Surface karst zone structure prediction method and device fused with multi-source data detection

Abstract

The invention discloses a multi-source data detection fused surface karst zone structure prediction method and device. The method comprises the steps of S1, obtaining apparent resistivity data detected by a high-density electrical method, electromagnetic wave data detected by a ground penetrating radar and remote sensing data of an overground environment; s2, extracting spatial distribution of surface environment factors based on the remote sensing image; s3, according to the apparent resistivity data, the electromagnetic wave data and the corresponding spatial distribution data, preliminarily determining soil and surface karst zone thickness data of the measured karst area; s4, the apparent resistivity data, the electromagnetic wave data, the surface environment factors and the corresponding thickness data form a training data set, and the model is trained to obtain a surface karst zone structure prediction model; and S5, inputting data obtained in real time into the surface karst zone structure prediction model to obtain a prediction result. The method has the advantages of being simple in implementation method, low in cost, high in prediction precision and efficiency and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of karst geological exploration, and in particular to a method and device for predicting surface karst zone structure by integrating multi-source data detection. Background Art

[0002] The epikarst zone is an important part of karst landforms. Its special binary hydrogeological structure, high heterogeneity and anisotropy make it difficult to quantitatively characterize its structure. As an important area for the exchange of groundwater and surface water, the accurate inversion of the epikarst zone structure in karst areas is of great significance for understanding karst hydrogeological processes, assessing geological disaster risks and guiding engineering construction. Traditional geological exploration methods use drilling, remote sensing and other methods. However, the dynamic changes of groundwater in the epikarst zone structure in karst areas are complex. Traditional exploration methods often have problems such as high labor costs, low efficiency and insufficient precision, which makes it difficult to meet the needs of high-precision and rapid acquisition of epikarst zone structure information.

[0003] Machine learning methods have powerful data processing and pattern recognition capabilities. They can automatically learn and extract hidden features and laws from massive data without or with only a small amount of human intervention, and can achieve deep mining and efficient processing of data. If machine learning methods can be applied to the prediction of surface karst zone structure, work efficiency can be significantly improved, exploration cycle can be shortened, and costs can be reduced. However, the accuracy of machine learning methods depends on the extraction of precise feature parameters and a large amount of data training process. The surface karst zone is underground and has a complex structure. It is difficult to directly extract feature parameters that can effectively characterize the structure of the surface karst zone. Therefore, the machine learning algorithm in the prior art cannot be applied to predict the structure of the surface karst zone. Summary of the invention

[0004] The technical problem to be solved by the present invention is: in view of the technical problems existing in the prior art, the present invention provides a method and device for predicting the structure of the surface karst zone by integrating multi-source data detection, which has a simple implementation method, low cost, high prediction accuracy and efficiency.

[0005] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0006] A method for predicting surface karst zone structure by integrating multi-source data detection, comprising the following steps:

[0007] Step S1. Multi-source detection data acquisition: obtaining the apparent resistivity data of the underground soil layer in the karst area under test detected by the high-density electrical method, obtaining the electromagnetic wave data of the underground karst zone in the karst area under test detected by the ground penetrating radar, and obtaining the remote sensing image data of the ground environment in the karst area under test, respectively obtaining the corresponding spatial distribution data according to the apparent resistivity data and the electromagnetic wave data, and obtaining the apparent resistivity spatial distribution data and the electromagnetic wave spatial distribution data;

[0008] Step S2. Comprehensive feature extraction: extracting surface environmental factors that affect the surface karst zone structure from the remote sensing image data of the ground environment;

[0009] Step S3. Preliminary structure identification: According to the apparent resistivity data, electromagnetic wave data and corresponding spatial distribution data, the layered interface of the rock and soil body, the interface between the surface karst zone and the bedrock are identified, and the soil thickness data of the measured karst area and the thickness data of the surface karst zone are preliminarily determined;

[0010] Step S4. Prediction model construction: The apparent resistivity data, the electromagnetic wave data, the surface environmental factors, and the corresponding soil thickness data and surface karst zone thickness data are used to form a training data set, and the deep learning model is trained using the training data set to obtain a surface karst zone structure prediction model, so as to establish a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness;

[0011] Step S5. Real-time structural prediction: Obtain real-time acquired apparent resistivity data, electromagnetic wave data and remote sensing data of the ground environment detected in the karst area to be measured, and extract real-time surface environmental factors, input the real-time acquired apparent resistivity data, electromagnetic wave data and surface environmental factors into the surface karst zone structure prediction model, obtain the soil thickness of the karst area to be measured and the prediction results of the thickness of the surface karst zone, and obtain the distribution state of the surface karst zone structure by combining the predicted soil thickness data with the surface karst zone thickness data.

[0012] Further, in step S1, the high-density electrical method uses a Wenner device for detection, and the A, M, N, and B electrodes of the Wenner device are arranged in a straight line with equal spacing, wherein A and B are power supply electrodes, and M and N are measuring electrodes. By controlling the power supply state of each electrode, the electrode spacing distance satisfies AM=MN=NB=n*x=a during each acquisition, wherein AM, MN, and NB are the distances between electrodes, n is the isolation coefficient, and a is the electrode spacing; when using the Wenner device for detection, AM=MN=NB=x is set, and A, M, N, and B are moved point by point along the measuring line to obtain the next profile, and then the distance between two adjacent electrodes is increased by a unit electrode spacing, and A, M, N, and B are moved point by point along the measuring line again to obtain the next profile, and the execution is repeated until an inverted trapezoidal section is finally obtained.

