A method for identifying geological factors of ground subsidence risk

Abstract

This invention relates to the field of engineering geological exploration technology, and in particular to a method for identifying geological factors of ground subsidence risk. It utilizes nested spatial data mining to obtain the spatial correlation between ground subsidence and geological information, and identifies the ranking of geological information indicators affecting ground subsidence. A natural discontinuity grading method is used to classify the level of ground subsidence risk based on the ranking of geological information indicators. A Bayesian algorithm is employed to divide the range of values ​​for geological factor indicators at different ground subsidence risk levels. The advantages of this invention are: (1) It uses measured data to mine risk factors and their level ranges, accurately assessing ground subsidence risk; (2) It does not determine risk levels and indicator ranges based on expert knowledge and prior assumptions, thus avoiding the decrease in accuracy caused by human judgment and differences between experience and actual engineering; (3) This invention advances risk factor identification from qualitative classification to quantitative identification.

Description

Technical Field

[0001] This invention relates to the field of engineering geological exploration technology, and in particular to a method for identifying geological factors of ground subsidence risk. Background Technology

[0002] Land subsidence is a common environmental geological hazard risk, often occurring in densely populated deltaic areas. The causes of regional land subsidence are mainly twofold: natural and anthropogenic factors. Natural factors primarily refer to the movement and secondary consolidation of geological strata, while anthropogenic factors mainly refer to excessive groundwater extraction and engineering construction.

[0003] Conducting risk assessments of land subsidence hazards can provide effective early warning of disaster risks. The general risk assessment process includes risk identification, selection of evaluation factors, risk value calculation, and risk zoning. Risk identification, based on mechanistic analysis, identifies all factors that may pose a risk and selects evaluation factors accordingly. Traditional risk identification methods typically rely on expert knowledge to determine the weights of each evaluation factor, such as expert scoring, analytic hierarchy process (AHP), fuzzy comprehensive evaluation, and statistical analysis. However, insufficient expert knowledge and discrepancies between experience and real-world engineering projects lead to subjectivity and a lack of specificity in the selection of evaluation factors. Therefore, traditional methods can only provide qualitative zoning descriptions (high-risk, medium-risk, low-risk areas) and lack quantitative analysis.

[0004] Patent document CN104133996A discloses a method for assessing land subsidence risk levels based on cloud models and data fields. It uses a cloud model algorithm to statistically analyze the distribution characteristics of PS points in PS-InSAR technology, and then uses these statistical characteristics to cluster indicators such as subsidence data, population density, and economic density in the data field to obtain the land subsidence risk level. However, this invention only addresses risk factors such as population density and economic density—factors with indirect influencing factors—to classify subsidence levels. It does not address risk factors with direct influencing factors such as geological factors, nor does it define risk factor levels or their indicator ranges.

[0005] Patent document CN110362866A discloses a method for land subsidence zoning based on different geological conditions. Based on screening characteristic parameters of geological indicators within boreholes, it determines the weight ratio of these characteristic parameters, indicator normalization, and a comprehensive evaluation index. The method then zons potential land subsidence zones according to the comprehensive evaluation index. This invention is a subsidence zoning prediction method that does not involve risk issues. The prediction is based on theoretical concepts rather than establishing correlations through actual measurements; instead, it employs a single indicator system of comprehensive evaluation indicators.

