Quantitative evaluation method and system for slag landslide risk considering spatial variability of slag

Abstract

The application provides a kind of considering the space variability of slag soil landslide risk quantitative evaluation method and system of slag soil, first, the application utilizes parameter thermal map and variation coefficient distribution diagram to determine complete variation intensity field, then, dynamic kernel function is assigned to each interpolation point in combination with variation intensity field, finally, interpolation is carried out on each interpolation point according to dynamic kernel function, specifically, for each interpolation point, the sample points that can participate in interpolation point calculation can be determined by kernel radius, the influence degree of each participating sample point is determined by weight attenuation coefficient, and then the characteristic parameters of interpolation point can be more accurately determined, then, the characteristic parameters of interpolation point and sample point are combined to form parameter space distribution atlas, compared with the atlas obtained by conventional interpolation, the atlas is more accurate, so that the spatial distribution of slag soil parameters can be accurately and completely presented, and the slag soil landslide risk can be accurately and quantitatively evaluated.

Description

Technical Field

[0001] This invention relates to the field of drilling in soil and rock layers, and particularly to a method and system for quantitatively assessing the risk of landslides caused by slag and soil that takes into account the spatial variability of slag and soil. Background Technology

[0002] In the field of quantitative risk assessment of construction waste landslides, accurately obtaining the spatial distribution characteristics of the physical and mechanical parameters of construction waste, such as cohesion, internal friction angle, and moisture content, is the core prerequisite for ensuring the reliability of the assessment. However, existing assessment methods are insufficient in determining the mechanical parameters of construction waste and cannot achieve high-precision assessment. Summary of the Invention

[0003] The key technical problem that this invention aims to solve is: for high-pressure, space-constrained muck transport pipelines in soil and rock drilling, the limited number of sensors installed leads to the inability to monitor local muck accumulation in the pipeline cross-section, thus making it impossible to accurately predict early blockages.

[0004] Therefore, the first aspect disclosed in this application provides a method for quantitatively assessing the risk of landslides caused by construction waste that considers the spatial variability of the waste soil, comprising the following steps:

[0005] Collect multi-source data of construction waste within the assessment area. The multi-source data includes measured data reflecting the physical and mechanical properties of the construction waste and image inversion data that helps characterize the distribution characteristics of the construction waste.

[0006] Based on the multi-source data, a parameter heatmap and a coefficient of variation distribution map of the evaluation area were drawn;

[0007] By integrating the aforementioned parameter heatmap and coefficient of variation distribution map, a global continuous variation intensity field is constructed;

[0008] Based on the variation intensity field, the dynamic kernel function parameters of each interpolation point are adaptively adapted and obtained. The dynamic kernel function parameters include the kernel radius and the weight decay coefficient.

[0009] Based on the dynamic kernel function parameters of each interpolation point, the dynamic weights of sample points that are less than or equal to the kernel radius at each interpolation point are calculated. Combined with the measured parameters in the multi-source data, the spatial interpolation of the physical and mechanical parameters of the entire waste soil is completed, and the parameter spatial distribution map is obtained.

[0010] Based on the spatial distribution map of the parameters, and combined with the slope stability analysis model, the risk level of slag and soil landslide in the assessment area is quantitatively calculated, and the risk assessment is completed.

[0011] Optionally, in step 1), multi-source data on the waste soil within the assessment area is collected, including the following sub-steps:

[0012] 1.1) Based on the preliminary survey results of the assessment area, sample points were set up, and measured data of cohesion, internal friction angle, moisture content and compaction degree of the slag were obtained through borehole sampling and in-situ testing;

[0013] 1.2) Collect UAV visible light images and ground-penetrating radar images of the assessment area, and obtain the rock block distribution, crack density, and dielectric constant correlation data of the slag soil through image feature extraction and inversion, which are used as the image inversion data;

[0014] 1.3) Perform preliminary screening on the measured data and image inversion data to remove obvious outliers and obtain purified multi-source data.

[0015] Optionally, in step 2), based on the multi-source data, a parameter heatmap and a coefficient of variation distribution map of the evaluation area are drawn, including the following sub-steps:

[0016] 2.1) Extract the measured parameters from the multi-source data, perform preliminary interpolation using ordinary kriging interpolation, and color the parameters according to their numerical ranges to obtain a parameter heatmap that reflects the spatial distribution of the absolute values ​​of the parameters.

[0017] 2.2) Divide the evaluation area according to the preset grid, calculate the standard deviation and mean of the measured parameters in each grid, and obtain the local coefficient of variation for each grid based on the standard deviation and mean;

[0018] 2.3) Color the data according to the numerical range of the local coefficient of variation to obtain a coefficient of variation distribution map that reflects the spatial distribution of the parameter variation intensity.

[0019] Optionally, in step 3), the parameter heatmap and the coefficient of variation distribution map are fused to construct a global continuous variation intensity field, including the following sub-steps:

[0020] 3.1) Unify the spatial resolution of the parameter heatmap and the coefficient of variation distribution map, and perform pixel-level alignment and fusion;

[0021] 3.2) For each pixel, extract the parameter gradient at the corresponding location from the parameter heatmap, and extract the local coefficient of variation at the corresponding location from the coefficient of variation distribution map;

[0022] 3.3) Based on the preset weight coefficient, the comprehensive variation intensity of each pixel is obtained through the comprehensive variation intensity calculation formula. The comprehensive variation intensity is calculated as follows: the comprehensive variation intensity is equal to the local variation coefficient multiplied by the first preset weight coefficient, plus the parameter gradient multiplied by the second preset weight coefficient, where the sum of the first preset weight coefficient and the second preset weight coefficient is 1.

[0023] 3.4) Based on the comprehensive variation intensity of all pixels, a global continuous variation intensity field is formed.

[0024] Optionally, in step 4), based on the variation intensity field, the dynamic kernel function parameters for each interpolation point are adaptively obtained, including the following sub-steps:

[0025] 4.1) Preset the initial range of kernel radius and weight decay coefficient, and establish a mapping rule between comprehensive variation intensity and kernel function parameters based on the measured data of the verification points;

[0026] 4.2) For each interpolation point in the aforementioned variation intensity field, extract its comprehensive variation intensity;

[0027] 4.3) Based on the mapping rules and the comprehensive variation intensity of the interpolation point, determine the kernel radius and weight decay coefficient corresponding to the interpolation point to form a point-by-point customized dynamic kernel function parameter field.

[0028] Optionally, in step 5), based on the dynamic kernel function parameters of each interpolation point, the dynamic weights of sample points whose distance from each interpolation point is less than or equal to the kernel radius are calculated, including the following sub-steps:

[0029] 5.1) For each interpolation point, search for all sample points within the corresponding kernel radius;

[0030] 5.2) Based on the spatial distance between the interpolation point and the sample point, the kernel radius of the interpolation point, and the weight decay coefficient, the weight of each sample point is obtained through dynamic weight calculation. The dynamic weight calculation method is as follows: the weight of each sample point is equal to the square of the distance between the sample point and the interpolation point divided by the product of the square of the kernel radius and the weight decay coefficient, the negative number is used to calculate the exponent value, and then the exponent value is multiplied by (1 and the sum of the comprehensive variation intensity), and the result is used as the numerator; the results of all sample points within the kernel radius calculated in the above way are summed, and the result is used as the denominator; the ratio of the numerator to the denominator is the weight of the sample point.

[0031] 5.3) Based on the weight of each sample point and the corresponding measured parameters, the parameter estimates of the interpolation points are obtained by weighted summation. All interpolation points are traversed to form a parameter spatial distribution map.