[0013] Furthermore, the apparent resistivity data of the underground structure in the karst area under test is obtained according to the following formula based on the data obtained by high-density electrical detection:

[0014]

[0015] Where ρ is the apparent resistivity, K is the device coefficient, ΔV is the potential difference, I is the supply current, and AN, BM, and BN are the distances between the electrodes, respectively;

[0016] The electromagnetic wave data of the underground structure of the karst area under test can be obtained by using the following formula:

[0017]

[0018] Among them, h represents the detection depth, v represents the propagation speed of electromagnetic waves, t represents the time required for the radar electromagnetic wave to be transmitted from the antenna to the receiving antenna, and x represents the distance between the transmitting antenna and the receiving antenna.

[0019] Furthermore, in step S2, the surface environmental factors affecting the surface karst zone structure are extracted from the remote sensing data of the ground environment, including:

[0020] Step S201. Select multiple environmental factors that affect the structure of the surface karst zone to construct an environmental factor data set, wherein the environmental factors include any of terrain factors, soil factors, rock factors, and vegetation factors;

[0021] Step S202: Screen the environmental factors in the environmental factor data set, select the final main controlling factor and use it as the surface environmental factor.

[0022] Furthermore, in step S202, the step of screening each environmental factor in the environmental factor data set includes:

[0023] Step S221. Preprocess the collected environmental factor data set to obtain a preprocessed independent variable matrix and a dependent variable matrix, wherein the independent variable is the environmental factor, and the dependent variable is the soil thickness and the surface karst zone thickness;

[0024] Step S222. Extracting principal components from the independent variables and the dependent variables;

[0025] Step S223. Calculate the VIP value of each principal component, and select the principal component with a VIP value greater than a preset threshold as the main control factor. The calculation formula of the VIP value is as follows:

[0026]

[0027] In the formula, p is the number of independent variables, m is the number of principal components of the independent variables, and R yh is the principal component t h Y * The explanation, R y (cum) is the sum of the principal components of the dependent variable Y * The cumulative explanation of W hj Yes X * With the principal component t h The correlation coefficient of .

[0028] Furthermore, the terrain factors include any one or more of altitude, slope aspect, slope, slope position, slope length factor, elevation variation coefficient, terrain undulation, surface roughness, slope variability, terrain moisture index, and surface cutting depth; the soil factors include any one or more corresponding indicators of soil temperature, soil moisture, soil type, soil porosity, soil texture, soil mineral composition, and soil organic matter; the rock factors include rock exposure rate index and lithology index, wherein the rock exposure rate index is the percentage of the exposed surface rock area in the measured area to the total area of ​​the area; the vegetation factors include any one or more of vegetation type, vegetation coverage, and vegetation index.

[0029] Furthermore, in step S3, the surface karst zone model based on the electromagnetic wave model is divided into two layers according to the spatial distribution of the electromagnetic wave data, that is, the upper part is the surface karst zone and the lower part is the bedrock; the surface karst zone model based on the resistivity model adopts a two-layer model, that is, the upper part is soil and the lower part is bedrock, and an inflection point is determined when the increase or decrease in the average resistivity of the first layer and the second layer exceeds a preset threshold, and the area above the determined inflection point is used as the soil layer, and the area below the inflection point is used as the bedrock layer.

[0030] Further, step S4 includes:

[0031] Step S401. A first data set is constructed from the apparent resistivity data, surface environmental factors and corresponding soil thickness, and a second data set is constructed from the electromagnetic wave data, surface environmental factors and corresponding surface karst zone thickness;

[0032] Step S402. Use the first data set and the second data set to train the machine learning model respectively to obtain a first model and a second model, wherein the first model inputs apparent resistivity data and surface environmental factors, and outputs predicted soil thickness, and the second model inputs electromagnetic wave data and surface environmental factor data, and outputs predicted surface karst zone thickness;

[0033] Step S403: Continuously adjust and optimize the first model and the second model obtained through training until the preset performance requirements are met, and finally establish a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness.

[0034] A device for predicting surface karst zone structure by integrating multi-source data detection comprises a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

[0035] A computer-readable storage medium storing a computer program, wherein the computer program implements the above method when executed by a processor.

[0036] Compared with the prior art, the advantages of the present invention are: the present invention realizes the prediction of the surface karst zone structure by combining physical detection methods with machine learning methods, firstly, the underground structure of the surface karst zone is detected by comprehensive physical detection means of ground penetrating radar and high-density electrical method, and the apparent resistivity data of the underground soil layer in the karst area and the electromagnetic wave data of the underground karst zone are obtained respectively, and at the same time, the surface environmental factors are extracted from the remote sensing data of the ground environment in the karst area, and the surface karst zone structure prediction model is trained based on the deep learning model by integrating the apparent resistivity data, the electromagnetic wave data and the surface environmental factors, which can fully explore the underground characteristics and the ground environmental characteristics of the surface karst zone structure, realize the rapid and high-precision inversion of the surface karst zone structure, and greatly improve the accuracy of data interpretation, and can also significantly improve the prediction efficiency and reduce the required cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the implementation process of the surface karst zone structure prediction method integrating multi-source data detection in this embodiment.

