect types, resulting in high long-term maintenance costs. Summary of the Invention

[0006] The purpose of the present invention is to develop a method for identifying internal defects in rock and soil based on wave field inversion. This method achieves leapfrog progress in defect identification efficiency, adaptability to complex forms, and environmental robustness through core technologies such as efficient coarse grid calculation, adaptive generation of dynamic templates, and strong verification of physical constraints. It also significantly reduces hardware and operation and maintenance costs, providing cost-effective solutions for underground engineering, geological exploration, and other fields, and has significant industrial application value.

[0007] The technical solution adopted by the present invention is as follows: a method for identifying internal defects of rock and soil based on wave field inversion, comprising the following steps: S1. Use a coarse-grid discretization model to perform wavefield numerical simulation of the target area and generate distorted wavefield data including numerical dispersion effects; S2. extracting the dispersion characteristic fragments associated with the defect from the distorted wavefield data; S3. Matching the dispersion feature fragments with pre-stored or dynamically generated defect feature templates to determine candidate defect areas; S4. When matching fails, decompose the scattering element based on the distorted wavefield residual data, generate a new defect template through parameterized folding and recombination, and update it to the template library; S5. Replace the coarse grid numerical solution with the analytical wavefield solution or the hybrid solution of the newly generated template in the candidate area and invert the defect parameters.

[0008] Wherein, the step S4 includes: a. Separate the direct wave, background scattered wave and unknown residual wave from the distorted wavefield residual; b. Extract the basic scattering units in the unknown residual wave, including inflection points, inflection lines and inflection surfaces; c. Perform mathematical origami-style reorganization of basic units according to the principle of minimum energy to generate fractal or surface defect morphologies; d. Verify the physical constraints of the reorganization defect, including energy conservation error ≤ 10% and travel time continuity error ≤ 1%.

[0009] The parameterized folding and reorganization in step c includes: Generate radial fractal lines with the inflection point as the center, with branch angles of 30°-150° and lengths decaying exponentially; Non-uniform rational B-splines (NURBS) are used to fit the discrete folded surfaces, and the curvature radius is adaptively adjusted.

[0010] The initial construction of the template library in step S3 includes: Generate benchmark wavefield data of regular defects through ultra-fine grid simulation; Generate the corresponding distorted wave field under the coarse grid and extract the time-frequency energy steep drop ratio and phase jump characteristics; Encode features as hash-indexed time-frequency matrix templates.

[0011] The hybrid solution replacement in step S5 includes: A transition zone is defined at the boundary of the defect candidate area, and the analytical solution and the numerical solution are spliced using a cosine weighting function; The width of the transition zone is 0.5-1 times the current wavelength, and the wave field energy difference after splicing is ≤5%.

[0012] A rock and soil defect detection system for implementing the method comprises: a sensor array for exciting and receiving wavefield signals; Edge computing module, which performs coarse-grid wavefield simulation, feature fragment extraction, and template matching; Dynamic template generator, including residual wave decomposition unit, basic unit extraction unit and parameterized folding and recombination unit; Hybrid wavefield reconstructor for analytical solution replacement and inversion calculation; Visual terminal displays defect parameters and confidence levels.

[0013] The dynamic template generator integrates a physical constraint verification unit to perform rapid verification of energy conservation, travel time continuity and self-consistency.

[0014] Wherein, the sensor array includes a positioning calibration module, including: The ultra-wideband (UWB) positioning tag at the source provides real-time feedback of the source coordinates; The micro strain gauge at the sensor end monitors the coupling contact status and compensates the signal transfer function.

[0015] The parameterized folding and recombination unit supports multi-level fractal generation, with a maximum number of iterations of 5 levels, and adjustable branch angles and length attenuation rates.

[0016] A computer-readable storage medium stores a computer program, which implements the steps of the method when executed by a processor.

[0017] The beneficial effects of the present invention include: 1. To ensure inversion accuracy, existing technologies generally require a grid size less than 1 / 20 of the minimum wavelength (Δx ≤ λ_min / 20). This results in an exponential increase in the computational complexity of 3D models, with single simulations often taking over several hours and failing to meet real-time engineering requirements. This invention innovatively incorporates coarse-grid distorted wavefield characteristics analysis and a dynamic template compensation mechanism, relaxing the grid size to λ_min / 6 and reducing the computational complexity by over 50 times. For example, at a maximum frequency of 10 kHz and v_min = 500 m / s, increasing the grid size from 0.5 mm to 10 mm reduces the single wavefield simulation time from 5 hours to 6 minutes. Furthermore, dynamic template generation (fractal / NURBS recombination) compensates for the numerical dispersion error caused by the coarse grid, maintaining defect location accuracy at ≤ 5 cm (comparable to ultra-fine grid methods), achieving both efficiency and accuracy improvements.

