Slope material identification method based on field electrical parameters

Abstract

The invention relates to a slope material identification method based on field electrical parameters, and the method comprises the following steps: obtaining field slope electrical measurement data, and preliminarily determining the boundary of a geotechnical material based on a clustering analysis K-Means algorithm; based on the preliminarily determined geotechnical material boundary, determining the target depth of the slope material interface through hierarchical probability statistical analysis; and fitting the electrical measurement parameters at the target depth of the interface of the slope material, determining a segmentation threshold value of the target slope material based on a peak value of a curve obtained by fitting, and dividing the slope material based on the segmentation threshold value. Compared with the prior art, the random uncertainty generated by noise data is considered, interference errors caused by the external environment in the electrical parameter testing process and personal errors in the observation process can be eliminated, accurate slope material recognition is achieved, and accurate preconditions are provided for engineering geological investigation, landslide disaster investigation and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of geotechnical engineering, and in particular to a slope material identification method based on in-situ electrical parameters. BACKGROUND

[0002] In-situ electrical parameter testing technology is widely used in geotechnical engineering investigation, but the resistivity data obtained by inversion is easily disturbed by measurement noise, inversion multi-solution and formation heterogeneity, resulting in fuzzy material division threshold. The traditional method relies on manual experience to set the threshold, which has strong subjectivity and insufficient resolution. The existing technology mainly discriminates lithology based on a single statistical parameter or a fixed threshold, which is difficult to effectively eliminate the influence of data errors on layering interpretation, especially in complex stratigraphic interfaces, which is prone to misjudgment. CN118914297A discloses a slope damage in-situ testing method, device and storage medium based on electrical measurement, which includes the following steps: obtaining the electrical parameter data and water content data of the slope obtained by electrical measurement at different periods; obtaining the orthographic image of the slope, constructing a digital terrain model, and obtaining the elevation data of the measurement line through gridding processing; based on the electrical parameter data and the elevation data of the measurement line, the resistivity data of the slope resistivity profile is obtained by smooth constrained least squares optimization; obtaining the relationship between soil damage variable, water content and soil resistivity; the slope resistivity profile at different periods is gridded, and the soil damage value is calculated based on the resistivity data and the water content data; the soil damage value is input into the digital terrain model to construct the damage variable profile, and the slope damage in-situ testing is realized. However, this method mainly focuses on the in-situ testing of slope damage, uses the smooth constrained least squares optimization method to analyze and process the electrical parameter data to obtain the resistivity data, and constructs the resistivity profile model based on this, which does not consider the complex noise influence in the obtained resistivity data, lacks the accurate division ability of the resistivity boundary of different materials, and cannot realize the high-precision identification of complex slope materials.

[0003] Therefore, there is an urgent need for a method that can eliminate noise interference in electrical measurement parameters under complex terrain conditions and accurately identify slope materials. SUMMARY

[0004] The purpose of the present application is to provide a slope material identification method based on in-situ electrical parameters, which combines K-Means clustering and layered probability statistical analysis to optimize the material classification threshold and significantly improve the objectivity and accuracy of rock-soil division.

[0005] The purpose of the present application can be achieved by the following technical solutions:

[0006] A slope material identification method based on in-situ electrical parameters, the method comprising the following steps:

[0007] Obtaining in-situ slope electrical measurement data, and preliminarily determining a rock-soil material boundary based on a clustering analysis K-Means algorithm;

[0008] Based on the preliminarily determined rock-soil material boundary, determining a target depth of a slope material interface through layered probability statistical analysis;

[0009] Fitting an electrical measurement parameter at the target depth of the slope material interface, determining a segmentation threshold of a target slope material based on a peak value of a curve obtained through the fitting, and dividing the slope material based on the segmentation threshold.

[0010] The in-situ slope electrical measurement data includes resistivity profile data obtained through high-density electrical testing.

[0011] The preliminarily determining the rock-soil material boundary based on the clustering analysis K-Means algorithm includes the following steps:

[0012] Integrating the resistivity profile data to obtain a resistivity sample data set, randomly generating a corresponding number of centroid sets based on a number of material types, determining initial centroids of the resistivity data, calculating distances between the resistivity sample data and each centroid, and distributing the resistivity sample data to a group with the smallest distance;

[0013] Recomputing clustering centroids based on the distributed data, iteratively calculating until a target function reaches a minimum value or remains unchanged, then stopping the iteration, and obtaining a resistivity clustering result.