[0038] Figure 2 It is a schematic diagram of the principle of realizing multi-source data detection of surface karst zone structure in this embodiment.

[0039] Figure 3It is a schematic diagram of the arrangement principle of the Wennan device when the high-density electrical detection is adopted in this embodiment.

[0040] Figure 4 It is a schematic diagram of the implementation process of filtering the detection radar data in this embodiment. DETAILED DESCRIPTION

[0041] The present invention is further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0042] Both high-density electrical method and ground-penetrating radar can realize the detection of underground structures. Among them, high-density electrical method can measure the distribution of underground electric fields through electrode arrays and current injection, and can provide high-resolution resistivity data by using mathematical models for interpretation, thereby revealing the details of underground structures. Ground-penetrating radar can obtain radar image data of underground media by emitting high-frequency electromagnetic waves and receiving signals reflected back by underground media, thereby revealing the distribution of underground media. The present invention detects the underground structure of the surface karst zone by comprehensively using the physical detection means of ground-penetrating radar and high-density electrical method, respectively obtains the apparent resistivity data of the underground soil layer in the karst area and the electromagnetic wave data of the underground karst zone, thereby obtaining the resistivity distribution data of the corresponding soil layer and the electromagnetic wave distribution data of the underground karst zone.

[0043] At the same time, considering that the structural state of the surface karst zone will be affected by the surface environment, for example, the topography, soil parent material, climate, vegetation growth, etc. of the karst area are closely related to the development of the surface karst zone, and thus by analyzing the remote sensing data of the surface environment, the environmental factors related to the structure of the surface karst zone can be determined, and then the apparent resistivity and electromagnetic wave data obtained by high-density electrical method and ground penetrating radar detection can be comprehensively obtained. The underground characteristics and ground environmental characteristics of the surface karst zone structure. However, the above data are often large and extremely complex, and traditional data analysis methods are often time-consuming and laborious in processing, and are easily affected by human factors, resulting in subjectivity and uncertainty in the interpretation results. The present invention applies machine learning methods to process and analyze the above-mentioned geophysical data, and integrates physical detection methods and machine learning methods to jointly realize the prediction of the surface karst zone structure, which can fully explore the underground characteristics and ground environmental characteristics of the surface karst zone structure, and realize the rapid and high-precision inversion of the surface karst zone structure, greatly improve the accuracy of data interpretation, and can also significantly improve the efficiency of prediction and reduce the required cost.

[0044] like Figure 1 As shown, the method for predicting the structure of the surface karst zone by integrating multi-source data detection in this embodiment includes the following steps:

[0045] Step S1. Multi-source detection data acquisition: obtain the apparent resistivity data of the underground soil layer in the measured karst area detected by the high-density electrical method, obtain the electromagnetic wave data of the underground karst zone in the measured karst area detected by the ground penetrating radar, and obtain the remote sensing data of the ground environment of the measured karst area, respectively obtain the corresponding spatial distribution data according to the apparent resistivity data and the electromagnetic wave data, and obtain the apparent resistivity spatial distribution data and the electromagnetic wave spatial distribution data.

[0046] like Figure 2 As shown in the figure, after planning the survey area in the surface karst zone, the high-density electrical method is used to obtain the resistivity data of the shallower soil layer, and the ground penetrating radar is used to emit high-frequency electromagnetic waves and receive the signals reflected by the underground medium to obtain the electromagnetic wave data of the deeper underground karst zone. At the same time, the remote sensing data collected from the ground survey of the surface layer is obtained, which can cover the underground soil layer, the underground karst zone and the surface environment for comprehensive detection.

[0047] In a specific application embodiment, when high-density electrical detection is used, a full-range preliminary survey of the study area can be conducted based on remote sensing images, historical data, etc., to determine the measured area, and a high-density resistivity instrument can be used for the measured area. An electrode array is arranged in the measured area, and underground apparent resistivity distribution data is obtained by automatic point arrangement, automatic pole running, automatic power supply, automatic observation, etc. When laying out the survey line, the differences in horizontal and vertical scales can be comprehensively considered to fully consider the survey line resolution. After the high-density electrical measurement is completed, the underground apparent resistivity distribution map is obtained, and the coordinate information of each sample point is recorded using the measuring equipment, and the collected high-density electrical data is subjected to pre-processing operations such as denoising and filtering to improve the data quality.

[0048] Taking the high-density electrical method using the SYSCAL resistivity meter as an example, the high-density electrical method parameters are initialized during field measurement, including the measurement point spacing, electrode device type, etc., wherein the high-density electrical method measurement line can be set to a spacing of 0 to 2 m. Considering that the smaller the spacing, the more detailed the detected underground structure, in order to obtain more underground structure information and taking into account the amount of data, this embodiment sets the measurement line spacing to 0.5 m, and the detectable underground depth is 6 m.

