Dam termite nest forward modeling method based on double random media, media and equipment

Abstract

The invention discloses a double random medium-based dam termite nest forward modeling method, medium and equipment, and relates to a dam termite nest geophysical exploration technology, the double random medium-based dam termite nest forward modeling method mainly comprises the following steps: measuring a dielectric constant of a local dam and a fungus garden gap ratio, creating a soil and termite main nest initial H5 matrix, and calculating a termite nest initial H5 matrix; the method comprises the following steps: establishing a random medium matrix of soil and a fungus garden, constructing a fungus garden gap by using an adjacent point fusion method, and finally simulating the generated dam termite nest model. According to the dam termite nest forward modeling method based on the double random media, the media and the equipment, accurate simulation of termite nest ground penetrating radar waveforms under different dam dielectric parameters can be achieved.

Description

Technical Field

[0001] The present invention relates to the geophysical detection technology of dam ant nests, and more specifically, to a forward modeling method, medium and device for termite nests in dams based on double random media. Background Art

[0002] As a risk-causing species of dam-damaging organisms in China, termite nests and channels can induce secondary risks such as seepage, piping, and collapse during the sudden rise of flood levels during the flood season. In severe cases, it may even lead to the instability and collapse of the dam body. The termite nest system significantly weakens the anti-seepage performance of the project by destroying the continuity of the dam body structure (see Jin Zhengya. Research progress on the interaction between termites, clay and ecological environment [J]. Scientia Silvae Sinicae, 2023.). Under the action of high water levels, it can induce a chain of risks from seepage to leakage to piping. Therefore, termite disasters have become the main biological diseases of reservoirs and dams in China (see Li Dong. On the differences and treatments between termite piping (leakage) and hydraulic piping [J]. Acta Entomologica Sinica, 2004.). Their wide distribution characteristics seriously restrict the safe operation of water conservancy projects.

[0003] Ground Penetrating Radar (GPR) is based on the differences in the electromagnetic characteristics of media. It radiates high-frequency short-pulse electromagnetic waves into the underground medium through a transmitting antenna, and uses a receiving antenna to capture the interface reflection signal, and inversely calculates the spatial coordinates and burial depth of the target object through time-frequency analysis. Facing the need for intelligent detection of termite nests in dams, a large number of waveforms of termite nests in dams need to be collected to train the deep learning network model. The sample data volume determines the accuracy of the deep learning model. Enriching the samples can improve the feature representation and generalization ability and accurately identify complex nest structures. In actual detection, the bottleneck in the collection efficiency of GPR termite nest data samples mainly lies in the excavation link: experienced nest-digging masters need to dig along the termite galleries to confirm the positions of the main nest and its secondary nests (burial depth 1.2 - 2.5 m). At the same time, affected by the Yangtze River flood season from May to September every year, dam excavation is prohibited, resulting in a long sample acquisition cycle.

[0004] The dielectric characteristics of dam media vary significantly in different provinces and cities across the country. This patent can establish a termite nest model that conforms to the local media characteristics, generate an adaptable waveform library, and further support the research and development of intelligent detection algorithms with local dam media characteristics. Forward modeling generates high-fidelity waveforms based on the physical laws of electromagnetic wave propagation, and has advantages such as clear principles and no need for training data compared with the GAN method. Based on this, this study proposes a double random medium forward algorithm to improve the simulation accuracy of complex nest structures through parametric modeling and achieve efficient dataset expansion.

[0005] The main dam medium materials such as clay, sandy soil and loam have significant heterogeneity and spatial variability. As a composite heterogeneous body, the fungus garden, mud skeleton and air gap of termite nests show non-uniform dielectric property distributions, and the random dielectric parameters of the fungus garden and mud skeleton are the key to modeling. The present invention models the dam medium and the fungus garden of termite nests using the random medium theory. At present, many scholars at home and abroad have carried out a large number of studies on the random medium theory and its numerical calculations and achieved certain results. Ikelle et al. proved that two-dimensional random media with ellipsoidal autocorrelation functions can simulate different small-scale non-uniformity distributions on the earth (see Ikelle L T, Yung S K, Daube F. 2-D random media with ellipsoidal autocorrelation functions[J]. Geophysics, 1993, 58(9): 1359-1372.). Yao Yao and Xi Xian et al. proposed an algorithm framework for constructing random media, demonstrated the characteristics of exponential and Gaussian correlation functions, and on this basis introduced a roughness factor to propose a mixed-type elliptical autocorrelation function (see Xi Xian. Simulation of random medium models and mixed-type random media[J]. Earth Science, 2002.). Chen Keyang et al. (see Chen Keyang. A construction method of random medium models[J]. Geophysical Prospecting and Geochemical Exploration, 2010.) further improved the algorithm implementation process, introduced two-dimensional random fields and random power spectra, enabling random medium models to more flexibly construct various geographical substances. Early research on the random medium theory mainly focused on modeling oil and gas fields, karst caves and volcanic rocks, etc., to facilitate the generation of seismic wave simulation data sets. Later, Jiang et al. applied the random medium theory and its construction algorithm to soil modeling to analyze the target response performance of ground-penetrating radar in complex background media, but there is still a lack of modeling research on constructing complex soil buried objects (see Jiang Z. Simulation and analysis of GPR signal based on stochastic media model with an ellipsoidal autocorrelation function[J]. Journal of Applied Geophysics, 2013.).