[0018] 2. Adaptive recognition capability of complex defect morphology Traditional methods rely on a library of pre-existing regular defect templates (e.g., circular holes and linear cracks). Their recognition rate for irregular, multi-scale composite defects (e.g., bifurcated cracks and honeycomb holes) is less than 60%, and their reconstruction error exceeds 20%. This paper proposes a residual-driven dynamic template generation technology, which achieves breakthrough improvements through three stages: Residual wave field decomposition: Direct wave window function interception, background scattering spatial filtering and 3σ residual extraction are used to accurately separate unknown defect scattering signals; Parametric reorganization of scattering units: Generates a multi-level fractal structure with the inflection point as the center (branch angle 30°–150°, length iterative attenuation rate 0.5–0.8), and uses NURBS surface fitting to achieve complex curvature adaptation, which can model arbitrary topological defects; Strong physical constraint verification: Enforces verification of energy conservation (error ≤ 10%) and travel time continuity (error ≤ 1%) to ensure that the generated template complies with the laws of wave propagation.

[0019] Actual measurements show that this method has increased the recognition rate of bifurcated cracks and multi-hole complexes to 95%, with a modeling error of ≤5%, breaking through the limitations of traditional geometric assumptions.

[0020] 3. Improved robustness in strong noise environments Existing methods experience a sharp performance degradation in low signal-to-noise ratio (SNR < 10 dB) environments, primarily due to fixed threshold filtering and sensor coupling interference. This invention achieves breakthroughs in anti-interference capabilities through three technological innovations: Dynamic compensation of coupling state: A micro strain gauge is integrated at the sensor end to monitor the contact state changes between the soil and the sensor in real time. When the strain fluctuation exceeds 10%, the signal transfer function is automatically corrected, reducing the signal distortion caused by poor contact by 90%; Wavelet-spatial joint denoising: Symlet wavelet basis is first used for multi-scale threshold denoising, and then 3×3×3 grid spatial filtering is performed on the residual wave field, which improves the environmental noise suppression capability by 40%; Enhanced extraction of dispersion features: Convert numerical dispersion distortion into quantitative indicators of time-frequency energy drop ratio (high frequency / low frequency ≥ 5) and phase jump, enhancing the noise robustness of defect features.

[0021] Experiments show that the system can still maintain a defect detection rate of more than 85% in a strong noise scenario with SNR=-10 dB, which is a significant improvement over the existing technology (detection rate of 80% when SNR≥0 dB).

[0022] 4. Engineering practicality and cost-effectiveness innovation Reduced hardware costs: Coarse grid computing enables edge devices to replace traditional supercomputing clusters, reducing hardware investment by 60%; Maintenance cost optimization: The dynamic template library is driven by residual self-updates, eliminating the need for manual intervention in template library expansion, reducing long-term operation and maintenance costs by 70%; Multi-scenario expansion capability: The modular design supports rapid access to multi-modal sensors such as infrared thermal imaging and acoustic emission. It has been successfully applied to scenarios such as tunnel lining crack detection (accuracy ±2 cm) and archaeological site cavity detection (resolution 0.1 m³). BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0025] See also Figure 1 A method for identifying internal defects of rock and soil based on wave field inversion includes the following steps: S1. Use a coarse-grid discretization model to perform wavefield numerical simulation of the target area and generate distorted wavefield data including numerical dispersion effects; S2. extracting the dispersion characteristic fragments associated with the defect from the distorted wavefield data; S3. Matching the dispersion feature fragments with pre-stored or dynamically generated defect feature templates to determine candidate defect areas; S4. When matching fails, decompose the scattering element based on the distorted wavefield residual data, generate a new defect template through parameterized folding and recombination, and update it to the template library; S5. Replace the coarse grid numerical solution with the analytical wavefield solution or the hybrid solution of the newly generated template in the candidate area and invert the defect parameters.