[0049] There are many types of electrode devices for high-density electrical method. According to the spatial position relationship between the power supply electrode and the measuring electrode, they can be divided into Wenner, dipole-dipole, monopole-dipole, and electrode-electrode devices. In this embodiment, the Wenner device is used for detection in combination with the underground structure of the surface karst zone. Figure 3As shown, the A, M, N, and B electrodes of the Wenner device are arranged in a straight line with equal spacing, where A and B are power supply electrodes, and M and N are measuring electrodes. By controlling the power supply state of each electrode, the electrode spacing distance satisfies AM=MN=NB=n*x=a during each acquisition, where AM, MN, and NB are the distances between electrodes, n is the isolation coefficient, and a is the electrode spacing; when the Wenner device is used for detection, AM=MN=NB=x is set, and A, M, N, and B are moved point by point along the measuring line to obtain the next profile, and then the distance between two adjacent electrodes is increased by a unit electrode spacing, and A, M, N, and B are moved point by point along the measuring line again to obtain the next profile, and the above steps are repeated until an inverted trapezoidal section is finally obtained.

[0050] In this embodiment, the apparent resistivity data of the underground structure in the karst area under test can be obtained according to the data obtained by high-density electrical detection according to the following formula:

[0051]

[0052] Where ρ is the apparent resistivity, K is the device coefficient, ΔV is the potential difference, I is the supply current, and AN, BM, and BN are the distances between the electrodes, respectively.

[0053] After the apparent resistivity data of the underground soil layer in the karst area is detected by high-density electrical method, the ground penetrating radar system is used on the original high-density electrical method line to transmit high-frequency electromagnetic waves to the high-density electrical method line and receive the signal reflected by the underground medium to obtain the electromagnetic wave data of the underground medium. For example, the MALA ProEx ground-penetrating three-dimensional geological radar instrument can be used for detection, and the ground-penetrating radar parameters are set as follows: select a 50MHz non-shielded antenna in frequency selection, and use point measurement data acquisition method. At the same time, in order to match the high-density electrical method data, the sampling interval is also set to 0.5m spacing, the sampling frequency is ~500~800MHz, the sampling time window is 1000ns, the number of superpositions of each data is 128 times, and the detection depth is 40m. After the ground-penetrating radar detection is completed, the collected electromagnetic wave data is filtered. Figure 4 As shown, Ground Vision data processing software is used to remove the zero drift of the data through one-dimensional filtering (removing DC drift); then static correction (movement start time) and gain (energy attenuation) are selected to amplify the deep signal, and two-dimensional filtering (extracting the average trace) is used to remove the horizontal part of the image; then, one-dimensional filtering / Butterworth bandpass filtering is performed on the corresponding frequency according to the antenna used, and finally two-dimensional filtering / sliding average is used to suppress noise, smooth the image, and obtain the underground radar electromagnetic wave distribution map through layer tracking.

[0054] In this embodiment, the electromagnetic wave data of the underground structure of the karst area under test can be obtained by using the ground penetrating radar detection data according to the following formula:

[0055]

[0056] Among them, h represents the detection depth, v represents the propagation speed of electromagnetic waves, t represents the time required for the radar electromagnetic wave to be transmitted from the antenna to the receiving antenna, and x represents the distance between the transmitting antenna and the receiving antenna.

[0057] In a specific application embodiment, a GPS instrument can also be used to record the position information of the survey line and the position information corresponding to each measuring point to better locate the target survey line; when selecting the survey line, a longer survey line can be arranged as much as possible according to the land use conditions of the survey to ensure that more underground structure information can be obtained. In order to obtain relatively stable resistivity values ​​and electromagnetic wave values, when laying out the survey line, a long sunny day can be selected according to weather conditions to avoid misjudgment of soil resistivity due to increased soil moisture content after precipitation.

[0058] In a specific application embodiment, the pre-processed apparent resistivity data and electromagnetic wave data are analyzed by using Res2dinv and Reflexw software respectively, and an underground apparent resistivity distribution map and an underground radar image are obtained.

[0059] Step S2. Comprehensive feature extraction: Obtain corresponding spatial distribution data according to apparent resistivity data and electromagnetic wave data, obtain apparent resistivity spatial distribution data and electromagnetic wave spatial distribution data, and extract surface environmental factors that affect the structure of the surface karst zone from the remote sensing data of the ground environment.

[0060] The topography, soil parent material, climate, and vegetation growth in karst areas are closely related to the development of the epikarst zone. This implementation selects environmental factors such as topography, soil, rock, and vegetation based on remote sensing image data to explore the impact of environmental factors on the structure of the epikarst zone. By combining multi-source remote sensing data such as DEM, vegetation index, and land use status, environmental factors such as geology, topography, and vegetation that may affect the structure of the epikarst zone are collected to construct an environmental factor library, and data fusion methods are used to extract the multi-dimensional spatial distribution characteristics above and below the ground that affect the structure of the epikarst zone.

[0061] In this embodiment, the following steps can be used to extract the surface environmental factors that affect the surface karst zone structure from the remote sensing data of the ground environment:

[0062] Step S201. Select multiple environmental factors that affect the structure of the surface karst zone to construct an environmental factor data set, where the environmental factors include terrain factors, soil factors, rock factors, vegetation factors, etc.