[0032] Optionally, in step 6), based on the spatial distribution map of the parameters and combined with the slope stability analysis model, the risk level of landslides in the assessment area is quantitatively calculated, including the following sub-steps:

[0033] 6.1) Import the parameter values ​​from the spatial distribution map into the slope stability analysis model and calculate the slope stability coefficient of each grid in the evaluation area;

[0034] 6.2) A preset stability coefficient threshold range is established, and based on the magnitude of the stability coefficient, the risk level of each grid is divided into low risk, medium risk, and high risk.

[0035] 6.3) Based on the risk level of all grids, draw a landslide risk heat map of the assessment area to complete the quantitative risk assessment.

[0036] Optionally, in step 1.3), after preliminary screening of the measured data and image inversion data, a data calibration sub-step is also included:

[0037] 1.3.1) Select a subset of sample points and simultaneously acquire the measured data and the corresponding image inversion data;

[0038] 1.3.2) Establish a calibration model for the measured data and the image inversion data, and use the calibration model to correct all the image inversion data;

[0039] 1.3.3) The corrected image inversion data and the filtered measured data are fused together to obtain the final multi-source data.

[0040] Optionally, after obtaining the parameter spatial distribution map in step 5), the process also includes a sub-step for result verification and optimization:

[0041] 5.4) Select verification points that did not participate in interpolation and obtain the measured parameter values ​​of the verification points;

[0042] 5.5) Calculate the error between the measured parameter values ​​of the verification point and the estimated parameter values ​​at the corresponding locations in the parameter spatial distribution map;

[0043] 5.6) If the error exceeds the preset acceptable range, adjust the mapping rules of the dynamic kernel function or the weighting coefficient of the comprehensive variation intensity, and return to step 4) to re-perform parameter adaptation and interpolation calculation until the error meets the requirements.

[0044] This invention also provides a quantitative assessment system for landslide risk of construction waste that considers the spatial variability of construction waste, comprising:

[0045] The data acquisition module is used to collect multi-source data of the construction waste in the assessment area. The multi-source data includes measured data reflecting the physical and mechanical properties of the construction waste and image inversion data that helps characterize the distribution characteristics of the construction waste.

[0046] The dual-map plotting module, connected to the data acquisition module, is used to plot the parameter heatmap and the coefficient of variation distribution map of the evaluation area based on the multi-source data.

[0047] The variation intensity field construction module is connected to the dual-map drawing module and is used to fuse the parameter heat map and the variation coefficient distribution map to construct a global continuous variation intensity field.

[0048] A dynamic kernel function adaptation module is connected to the mutation intensity field construction module. It is used to adaptively adapt the dynamic kernel function parameters of each interpolation point based on the mutation intensity field. The dynamic kernel function parameters include the kernel radius and the weight decay coefficient.

[0049] The dynamic interpolation module is connected to the data acquisition module and the dynamic kernel function adaptation module respectively. It is used to calculate the dynamic weight of sample points that are less than or equal to the kernel radius from each interpolation point based on the dynamic kernel function parameters of each interpolation point. Combined with the measured parameters in the multi-source data, it completes the spatial interpolation of the physical and mechanical parameters of the entire waste soil and obtains the parameter spatial distribution map.

[0050] The risk assessment module, connected to the dynamic interpolation module, is used to quantitatively calculate the risk level of slag landslide in the assessment area based on the parameter spatial distribution map and the slope stability analysis model, thereby completing the risk assessment.

[0051] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0052] This application provides a method and system for quantitatively assessing the risk of landslides caused by construction waste, taking into account the spatial variability of the waste soil. First, a complete variability intensity field is determined using a parameter heatmap and a coefficient of variation distribution map. Then, a dynamic kernel function is assigned to each interpolation point based on the variability intensity field. Finally, interpolation is performed on each interpolation point according to the dynamic kernel function. Specifically, for each interpolation point, the sample points that can participate in the interpolation calculation can be determined by the kernel radius, and the influence degree of each participating sample point can be determined by the weight attenuation coefficient. This allows for a more accurate determination of the characteristic parameters of the interpolation point. Then, the characteristic parameters of the interpolation point and the sample points are combined to form a parameter spatial distribution map. This map is more accurate than maps obtained through conventional interpolation, thus accurately and completely presenting the spatial distribution of construction waste parameters, and thereby accurately and quantitatively assessing the risk of landslides caused by construction waste. Attached Figure Description

[0053] Figure 1 This application provides a flowchart illustrating a method for quantitatively assessing the risk of landslides caused by construction waste that takes into account the spatial variability of construction waste.

[0054] Figure 2 This is a parametric heatmap shown as an example in this application;

[0055] Figure 3 This is a graph of the coefficient of variation shown as an example in this application;

[0056] Figure 4 This is a variation intensity field diagram shown as an example in this application;

[0057] Figure 5This is a spatial distribution map of parameters presented as an example in this application;

[0058] Figure 6 This is a schematic diagram of the structure of the quantitative assessment system for landslide risk of construction waste that considers the spatial variability of construction waste, provided in the embodiments of this application. Detailed Implementation

[0059] To better understand this application, various aspects of this application will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are merely illustrative of exemplary embodiments of this application and are not intended to limit the scope of this application in any way. Throughout the specification, the same reference numerals refer to the same elements. Descriptions and / or include any and all combinations of one or more of the associated listed items.

[0060] It should be noted that, where there is no conflict, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0061] In the quantitative assessment of the risk of construction waste landslides, most current methods rely on a single type of data, such as using discrete borehole measurement data to characterize parameters. However, the inventors of this invention have discovered that discrete data itself is not conducive to the quantitative assessment of construction waste landslide risk. This is because the biggest cause of construction waste landslides is the difference in the physical and mechanical parameters of the construction waste in space. The physical and mechanical parameters of construction waste in space do not have a regularity or trend. Discrete data cannot accurately express the differences in spatial distribution (i.e., the spatial variability of this application), thus leading to inaccurate assessment.

[0062] In this application embodiment, a method for quantitatively assessing the risk of landslides caused by construction waste, considering the spatial variability of the waste soil, is provided, such as... Figure 1 As shown, it includes the following steps:

[0063] Step 1) Collect multi-source data of the construction waste in the assessment area. The multi-source data includes measured data reflecting the physical and mechanical properties of the construction waste and image inversion data that helps characterize the distribution characteristics of the construction waste.

[0064] It should be noted that the multi-source data includes measured data, such as cohesion, internal friction angle, water content, and compaction degree; and also image inversion data, such as rock block distribution, fracture density, and dielectric constant correlation data. Specifically, measured data are obtained by setting up sample points according to a grid, measuring cohesion and internal friction angle in the laboratory after borehole sampling, and obtaining water content and compaction degree through in-situ testing. Image inversion data is obtained by extracting rock block distribution and fracture density through UAV aerial photography, and inverting dielectric constant correlation data through ground penetrating radar detection. The above acquisition methods and approaches are all existing technologies, and this application will not elaborate on them.

[0065] Step 2) Based on the multi-source data, draw a parameter heat map and a coefficient of variation distribution map of the evaluation area.