[0006] The above content is only used to assist in understanding the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0007] The object of the present invention is to provide a forward modeling method, medium and device for termite nests in dams based on double random media, which can accurately simulate the ground-penetrating radar waveforms of termite nests under different dam dielectric parameters.

[0008] The present invention provides a forward modeling method for termite nests in dikes based on double random media, comprising the following steps: S1: Measuring the dielectric constant of the local dike and the proportion of the gap of the fungus garden; S2: Creating an initial H5 matrix of the soil and the main termite nest according to the pixel map of the termite nest-dike main body medium system; S3: Creating a random medium matrix of the soil and the fungus garden according to the dielectric constant and the initial H5 matrix of the soil and the main termite nest; S4: Constructing the gap of the fungus garden by using the adjacent point fusion method according to the proportion of the gap of the fungus garden and the random medium matrix of the soil and the fungus garden to form the final dielectric constant spatial distribution map; S5: Forward modeling the final dielectric constant spatial distribution map by using the finite-difference time-domain method to generate the waveform of the ground-penetrating radar dike main body medium and the termite nest.

[0009] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above forward modeling method for termite nests in dikes based on double random media are implemented.

[0010] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, the steps of the above forward modeling method for termite nests in dikes based on double random media are implemented.

[0011] The present invention also provides a computer program product, comprising a computer program, and when the computer program is executed by a processor, the steps of the above forward modeling method for termite nests in dikes based on double random media are implemented.

[0012] Implementing the forward modeling method for termite nests in dikes based on double random media, the medium and the device provided by the present invention has the following beneficial effects: Based on the classical random medium soil model, the present invention establishes a joint modeling method of double random media for the dike main body medium and the termite nest; based on the random medium theory, a double random medium algorithm is proposed for complex modeling of the soil and its termite nest; Compared with the traditional ground-penetrating radar modeling, which is relatively ideal, cannot reflect the real soil echo and stacking, and cannot simulate the characteristics and states of complex termite nests, the present invention innovatively proposes to form an H5 file with pixel points as the dielectric constant space identifier, and then map out the complex space dielectric constant distribution state map in the form of a mask; Compared with the traditional modeling scheme, it is not only relatively ideal, but also cannot reflect the electrical parameters of the dike main body media across the country, cannot selectively and specifically generate the ground-penetrating radar waveforms of termite nests in dikes that conform to the local dike electrical parameters, and is not conducive to constructing a termite nest waveform data set close to the real situation locally; The present invention in-situ obtains the key parameters of the dam medium (dielectric constant, fungus garden porosity, electromagnetic wave travel time at the target depth), constructs a pixel topological mapping of the ant nest-soil system based on the measured data, calibrates the anisotropic ellipse parameters of the random medium and the gap space distribution, relies on the GPRMAX open-source platform to realize the electromagnetic response characteristic simulation, generates a regional-specific radar waveform database, and constructs a region-adapted dam termite nest simulation model; The present invention reconstructs the fungus garden gap structure of the ant nest through the adjacent point fusion algorithm, and realizes the accurate simulation of the ground penetrating radar waveform of the termite nest under different dam dielectric parameters; this method has the potential for extended application in the electromagnetic forward modeling of complex underground targets. Brief Description of the Drawings

[0013] The present invention will be further described below in conjunction with the drawings and embodiments. In the drawings: Figure 1 is the flow chart of the forward method for dam termite nests based on double random media provided by the present invention; Figure 2 is the main termite nest structure diagram provided by the present invention; Figure 3 is the schematic diagram of measuring the dielectric constant of the main dam medium provided by the present invention. Among them, Figure 3 in (a) is the schematic diagram of the sounding experiment, Figure 3 in (b) is the sounding radar waveform diagram; Figure 4 is the schematic diagram of the termite fungus garden provided by the present invention; Figure 5 is the schematic diagram of the implementation process of the pixel ratio method provided by the present invention; Figure 6 is the pixel map of the termite nest and soil provided by the present invention; Figure 7 is the schematic diagram of some initial H5 matrix values provided by the present invention; Figure 8 is the flow chart of constructing the random medium provided by the present invention; Figure 9 is the distribution diagram of the soil dielectric constant identifier provided by the present invention; Figure 10 is the distribution diagram of the soil dielectric constant identifier provided by the present invention; Figure 11 is the schematic diagram of the distribution state of the dielectric constant identifier of the termite main nest simulation model without fungus garden gaps provided by the present invention; Figure 12 is the schematic diagram of the distribution state of the dielectric constant identifier of the termite main nest simulation model with fungus garden gaps provided by the present invention; Figure 13 is the schematic diagram of some dielectric constants provided by the present invention; Figure 14 It is the simulated waveform diagram of termite nests in the Yangtze River levee in a certain place provided by the present invention; Figure 15 It is the measured waveform diagram of the main termite nest provided by the present invention; Figure 16 It is the structural block diagram of the computer device provided by the present invention. Detailed implementation manners

[0014] For a clearer understanding of the technical features, objectives, and effects of the present invention, the detailed implementation manners of the present invention will now be described in detail with reference to the accompanying drawings.