[0006] Patent document CN112070129 A discloses a method, device, and system for identifying land subsidence risk. Based on multiple SAR images from various time periods, a dataset containing land subsidence information and evaluation factor information is obtained. The land subsidence information indicates that the subsidence rate exceeds a preset subsidence threshold, while the evaluation factor information indicates information about each preset land subsidence hazard evaluation factor. The weights of each preset land subsidence hazard evaluation factor are calculated using a preset fuzzy hierarchical analysis. Based on Bayes' theorem and the dataset, the posterior probability of each preset land subsidence hazard evaluation factor is calculated. Based on the weights and posterior probabilities of each preset land subsidence hazard evaluation factor, the land subsidence risk probability is calculated, and then the land subsidence risk is identified based on the land subsidence risk probability. However, this invention is still a single-factor risk assessment method, and the assessment system still uses traditional prior-based pre-defined risk levels, and cannot classify risk levels based on measured data. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by providing a method for identifying geological factors contributing to land subsidence risk. This method utilizes nested spatial data mining to obtain the spatial correlation between land subsidence and geological information, identifying and ranking geological information indicators that influence land subsidence. It employs a natural discontinuity grading method to classify land subsidence risk levels based on the ranking of these geological information indicators. Furthermore, it uses a Bayesian algorithm to define the value ranges of geological factor indicators for different land subsidence risk levels. This invention achieves a method for identifying geological factors associated with land subsidence risk and determining the range of indicator levels based on geological survey results and measured land subsidence data.

[0008] The objective of this invention is achieved through the following technical solutions:

[0009] A method for identifying geological factors of land subsidence risk, characterized by the following steps:

[0010] S1. Obtain ground subsidence data and maps of the study area;

[0011] S2. Register the ground subsidence data map of the study area with the corresponding latitude and longitude coordinates, and use cluster analysis to classify the ground subsidence data into levels based on the time domain and spatial domain to obtain a time-series zoning map of ground subsidence levels.

[0012] S3. Collect geological borehole exploration data of the study area, match borehole location information with corresponding latitude and longitude coordinates, and obtain geological information and corresponding indicators of strata at different depths of each borehole in the study area.

[0013] S4. According to the soil classification standard, the geological information indicators of the same stratum depth and the same soil type are weighted and averaged to divide the three-dimensional distribution map of geological information indicators into planar spatial domain and depth spatial domain.

[0014] S5. Perform coordinate registration between the ground settlement level time series zoning map and the geological information index three-dimensional distribution map, that is, register the two in the same coordinate system;

[0015] S6. Using a spatial correlation algorithm, obtain the correlation ranking of soil types and corresponding geological indicators related to ground settlement in three-dimensional space;

[0016] S7. Using the natural discontinuity grading method, the correlation and ranking relationship between soil type indicators and geological indicators is graded, and the soil type indicators and geological indicators from the first level, the second level, the third level to the nth level are found as indicators corresponding to different risk levels.

[0017] S8. Using the first-level soil type index and geological index as geological factors for ground subsidence risk, the Bayesian algorithm is used to analyze and calculate the index range corresponding to different risk levels of soil type and geological factors for ground subsidence risk.

[0018] The ground subsidence data of the study area is obtained by acquiring SAR image data of the study area within a certain time period and performing InSAR calculation of the ground deformation rate of the study area based on the SAR image data.

[0019] The SAR image data refers to SAR image data of the study area over a period of more than 20 months.

[0020] The registration coordinate system is the WGS84 coordinate system.

[0021] The association ranking of soil types and corresponding geological indicators associated with ground subsidence in the three-dimensional space is achieved by: carrying out nested spatial association data mining in two directions, namely, in the planar spatial domain and the depth spatial domain. Specifically, in the depth spatial domain, the association between ground subsidence data and soil type indicators is mined, and in the planar spatial domain, the association between ground subsidence data and geological indicators is mined.

[0022] The soil type indicators include silty clay, silt, silty clay, fine sand, and silty sand.

[0023] The geological parameters include numerically weighted average water content, void ratio, compressibility coefficient, compressibility modulus, wet density, and aquifer level.

[0024] The advantages of this invention are:

[0025] (1) Based on measured data, risk factors and their level ranges are mined to accurately assess the risk of ground subsidence;

[0026] (2) The risk level and index range are not determined according to expert knowledge and prior assumptions, thereby avoiding the decrease in accuracy caused by human judgment and the difference between experience and actual engineering.