[0066] In this embodiment, the parameter heatmap is a graphic representation of the spatial distribution of absolute values ​​of soil parameters. By interpolating and visualizing the measured parameters from multi-source data, discrete parameter points are transformed into a continuous spatial distribution. For example, the drawing process involves two key steps: The first step is parameter interpolation, extracting filtered measured parameters from the multi-source data. Since the measured parameters are only distributed at sample points and cannot directly reflect the overall situation, this application uses an interpolation method to fill the parameter gaps. In this embodiment, ordinary kriging interpolation is used. This allows for full consideration of the spatial correlation of parameters based on geostatistical principles, calculating the parameter estimate for any location within the evaluation area using the parameter values ​​of discrete sample points. The second step is visualization and coloring, dividing the parameter estimate into several numerical intervals and assigning a differentiated color to each interval (e.g., red for high-value areas and blue for low-value areas), ultimately forming the parameter heatmap. This parameter heatmap clearly marks the locations of key areas such as high-cohesion rock blocks and low-cohesion, water-rich areas, providing a spatial positioning benchmark for subsequent variation analysis.

[0067] It should be noted that the coefficient of variation distribution map is a graphic used to quantitatively present the spatial distribution of the variation intensity of soil parameters. By combining statistical analysis and visualization, it transforms the dispersion of parameters into an intuitive spatial distribution. For example, the drawing process should follow the following logic: First, divide the evaluation area into several calculation units according to a preset grid, ensuring that each unit can cover a sufficient number of sample points; then, for each unit, collect the measured parameters within a certain range around it, and calculate the standard deviation of the set of parameters. The mean-standard deviation reflects the dispersion of the parameters, and the mean reflects the average level of the parameters. The ratio of the two is the local coefficient of variation for that unit; finally, divide the intervals according to the magnitude of the local coefficient of variation and color them to form the coefficient of variation distribution map, which can accurately locate high-variance areas (such as the mixed area of ​​rock and loose soil) and low-variance areas (such as the area of ​​uniformly compacted cohesive soil).

[0068] Step 3) Integrate the parameter heatmap and the coefficient of variation distribution map to construct a global continuous variation intensity field.

[0069] It should be noted that the variation intensity field is a data field that can reflect the variation characteristics of soil parameters across the entire domain continuously. It is obtained by fusing parameter heat maps and variation coefficient distribution maps. In this embodiment, the construction process needs to be carried out in four steps:

[0070] The first step is image preprocessing. Due to the spatial resolution difference between the parameter heatmap and the coefficient of variation distribution map, the resolution of the two is unified first. The low-resolution image is adjusted to match the high-resolution image, and pixel-level alignment is performed to ensure that information at the same spatial location can be accurately matched. The second step is feature extraction. For each aligned pixel, its parameter gradient is calculated from the parameter heatmap, and its local coefficient of variation is extracted from the coefficient of variation distribution map. The third step is comprehensive calculation. Based on preset weight coefficients, the local coefficient of variation and parameter gradient are weighted and summed to obtain the comprehensive variation intensity of each pixel. The fourth step is global integration. The comprehensive variation intensity of all pixels is arranged according to spatial location to form a continuous data field covering the entire evaluation area, i.e., the variation intensity field.

[0071] Step 4) Based on the variation intensity field, adaptively adapt to obtain the dynamic kernel function parameters for each interpolation point. The dynamic kernel function parameters include the kernel radius and the weight decay coefficient.

[0072] In this embodiment, the dynamic kernel function parameters are interpolation strategy parameters that are dynamically adjusted according to the interpolation point variation intensity, including the kernel radius and weight decay coefficient.

[0073] For example, the adaptive adaptation process involves three key steps:

[0074] The first step is to establish mapping rules. First, the initial range of the kernel radius and weight decay coefficient is preset (the initial range can be determined based on similar engineering experience or pre-test results). Then, some verification points are selected, and the interpolation error of the verification points is calculated through different parameter combinations. With the goal of minimizing the error, the correspondence between the comprehensive variation intensity and the kernel function parameters is established.

[0075] The second step is to extract the variation intensity. For each interpolation point in the variation intensity field, the corresponding comprehensive variation intensity is extracted.

[0076] The third step is parameter allocation. Based on the established mapping rules, a kernel radius and a weight decay coefficient are assigned to each interpolation point. The kernel radius controls the range of sample points participating in the interpolation, while the weight decay coefficient controls the rate at which the weights of the sample points decay with distance. This ultimately forms a point-by-point customized dynamic kernel function parameter field, ensuring that the interpolation logic of each interpolation point matches the variation characteristics of its region.

[0077] Step 5) Based on the dynamic kernel function parameters of each interpolation point, calculate the dynamic weights of sample points that are less than or equal to the kernel radius at each interpolation point. Combined with the measured parameters in the multi-source data, complete the spatial interpolation of the physical and mechanical parameters of the entire waste soil and obtain the parameter spatial distribution map.

[0078] It should be noted that the dynamic weight can reflect the coefficient of the contribution of the sample point to the parameter estimation of the interpolation point, and can be calculated based on the relationship between the dynamic kernel function parameters and the position of the sample point. In the embodiment of this application, the calculation process needs to be carried out in three steps: The first step is sample point screening, which selects only sample points whose distance from the interpolation point is less than or equal to the kernel radius based on the kernel radius of the interpolation point; the second step is weight calculation, which calculates the spatial distance between each selected sample point and the interpolation point, and then calculates the exponential value by dividing the square of the distance by the product of the square of the kernel radius and the weight attenuation coefficient, taking the negative number, and then multiplying it by (1 + comprehensive variation intensity value). Finally, the dynamic weight is obtained by comparing this result with the sum of the calculation results of all sample points; the third step is parameter estimation, which multiplies the dynamic weight of each sample point by its measured parameters and sums them to obtain the parameter estimate of the interpolation point.

[0079] This application is not limited to this. The parameter spatial distribution map is a diagram that presents the distribution of parameters of the entire waste soil area. For example, all interpolation points in the evaluation area can be traversed first to collect the parameter estimates of each point and form a global parameter dataset. Then, the dataset can be transformed into an intuitive map through visualization methods such as contour line drawing and color rendering.

[0080] Step 6) Based on the spatial distribution map of the parameters and combined with the slope stability analysis model, quantitatively calculate the risk level of slag and soil landslide in the assessment area to complete the risk assessment.

[0081] In this embodiment, the slope stability analysis model is a professional calculation model based on rock and soil mechanics theory. Through theoretical derivation and engineering verification, it can calculate the slope stability based on soil parameters. In this embodiment, either the limit equilibrium method or the finite element method can be used. The quantitative calculation of the risk level involves three steps: First, parameter import, where parameter values ​​in the spatial distribution map are divided into grids, and parameters such as cohesion and internal friction angle of each grid are extracted and imported into the slope stability analysis model; second, stability calculation, where the stability coefficient of each grid is calculated using the model; and third, risk classification, where a preset stability coefficient threshold range (specifically, the preset stability coefficient threshold range can be determined based on engineering safety standards) is used to map the stability coefficient to three levels: low risk, medium risk, and high risk, thus completing the risk assessment.

[0082] This application first uses parameter heatmaps and coefficient of variation distribution maps to determine the complete variation intensity field. Then, it assigns a dynamic kernel function to each interpolation point based on the variation intensity field. Finally, it performs interpolation on each interpolation point according to the dynamic kernel function. Specifically, for each interpolation point, the kernel radius can be used to determine the sample points that can participate in the interpolation calculation, and the weight attenuation coefficient can be used to determine the influence of each participating sample point. This allows for a more accurate determination of the characteristic parameters of the interpolation point. Then, by combining the characteristic parameters of the interpolation point and the sample points, a parameter spatial distribution map is formed. This map is more accurate than the map obtained through conventional interpolation, thus accurately and completely presenting the spatial distribution of the soil parameters, and thus enabling accurate quantitative assessment of the risk of soil landslides.

[0083] The following is a detailed description of this application.