[0015] Figure 1 It shows a schematic diagram of the forward modeling method of termite nests in levees based on double random media in this embodiment. In this embodiment, the forward modeling method of termite nests in levees based on double random media includes the following steps: S1: Measure the dielectric constant and the proportion of mycelium bed gaps of the local levee; In an exemplary embodiment, step S1 specifically includes: S11: Conduct a sounding experiment on the local levee to obtain a detection image, and based on the detection image, obtain the dielectric constant of the local levee; In an exemplary embodiment, obtaining the dielectric constant of the local levee based on the detection image, such as the formula: , , where T is the travel time, D is the depth of the underground target, C is the propagation speed of electromagnetic waves in the medium, is the speed of light in a vacuum, is the relative dielectric constant of the medium; S12: Based on the mycelium bed samples of the local levee, use the pixel ratio method to obtain the proportion of mycelium bed gaps; In an exemplary embodiment, step S12 specifically includes: Based on the mycelium bed samples, obtain the number of pixels in the mycelium bed area; Based on the number of pixels in the mycelium bed area and a preset gray threshold, obtain the number of pixels in the mycelium bed gaps; Based on the number of pixels in the mycelium bed area and the number of pixels in the mycelium bed gaps, obtain the gap ratio; Average the gap ratios of multiple mycelium bed samples to obtain the proportion of mycelium bed gaps; S2: Create an initial H5 matrix of soil and the main termite nest according to the pixel map of the termite nest - levee main body medium system; S3: Create a random medium matrix of soil and mycelium bed according to the dielectric constant and the initial H5 matrix of soil and the main termite nest; In an exemplary embodiment, step S3 specifically includes: creating a random medium matrix of soil and fungus combs according to the dielectric constant, the initial H5 matrix of soil and the main termite nest, as shown in the formula: , , , , , , = , , wherein, is the ellipsoidal autocorrelation function, , represent the spatial coordinates in two orthogonal directions describing the characteristics of the medium, a and b are the autocorrelation lengths, r is the roughness factor, is the power spectral density function, is the transverse frequency component in the spatial domain, is the longitudinal frequency component in the spatial domain, is the smoothed random power spectrum, is the power spectral density function, and are respectively the mean and variance of the random perturbation field, is the relative perturbation, is the preset perturbation standard deviation, is the actual standard deviation of generating the random field f(x,z), is the spatial distribution map of the dielectric constant of the random medium, represents large-scale inhomogeneity, is the random perturbation ratio of the dielectric constant of the medium relative to the background value, represents small-scale inhomogeneity; S4: According to the gap ratio of the fungus combs and the random medium matrix of soil and fungus combs, use the adjacent point fusion method to construct the gaps of the fungus combs; In an exemplary embodiment, the using the adjacent point fusion method to construct the gaps of the fungus combs includes: S41: Select the fungus comb part in the model and determine the diameter and ratio density of the gaps; It should be noted that the ratio density is the ratio of the gaps of the fungus combs; S42: Determine the local maximum points according to the random medium matrix of soil and fungus combs; In an exemplary embodiment, step S42 specifically includes: determining local maximum points according to the random medium matrix of the soil and the fungus garden, and constructing the gaps of the fungus garden by using the adjacent point fusion method. This method effectively simulates the actual distribution characteristics of the pores of the termite fungus garden by generating a large number of irregular random regions, as shown in the formula: , where is the random medium matrix of the continuous soil and the fungus garden, and are matrix regions of size in the fungus garden or soil model, and are local maximum points, is the radius of the gap; S43: According to the local maximum points and the diameter of the gap, gradually fuse adjacent points at a preset discrete space step until the gap density in the termite fungus garden area reaches the proportion of the fungus garden gap, to obtain the fungus garden gap; As an exemplary embodiment, the preset discrete space step is 0.005 m; In an exemplary embodiment, step S43 specifically includes: The initial fungus garden gap only contains one point , and the set composed of the adjacent points of is denoted as , where has no common point with ; Find the maximum value in the fungus garden model such that: (18) Then add the point to S0, that is: S0 = S0 + { }; If the gap proportion reaches 0.09, end this process; otherwise, go to process b; Through this method, a large number of fungus garden gaps with a diameter of 0.5 cm or greater than 0.5 cm can be constructed until the gap proportion reaches the set threshold; S5: Use the finite-difference time-domain method to forward model the final dielectric constant spatial distribution map, and generate the ground-penetrating radar waveforms of the dam body medium background and termite nests.