[0014] Determining a resistivity threshold between each group based on the clustering result, and preliminarily determining a material boundary based on the resistivity threshold.

[0015] The number of material types is obtained based on layered probability statistical analysis of the slope as a whole.

[0016] The recomputing the clustering centroids based on the distributed data specifically includes updating the clustering points to average values of resistivity sample data of the group.

[0017] The target function is to minimize a sum of distances between resistivity sample data of each group and the clustering centroids.

[0018] The determining the target depth of the slope material interface through layered probability statistical analysis includes the following steps:

[0019] According to the preliminarily determined rock-soil material boundary, dividing the rock-soil material into layers of the same thickness within the rock-soil material boundary according to an electrode spacing, and performing layered probability statistical analysis on resistivity at different depths.

[0020] The resistivity logarithm value is calculated, and the frequency of the resistivity logarithm value in each interval is counted at a preset interval, the frequency being the ratio of the resistivity number in the interval to the total number, to obtain a probability density of the resistivity logarithm value, the probability density is subjected to multi-peak regression fitting in each layer to obtain a peak value of the probability density and a mathematical expectation of the resistivity logarithm value, so as to calculate a resistivity standard deviation of each layer, and a target depth of a material interface is determined based on the resistivity standard deviation.

[0021] The method for determining the target depth of the material interface based on the resistivity standard deviation is that when the target is located at the interface between high-resistivity and low-resistivity media, the standard deviation appears a minimum value, and the target depth of the material interface is determined based on the minimum value of the standard deviation corresponding to the layered position.

[0022] The determination of the segmentation threshold of the target slope material comprises the following steps: fitting the resistivity data at the target depth into two or more normal distribution functions, the functions having two or more separated peak values, and taking the value corresponding to the valley bottom between the peaks as the segmentation threshold of the slope material.

[0023] The method further comprises: making a distribution profile of the slope material according to the division result of the slope material.

[0024] Compared with the prior art, the present application has the following beneficial effects:

[0025] 1. The present application proposes a slope material identification method based on field electrical parameters, which combines K-Means clustering and layered probability density analysis to overcome the limitation that traditional single statistical model is sensitive to noise, reduces data error propagation through spatial-statistical double constraints, and improves the robustness of resistivity threshold division.

[0026] 2. The present application proposes a clustering and layered probability statistical joint analysis, which can more accurately divide the resistivity boundaries of different materials and avoid the situation that a large number of boreholes are needed in exploration to cause soil disturbance, thereby greatly saving the time cost and economic cost required for proving the state of soil layers.

[0027] 3. The present application introduces a dynamic layering strategy and probability distribution fitting to quantify the statistical difference of each electrical layer, breaks through the homogenization assumption of the fixed threshold method, and significantly improves the interface identification accuracy of thin interbedded and gradually changing strata. Based on the clustering objective function and statistical test results, the material grouping threshold is automatically iteratively optimized to reduce the dependence on artificial experience and adapt to the complexity of resistivity distribution in different geological scenarios.

[0028] 4. The present application has important significance for the fields of geotechnical engineering survey, disaster-causing mechanism research of natural disasters, and risk assessment and early warning, and provides an important premise for further effectively reducing the risk of natural disasters and improving the ability to respond to emergencies. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 Flow chart of the method of the present application;

[0030] Figure 2 Illustration of the threshold division based on cluster analysis of the present application;

[0031] Figure 3 Illustration of the identification of the boundary of the rock-soil material of the present application;

[0032] Figure 4 Illustration of the stratified probability statistical analysis of the present application;

[0033] Figure 5 Illustration of the probability statistics of the resistivity data at different depths of the present application;

[0034] Figure 6 Illustration of the determination of the target depth of the interface of the slope material of the present application;

[0035] Figure 7 Illustration of the accurate determination of the threshold of the slope material of the present application;

[0036] Figure 8 Illustration of the distribution profile of the slope material of the present application. DETAILED DESCRIPTION

[0037] The present application will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments are implemented on the premise of the technical solutions of the present application, and detailed implementation manners and specific operation processes are given, but the protection scope of the present application is not limited to the following embodiments.

[0038] The present embodiment provides a slope material identification method based on field electrical parameters, which realizes the noise reduction processing of the interference error, observation error, accidental error and system error contained in the field slope electrical measurement data, so as to determine the material group of different resistivity ranges. Specifically, as shown in the figure, Figure 1 The method comprises the following steps:

[0039] S1, obtaining field slope electrical measurement data, and preliminarily determining the boundary of the rock-soil material based on the cluster analysis K-Means algorithm.