[0084] The specific sub-steps for collecting multi-source data on construction waste within the assessment area include:

[0085] 1.1) Based on the preliminary survey results of the assessment area, sample points can be set up using a strategy of grid layout combined with densification in areas with high variability. Through borehole sampling and in-situ testing, measured data on the cohesion, internal friction angle, moisture content, and compaction degree of the slag can be obtained.

[0086] 1.2) Collect UAV visible light images and ground penetrating radar images of the assessment area, and obtain the rock block distribution, crack density, and dielectric constant correlation data of the slag soil through image feature extraction and inversion, which are used as the image inversion data.

[0087] For example, for UAV visible light images, edge detection algorithms are used to extract rock contours and crack orientations, and the area ratio of rock blocks and the number of cracks are statistically analyzed. For ground-penetrating radar images, filtering algorithms are used to remove clutter and extract features such as amplitude and travel time of reflected signals. Image inversion is the process of converting features into attribute data. A mapping relationship between ground-penetrating radar signal features and dielectric constant is established, and dielectric constant correlation data is derived. The final rock block distribution, crack density, and dielectric constant correlation data constitute the image inversion data, which will not be elaborated upon in this application.

[0088] 1.3) Perform preliminary screening on the measured data and image inversion data to remove obvious outliers and obtain purified multi-source data.

[0089] It should be noted that obvious outliers are data that deviate greatly from the majority of data and do not conform to the actual situation. They are caused by sampling disturbances, equipment errors, or image interference. This application does not impose any restrictions on them.

[0090] The specific sub-steps for drawing the parameter heatmap and coefficient of variation distribution map of the evaluation area include:

[0091] 2.1) Extract the measured parameters from the multi-source data, perform preliminary interpolation using ordinary kriging interpolation, and color the parameters according to their numerical ranges to obtain a parameter heatmap that reflects the spatial distribution of the absolute values ​​of the parameters.

[0092] In this embodiment, the ordinary kriging interpolation method is an interpolation method based on geostatistics. Specifically, it utilizes the spatial correlation of parameters to transform discrete sample points into a continuous parameter distribution. By calculating the variability function between sample points, a parameter spatial distribution model is established, thereby obtaining parameter estimates at any location. Then, through color processing, the interpolated parameter estimates are assigned colors according to numerical intervals to form a parameter heatmap.

[0093] 2.2) Divide the evaluation area according to the preset grid, calculate the standard deviation and mean of the measured parameters in each grid, and obtain the local coefficient of variation of each grid based on the standard deviation and mean.

[0094] In this embodiment, the preset grid is a calculation unit divided according to the size of the evaluation area and the data accuracy requirements. Specifically, the entire area is divided into several small areas.

[0095] 2.3) Color the data according to the numerical range of the local coefficient of variation to obtain a coefficient of variation distribution map that reflects the spatial distribution of the parameter variation intensity.

[0096] The sub-step of constructing a globally continuous variation intensity field by integrating the aforementioned parameter heatmap and variation coefficient distribution map specifically includes:

[0097] 3.1) Unify the spatial resolution of the parameter heatmap and the coefficient of variation distribution map, and perform pixel-level alignment and fusion.

[0098] In this embodiment, unifying spatial resolution means adjusting a low-resolution graphic to match a high-resolution graphic, so that pixels at the same spatial position in the two graphics correspond one-to-one.

[0099] 3.2) For each pixel, extract the parameter gradient at the corresponding position from the parameter heatmap and extract the local coefficient of variation at the corresponding position from the coefficient of variation distribution map.

[0100] It should be noted that the parameter gradient is obtained by calculating the difference between the parameters of a pixel and its surrounding pixels, thus obtaining the rate of change of the parameter in space.

[0101] 3.3) Based on the preset weight coefficient, the comprehensive variation intensity of each pixel is obtained through the comprehensive variation intensity calculation formula. The comprehensive variation intensity is calculated as follows: the comprehensive variation intensity is equal to the local variation coefficient multiplied by the first preset weight coefficient, plus the parameter gradient multiplied by the second preset weight coefficient, wherein the sum of the first preset weight coefficient and the second preset weight coefficient is 1. This application will not elaborate on this.

[0102] In this embodiment, the preset weighting coefficient can be set according to engineering requirements, and this application does not impose any restrictions on it.

[0103] 3.4) Based on the comprehensive variation intensity of all pixels, a global continuous variation intensity field is formed.

[0104] It should be noted that the global continuous variation intensity field is a data field formed by arranging the comprehensive variation intensity of all pixels according to their spatial positions.

[0105] The sub-step of adaptively adapting to obtain the dynamic kernel function parameters for each interpolation point based on the aforementioned variation intensity field specifically includes:

[0106] 4.1) Preset the initial range of kernel radius and weight decay coefficient, and establish the mapping rule between comprehensive variation intensity and kernel function parameters based on the measured data of the verification points.

[0107] In this embodiment, the initial value range is a parameter range set based on similar engineering experience. The measured data of the verification points are sample point data that have not participated in the preprocessing. By calculating the interpolation error of different parameter combinations, the correspondence between the comprehensive variation intensity and the kernel function parameters can be established, i.e., the mapping rule.

[0108] It should be noted that in the embodiments of this application, the sample points are collected or inverted from the aforementioned embodiments. The interpolation points and sample points together constitute all the location points of the variation intensity field. In the three-dimensional view, each location point of the variation intensity field corresponds to a pixel.

[0109] 4.2) For each interpolation point in the said variation intensity field, extract its comprehensive variation intensity.

[0110] It should be noted that extracting the comprehensive variation intensity involves obtaining the quantized value corresponding to each interpolation point from the variation intensity field, which will not be elaborated upon in this application.

[0111] 4.3) Based on the mapping rules and the comprehensive variation intensity of the interpolation point, determine the kernel radius and weight decay coefficient corresponding to the interpolation point to form a dynamic kernel function parameter field.

[0112] In this embodiment, the kernel function parameters are determined by assigning a corresponding kernel radius and weight decay coefficient to each interpolation point according to the mapping rules, thus forming a dynamic kernel function parameter field.

[0113] The sub-step of calculating the dynamic weights of sample points whose distance from each interpolation point is less than or equal to the kernel radius, based on the dynamic kernel function parameters of each interpolation point, specifically includes:

[0114] 5.1) For each interpolation point, search for all sample points within the range of its corresponding kernel radius.

[0115] It should be noted that the search sample points are only selected from those whose distance from the interpolation point is less than or equal to the kernel radius.

[0116] 5.2) Based on the spatial distance between the interpolation point and the sample point, the kernel radius of the interpolation point, and the weight decay coefficient, the weight of each sample point is obtained through dynamic weight calculation. The dynamic weight calculation method is as follows: the weight of each sample point is equal to the square of the distance between the sample point and the interpolation point divided by the product of the square of the kernel radius and the weight decay coefficient, the negative number is used to calculate the exponent value, and then the exponent value is multiplied by (1 and the sum of the comprehensive variation intensity), and the result is used as the numerator; the results of all sample points within the kernel radius calculated in the above way are summed, and the result is used as the denominator; the ratio of the numerator to the denominator is the weight of the sample point.

[0117] 5.3) Based on the weight of each sample point and the corresponding measured parameters, the parameter estimates of the interpolation points are obtained by weighted summation. All interpolation points are traversed to form a parameter spatial distribution map.

[0118] It should be noted that the weighted summation is to multiply the sample point weights by their measured parameters and then sum them to obtain the parameter estimates of the interpolation points; by traversing all interpolation points, the global parameter estimates are obtained, forming a parameter spatial distribution map.