[0027] (3) This invention advances the identification of risk factors from qualitative classification to quantitative identification. Attached Figure Description

[0028] Figure 1 This is a flowchart of the method of the present invention;

[0029] Figure 2 This is a diagram illustrating the association rule mining steps in this invention. Detailed Implementation

[0030] The features and other related features of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments, so as to facilitate understanding by those skilled in the art:

[0031] Example: Figure 1 As shown, the method for identifying geological factors of ground subsidence risk in this embodiment is a ground subsidence risk assessment that combines spatial rules and Bayesian algorithms, and includes the following steps:

[0032] (1) Obtain SAR imagery data for more than 20 months within the study area, perform InSAR calculations on the ground deformation rate within the study area, and obtain ground subsidence data for the study area. The study area refers to the region where the ground subsidence risk needs to be studied.

[0033] In this embodiment, more than 20 SAR image data are required (generally one per month) to ensure the accuracy of the observation. The influencing data are images of the same study area at different times. Then, InSAR technology is used to process and calculate the ground deformation rate of the study area, generate a ground deformation image, and obtain the corresponding temporal changes.

[0034] (2) The ground subsidence data map obtained in step (1) is registered with the corresponding latitude and longitude coordinates. The cluster analysis method is used to classify the ground subsidence data into levels based on the time domain and spatial domain, and a time-series zoning map of ground subsidence levels is obtained.

[0035] To study the deformation and latitude-longitude coordinate registration of the study area, it is necessary to obtain geographic images in the same coordinate system as those calculated by InSAR technology. The grading method can be either the equal-interval grading method or the natural discontinuity grading method, that is, to use clustering to divide different areas into subsidence zones. This step can be achieved in ArcGIS through different grading methods.

[0036] (3) Collect geological borehole exploration data in the study area, match the borehole location information with the corresponding latitude and longitude coordinates, and obtain the geological information (soil type) and corresponding indicators (water content, void ratio, compression coefficient, compression modulus, wet density, aquifer, etc.) of the strata at different depths in each borehole in the study area.

[0037] Geological and stratigraphic information for the study area can be obtained from existing exploration data, while physical and mechanical indicators can be obtained from indoor geotechnical experiments. Then, all physical indicators that may affect the ground settlement rate and their values ​​are listed.

[0038] (4) Based on the soil classification standard in the Code for Geotechnical Investigation GB 50021-2001 (2009 Edition), the geological information indicators of the same stratum depth and the same soil type are weighted and averaged to divide the three-dimensional distribution map of geological information indicators in the planar spatial domain and the depth spatial domain.

[0039] The formula for the weighted average method is:

[0040] In the formula, x represents the average value of geological information indicators; k f represents the thickness of strata with the same soil type in each layer; k is the corresponding value of the geological information index; n is the number of samples.

[0041] The depth spatial domain contains geological information (soil type, such as silty clay, silt, fine sand, silty sand) of different strata at various borehole depths. The planar spatial domain contains corresponding indicators of soil type (moisture content, void ratio, compression coefficient, compression modulus, wet density, aquifer, etc.). The data is then uniformly imported into a GIS database to create a three-dimensional distribution map of geological information indicators.

[0042] (5) The time-series zoning map of ground settlement level is registered with the three-dimensional distribution map of geological information indicators. The registered coordinate system is the WGS84 coordinate system.

[0043] A coordinate system is an essential component of spatial data, and a complete spatial data transformation should also include coordinate system transformation. Settlement information distribution maps and physical index information distribution maps need to be in the same coordinate system, which can be achieved through coordinate system transformation. Here, the unified WGS84 coordinate system is chosen for coordinate registration of the two types of data.

[0044] (6) Using a spatial correlation algorithm, nested spatial correlation data mining is carried out in two directions: planar spatial domain and depth spatial domain. In the depth spatial domain, the correlation between ground settlement data and soil type indicators (such as silty clay, silt, fine sand, and silty sand) is mined. In the planar spatial domain, the correlation between ground settlement data and geological indicators (numerical weighted average water content, void ratio, compression coefficient, compression modulus, wet density, and aquifer) is mined. The correlation ranking of soil types and corresponding geological indicators associated with ground settlement in three-dimensional space is found.