[0040] S11, obtaining resistivity profile data by high-density electrical method test.

[0041] S12, integrating the resistivity profile data to obtain resistivity sample data set X(x1, x2, ···, xn), n is the sample number, and based on the stratified probability statistical analysis of the overall slope, the number of material types is k (i.e. the number of grouping groups is k), and a corresponding number of centroid sets C(c1, c2, ···, ck) are randomly generated based on the number of material types k. n k ​), determine the initial centroid of resistivity data, calculate the distance d of resistivity sample data and the clustering centroid of the jth group j , assign the resistivity sample data to the group with the minimum distance. Wherein the distance d of resistivity sample data and the clustering centroid of the jth group j satisfies the following equation:

[0042]

[0043] Wherein, d j is the distance (m) of resistivity sample data and clustering centroid j; x i is the ith resistivity sample data; c j is the clustering centroid of the jth group.

[0044] S13, recalculate the clustering centroid based on the assigned data, that is, update the clustering point to the average value of the resistivity sample data in the group, and the clustering centroid of the jth group satisfies the following equation:

[0045]

[0046] Wherein, c' j is the recalculated clustering centroid of the jth group; x j is the resistivity sample data of the jth group; n j is the number of resistivity sample data of the jth group.

[0047] Iterative calculation is performed until the objective function d reaches a minimum value or remains unchanged, and then the iteration is stopped, and the resistivity clustering result is obtained, wherein the objective function is expressed as:

[0048]

[0049] Wherein, d j is the distance (m) of resistivity sample data and clustering centroid j; d is the objective function, and min d represents minimizing the objective function.

[0050] S14, determine the resistivity threshold value between each group based on the clustering result (as shown in Figure 2 ), and preliminarily determine the material boundary based on the resistivity threshold value (as shown in Figure 3 ).

[0051] S2, based on the preliminarily determined rock-soil material boundary, determine the target depth of the slope material interface through stratified probability statistical analysis.

[0052] Specifically, according to the preliminarily determined rock-soil material boundary, the stratification thickness δh is determined according to the electrode spacing in the rock-soil material boundary, and for the standard symmetric Wenner device (AMNB), there is an empirical formula between the detection depth h and the electrode spacing a as follows:

[0053]

[0054] Beyond the depth of detection, the current density and electric field intensity decay significantly, the contribution from deeper formations becomes weak, thus the depth of detection is taken as the layer thickness. The rock-soil material is evenly divided into layers with the same thickness of δh, and the resistivity at different depths is statistically stratified, as shown in FIG. 1. Figure 4

[0055] The resistivity logarithm value is calculated, and the frequency of the resistivity logarithm value in each interval is statistically calculated at a preset interval of 0.1, the frequency is the ratio of the number of resistivity in the interval to the total number, the probability density of the resistivity logarithm value is obtained, and the probability density multi-peak value regression fitting is performed in each layer, as shown in FIG. 2, the mathematical expectation of the probability density peak value and the resistivity logarithm value of each layer is obtained, so as to calculate the standard deviation of the resistivity of each layer, and the standard deviation formula is as follows: Figure 5

[0056]

[0057] Where N is the total number of data points in the population, x i is the ith data, and μ is the population mean. Based on the resistivity standard deviation, the target depth of the material interface is determined. When the target is located in the high-resistance medium one and the high-resistance medium three (points A and E), the average property values are ρ1 and ρ3 respectively, and the standard deviations are σ1 and σ3; when the target is located in the low-resistance medium two (point C), the average property value is ρ2, and the standard deviation is σ2, wherein ρ1>ρ2; and when the target is located at the interface of high and low resistance media (points B and D), the standard deviation appears a minimum value, and the target depth of the material interface is determined based on the layer position corresponding to the minimum standard deviation, as shown in FIG. 3. Figure 6

[0058] S3, fitting the electrical measurement parameters at the target depth of the material interface of the slope, determining the segmentation threshold of the target slope material based on the peak value of the fitted curve, and dividing the slope material based on the segmentation threshold.