[0119] In some embodiments, the sub-step of quantitatively calculating the risk level of landslides in the assessment area based on the spatial distribution map of the parameters and in conjunction with the slope stability analysis model specifically includes:

[0120] 6.1) Import the parameter values ​​from the parameter spatial distribution map into the slope stability analysis model and calculate the slope stability coefficient of each grid in the evaluation area.

[0121] In this embodiment, the imported parameter values ​​are obtained by importing parameters such as cohesion and internal friction angle of each grid in the parameter spatial distribution map into the slope stability analysis model. The stability coefficient is calculated by determining the stability of each grid using the model.

[0122] 6.2) A preset stability coefficient threshold range is set, and the risk level of each grid is divided based on the magnitude of the stability coefficient. The risk level includes low risk, medium risk and high risk.

[0123] It should be noted that the preset threshold range can be set based on engineering safety standards, and the risk level division can map the stability coefficient of each grid to a specific risk level.

[0124] 6.3) Based on the risk level of all grids, draw a landslide risk heat map of the assessment area to complete the quantitative risk assessment.

[0125] In this embodiment of the application, drawing a risk heat map can distinguish the risk level of each grid by color, forming an intuitive risk distribution map.

[0126] In some embodiments, the sub-step of data calibration after preliminary screening of the measured data and image inversion data specifically includes:

[0127] 1.3.1) Select a subset of sample points and simultaneously acquire the measured data and the corresponding image inversion data.

[0128] In this embodiment of the application, selecting a subset of sample points involves selecting a small number of representative points (covering different variation regions) from the sample points, and simultaneously acquiring data involves collecting measured data and image inversion data at these points at the same time.

[0129] 1.3.2) Establish a calibration model for the measured data and the image inversion data, and correct all the image inversion data through the calibration model.

[0130] It should be noted that establishing a calibration model involves using statistical analysis (such as linear regression) to establish the correspondence between measured data and image inversion data. Correcting the image inversion data involves using this model to adjust all image inversion data to make them closer to the accuracy of the measured data. This application will not elaborate on this, but commonly used calibration and correction methods can be used.

[0131] 1.3.3) The corrected image inversion data and the filtered measured data are fused together to obtain the final multi-source data.

[0132] In this embodiment of the application, the fused data is formed by integrating the corrected image inversion data with the filtered measured data to form the final multi-source data.

[0133] The specific sub-steps for result verification and optimization after obtaining the parameter spatial distribution map include:

[0134] 5.4) Select the verification points that did not participate in the interpolation and obtain the measured parameter values ​​of the verification points.

[0135] It should be noted that the verification points are sample points that did not participate in the previous interpolation.

[0136] 5.5) Calculate the error between the measured parameter values ​​of the verification point and the estimated parameter values ​​at the corresponding locations in the parameter spatial distribution map.

[0137] In this embodiment, the error is calculated by using a statistical formula (such as mean absolute error) to calculate the difference between the measured value and the estimated value of the verification point, which can be used to evaluate the accuracy of the interpolation result.

[0138] 5.6) If the error exceeds the preset acceptable range, adjust the mapping rules of the dynamic kernel function or the weighting coefficient of the comprehensive variation intensity, and return to step 4) to re-perform parameter adaptation and interpolation calculation until the error meets the requirements.

[0139] It should be noted that adjusting parameters is done when the error exceeds the limit. This involves optimizing the dynamic kernel function mapping rule or the comprehensive variation intensity weight coefficient, and recalculating the parameter adaptation step until the error meets the requirements. This ensures the accuracy of the interpolation results.

[0140] For example, a quantitative risk assessment was conducted on a slope of a construction waste dump in a suburban city. This waste dump was formed by the long-term accumulation of waste soil generated from recent building demolitions and road construction. The slope morphology is irregular due to natural settlement and artificial filling, with the slope height and top width varying with the region, covering a total area of ​​tens of thousands of square meters. The waste soil has a complex composition, mainly including gravel, clay, bricks, and concrete blocks, and its spatial distribution is extremely uneven, forming distinct zoning areas: rock accumulation zone (large-diameter solid waste is concentrated, mostly located in the upper and middle parts of the slope), crack development zone (due to rainwater infiltration and the slope's own weight, through or non-through cracks easily appear at the top edge and middle of the slope), and water-rich zone (low-lying areas at the toe of the slope and areas with higher groundwater levels, prone to water accumulation after rain, with soil moisture significantly higher than other areas). The area surrounding the spoil heap is dotted with suburban roads, small industrial parks, and a few residential areas. Historical monitoring has shown localized surface landslides, posing a potential threat to the safety of people and property in the surrounding area. Accurate risk assessment is urgently needed to guide the design of subsequent reinforcement and long-term monitoring plans.

[0141] Based on this, this application provides specific implementation scenarios; please refer to them. Figures 2 to 5 As shown, Figure 2 Displaying parameter heatmaps, Figure 3 The distribution of the coefficient of variation is shown. Figure 4 Showing the variation intensity field map, Figure 5 The spatial distribution map of the parameters is shown. It should be noted that... Figures 2 to 5 Specifically, the drawing is generated using GeoStudio software. Of course, other software, such as ANSYS / ABAQUS or Tecplot, can also be used in this application. Furthermore, the drawing tools in this application can also be custom-developed using Matlab or Python programming. This application does not impose any restrictions on this.

[0142] It should be noted that, among them Figure 3 Each coordinate axis in the figure represents a spatial coordinate axis. The x-axis represents the horizontal x-direction in the geodetic coordinate system, the y-axis represents the horizontal y-direction in the geodetic coordinate system, and the z-axis represents the vertical z-direction in the geodetic coordinate system. This application will not elaborate on these aspects.

[0143] It should be further explained that, Figures 2 to 5 Taking cohesion as an example, the same applies to other parameters, and this application will not elaborate on them.

[0144] The specific implementation scenarios of this application are described in detail below:

[0145] 1. Multi-source data acquisition

[0146] 1.1 Data Acquisition

[0147] Sampling Strategy: A composite sampling strategy combining grid-based sampling and densification in high-variability areas is adopted. First, based on the topography of the assessment area and the preliminary survey results, basic sample points are laid out according to the preset grid spacing to ensure uniform coverage of the entire area. Then, combined with special areas such as rock clusters, fractured areas, and water-rich areas identified by on-site reconnaissance, the grid spacing is reduced to increase the density of sampling and improve the representativeness of data in high-variability areas.

[0148] Test method:

[0149] Drilling and sampling: Vertical boreholes are drilled at pre-set sample points using professional drilling equipment, and soil samples are collected in layers at different depths to avoid structural disturbance or compositional changes during collection and transportation. The collected samples are sent to the laboratory, where direct shear tests are conducted using professional mechanical testing instruments to determine the cohesion, internal friction angle, and other core physical and mechanical parameters of the soil.

[0150] In-situ testing: In-situ testing is conducted at the sample points. The soil moisture content is directly obtained using a moisture meter, and the soil compaction degree is determined using a dynamic or static cone penetrometer. The testing process strictly follows relevant engineering specifications to ensure data accuracy.

[0151] Data types: Measured data include core parameters that directly reflect the mechanical properties and physical state of slag and soil, such as cohesion, internal friction angle, moisture content, and compaction degree.

[0152] 1.2 Image Inversion Data Acquisition

[0153] UAV Visible Light Imagery: High-definition imaging UAVs were selected, and aerial photography was conducted according to a preset flight route, altitude, and speed to ensure sufficient overlap between adjacent images for subsequent stitching. High-definition images of the spoil heap surface were acquired through aerial photography. Image processing algorithms were used to extract surface texture features, rock block contour information, and crack distribution characteristics, further quantifying parameters such as rock block distribution density and crack length density.