[0063] Terrain factors can be extracted from remote sensing data using ArcGIS 10.7 software and retrograde extraction methods, including altitude, slope aspect, slope, slope position, slope length factor, elevation variation coefficient, terrain undulation, surface roughness, slope variability, terrain moisture index, surface cutting depth, etc. Soil factors Soil factors can be obtained through long-term site monitoring and indoor experiments. Site monitoring indicators can include soil temperature, soil moisture, etc., and indoor sampling and analysis experimental indicators can include soil type, soil porosity, soil texture, soil mineral composition, soil organic matter, etc. Rock indicators mainly include lithology and rock exposure rate, among which the rock exposure rate is the percentage of the surface rock exposed area in the region to the total area of ​​the region. This value is the most intuitive manifestation of the rocky desertification landscape phenomenon and an important factor leading to the generation and change of rocky desertification, that is, the rock exposure rate is also closely related to the structure of the surface karst zone.

[0064] The rock exposure rate can be calculated using a pixel binary model based on the Normalized Difference Rock Index (NDRI). For example, the calculation formula for NDRI can be expressed as:

[0065] NDRI=(TM7-TM4) / (TM7+TM4) (4)

[0066] f r =(NDRI-NDRI0) / (NDRI r -NDRI0) (5)

[0067] Among them, TM7 is the reflection value of the short-wave infrared band, TM4 is the reflection value of the near-infrared band, and f r is the rock exposure rate, NDRI is the information obtained by the sensor, r It is the NDRI value when it is composed entirely of bare rock, and NDRI0 is the NDRI value when there is no exposed rock.

[0068] Vegetation factors include vegetation type and vegetation coverage, among which vegetation coverage can be estimated using a pixel binary model based on the Normalized Difference Vegetation Index (NDVI). NDVI comes from Landsat 8TM remote sensing images. The annual average value of NDVI is calculated using the annual remote sensing impact of a specified time, and then normalized. The calculation formulas for vegetation index and vegetation coverage can be expressed as follows:

[0069]

[0070] fc c =(NDVI-NDVI soil) / (NDVI veg -NDVI soil ) (7)

[0071] Among them, NIR is the reflectance value of the near infrared band, R is the reflectance value of the red light band; NDVI soil is the NDVI value of the soil part, NDVI veg is the NDVI value of the vegetation part, f c is the vegetation coverage.

[0072] Step S202: Screen the environmental factors in the environmental factor data set, select the final main controlling factor and use it as the surface environmental factor.

[0073] In this embodiment, VIP (Variable Importance in Projection) importance analysis can be used to screen the environmental factors in the environmental factor data set. VIP importance analysis is a variable screening method based on partial least squares. The value of VIP can effectively identify the predictor variable that best explains the dependent variable among all independent variables. For example, environmental factors with higher VIP values ​​are considered to be the main controlling factors, and their impact on the surface environment is the most significant.

[0074] In this embodiment, the steps of screening each environmental factor in the environmental factor data set using VIP importance analysis include:

[0075] Step S221. Data preprocessing: preprocess the collected environmental factor data set to obtain the preprocessed independent variable matrix and dependent variable matrix, where the independent variables are environmental factors and the dependent variables are soil thickness and surface karst zone thickness.

[0076] Specifically, preprocessing includes data cleaning (removing duplicate, invalid or outlier values), missing value processing (such as interpolation or deletion of severely missing samples) and standardization (such as normalization or standardization to eliminate the impact of different dimensions and numerical ranges on the model). ij )n×k and dependent variables (soil thickness, surface karst zone thickness) matrix Y=(y i )n×1 for standardization to obtain the standardized variable matrix X * and Y * .

[0077] Step S222. From the independent variable X * (environmental factors) and dependent variable Y * (soil thickness, surface karst zone thickness) and extract the main component t h .

[0078] Specifically, principal components (t=1,...h) can be extracted from independent variables and dependent variables through PLS (partial least squares regression) algorithm, etc. These principal components can carry the variation information in their respective data sets to the maximum extent, and the correlation between the principal components is maximized.

[0079] Step S223. Calculate the VIP value of each principal component, and select the principal component with a VIP value greater than a preset threshold as the main control factor. The calculation formula of the VIP value is as follows:

[0080]

[0081] In the formula, p is the number of independent variables, m is the number of principal components of the independent variables, and R yh is the principal component t h Y * The explanation, R y (cum) is the sum of the principal components of the dependent variable Y * The cumulative explanation of W hj Yes X * With the principal component t h The correlation coefficient of .

[0082] In a specific application embodiment, it can be considered that when VIP>1, the independent variable is significantly correlated with the dependent variable, and the principal component with VIP>1 is taken as the main control factor. When VIP is smaller, it means that the variable contributes little to the model and can be considered to be eliminated to improve the stability of the model.

[0083] This embodiment uses the factor importance calculated by the partial least squares model combined with the correlation matrix table of all explanatory variables to further screen the environmental factors, which can effectively screen out the main surface environmental factors that can effectively characterize the structure of the surface karst zone.

[0084] Step S3. Preliminary structure identification: Based on the apparent resistivity data, electromagnetic wave data and the corresponding spatial distribution data, the stratified interface of the rock and soil body, the interface between the surface karst zone and the bedrock are identified, and the soil thickness data of the measured karst area and the thickness data of the surface karst zone are preliminarily determined.

[0085] In a specific application embodiment, data analysis tools such as MATLAB can be used to analyze the preprocessed apparent resistivity data and electromagnetic wave data, and geological prior knowledge can be combined to preliminarily identify the stratification interface of the rock and soil body, the interface between the surface karst zone and the bedrock, determine the soil thickness range, and then combine the radar image data of the underground medium to determine the thickness range of the surface karst zone.