[0045] The task of association rule mining is to find strong association rules in a database that have the minimum support and minimum confidence given by the user. The itemset corresponding to a strong association rule must be a frequent itemset, and the confidence of the association rule derived from the frequent itemset is calculated from the support of the frequent itemset. Therefore, association rule mining can be divided into two steps: First, search the dataset for all frequent itemsets that satisfy the minimum support threshold, and then generate association rules that satisfy the minimum confidence threshold from all frequent itemsets. This first step is the standard for evaluating the effectiveness of an association rule mining algorithm and is the core issue in the entire association rule mining technology. The second step is an enumeration exploration process for generating strong rules from frequent itemsets. Specifically, for each element in a frequent itemset (i.e., a certain frequent itemset in the dataset), generate all its non-empty subsets. For each non-empty subset in the dataset, output the strong rule. The association rule mining steps are as follows: Figure 2 As shown.

[0046] In the planar spatial domain, the ranking of indicators is: compressibility coefficient > moisture content > water level > compression modulus > void ratio > wet density. In the depth spatial domain, the ranking of indicators is: silty clay > silty clay > clay > silt > fine sand > silty sand. Among all indicators, the soil's compressibility coefficient has the strongest correlation with the magnitude of settlement.

[0047] (7) The natural discontinuity grading method is used to classify the correlation between soil type indicators and geological indicators, and find the first level, second level, third level, etc. of soil type indicators and geological indicators as indicators corresponding to different risk levels.

[0048] The soil is classified into two levels: one for planar spatial domain and one for depth spatial domain. The primary indicators are: compressibility coefficient, water content, silty clay, and silty clay; the secondary indicators are: compression modulus, water level, clay, and silt; and the tertiary indicators are: void ratio, wet density, fine sand, and silt. These indicators are then classified using the natural discontinuity method, a statistical method based on the statistical distribution of numerical values, which maximizes the differences between classes.

[0049] (8) The first-level soil type index and geological index are used as geological factors for ground subsidence risk. The Bayesian algorithm is used to analyze and calculate the index range corresponding to different risk levels of soil type and geological factor for ground subsidence risk.

[0050] The basic idea of ​​this embodiment is as follows: assuming the research area is divided into several sub-regions, if the sum of the variances of the sub-regions is less than the total variance of the region, then spatial heterogeneity exists; if the spatial distributions of two variables tend to be consistent, then there is a statistical correlation between them. The q-statistic is a set of statistical methods used to measure spatial heterogeneity, detect explanatory factors, analyze interactions between variables, and reveal the underlying driving forces. Its core idea is based on the assumption that if an independent variable has a significant impact on a dependent variable, then the spatial distributions of the independent and dependent variables should be similar.

[0051] Differentiation and Factor Detection: Used to detect the spatial differentiation of land subsidence and the magnitude of the influence of various evaluation indicators on subsidence changes. Its metric is q, and the corresponding relationship is:

[0052]

[0053] In the formula, L represents the stratification of variable Y or factor X; N, N h SSW represents the number of cells in the region and layer h; SSW represents the intra-layer variance; SST represents the total variance of the entire region; σ h 2 and σ 2 , respectively, represent the variances of the Y values ​​for layer h and the entire region. In the formula, q represents the influence of the evaluation index on the magnitude of settlement, ranging from 0 to 1. A larger value indicates a stronger influence of the evaluation index; conversely, a smaller value indicates a weaker influence.

[0054] In this embodiment, the compressibility coefficient (q) has the largest value among the planar spatial domain indices, at 0.6898. Therefore, the compressibility coefficient is the most significant factor affecting ground settlement. Relatively speaking, the q values ​​of moisture content and compressibility modulus are quite similar and contribute significantly, while the influence of wet density is 0.3857, indicating that it has a certain impact on ground settlement. Among the depth spatial domain indices, silty clay has the largest q value, at 0.5963.