[0154] Ground-penetrating radar (GPR) images: Using specialized GPR equipment, survey lines are laid out along directions parallel and perpendicular to the slope direction to ensure coverage of the entire assessment area. The GPR emits high-frequency electromagnetic waves into the ground and receives reflected signals from interfaces between different media, forming radar images reflecting the internal structure of the spoil heap. Specialized data processing software is used to filter and denoise the radar images, and the dielectric constant of the underground media is retrieved. This parameter is significantly correlated with physical properties such as moisture content and density of the spoil and can be used as an auxiliary characterization parameter.

[0155] 1.3 Data Preprocessing

[0156] Outlier removal: Box plot method is used to filter measured data and image inversion data. By calculating the quartiles (lower quartile Q1, upper quartile Q3) and interquartile range (IQR = Q3 - Q1), a reasonable range for the data is set as [Q1 - 1.5IQR, Q3 + 1.5IQR]. Extreme data outside this range are identified as outliers and removed to avoid data distortion caused by sampling disturbances, equipment errors, environmental interference and other factors.

[0157] Data calibration: Select a subset of synchronous sample points (simultaneously acquiring measured data and corresponding image inversion data), and establish calibration models for the measured data and image inversion data based on statistical analysis methods. For example, linear or nonlinear calibration models are established for water content and dielectric constant. The measured data is used to correct the image inversion data, improving the quantitative accuracy of the image inversion data. Finally, the corrected image inversion data and the selected measured data are fused to form a complete multi-source dataset.

[0158] 2.1 Parameter Heat Map

[0159] Interpolation method: Ordinary Kriging interpolation is used. This method is based on geostatistical principles, fully considers the spatial correlation of parameters, and can effectively utilize discrete sample point data to generate a continuous spatial distribution of parameters. The core formula is:

[0160]

[0161] in:

[0162] : The parameter estimate at the interpolation point (x0, y0);

[0163] The weight coefficient of the i-th sample point reflects the contribution of the sample point to the parameter estimation of the interpolation point. It satisfies the constraint that the sum of the weights is 1, and its value is determined by the spatial correlation characteristics of the parameters.

[0164] : The measured parameter value of the i-th sample point;

[0165] n: The number of sample points involved in the parameter estimation of this interpolation point, determined based on the range of parameter spatial correlation.

[0166] Variation function model: A spherical model is selected to describe the spatial variation structure of the parameters. This model can fit the spatial correlation characteristics of most soil and rock parameters well. The formula is:

[0167]

[0168] in:

[0169] The semivariance when the lag distance is h is used to quantify the degree of parameter difference between sample points at different distances.

[0170] h: Spatial distance between two sample points;

[0171] The nugget value reflects parameter fluctuations caused by small-scale random variations.

[0172] C: Arch height, reflecting structural variations caused by systematic factors such as geological structure and material composition;

[0173] a: Range, which reflects the effective range of spatial correlation of parameters. When the distance between two sample points exceeds the range, there is no obvious spatial correlation between the parameters.

[0174] Visualization: The interpolated global parameter estimates are divided into several intervals according to their numerical values, and each interval is assigned a different color (for example, when visualizing, the color changes from blue to red as the values ​​increase from low to high). A parameter heatmap is then generated using plotting software.

[0175] 2.2 Distribution of Coefficient of Variation

[0176] Local coefficient of variation calculation: The evaluation area is divided into several uniform calculation units according to a preset grid, ensuring that each unit contains a sufficient number of sample points. For each calculation unit, the standard deviation and mean of the measured parameters within the unit are first calculated, and then the local coefficient of variation is calculated using the following formula:

[0177]

[0178] in:

[0179] Local coefficient of variation, dimensionless, is used to quantify the dispersion of parameters within a unit; the larger the value, the more significant the difference in parameters within the unit.

[0180] The standard deviation of the measured parameters within the calculation unit reflects the degree to which the parameter values ​​deviate from the mean.

[0181] : Calculate the mean of the measured parameters within the unit, reflecting the average level of the parameters within the unit.

[0182] Visualization: The local coefficient of variation of each calculation unit is divided into intervals according to its numerical value, and corresponding colors are assigned to generate a coefficient of variation distribution map, which intuitively presents the spatial distribution differences of parameter variation intensity within the evaluation area.

[0183] 3. Construction of the variation intensity field

[0184] It should be noted that the variation intensity field is a global continuous data field that integrates the distribution of absolute parameter values ​​and the distribution of variation intensity. The specific construction steps are as follows:

[0185] Step 1: Perform pixel-level coordinate alignment to ensure that pixels at the same spatial position in the two maps correspond one-to-one.

[0186] Step 2: Feature Extraction. For each aligned pixel, extract the parameter values ​​of that pixel and its surrounding pixels from the parameter heatmap, and calculate the parameter gradient using spatial derivatives.

[0187]

[0188] in:

[0189] The parameter gradient reflects the rate of change of the parameter in space. The larger the value, the more drastic the change of the parameter in the local range at that point, that is, the more obvious the local abrupt change characteristics.

[0190] : These are the partial derivatives of the parameter in the x-axis and y-axis directions, respectively, used to quantify the changing trend of the parameter in different directions.

[0191] At the same time, the local coefficient of variation corresponding to the pixel is directly extracted from the coefficient of variation distribution map, reflecting the overall variation intensity of the region where the pixel is located.

[0192] Step 3: Calculate the overall mutation intensity. Based on preset weighting coefficients, the local mutation coefficients and parameter gradients are weighted and summed to obtain the overall mutation intensity of each pixel:

[0193]

[0194] in:

[0195] I(x,y): The overall variation intensity of pixel (x,y), dimensionless, comprehensively reflecting the overall variation and local mutation characteristics of the region at that point;

[0196] The weighting coefficient of the local variation coefficient is used to adjust the contribution of the overall regional variation to the comprehensive variation intensity.

[0197] The weighting coefficients of the parameter gradient are used to adjust the contribution of local mutation features to the overall mutation intensity.

[0198] Must meet The specific value can be determined based on the variation characteristics of the evaluation area and engineering experience.

[0199] Step 4: Global Integration. Arrange and combine the overall variation intensity of all pixels according to their spatial coordinates to form a continuous data field covering the entire evaluation area, i.e., the variation intensity field. This data field can fully represent the variation characteristics of each spatial location.

[0200] 4. Dynamic kernel function parameter adaptation

[0201] In this embodiment, the dynamic kernel function parameters include the kernel radius and the weight decay coefficient, whose values ​​are dynamically adjusted according to the overall variation intensity of the interpolation points. The specific adaptation process is as follows:

[0202] Step 1: Establish mapping rules. First, based on similar engineering experience or pre-test results, preset the initial value range of the kernel radius and weight decay coefficient; then, select some validation points that did not participate in the previous data processing, and perform parameter interpolation on the validation points through different combinations of kernel function parameters, calculating the interpolation error under each combination; with the goal of minimizing the interpolation error, fit the mapping rules between the comprehensive variation intensity and the kernel function parameters:

[0203]

[0204]

[0205] in:

[0206] r: Kernel radius, used to control the range of sample points participating in the interpolation calculation, that is, only sample points whose distance from the interpolation point is less than the kernel radius are selected to participate in the calculation;

[0207] Weight decay coefficient, used to control the rate at which the weight of a sample point decays with distance;

[0208] I: Overall variation intensity of the interpolation point;

[0209] f(I), g(I): Mapping functions, determined based on the error fitting results. The higher the overall variation intensity, the smaller the kernel radius and the larger the weight decay coefficient; the lower the overall variation intensity, the larger the kernel radius and the smaller the weight decay coefficient.