[0026] Furthermore, the step S4 includes: a. Separate the direct wave, background scattered wave and unknown residual wave from the distorted wavefield residual; b. Extract the basic scattering units in the unknown residual wave, including inflection points, inflection lines and inflection surfaces; c. Perform mathematical origami-style reorganization of basic units according to the principle of minimum energy to generate fractal or surface defect morphologies; d. Verify the physical constraints of the reorganization defect, including energy conservation error ≤ 10% and travel time continuity error ≤ 1%.

[0027] Furthermore, the parameterized folding and reorganization in step c includes: Generate radial fractal lines with the inflection point as the center, with branch angles of 30°-150° and lengths decaying exponentially; Non-uniform rational B-splines (NURBS) are used to fit the discrete folded surfaces, and the curvature radius is adaptively adjusted.

[0028] Furthermore, the initial construction of the template library in step S3 includes: Generate benchmark wavefield data of regular defects through ultra-fine grid simulation; Generate the corresponding distorted wave field under the coarse grid and extract the time-frequency energy steep drop ratio and phase jump characteristics; Encode features as hash-indexed time-frequency matrix templates.

[0029] Furthermore, the hybrid solution replacement in step S5 includes: A transition zone is defined at the boundary of the defect candidate area, and the analytical solution and the numerical solution are spliced using a cosine weighting function; The width of the transition zone is 0.5-1 times the current wavelength, and the wave field energy difference after splicing is ≤5%.

[0030] A rock and soil defect detection system for implementing the method comprises: a sensor array for exciting and receiving wavefield signals; Edge computing module, which performs coarse-grid wavefield simulation, feature fragment extraction, and template matching; Dynamic template generator, including residual wave decomposition unit, basic unit extraction unit and parameterized folding and recombination unit; Hybrid wavefield reconstructor for analytical solution replacement and inversion calculation; Visual terminal displays defect parameters and confidence levels.

[0031] Furthermore, the dynamic template generator integrates a physical constraint verification unit to perform rapid verification of energy conservation, travel time continuity and self-consistency.

[0032] Furthermore, the sensor array includes a positioning calibration module, including: The ultra-wideband (UWB) positioning tag at the source provides real-time feedback of the source coordinates; The micro strain gauge at the sensor end monitors the coupling contact status and compensates the signal transfer function.

[0033] Furthermore, the parameterized folding and recombination unit supports multi-level fractal generation, with a maximum number of iterations of 5 levels, and adjustable branch angles and length decay rates.

[0034] A computer-readable storage medium stores a computer program, which implements the steps of the method when executed by a processor.

[0035] The specific application process is as follows: Step S1: Coarse grid wave field simulation S11 model construction: The rock and soil area to be inspected (such as underground structures or tunnel linings) is discretized into a coarse grid model. The grid size is determined according to the minimum wave velocity of the rock and soil (v_min) and the maximum frequency of the source signal (f_max), satisfying the following: spatial step size Δx ≤ v_min / (6×f_max).

[0036] For example: If v_min = 500 m / s (loose soil layer) and f_max = 10 kHz, then Δx ≤ 500 / (6×10,000) = 0.0083 m, and Δx = 10 mm.

[0037] The time step Δt is determined according to the CFL stability condition and satisfies: Δt ≤ 0.8×Δx / (v_max×√3), where v_max is the maximum wave velocity of the rock mass (e.g., v_max = 4000 m / s for dense rock). Then, Δt ≤ 0.8×0.01 / (4000×1.732)≈1.15 μs, and Δt = 1 μs is taken.

[0038] S12 wavefield solution: The explicit finite difference method (FDTD) of the velocity-stress elastic wave equation is used for numerical solution. An absorbing boundary condition (PML layer) with a thickness of three times the grid size (i.e., 30 mm) is set to simulate an infinite domain. The source signal is a Ricker wavelet with a dominant frequency of 5 kHz, which is applied to the model surface or the preset drill hole location.

[0039] S13 data output: records the waveform data of velocity components (v_x, v_y, v_z) and stress components (σ_xx, σ_yy, σ_zz, etc.) of each receiving point as they change with time.

[0040] Step S2: Extraction of dispersion feature fragments S21 signal preprocessing: Perform wavelet denoising on the received signal (using the Symlet wavelet basis with 5 decomposition layers) to remove high-frequency noise; S22 normalization processing: scale the signal amplitude to the [-1,1] range.