[0086] In order to calculate the soil thickness distribution in the measured area, this embodiment defines the stratification interface of the rock and soil body, that is, the surface karst zone based on the resistivity model is only considered as a simple two-layer model, in which the upper part is soil and the lower part is bedrock. The first inflection point is determined only when the average resistivity of the first layer (soil) and the second layer (bedrock) significantly increases or decreases to a preset threshold, and the area above the inflection point is taken as the soil layer, and the area below the inflection point is taken as the bedrock layer.

[0087] Step S4. Prediction model construction: The apparent resistivity data, electromagnetic wave data, surface environmental factors, and corresponding soil thickness data and surface karst zone thickness data constitute a training data set, and the training data set is used to train a deep learning model to obtain a surface karst zone structure prediction model to establish a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness.

[0088] The preprocessed apparent resistivity data, electromagnetic wave data and extracted environmental factors are used as input, and the surface karst zone structure prediction model is obtained through model training and verification. It can quickly process complex geological data and establish nonlinear mapping relationships between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness.

[0089] In this embodiment, the construction of the surface karst zone structure prediction model can adopt the following steps:

[0090] Step S301. A first data set is constructed from apparent resistivity data, surface environmental factors and corresponding soil thickness, and a second data set is constructed from electromagnetic wave data, surface environmental factors and corresponding surface karst zone thickness;

[0091] Step S302. Use the first data set and the second data set to train the machine learning model respectively to obtain a first model and a second model, wherein the first model inputs apparent resistivity data and surface environmental factors, and outputs predicted soil thickness, and the second model inputs electromagnetic wave data and surface environmental factor data, and outputs predicted surface karst zone thickness;

[0092] Step S303: Continuously adjust and optimize the first model and the second model obtained through training until the preset performance requirements are met, and finally establish a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness.

[0093] Specifically, a first data set can be constructed by apparent resistivity, ground environmental factors and corresponding soil thickness, and a second data set can be constructed by electromagnetic waves, ground environmental factors and corresponding surface karst zone thickness. The constructed data sets are then used to train the machine learning model, and the prediction performance is optimized by adjusting the model parameters to establish a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness.

[0094] The machine learning model can specifically adopt an artificial neural network model (ANN) or a support vector machine model. ANN is a model that simulates the mechanism of biological neural networks and realizes information processing in combination with their behavioral characteristics. It has the characteristics of extremely high parallel implementation capability, self-organization, nonlinear processing capability and adaptability. The present embodiment establishes a three-layer forward neural network model of a typical BPA algorithm to predict the surface karst zone structure. The surface environmental factors screened by the factor screening model, the apparent resistivity data, and the electromagnetic wave data are used as input vectors. A neuron is used in the output layer to predict the surface karst zone structure. The neural network structure can also be further determined by determining the ratio of input and output data and selecting input and output variables, and the optimal number of hidden neurons (N) is determined by trial and error. The number of hidden neurons is selected in combination with the model accuracy verification method, and finally the neural network model with the best performance is obtained.

[0095] Support Vector Machine (SVM) is based on the principle of linear separability and structural risk minimization. For linear situations, the support vector machine function fitting first considers the use of linear regression function f(x). When linearly inseparable, the low-dimensional linear inseparable can be converted into high-dimensional linear separable; at the same time, it can seek the optimal solution between the complexity of the model and the learning ability of the model based on limited sample information, and ensure that the final model has good generalization ability. The basic idea of ​​nonlinear SVR is to map the input vector to a high-dimensional feature space (Hilbert space) through a pre-determined nonlinear mapping, and then perform linear regression in this high-dimensional space to achieve the effect of nonlinear regression in the original space. The expression of the nonlinear fitting function is:

[0096]

[0097] In a specific application embodiment, by selecting a machine learning algorithm suitable for processing complex geological data, such as the above-mentioned support vector machine (SVM), artificial neural network (ANNs), etc., the first model input includes pre-processed apparent resistivity data and ground environmental factor data, and the output is the predicted soil thickness, the second model input includes pre-processed electromagnetic wave data and ground environmental factor data, and the output is the predicted surface karst zone thickness, and the first model and the second model are combined to form a surface karst zone structure prediction model. Further, the prediction results can be cross-validated with the actual measurement results, and the model parameters can be continuously adjusted and optimized to ensure the applicability of the model in unknown areas and improve the generalization ability and prediction accuracy of the model.

[0098] Step S4. Real-time structural prediction: obtain the real-time acquired apparent resistivity data, electromagnetic wave data and remote sensing data of the ground environment detected in the karst area to be measured, and extract the real-time surface environmental factors, and input the real-time acquired apparent resistivity data, electromagnetic wave data and surface environmental factors into the surface karst zone structure prediction model to obtain the predicted results of the soil thickness of the karst area to be measured and the thickness of the surface karst zone.

[0099] By inputting the real-time detected apparent resistivity data, electromagnetic wave data and surface environmental factors of the karst area under test into the trained surface karst zone structure prediction model, the soil thickness and surface karst zone thickness of the karst area under test can be predicted, thus realizing the rapid and high-precision inversion of the surface karst zone structure.