[0016] In some embodiments, the above forward modeling method for termite nests in dams based on double random media can also be implemented in the following manner.

[0017] To clearly demonstrate the simulation process of this embodiment, this embodiment takes the electrical parameters of the main medium of the Yangtze River levee in a certain place as an example to carry out the forward simulation of ground-penetrating radar for detecting termite nests in the levee. There are various types of main termite nest structures, which are mainly affected by the environment and the medium beside the nest. Generally, they can be divided into multi-comb main nests and single-comb main nests. The single-comb main nest is composed of a single comb, a mud skeleton, and an air layer above. Compared with the single-comb main nest, the structure of the multi-comb main nest is more complex. The structure of the multi-comb termite main nest is as shown in Figure 2 So, this embodiment will adopt the multi-comb main nest.

[0018] In this embodiment, the forward method for termite nests in the levee based on double random media mainly includes the following four steps: Step 1: Measure the dielectric constant of the local levee and the proportion of the comb gap. To accurately measure the dielectric properties of the main medium of the levee, three test boreholes are excavated in the levee section (as shown in (a) of Figure 3 ). The depth of the boreholes is 0.6 m (the width of the ground-penetrating radar antenna is 0.4 m), and the vertical heights from the levee slope are 1.1 m, 0.45 m, and 0.65 m respectively.

[0019] Set the initial dielectric constant of the main medium of this levee to 9, and let the ground-penetrating radar move and scan along the slope directly above the three boreholes to obtain the detection image as shown in (b) of Figure 3 . At the 16 ns time course, an isolated hyperbolic reflection signal can be seen. Combining the lateral positioning information, it is determined that the target originates from borehole b. The relationship between the propagation time and speed of electromagnetic waves in the medium is shown in Formulas 1 and 2, (1) (2) where T is the travel time, D is the depth of the underground target, C is the propagation speed of electromagnetic waves in the medium, is the speed of light in vacuum, is the relative dielectric constant of the medium. It can be calculated from Formulas (1) and (2) that the dielectric constant of the main medium of this levee is 28.4. Natural termite combs have a significant non-uniform porous structure (including dense activity gaps, heterogeneous media). Figure 4 Six termite combs excavated from a levee in a certain place are shown in Figure 4 . This embodiment conducts a gap rate quantitative analysis on the comb samples shown in

[0020] The implementation process of the pixel ratio method is as follows: First, determine the number of pixels S1 existing in the area where pixels need to be extracted in the figure (here it is the nursery area of termite nests). Then, determine the number of gap pixels S2 of the termite nursery in the area where pixels need to be extracted according to the pixel color depth threshold. Then, the proportion P of the termite nursery gap is S2 to S1. The calculation process of the pixel ratio method is as follows: (1) Extract the total number of pixels (S1) in the nursery area of the target image; (2) Based on the preset gray threshold, identify the number of pixels (S2) that conform to the gap characteristics from S1; (3) Calculate the gap rate: P = S1 / S2 × 100%.

[0021] For Figure 4 the process of calculating the gap proportion of the nursery in Figure 5 is as follows

[0022] Through the pixel ratio method for Figure 4 the six nursery samples shown in the figure are quantitatively analyzed, and the measured gap rates are 8%, 13%, 10%, 5%, 9% and 7% respectively. Further, the average value is obtained as 9%. In this embodiment, the average gap rate is selected as the gap rate reference parameter of the three-dimensional model of the termite nest in the dam.

[0023] Step 2: Create the initial H5 matrix of the soil and the main termite nest; Based on the pixel map of the termite nest-dam main body medium system ( Figure 6 ), each pixel point is mapped to a preset material identifier through an image processing tool to construct the first H5 matrix file. After the matrix is parsed, it is imported into the algorithm program to generate the initial dielectric constant distribution field. The pixel-material mapping relationship is defined as: air (blue, identifier 0), soil (orange, 1), termite nursery (red, 2), main nest mud skeleton (purple, 3). Figure 7 Shows part of the data of the converted H5 matrix. The pixel point information stored is the corresponding result after the original image is rotated 90° clockwise.

[0024] Step 3: Create a random medium matrix of the soil and the nursery; The random medium model encompasses the inhomogeneity characteristics of both large and small scales: the large-scale inhomogeneous structure characterizes the macroscopic statistical properties of the medium (such as the average characteristics of the earth's medium), while the small-scale inhomogeneity is defined as the local random perturbation field superimposed on the macroscopic background field, and its spatial distribution is extensive and irregular. Given the significant impact of such small-scale perturbations on the inversion of underground medium characteristics, it is necessary to improve the accuracy of medium modeling by quantifying its spatial variation law. (See Shi‐Li G. Study on Multiphase Discrete Random Medium Model and its GPR Wave Field Characteristics[J]. Chinese Journal of Geophysics, 2015.).