[0041] S23 Time-Frequency Analysis (S Transform): Perform an S transform on the preprocessed signal to obtain a time-frequency energy matrix with a frequency resolution of 100 Hz and an adaptive time window length. S24 extracts the time-frequency energy steep drop ratio: calculates the energy ratio of the high frequency band (8–10 kHz) to the low frequency band (0–2 kHz), and marks it as a candidate feature when the ratio is ≥5; S25 detects phase jump: If the phase derivative exceeds π / (2Δt) (the threshold is 1.57×10 6 rad / s), which is determined as the phase mutation point.

[0042] S26 hash coding: quantizes the time-frequency matrix into 10 energy levels (0–9) and generates a 64-bit binary hash index. For example, if the energy of a time-frequency point is level 7, the corresponding binary segment is "0111", and the entire matrix is spliced into a 64-bit hash value.

[0043] Step S3: Template matching and candidate region positioning S31 benchmark template generation: Use an ultra-fine grid (Δx = 0.5 mm) to simulate the wavefield response of regular defects (such as a spherical hole with a radius of 0.2 m and a crack with a length of 1 m). Generate a distorted wavefield using the same coarse grid (Δx = 10 mm), extract the time-frequency features, and encode them into a hash template.

[0044] S32 template storage: Hash table structure storage, the key is a 64-bit hash value, and the value is the defect type, size, and coordinates.

[0045] S33 matching process: Calculate the Hamming distance (number of different bits) between the hash value of the feature to be tested and the template library; if the Hamming distance is ≤3, it is determined to be a successful match, and the corresponding area is marked as a defect candidate area (such as the coordinate (x, y, z) ±0.1 m range).

[0046] Step S4: Dynamic template generation and update S41: Residual wave decomposition Direct wave separation: Extract the signal within the theoretical direct wave time window (±0.1ms) based on the geometric relationship between the source location and the receiving point; Background scatter removal: Perform spatial average filtering (window size 3×3×3 grid) on the remaining signal to remove background noise; Unknown residual extraction: Signals with amplitudes exceeding three times the standard deviation are retained as unknown scattered components.

[0047] S42: Scattering cell extraction Breakpoint detection: marking the amplitude mutation point in the residual signal (change rate ≥ 10³ Pa / μs); Polyline tracing: An improved ant colony algorithm is used to connect adjacent polylines, with the path curvature constraint of κ < 0.1 rad / m; Surface generation: Perform Delaunay triangulation on closed polylines to form a three-dimensional defect surface.

[0048] S43: Parametric Restructuring Fractal generation: Generates radial fractal lines with the inflection point as the center, with a first-level branch length of 0.1 m and an angle of 30°–150°; the length of the second-level branches is gradually reduced at a decay rate of 0.5–0.8, with a maximum number of iterations of 5 levels; NURBS surface fitting: Non-uniform rational B-spline fitting is performed on discrete folded surfaces, with the curvature radius adaptively adjusted according to the current wavelength λ (R ≥ λ / 4); the node vectors are uniformly distributed, and the weight coefficients are determined by the minimum energy principle.

[0049] S44: Physical Verification Energy conservation: The difference between input energy (source) and output energy (scattering + absorption) is ≤ 10%; Travel time continuity: The wave field arrival time error at the defect edge is ≤1%.

[0050] Step S5: Hybrid inversion and parameter optimization S51 transition zone stitching: A transition zone is set at the boundary of the defect candidate area, with a width of 0.5–1 times the current wavelength λ (λ=vp / f_dominant, where vp is the longitudinal wave velocity). The analytical solution and the numerical solution are merged using a cosine weighting function (weight w=cos²(πd / (2W)), where d is the distance from the boundary). After stitching, the wavefield energy jump is ≤5%.

[0051] S52 inversion optimization: Objective function: Minimize the difference between the hybrid wave field and the measured data, and constrain the smoothness of the inversion parameters; Optimization variables: defect location (x, y, z), size (radius r or length L); Algorithm: Levenberg-Marquardt iterative algorithm, maximum number of iterations 100 times, residual convergence threshold 1e-4.

[0052] Further, system implementation details include: Sensor array: 12 piezoelectric sensors (frequency response 0.1 Hz–100 kHz) are arranged in an inner layer (0.5 m spacing), a middle layer (1 m spacing), and an outer layer (2 m spacing); Positioning and calibration module: The source end integrates a UWB tag (Decawave DW1000 chip, accuracy ±1 cm), and the sensor end is attached with a micro strain gauge to monitor the coupling status.

[0053] Edge computing module: Hardware configuration: Intel Core i7-1185G7 processor, NVIDIA RTX A2000 graphics card, 32GB memory; Real-time processing: wave field simulation (Devito library), feature extraction (FFTW acceleration), template matching (Hamming distance calculation).