[0059] Specifically, the resistivity data at the target depth is fitted into two or more normal distribution functions, and the functions have two or more separate peak values. In one embodiment, the value corresponding to the valley between the peaks of the normal distribution function is taken as the material segmentation threshold, as shown in FIG. 4, and the resistivity is considered as three different rock-soil materials in the ranges of 0<ρ t1 , ρ t1 t2 , and greater than ρ t2 . Figure 7

[0060] ​​​​​Based on the above results of the slope material classification, a slope material distribution profile is made, as shown in Figure 8

[0061] The preferred embodiments of the present application have been described in detail. It should be understood that modifications and variations can be resorted to without departing from the spirit of this application, as those skilled in the art will readily appreciate. It is intended that the specification and claims be construed in a non-limiting fashion. Thus, the scope of the technology should be determined by the appended claims and their legal equivalents rather than by the description of the preferred embodiments.​

Claims

1. A slope material identification method based on field electrical parameters, characterized in that: The method comprises the following steps: Obtain on-site slope electrical measurement data and preliminarily determine the geotechnical material boundaries based on cluster analysis K-Means algorithm; Based on the preliminarily determined geotechnical material boundaries, the target depth of the slope material interface is determined through layered probability statistical analysis; The electrical measurement parameters at the target depth of the slope material interface are fitted, a segmentation threshold of the target slope material is determined based on the peak value of the fitted curve, and the slope material is divided based on the segmentation threshold.

2. The slope material identification method based on on-site electrical parameters according to claim 1 is characterized in that: The on-site slope electrical measurement data includes resistivity profile data obtained through high-density electrical testing.

3. The slope material identification method based on on-site electrical parameters according to claim 2 is characterized in that: The method of preliminarily determining the geotechnical material boundary based on the cluster analysis K-Means algorithm includes the following steps: The resistivity profile data are integrated to obtain a resistivity sample data set. A corresponding number of centroid sets are randomly generated based on the number of material types. The initial centroid of the resistivity data is determined. The distance between the resistivity sample data and each centroid is calculated, and the resistivity sample data is assigned to the group with the minimum distance. The cluster centroid is recalculated based on the allocated data and iterative calculation is performed until the objective function reaches the minimum value or remains unchanged, then the iteration is stopped to obtain the resistivity clustering result; A resistivity threshold between each group is determined based on the clustering result, and a material boundary is preliminarily determined based on the resistivity threshold.

4. The slope material identification method based on on-site electrical parameters according to claim 3 is characterized in that: The number of material types is obtained based on statistical analysis of the overall stratification probability of the slope.

5. The slope material identification method based on on-site electrical parameters according to claim 3 is characterized in that: The recalculation of the cluster centroid based on the allocated data specifically includes: updating the cluster centroid to the average value of the resistivity sample data of the group.

6. The slope material identification method based on on-site electrical parameters according to claim 3 is characterized in that: The objective function is to minimize the sum of the distances between each group of resistivity sample data and the cluster centroid.

7. The slope material identification method based on on-site electrical parameters according to claim 1 is characterized in that: The method of determining the target depth of the slope material interface by layered probability statistical analysis includes the following steps: According to the preliminarily determined geotechnical material boundary, within the geotechnical material boundary, the layer thickness is determined according to the electrode spacing, the geotechnical material is evenly divided into layers of equal thickness, and the resistivity at different depths is statistically analyzed for layer probability: The logarithmic resistivity value is calculated, and the frequency of the logarithmic resistivity value in each interval is counted at preset intervals. The frequency is the ratio of the number of resistivities in the interval to the total number, and the probability density of the logarithmic resistivity value is obtained. A multi-peak regression fitting of the probability density is performed on each layer to obtain the mathematical expectation of the probability density peak and the logarithmic resistivity value of each layer, thereby calculating the resistivity standard deviation of each layer, and determining the target depth of the material interface based on the resistivity standard deviation.

8. The slope material identification method based on on-site electrical parameters according to claim 7 is characterized in that: The method for determining the target depth of the material interface based on the resistivity standard deviation is as follows: when the target is located at the interface of high-resistance and low-resistance media, the standard deviation reaches a minimum value, and the target depth of the material interface is determined based on the layer position corresponding to the minimum standard deviation.

9. The slope material identification method based on on-site electrical parameters according to claim 1 is characterized in that: Determining the segmentation threshold of the target slope material includes the following steps: fitting the resistivity data at the target depth into two or more normal distribution functions, wherein the functions have two or more separated peaks, and using the value corresponding to the valley between the peaks as the slope material segmentation threshold.

10. The slope material identification method based on on-site electrical parameters according to claim 1, characterized in that: The method further comprises: making a slope material distribution profile diagram according to the slope material division result.

Images