[0025] In electromagnetics and geophysics, the dielectric constant is a key parameter for describing the response of a medium to an electric field, and its accuracy and complexity are crucial for understanding and predicting the electromagnetic behavior of underground media. The variation of the dielectric constant in space is highly random. To capture this spatial variation of inhomogeneity, in this embodiment, M(X) is introduced to represent the dielectric constant, where X = (x, z) is a vector in two-dimensional space. M(X) contains large-scale and small-scale inhomogeneities. As a random medium, M(X) can be decomposed into, (3) where M(0) represents the large-scale inhomogeneity, is the random perturbation ratio of the medium's dielectric constant relative to the background value. represents the small-scale inhomogeneity. The small-scale inhomogeneities are numerous and irregularly distributed, and can be represented by statistical characteristics, (4) Spatial random (relative) perturbation is a mathematical modeling method that introduces randomness adjustment in the spatial domain. Assuming that the spatial random (relative) perturbation is a spatial stationary random process function expression with zero mean, a certain autocorrelation function, and variance. From equations (3) and (4), we can obtain, (5) Relative perturbation satisfies a mean of zero and a variance of

[0026] （6） （7） In the early stage, many scholars at home and abroad mainly used mixed autocorrelation functions (including exponential and Gaussian types) to construct earth medium models such as soil and random hole media to describe the small-scale inhomogeneity of underground media (see Jiang Z. Simulation and analysis of GPR signal based on stochastic media model with an ellipsoidal autocorrelation function[J]. Journal of Applied Geophysics, 2013.). However, the exponential and Gaussian autocorrelation functions have problems of ignoring anisotropy and its non-stationarity and being unable to characterize complex multi-scale structures. The ellipsoidal autocorrelation function can effectively simulate different distributions of small-scale inhomogeneity inside the earth, and combined with multi-scale stochastic modeling technology, it provides a more accurate description of underground medium characteristics. Based on this, in this embodiment, the ellipsoidal autocorrelation function will be used as the autocorrelation function in the construction of stochastic media: （8） where a and b are autocorrelation lengths, is considered ellipsoidal because x and z have different scaling factors a and b respectively. r is the roughness factor. When r = 0, is a Gaussian ellipsoid. When r = 1, is an exponential ellipsoid. When 0 < r < 1, is a mixed ellipsoidal autocorrelation function.

[0027] In this embodiment, the ellipsoidal autocorrelation function is the spatial distribution map of dielectric constant. According to the Wiener-Khinchin theorem, the power spectrum of the spatial distribution map of dielectric constant and its autocorrelation function are a Fourier transform pair. Thus, the power spectrum of the spatial distribution map of dielectric constant can be calculated first, and then its two-dimensional inverse Fourier transform is performed to obtain its autocorrelation function image, and then normalization and binarization processing are carried out to extract characteristic parameters such as the autocorrelation length and autocorrelation angle of the autocorrelation function in different directions. (See Shi-Li G. Study on Multiphase Discrete Random Medium Model and its GPR Wave Field Characteristics[J]. Chinese Journal of Geophysics, 2015.) The following is the process of generating stochastic media: (1) After the two-dimensional Fourier transform of the ellipsoidal autocorrelation function, the autocorrelation equation will be used as the power spectral density function: (9) (2) Multiply the power spectrum with the two-dimensional random field ( ) to create a random power spectrum. The two-dimensional random field ( ) is a set of distributions of uncorrelated random numbers within [0, 2π] generated by a random number generator. At this time, the obtained random power spectrum is a series of discrete data, and it is smoothed as follows: (10) (3) A random perturbation field can be generated through the inverse Fourier transform of the random power spectrum

[0028] (11) The mean and variance of (12) (13) Then the relative perturbation : = (14) According to Equation (5), the expression of the dielectric property function of the random medium model can be obtained: (15) is the spatial distribution map of the dielectric constant of the random medium. The construction process of the random medium is as Figure 8 shown.

[0029] The selection of the autocorrelation function parameters is an important part of constructing a realistic random medium model. When constructing a random medium model, the autocorrelation lengths a, b and the roughness factor r are selected as key parameters, mainly because these three parameters can directly affect the microscopic and macroscopic characteristics of the model, and thus have a significant impact on the simulation results of the ground penetrating radar signal. In the dielectric constant random medium model, due to its small random perturbation, it is not easy to observe the differences. In order to conveniently observe and compare the influence of the transformation of the autocorrelation length and the roughness factor on the random model, in this embodiment, the dielectric constant identifier distribution map is selected to replace the dielectric constant spatial distribution map. In order to determine the selection of the elliptical parameters a, b and r, two groups of parameters a = 1, b = 50, r = 0.5 and a = 50, b = 1, r = 0.5 are set to demonstrate the influence brought by different directions of the self-length; a = 100, b = 100, r = 0.5 is set to demonstrate the influence brought by a large value of the autocorrelation length; a = 50, b = 50, r = 0, a = 50, b = 50, r = 0.5, a = 50, b = 50, r = 1 are set to demonstrate the influence of the value of r.