[0100] Furthermore, the predicted soil thickness data can be combined with the surface karst zone thickness data to obtain the surface karst zone structure distribution status, and the inverted soil thickness data can be combined with the surface karst zone thickness data to obtain the surface karst zone structure distribution status.

[0101] In a specific application embodiment, the trained surface karst zone structure prediction model is applied to high-density electrical data, ground penetrating radar data and ground environmental factor data in the actual exploration area, and the model reasoning process is executed to obtain the prediction results of the soil thickness and the surface karst zone thickness in the area. Then, based on the prediction results, the soil thickness and the surface karst zone thickness in the karst area are accurately interpreted. The soil thickness distribution map, the surface karst zone thickness distribution map and the surface karst zone structure profile map are generated by Surfer software and geological interpretation is performed, which can be flexibly applied to various application scenarios such as engineering construction and geological disaster warning.

[0102] like Figure 2As shown, this embodiment first collects geophysical data, selects a measured area from the study area, uses high-density electrical method and ground penetrating radar to detect the underground structure of the surface karst zone, obtains apparent resistivity and electromagnetic wave data, then extracts features, selects environmental factors, and establishes a surface karst zone influencing factor data set through ground environmental investigation and remote sensing data processing, and uses a main control factor screening model to determine the influencing factors; then, the surface karst zone structure is identified, and based on the pre-processed apparent resistivity data and electromagnetic wave data and their spatial distribution, the soil thickness data and the surface karst zone thickness data are preliminarily identified; then, the pre-processed apparent resistivity data, electromagnetic wave data and ground environmental factors are used as input, and training is performed respectively. The machine learning model is trained to form a prediction model. Through model training and verification, a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and the thickness of the surface karst zone is established. Finally, the trained machine learning model is applied to actual exploration data to generate a spatial distribution map of soil thickness and surface karst zone thickness in the target area. It can integrate physical detection methods and machine learning methods to achieve rapid and high-precision inversion of the surface karst zone, improve the depth and breadth of multi-source detection data processing, fully integrate multi-source detection data to fully explore the characteristics of the surface karst zone structure, thereby significantly improving the accuracy and efficiency of surface karst zone prediction, and then realizing geological research, engineering construction and disaster warning in karst areas.

[0103] This embodiment further provides a surface karst zone structure prediction device that integrates multi-source data detection, including a processor and a memory, the memory is used to store computer programs, and the processor is used to execute the computer program to perform the above method.

[0104] It is understandable that the above method of this embodiment can be executed by a single device, such as a computer or server, etc., and can also be applied to a distributed scenario and completed by multiple devices in cooperation with each other. In the case of a distributed scenario, one of the multiple devices can only execute one or more steps in the above method of this embodiment, and multiple devices interact to complete the above method. The processor can be implemented in the form of a general-purpose CPU, a microprocessor, an application-specific integrated circuit, or one or more integrated circuits, etc., for executing related programs to implement the above method of this embodiment. The memory can be implemented in the form of a read-only memory ROM, a random access memory RAM, a static storage device, and a dynamic storage device. The memory can store an operating system and other applications. When the above method of this embodiment is implemented by software or firmware, the relevant program code is stored in the memory and called and executed by the processor.

[0105] This embodiment further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above method is implemented.

[0106] Those skilled in the art will appreciate that the above-mentioned embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0107] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for predicting the structure of a surface karst zone by integrating multi-source data detection, characterized in that the steps include: Step S1. Multi-source detection data acquisition: obtaining the apparent resistivity data of the underground soil layer in the karst area under test detected by the high-density electrical method, obtaining the electromagnetic wave data of the underground karst zone in the karst area under test detected by the ground penetrating radar, and obtaining the remote sensing image data of the ground environment in the karst area under test, respectively obtaining the corresponding spatial distribution data according to the apparent resistivity data and the electromagnetic wave data, and obtaining the apparent resistivity spatial distribution data and the electromagnetic wave spatial distribution data; Step S2. Comprehensive feature extraction: extracting surface environmental factors that affect the structure of the surface karst zone from the remote sensing image data of the ground environment; Step S3. Preliminary structure identification: According to the apparent resistivity data, electromagnetic wave data and corresponding spatial distribution data, the layered interface of the rock and soil body, the interface between the surface karst zone and the bedrock are identified, and the soil thickness data of the measured karst area and the thickness data of the surface karst zone are preliminarily determined; Step S4. Prediction model construction: The apparent resistivity data, the electromagnetic wave data, the surface environmental factors, and the corresponding soil thickness data and surface karst zone thickness data are used to form a training data set, and the deep learning model is trained using the training data set to obtain a surface karst zone structure prediction model, so as to establish a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness; Step S5. Real-time structural prediction: Obtain real-time acquired apparent resistivity data, electromagnetic wave data and remote sensing data of the ground environment detected in the karst area to be measured, and extract real-time surface environmental factors, input the real-time acquired apparent resistivity data, electromagnetic wave data and surface environmental factors into the surface karst zone structure prediction model, obtain the soil thickness of the karst area to be measured and the prediction results of the thickness of the surface karst zone, and obtain the distribution state of the surface karst zone structure by combining the predicted soil thickness data with the surface karst zone thickness data.