[0055] Risk detection: Used to determine whether there is a significant difference in the attribute mean values ​​in the classification regions of two factors. It can be used to search for regions with larger values ​​of the research target and is tested using the t-statistic.

[0056]

[0057] In the formula, n is the mean of the attributes within subregion h; h is the number of samples in subregion h; Var is the variance.

[0058] In the assessment of the impact of primary driving factors obtained from risk detection on the magnitude of settlement, the soil compressibility coefficient corresponding to areas with severe ground settlement was 0.58-0.66 MPa. -1 The average ground settlement was 17.398 mm for soils with a moisture content of 34%-38%; the average ground settlement was 17.414 mm for soils with a moisture content of 60.56%; the average ground settlement was 17.880 mm for soils with a moisture content of 30.16%; and the average ground settlement was 17.483 mm for soils with a moisture content of 34%-38%.

[0059] The range of indicators corresponding to different risk levels of soil types and geological factors for ground subsidence risk was compiled.

[0060] Although the above embodiments have described the concept and embodiments of the present invention in detail with reference to the accompanying drawings, those skilled in the art will recognize that various improvements and modifications can still be made to the present invention without departing from the scope of the claims, and therefore will not be elaborated here.

Claims

1. A method for identifying geological factors of ground subsidence risk, characterized in that: The method includes the following steps: S1. Obtain ground subsidence data and maps of the study area; S2. Register the ground subsidence data map of the study area with the corresponding latitude and longitude coordinates, and use cluster analysis to classify the ground subsidence data into levels based on the time domain and spatial domain to obtain a time-series zoning map of ground subsidence levels. S3. Collect geological borehole exploration data of the study area, match borehole location information with corresponding latitude and longitude coordinates, and obtain geological information and corresponding indicators of strata at different depths of each borehole in the study area. S4. According to the soil classification standard, the geological information indicators of the same stratum depth and the same soil type are weighted and averaged to divide the three-dimensional distribution map of geological information indicators into planar spatial domain and depth spatial domain. S5. Perform coordinate registration between the ground settlement level time series zoning map and the geological information index three-dimensional distribution map, that is, register the two in the same coordinate system; S6. Using a spatial association algorithm, obtain the association ranking of soil types and corresponding geological indicators associated with ground subsidence in three-dimensional space. The association ranking of soil types and corresponding geological indicators associated with ground subsidence in three-dimensional space is achieved by: carrying out nested spatial association data mining in two directions, planar spatial domain and depth spatial domain. In the depth spatial domain, data mining is performed to explore the association between ground subsidence data and soil type indicators, and in the planar spatial domain, data mining is performed to explore the association between ground subsidence data and geological indicators. S7. Using the natural discontinuity grading method, the correlation and ranking relationship between soil type indicators and geological indicators is graded, and the soil type indicators and geological indicators from the first level, the second level, the third level to the nth level are found as indicators corresponding to different risk levels. S8. Using the first-level soil type index and geological index as geological factors for ground subsidence risk, the Bayesian algorithm is used to analyze and calculate the index range corresponding to different risk levels of soil type and geological factor for ground subsidence risk.

2. The method for identifying geological factors of ground subsidence risk according to claim 1, characterized in that: The ground subsidence data of the study area is obtained by acquiring SAR image data of the study area within a certain time period and performing InSAR calculation of the ground deformation rate of the study area based on the SAR image data.

3. The method for identifying geological factors of ground subsidence risk according to claim 2, characterized in that: The SAR image data refers to SAR image data of the study area over a period of more than 20 months.

4. The method for identifying geological factors of ground subsidence risk according to claim 1, characterized in that: The registration coordinate system is the WGS84 coordinate system.

5. The method for identifying geological factors of ground subsidence risk according to claim 1, characterized in that: The soil type indicators include silty clay, silt, silty clay, fine sand, and silty sand.

6. The method for identifying geological factors of ground subsidence risk according to claim 1, characterized in that: The geological parameters include numerically weighted average water content, void ratio, compressibility coefficient, compressibility modulus, wet density, and aquifer level.

Images