[0030] As shown in Equation (8) and Figure 9 shown, the autocorrelation length parameters a and b respectively control the average scale of inhomogeneities of the random medium in the horizontal and vertical directions. The measured soil medium has similar inhomogeneity characteristics in the horizontal and vertical dimensions (a≈b), and their values are positively correlated with the radar signal scattering intensity: when the values of a and b increase, the spatial extensibility of the inhomogeneities in the corresponding directions increases, resulting in a significant enhancement of the electromagnetic wave scattering and diffraction effects (such as the large-scale scattering characteristics shown in (c) of Figure 9 ). It is recommended that the values of a and b be less than 100 in actual modeling to balance the model accuracy and physical authenticity. The roughness factor r characterizes the irregularity of the medium interface. An increase in it will cause strong scattering loss and bandwidth broadening effects, resulting in increased signal attenuation; while a low r value (0.1≤r≤0.9) can improve the detection resolution, but will overly simplify the random perturbation characteristics of the medium. Therefore, it is necessary to achieve the adaptability modeling of the detection scenario through constraint parameters (a, b<100, r∈[0.1,0.9]).

[0031] The simulation verification model based on the measured parameters of a certain dam (soil dielectric constant 28.4, electromagnetic wave travel time at a depth of 6m is 217.38ns) is as shown in Figure 10 shown: Figure 10 In the model of (a), the blue area is the air layer; the orange shaded area is a 6m thick random medium layer (average dielectric constant 28.4); the red mark is the radar antenna transceiver assembly, and the transceiver spacing is 0.1m. Figure 10 In (b), the waveform of the single-channel A-SCAN signal after gain processing is shown, and the red dot spacing corresponds to the thickness of the medium layer. To eliminate the interference of random parameters, the elliptical parameters of each group need to be cross-validated through the travel times of 5 A-SCAN signals to ensure that the time delay calculation error is controllable.

[0032] The simulation error ratio M is introduced to evaluate the selection of elliptical parameters. The simulation error ratio is the simulated propagation time value T m divided by the true propagation time value T t : M = T m / T t (16) The simulation error ratios of each group of parameters are calculated. Table 1 shows the simulation error percentages under different combinations of elliptical parameters and radii. As the autocorrelation length and roughness factor increase, the simulated electromagnetic wave propagation time increases, that is, the electromagnetic wave velocity decreases.

[0033] Table 1 Simulation error ratio of elliptical parameter selection

[0034] The time delay of electromagnetic wave propagation in soil is regulated by multiple factors such as particle gradation, degree of cementation, and hydrological model parameters (see Oguchi T. Electromagnetic wave propagation and scattering in rain and other hydrometeors[J]. Proceedings of the IEEE, 1983, 71(9): 1029-1078.). Fine-grained modeling requires high computing power support. However, as the degree of modeling refinement increases, the improvement in model accuracy gain is limited. In this embodiment, a parameter simplification strategy is adopted to balance computational efficiency and model reliability. Based on the error analysis results in Table 1 (the minimum simulation error ratio is 99.8619%), the isotropic autocorrelation length (a = b = 60) and roughness factor (r = 0.3) are determined as the optimal parameter combination. The random medium model constructed accordingly can effectively reproduce the dielectric properties of the main medium material of a certain dam.

[0035] The calculation domain of the simulation model is set to 3.5m × 2.5m × 0.005m, the spatial discretization step is 0.005m, and the frequency of the time-harmonic excitation source is 600MHz. The central burial depth of the main nest of the termite multi-comb garden is 1.5m. Considering that the comb medium is composed of a mixture of termite excrement, wood chips, and soil, and is rich in fungal hyphae inside, with delicate medium characteristics (see Kreuzenbeck N B. Comparative genomic and metabolomic analysis of Termitomyces species provides insights into the terpenome of the fungal cultivar and the characteristic odor of the fungus garden of Macrotermes natalensis termites[J]. Msystems, 2022, 7(1): e01214-21.), combining computational efficiency and the matching degree of soil medium characteristics, the random medium parameters of the comb are finally selected as a = 20, b = 20, r = 0.2. The compactness of the microstructure of the mud skeleton of the main termite nest is significantly improved, and its dielectric constant may be increased to 5-15. In this embodiment, a dielectric constant of 34 is used as its value. At the same time, because its structure is delicate and almost a single medium, it can be regarded as a quasi-homogeneous medium.

[0036] The random medium soil matrix and the termite main nest matrix are directionally replaced with the corresponding dielectric constant identifiers in the initial H5 file through the masking algorithm. Figure 11Shows the distribution characteristics of the dielectric constant of the main nest model of termites with a gapless fungus garden: the spatial distribution of the air medium (purple) simulates the dynamic changes of the radar detection path under the condition of the undulating terrain of the dam slope; the main nest mud skeleton (green) and the fungus garden (yellow) present a coupled structure, and its discrete distribution pattern truly reproduces the occurrence state of the mud skeleton debris in the natural fungus garden.