2. The method for predicting surface karst zone structure by integrating multi-source data detection according to claim 1, characterized in that: In step S1, the high-density electrical method uses a Wenner device for detection, and the A, M, N, and B electrodes of the Wenner device are arranged in a straight line with equal spacing, wherein A and B are power supply electrodes, and M and N are measuring electrodes. By controlling the power supply state of each electrode, the electrode spacing distance satisfies AM=MN=NB=n*x=a during each acquisition, wherein AM, MN, and NB are the distances between electrodes, n is the isolation coefficient, and a is the electrode spacing; when using the Wenner device for detection, AM=MN=NB=x is set, and A, M, N, and B are moved point by point along the measurement line to obtain the next profile, and then the distance between two adjacent electrodes is increased by a unit electrode spacing, and A, M, N, and B are moved point by point along the measurement line again to obtain the next profile, and the execution is repeated until an inverted trapezoidal section is finally obtained.

3. The method for predicting surface karst zone structure by integrating multi-source data detection according to claim 2, characterized in that: Based on the data obtained from high-density electrical detection, the apparent resistivity data of the underground structure in the karst area under test can be obtained according to the following formula: Where ρ is the apparent resistivity, K is the device coefficient, ΔV is the potential difference, I is the supply current, and AN, BM, and BN are the distances between the electrodes, respectively; The electromagnetic wave data of the underground structure of the karst area under test can be obtained by using the data obtained from ground penetrating radar detection according to the following formula: Among them, h represents the detection depth, v represents the propagation speed of electromagnetic waves, t represents the time required for the radar electromagnetic wave to be transmitted from the antenna to the receiving antenna, and x represents the distance between the transmitting antenna and the receiving antenna.

4. The method for predicting surface karst zone structure by integrating multi-source data detection according to claim 1, characterized in that: Step S2 includes: Step S201. Select multiple environmental factors that affect the structure of the surface karst zone to construct an environmental factor data set, wherein the environmental factors include any of terrain factors, soil factors, rock factors, and vegetation factors; Step S202: Screen the environmental factors in the environmental factor data set, select the final main controlling factor and use it as the surface environmental factor.

5. The method for predicting surface karst zone structure by integrating multi-source data detection according to claim 4 is characterized in that: In step S202, the step of screening each environmental factor in the environmental factor data set includes: Step S221. Preprocess the collected environmental factor data set to obtain a preprocessed independent variable matrix and a dependent variable matrix, wherein the independent variable is the environmental factor, and the dependent variable is the soil thickness and the surface karst zone thickness; Step S222. Extracting principal components from the independent variables and the dependent variables; Step S223. Calculate the VIP value of each principal component, and select the principal component with a VIP value greater than a preset threshold as the main control factor. The calculation formula of the VIP value is as follows: In the formula, p is the number of independent variables, m is the number of principal components of the independent variables, and R yh is the principal component t h Y * The explanation, R y (cum) is the sum of the principal components of the dependent variable Y * The cumulative explanation of W hj Yes X * With the principal component t h The correlation coefficient of .

6. The method for predicting surface karst zone structure by integrating multi-source data detection according to claim 4, characterized in that: The terrain factors include any one or more of altitude, slope aspect, slope, slope position, slope length factor, elevation variation coefficient, terrain undulation, surface roughness, slope variability, terrain moisture index, and surface cutting depth; the soil factors include any one or more corresponding indicators of soil temperature, soil moisture, soil type, soil porosity, soil texture, soil mineral composition, and soil organic matter; the rock factors include rock exposure rate index and lithology index, wherein the rock exposure rate index is the percentage of the surface rock exposed area in the measured area to the total area of ​​the area; the vegetation factors include any one or more of vegetation type, vegetation coverage, and vegetation index.

7. The method for predicting surface karst zone structure by integrating multi-source data detection according to any one of claims 1 to 5, characterized in that: In step S3, the surface karst zone model based on the electromagnetic wave model is divided into two layers according to the spatial distribution of the electromagnetic wave data, that is, the upper part is the surface karst zone and the lower part is the bedrock; the surface karst zone model based on the resistivity model adopts a two-layer model, that is, the upper part is soil and the lower part is bedrock, and the inflection point is determined when the increase or decrease in the average resistivity of the first layer and the second layer exceeds a preset threshold, and the area above the determined inflection point is used as the soil layer, and the area below the inflection point is used as the bedrock layer.

8. The method for predicting surface karst zone structure by integrating multi-source data detection according to any one of claims 1 to 5, characterized in that: Step S4 includes: Step S401. A first data set is constructed from the apparent resistivity data, surface environmental factors and corresponding soil thickness, and a second data set is constructed from the electromagnetic wave data, surface environmental factors and corresponding surface karst zone thickness; Step S402. Use the first data set and the second data set to train the machine learning model respectively to obtain a first model and a second model, wherein the first model inputs apparent resistivity data and surface environmental factors, and outputs predicted soil thickness, and the second model inputs electromagnetic wave data and surface environmental factor data, and outputs predicted surface karst zone thickness; Step S403: Continuously adjust and optimize the first model and the second model obtained through training until the preset performance requirements are met, and finally establish a nonlinear mapping relationship between apparent resistivity and soil thickness, and between electromagnetic waves and surface karst zone thickness.

9. A device for predicting surface karst zone structure by integrating multi-source data detection, comprising a processor and a memory, wherein the memory is used to store a computer program, characterized in that: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images