[0210] Step 2: Point-by-point parameter allocation. Traverse each interpolation point in the mutation intensity field, extract the comprehensive mutation intensity I at that point, substitute it into the mapping rule described above, and calculate the kernel radius r and weight decay coefficient corresponding to that interpolation point. This ultimately forms a point-by-point customized dynamic kernel function parameter field.

[0211] 5. Dynamic weight interpolation and parameter correction

[0212] 5.1 Dynamic Weight Calculation

[0213] For each interpolation point, the weights of all sample points within the kernel radius are calculated based on the corresponding dynamic kernel function parameters, using the following formula:

[0214]

[0215] in:

[0216] : The dynamic weight of the i-th sample point relative to the current interpolation point;

[0217] : Spatial distance between the i-th sample point and the interpolation point;

[0218] r: Kernel radius of the current interpolation point;

[0219] : Weight decay coefficient of the current interpolation point;

[0220] I: The overall variation intensity of the current interpolation point, (1+I) is the variation intensity enhancement term, which is used to further enhance the weight contribution of sample points in high variation regions;

[0221] m: The total number of sample points participating in the interpolation calculation within the kernel radius.

[0222] 5.2 Initial Parameter Estimation

[0223] Based on the calculated dynamic weights and the measured parameters of the sample points, the initial parameter estimates of the interpolation points are obtained by weighted summation:

[0224]

[0225] in:

[0226] : Initial parameter estimates for the interpolation point (x0, y0);

[0227] The meanings of the other parameters are the same as those described above.

[0228] 5.4 Generation of Parameter Spatial Distribution Map

[0229] Traverse all interpolation points within the evaluation area, collect the final parameter estimates for each point, and integrate them according to spatial coordinates to form a global parameter dataset; process the dataset using professional visualization software to generate a parameter spatial distribution map.

[0230] 6. Slope stability calculation and risk assessment

[0231] 6.1 Stability Calculation Model

[0232] This application uses a simplified Bishop method to calculate the slope stability coefficient, which is applicable to slopes made of granular materials such as spoil heaps. The formula is:

[0233]

[0234] in:

[0235] The slope stability coefficient is a dimensionless indicator that is the core indicator for evaluating the stability of a slope. It needs to be obtained through iterative solution. The iteration terminates when the error between the two calculated stability coefficients is less than a preset threshold.

[0236] n: The number of strips into which the sliding body is divided along the sliding direction. The division of strips needs to take into account both calculation accuracy and efficiency.

[0237] : The cohesion corresponding to the i-th strip is taken from the parameter space distribution map;

[0238] : The length of the sliding surface of the i-th strip;

[0239] The self-weight of the i-th block is calculated from the block volume and the unit weight of the slag.

[0240] The pore water pressure on the sliding surface of the i-th block is related to the soil moisture content; the higher the moisture content, the greater the pore water pressure.

[0241] The internal friction angle corresponding to the i-th strip is taken from the parameter space distribution map;

[0242] : The inclination angle of the sliding surface of the i-th block.

[0243] Of course, it should be noted that the stability calculation model in this application embodiment is based on the characteristics of the slag and soil scenario. Other models can be used in other embodiments. This application uses the Bishop method to calculate the slope stability coefficient because the Bishop method is more suitable for the physical characteristics of granular material slopes. For other slag and soil environments, other more suitable stability calculation models can also be used. This application does not limit or elaborate on this.

[0244] In the embodiments of this application, such as Figure 6 As shown, a quantitative assessment system for landslide risk of construction waste that considers the spatial variability of construction waste includes:

[0245] Data acquisition module 11 is used to collect multi-source data of the construction waste in the assessment area. The multi-source data includes measured data reflecting the physical and mechanical properties of the construction waste and image inversion data that helps characterize the distribution characteristics of the construction waste.

[0246] The dual-map drawing module 12 is connected to the data acquisition module and is used to draw a parameter heat map and a coefficient of variation distribution map of the evaluation area based on the multi-source data.

[0247] The variation intensity field construction module 13 is connected to the dual-map drawing module and is used to fuse the parameter heat map and the variation coefficient distribution map to construct a continuous variation intensity field over the entire domain.

[0248] The dynamic kernel function adaptation module 14 is connected to the mutation intensity field construction module and is used to adaptively adapt to obtain the dynamic kernel function parameters of each interpolation point based on the mutation intensity field. The dynamic kernel function parameters include the kernel radius and the weight decay coefficient.

[0249] The dynamic interpolation module 15 is connected to the data acquisition module and the dynamic kernel function adaptation module respectively. It is used to calculate the dynamic weight of sample points that are less than or equal to the kernel radius at each interpolation point based on the dynamic kernel function parameters of each interpolation point. Combined with the measured parameters in the multi-source data, it completes the spatial interpolation of the physical and mechanical parameters of the entire waste soil and obtains the parameter spatial distribution map.

[0250] The risk assessment module 16, connected to the dynamic interpolation module, is used to quantitatively calculate the risk level of slag landslide in the assessment area based on the parameter spatial distribution map and the slope stability analysis model, and to complete the risk assessment.

[0251] Since the principle by which this system solves the problem is similar to the methods and systems described above, the implementation of this system can be found in the implementation of the methods and systems described above, and will not be repeated here.

[0252] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for quantitatively evaluating a risk of a soil landslide considering spatial variability of a soil, characterized by, Includes the following steps: Step 1): Collect multi-source data of the construction waste in the assessment area. The multi-source data includes measured data reflecting the physical and mechanical properties of the construction waste and image inversion data that helps characterize the distribution characteristics of the construction waste. Step 2): Based on the multi-source data, draw a parameter heatmap and a coefficient of variation distribution map of the evaluation area; Step 3): Integrate the parameter heatmap and the coefficient of variation distribution map to construct a global continuous variation intensity field; The step of fusing the parameter heatmap and the coefficient of variation distribution map to construct a global continuous variation intensity field includes the following sub-steps: 3.1) Unify the spatial resolution of the parameter heatmap and the coefficient of variation distribution map, and perform pixel-level alignment and fusion; 3.2) For each pixel, extract the parameter gradient at the corresponding location from the parameter heatmap, and extract the local coefficient of variation at the corresponding location from the coefficient of variation distribution map; 3.3) Based on the preset weight coefficient, the comprehensive variation intensity of each pixel is obtained through the comprehensive variation intensity calculation formula. The comprehensive variation intensity is calculated as follows: the comprehensive variation intensity is equal to the local variation coefficient multiplied by the first preset weight coefficient, plus the parameter gradient multiplied by the second preset weight coefficient, where the sum of the first preset weight coefficient and the second preset weight coefficient is 1. 3.4) Based on the comprehensive variation intensity of all pixels, a global continuous variation intensity field is formed; Step 4): Based on the variation intensity field, adaptively adapt and obtain the dynamic kernel function parameters for each interpolation point. The dynamic kernel function parameters include the kernel radius and the weight decay coefficient. Step 5): Based on the dynamic kernel function parameters of each interpolation point, calculate the dynamic weights of sample points that are less than or equal to the kernel radius at each interpolation point. Combined with the measured parameters in the multi-source data, complete the spatial interpolation of the physical and mechanical parameters of the entire waste soil and obtain the parameter spatial distribution map. Step 6): Based on the spatial distribution map of the parameters and combined with the slope stability analysis model, quantitatively calculate the risk level of slag and soil landslide in the assessment area to complete the risk assessment.