[0037] Step 4: Create fungus garden gaps by the adjacent point fusion method; According to the results in Step 1, assume that the proportion of the fungus garden gap of a certain example of the Yangtze River dam in this embodiment is 9%. In this embodiment, the adjacent point fusion method is used to construct gaps in the termite fungus garden. The adjacent point fusion method will generate a large number of irregular and random areas, which can better simulate the distribution of termite fungus garden gaps. The basic implementation process of the adjacent point fusion method is as follows: (1) Select the fungus garden part in the model, and determine the diameter and proportion density of the gap. The diameter of the gap is taken as 0.5 cm, and the density is taken as 0.09.

[0038] (2) In the continuous random medium termite fungus garden model determine the local maximum points

[0039] (17) (3) Take each maximum value as the center of a fungus garden gap, and the area of each fungus garden gap is πR 2 . The discrete space step size of the model construction is 0.005 m. Take the point set with a step size of 1 centered on the maximum point, and the diameter of the gap is 0.01 m. Gradually fuse adjacent points step by step according to the following steps until the gap density P in the termite fungus garden area reaches 0.09: a) The initial fungus garden gap only contains one point S 0 ={ } , and denote the set composed of the adjacent points of S0 as N(S 0 ) , where N(S 0 ) and S 0 have no common points.

[0040] b) Find the maximum value in the fungus garden model such that (18) Then add the point to S0, that is, S0 = S0 + { }.

[0041] c) If the gap proportion reaches 0.09, end this process; otherwise, go to process b.

[0042] A large number of fungal garden gaps with a diameter of 0.5 cm or greater than 0.5 cm can be constructed through this method until the gap ratio reaches the set threshold.

[0043] Based on the simulation domain model parameters in step 3, the adjacent point fusion method is used to construct the fungal garden gaps of the random medium model of termite fungal gardens. The medium in the gaps is air, and the distribution state of the dielectric constant identifiers of the main termite nest is as Figure 12 shown.

[0044] Then, a new H5 file is generated for the distribution state of the dielectric constant identifiers of the termite main nest simulation model containing fungal garden gaps. Several random dielectric constants are created using a random number seed and saved in.txt format. Some example dielectric constants are as Figure 13 shown. The number of random dielectric constants is equal to the number of image pixels. For example, the simulation domain of a certain section of the Yangtze River levee is 1000 * 600 = 600000 (5 m * 3 m, with a step size of 0.005, that is, 5 / 0.005 = 1000, 3 / 0.005 = 600). The types are divided into air, soil, termite fungal gardens, and the mud skeleton of the termite main nest, and the dielectric constants are random values of each type of medium within the range in Equation (15) range.

[0045] Based on the above method, the GPRMAX parameters are configured, and its related parameters conform to the measured electrical parameters of a certain section of the Yangtze River levee. Finally, the simulated waveform of the termite nest in the levee is generated as Figure 14 shown.

[0046] To verify that this embodiment has practical engineering value, the measured waveform and the random medium waveform are selected for comparison. The measured waveform is as Figure 15 shown. By comparing the measured waveform of the termite main nest with the simulated waveform constructed in this embodiment ( Figure 14 ), it can be seen that there is a significant consistency between the two in terms of the soil medium response and the ant nest signal characteristics: the simulation error of the soil propagation path conforms to the attenuation law of electromagnetic waves in real media; although there are deviations in the waveform amplitude due to differences in antenna frequency, soil moisture content, and ant nest structure, the core characteristics are all asymmetric hyperbolic anomalies (main nest marker signals) presented by multi-phase superposition. The results show that the forward ground penetrating radar method based on the random medium theory can effectively construct a waveform database of termite nests in the levee with regional adaptability, and the generated data has a strong correlation with the measured waveform in terms of dielectric characteristics and structural response.

[0047] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned forward modeling method for termite nests in dikes based on double random media are implemented. Among them, the storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD), or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memories.

[0048] This embodiment provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the above-mentioned forward modeling method for termite nests in dikes based on double random media are implemented.

[0049] As Figure 16 shown, the computer device 120 may include: at least one processor 121, such as a central processing unit (CPU), at least one communication interface 123, a memory 124, and at least one communication bus 122. Among them, the communication bus 122 is used to realize the connection and communication between these components. Among them, the communication interface 123 may include a display screen (Display), a keyboard (Keyboard), and optionally, the communication interface 123 may further include a standard wired interface and a wireless interface. The memory 124 may be a high-speed random access memory (RAM), or a non-volatile memory, such as at least one disk memory. Optionally, the memory 124 may also be at least one storage device located far from the aforementioned processor 121. Among them, an application program is stored in the memory 124, and the processor 121 calls the program code stored in the memory 124 to execute any of the above method steps. Among them, the communication bus 122 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The communication bus 122 can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 16It is represented by only one line, but it does not mean that there is only one bus or one type of bus. Among them, the memory 124 may include volatile memory, such as random-access memory (RAM); the memory may also include non-volatile memory, such as flash memory, hard disk drive (HDD) or solid-state drive (SSD); the memory 124 may also include a combination of the above types of memory. Among them, the processor 121 may be a central processing unit (CPU), a network processor (NP), or a combination of a CPU and an NP. Among them, the processor 121 may further include a hardware chip. The above hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof. Optionally, the memory 124 is further configured to store program instructions. The processor 121 may call the program instructions to implement the forward modeling method for termite nests in dikes based on double random media as in this embodiment.