[0054] Dynamic template generator: Residual decomposition unit: GPU-accelerated Wiener filtering algorithm; Fractal generation unit: Based on L-system syntax rules, supports parameter configuration interface; Physical verification unit: Parallel calculation of energy conservation and travel time error.

[0055] Visualization terminal: Outputs defect 3D model (STL format), confidence heat map and inversion parameter table; supports touch interaction to adjust viewing angle and profile analysis.

[0056] Example 1: Concrete pile foundation hole detection The pile foundation size is 1 m × 10 m, the hole radius is 0.15 m, and the center position is (2 m, 3 m, 5 m); Results: Positioning error ±0.02 m, dimension inversion error 3%, calculation time 120 ms.

[0057] Example 2: Identification of Hidden Rock Cracks Parameters: crack length 2 m, width 0.05 m, inclination angle 45°; Results: Tracking error <5°, confidence level 92%, system power consumption <50 W.

[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for identifying internal defects of rock and soil based on wave field inversion, characterized in that: The following steps are involved: S1. Use a coarse-grid discretization model to perform wavefield numerical simulation of the target area and generate distorted wavefield data including numerical dispersion effects; S2. extracting the dispersion characteristic fragments associated with the defect from the distorted wavefield data; S3. Matching the dispersion feature fragments with pre-stored or dynamically generated defect feature templates to determine candidate defect areas; S4. When matching fails, decompose the scattering element based on the distorted wavefield residual data, generate a new defect template through parameterized folding and recombination, and update it to the template library; S5. Replace the coarse grid numerical solution with the analytical wavefield solution or the hybrid solution of the newly generated template in the candidate area and invert the defect parameters.

2. The method according to claim 1, characterized in that The step S4 comprises: a. Separate the direct wave, background scattered wave and unknown residual wave from the distorted wavefield residual; b. Extract the basic scattering units in the unknown residual wave, including inflection points, inflection lines and inflection surfaces; c. Perform mathematical origami-style reorganization of basic units according to the principle of minimum energy to generate fractal or surface defect morphologies; d. Verify the physical constraints of the reorganization defect, including energy conservation error ≤ 10% and travel time continuity error ≤ 1%.

3. The method according to claim 2, characterized in that The parameterized folding and reorganization in step c includes: Generate radial fractal lines with the inflection point as the center, with branch angles of 30°-150° and lengths decaying exponentially; Non-uniform rational B-splines (NURBS) are used to fit the discrete folded surfaces, and the curvature radius is adaptively adjusted.

4. The method according to claim 1, wherein The initial construction of the template library in step S3 includes: Generate benchmark wavefield data of regular defects through ultra-fine grid simulation; Generate the corresponding distorted wave field under the coarse grid and extract the time-frequency energy steep drop ratio and phase jump characteristics; Encode features as hash-indexed time-frequency matrix templates.

5. The method according to claim 1, wherein The hybrid solution replacement in step S5 includes: A transition zone is defined at the boundary of the defect candidate area, and the analytical solution and the numerical solution are spliced using a cosine weighting function; The width of the transition zone is 0.5-1 times the current wavelength, and the wave field energy difference after splicing is ≤5%.

6. A rock and soil defect detection system for implementing the method according to any one of claims 1 to 5, characterized in that: include: a sensor array for exciting and receiving wavefield signals; Edge computing module, which performs coarse-grid wavefield simulation, feature fragment extraction, and template matching; Dynamic template generator, including residual wave decomposition unit, basic unit extraction unit and parameterized folding and recombination unit; Hybrid wavefield reconstructor for analytical solution replacement and inversion calculation; Visual terminal displays defect parameters and confidence levels.

7. The system according to claim 6, characterized in that The dynamic template generator integrates a physical constraint verification unit to perform rapid verification of energy conservation, travel time continuity and self-consistency.

8. The system according to claim 6, wherein: The sensor array includes a positioning calibration module, including: The ultra-wideband (UWB) positioning tag at the source provides real-time feedback of the source coordinates; The micro strain gauge at the sensor end monitors the coupling contact status and compensates the signal transfer function.

9. The system according to claim 6, wherein: The parameterized folding and recombination unit supports multi-level fractal generation, with a maximum number of iterations of 5 levels, and adjustable branch angles and length decay rates.

10. A computer-readable storage medium storing a computer program, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images