2. The method of claim 1, wherein, Step 1) involves collecting multi-source data on the waste soil within the assessment area, including the following sub-steps: 1.1) Based on the preliminary survey results of the assessment area, sample points were set up, and measured data of cohesion, internal friction angle, moisture content and compaction degree of the slag were obtained through borehole sampling and in-situ testing; 1.2) Collect UAV visible light images and ground-penetrating radar images of the assessment area, and obtain the rock block distribution, crack density, and dielectric constant correlation data of the slag soil through image feature extraction and inversion, which are used as the image inversion data; 1.3) Perform preliminary screening on the measured data and image inversion data to remove obvious outliers and obtain purified multi-source data.

3. The method of claim 1, wherein, In step 2), based on the multi-source data, a parameter heatmap and a coefficient of variation distribution map of the evaluation area are drawn, including the following sub-steps: 2.1) Extract the measured parameters from the multi-source data, perform preliminary interpolation using ordinary kriging interpolation, and color the parameters according to their numerical ranges to obtain a parameter heatmap that reflects the spatial distribution of the absolute values ​​of the parameters. 2.2) Divide the evaluation area according to the preset grid, calculate the standard deviation and mean of the measured parameters in each grid, and obtain the local coefficient of variation for each grid based on the standard deviation and mean; 2.3) Color the data according to the numerical range of the local coefficient of variation to obtain a coefficient of variation distribution map that reflects the spatial distribution of the parameter variation intensity.

4. The method according to claim 1, characterized in that, In step 4), based on the variation intensity field, the dynamic kernel function parameters for each interpolation point are adaptively obtained, including the following sub-steps: 4.1) Preset the initial range of kernel radius and weight decay coefficient, and establish a mapping rule between comprehensive variation intensity and kernel function parameters based on the measured data of the verification points; 4.2) For each interpolation point in the aforementioned variation intensity field, extract its comprehensive variation intensity; 4.3) Based on the mapping rules and the comprehensive variation intensity of the interpolation point, determine the kernel radius and weight decay coefficient corresponding to the interpolation point to form a point-by-point customized dynamic kernel function parameter field.

5. The method according to claim 1, characterized in that, In step 5), based on the dynamic kernel function parameters of each interpolation point, the dynamic weights of sample points whose distance from each interpolation point is less than or equal to the kernel radius are calculated, including the following sub-steps: 5.1) For each interpolation point, search for all sample points within the corresponding kernel radius; 5.2) Based on the spatial distance between the interpolation point and the sample point, the kernel radius of the interpolation point, and the weight decay coefficient, the weight of each sample point is obtained through dynamic weight calculation. The dynamic weight calculation method is as follows: the weight of each sample point is equal to the square of the distance between the sample point and the interpolation point divided by the product of the square of the kernel radius and the weight decay coefficient, the negative number is used to calculate the exponent value, and then the exponent value is multiplied by (1 and the sum of the comprehensive variation intensity), and the result is used as the numerator; the results of all sample points within the kernel radius calculated in the above way are summed, and the result is used as the denominator; the ratio of the numerator to the denominator is the weight of the sample point. 5.3) Based on the weight of each sample point and the corresponding measured parameters, the parameter estimates of the interpolation points are obtained by weighted summation. All interpolation points are traversed to form a parameter spatial distribution map.

6. The method according to claim 1, characterized in that, In step 6), based on the spatial distribution map of the parameters and combined with the slope stability analysis model, the risk level of landslides in the assessment area is quantitatively calculated, including the following sub-steps: 6.1) Import the parameter values ​​from the spatial distribution map into the slope stability analysis model and calculate the slope stability coefficient of each grid in the evaluation area; 6.2) A preset stability coefficient threshold range is established, and based on the magnitude of the stability coefficient, the risk level of each grid is divided into low risk, medium risk, and high risk. 6.3) Based on the risk level of all grids, draw a landslide risk heat map of the assessment area to complete the quantitative risk assessment.

7. The method according to claim 2, characterized in that, In step 1.3), after preliminary screening of the measured data and image inversion data, a data calibration sub-step is also included: 1.3.1) Select a subset of sample points and simultaneously acquire the measured data and the corresponding image inversion data; 1.3.2) Establish a calibration model for the measured data and the image inversion data, and use the calibration model to correct all the image inversion data; 1.3.3) The corrected image inversion data and the filtered measured data are fused together to obtain the final multi-source data.

8. The method according to claim 1, characterized in that, Step 5) After obtaining the spatial distribution map of the parameters, the process also includes sub-steps for result verification and optimization: 5.4) Select verification points that did not participate in interpolation and obtain the measured parameter values ​​of the verification points; 5.5) Calculate the error between the measured parameter values ​​of the verification point and the estimated parameter values ​​at the corresponding locations in the parameter spatial distribution map; 5.6) If the error exceeds the preset acceptable range, adjust the mapping rules of the dynamic kernel function or the weighting coefficient of the comprehensive variation intensity, and return to step 4) to re-perform parameter adaptation and interpolation calculation until the error meets the requirements.

9. A quantitative assessment system for landslide risk of construction waste considering the spatial variability of construction waste, characterized in that, include: The data acquisition module is used to collect multi-source data of the construction waste in the assessment area. The multi-source data includes measured data reflecting the physical and mechanical properties of the construction waste and image inversion data that helps characterize the distribution characteristics of the construction waste. The dual-map plotting module, connected to the data acquisition module, is used to plot the parameter heatmap and the coefficient of variation distribution map of the evaluation area based on the multi-source data. The variation intensity field construction module is connected to the dual-map drawing module and is used to fuse the parameter heat map and the variation coefficient distribution map to construct a global continuous variation intensity field. The step of fusing the parameter heatmap and the coefficient of variation distribution map to construct a global continuous variation intensity field includes the following sub-steps: 3.1) Unify the spatial resolution of the parameter heatmap and the coefficient of variation distribution map, and perform pixel-level alignment and fusion; 3.2) For each pixel, extract the parameter gradient at the corresponding location from the parameter heatmap, and extract the local coefficient of variation at the corresponding location from the coefficient of variation distribution map; 3.3) Based on the preset weight coefficient, the comprehensive variation intensity of each pixel is obtained through the comprehensive variation intensity calculation formula. The comprehensive variation intensity is calculated as follows: the comprehensive variation intensity is equal to the local variation coefficient multiplied by the first preset weight coefficient, plus the parameter gradient multiplied by the second preset weight coefficient, where the sum of the first preset weight coefficient and the second preset weight coefficient is 1. 3.4) Based on the comprehensive variation intensity of all pixels, a global continuous variation intensity field is formed; A dynamic kernel function adaptation module is connected to the mutation intensity field construction module. It is used to adaptively adapt the dynamic kernel function parameters of each interpolation point based on the mutation intensity field. The dynamic kernel function parameters include the kernel radius and the weight decay coefficient. The dynamic interpolation module is connected to the data acquisition module and the dynamic kernel function adaptation module respectively. It is used to calculate the dynamic weight of sample points that are less than or equal to the kernel radius from each interpolation point based on the dynamic kernel function parameters of each interpolation point. Combined with the measured parameters in the multi-source data, it completes the spatial interpolation of the physical and mechanical parameters of the entire waste soil and obtains the parameter spatial distribution map. The risk assessment module, connected to the dynamic interpolation module, is used to quantitatively calculate the risk level of slag landslide in the assessment area based on the parameter spatial distribution map and the slope stability analysis model, thereby completing the risk assessment.

Images