[0050] This embodiment provides a computer program product, including a computer program, which implements the steps of the above-mentioned forward modeling method for termite nests in dikes based on double random media when executed by a processor.

[0051] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit and scope protected by the claims of the present invention. All of these are within the protection scope of the present invention.

Claims

1. A forward modeling method for termite nests in dikes based on double random media, characterized in that Including the following steps: S1: Measure the dielectric constant of the local dike and the proportion of the fungus garden gap; S2: Create the initial H5 matrix of the soil and the main termite nest according to the pixel map of the termite nest-dike main body medium system; S3: Create a random medium matrix of the soil and the fungus garden according to the dielectric constant and the initial H5 matrix of the soil and the main termite nest; S4: According to the proportion of the fungus garden gap and the random medium matrix of the soil and the fungus garden, use the adjacent point fusion method to construct the fungus garden gap and form the final dielectric constant spatial distribution map; S5: Use the finite-difference time-domain method to forward model the final dielectric constant spatial distribution map and generate the ground-penetrating radar waveforms of the dike main body medium and the termite nest; 2. The forward modeling method for termite nests in dikes based on double random media according to claim 1, characterized in that, Step S1 specifically includes: S11: Conduct a sounding experiment on the local dike to obtain a detection image, and obtain the dielectric constant of the local dike according to the detection image; S12: Obtain the proportion of the fungus garden gap according to the fungus garden sample of the local dike by using the pixel ratio method; 3. The forward modeling method for termite nests in dikes based on double random media according to claim 2, characterized in that The obtaining of the dielectric constant of the local dike according to the detection image is as shown in the formula: ， ， Where T is the travel distance, D is the depth of the underground target, and C is the propagation speed of electromagnetic waves in the medium, is the speed of light in vacuum, is the relative permittivity of the medium.

4. The forward modeling method for termite nests in dikes based on double random media according to claim 2, wherein, Step S12 specifically includes: obtaining the number of pixels in the fungus garden area according to the fungus garden sample; obtaining the number of pixels in the fungus garden gap according to the number of pixels in the fungus garden area and a preset gray threshold; obtaining the gap ratio according to the number of pixels in the fungus garden area and the number of pixels in the fungus garden gap; averaging the gap ratios of multiple fungus garden samples to obtain the proportion of the fungus garden gap; 5. The forward modeling method for termite nests in dikes based on double random media according to claim 1, characterized in that Step S3 specifically includes: creating a random medium matrix of the soil and the fungus garden according to the dielectric constant and the initial H5 matrix of the soil and the main termite nest, as shown in the following formula: ， ， ， ， ， ， = ， ， Among them, is the ellipsoidal autocorrelation function, , represent the spatial coordinates of two orthogonal directions describing the characteristics of the medium, a and b are the autocorrelation lengths, and r is the roughness factor. is the power spectral density function, is the transverse frequency component in the spatial domain, is the longitudinal frequency component in the spatial domain, is the smoothed random power spectrum, is the power spectral density function, and are the mean and variance of the random perturbation field respectively, is the relative perturbation, is the preset perturbation standard deviation, is the actual standard deviation of generating the random field f(x,z), is the spatial distribution map of the dielectric constant of the random medium, represents large-scale inhomogeneity, is the random perturbation ratio of the dielectric constant of the medium relative to the background value, represents small-scale inhomogeneity.

6. The forward modeling method for termite nests in dikes based on double random media according to claim 1, characterized in that The use of the adjacent point fusion method to construct the fungus garden gap includes: S41: Select the fungus garden part in the model and determine the diameter and proportion density of the gap; S42: Determine the local maximum points according to the random medium matrix of the soil and the fungus garden; [[ID=z15]]S43: According to the local maximum points and the diameter of the gap, gradually fuse adjacent points at a preset discrete space step until the gap density in the termite fungus garden area reaches the proportion of the fungus garden gap to obtain the fungus garden gap; 7. The forward modeling method for termite nests in dikes based on double random media according to claim 1, characterized in that Step S42 specifically includes: using the adjacent point fusion method to construct the fungus garden gap according to the random medium matrix of the soil and the fungus garden. This method effectively simulates the actual distribution characteristics of the pores in the termite fungus garden by generating a large number of irregular random regions, as shown in the formula: ， Among them, is a random medium matrix of continuous soil and mycelium nursery, and are matrix regions of size in the mycelium nursery or soil model, and are local maximum points, is the radius of the gap.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it realizes the steps of the forward modeling method for termite nests in dikes based on double random media according to any one of claims 1-7; 9. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it realizes the steps of the forward modeling method for termite nests in dikes based on double random media according to any one of claims 1-7; 10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it realizes the steps of the forward modeling method for termite nests in dikes based on double random media according to any one of claims 1-7;

Images