County geological disaster risk analysis method and system

Abstract

The invention relates to the technical field of geological disaster analysis, and discloses a county geological disaster risk analysis method and system, and the method comprises the steps: dividing slope units of a geological disaster region, screening susceptibility analysis factors in the slope units, and selecting a training set and a test set from the geological disaster region and a non-geological disaster region; establishing an initial analysis model of the training set, performing hyper-parameter optimization on the initial analysis model, outputting a susceptibility classification result of the training set by using the optimized analysis model, analyzing the classification precision of the susceptibility classification result, and determining a susceptibility analysis result of the slope unit; calculating a risk index of the slope unit, and dividing risk subareas in the geological disaster area by using the risk index; screening vulnerability analysis factors in the slope units, and performing hierarchical analysis on the vulnerability analysis factors by using the judgment matrix; and calculating the vulnerability index of the slope unit. The real geological disaster risk condition is matched by improving the accuracy of the geological disaster analysis result.

Description

Technical Field

[0001] The invention relates to a method and system for analyzing the risk of geological disasters in a county, and belongs to the technical field of geological disaster analysis. Background Art

[0002] Geological disasters usually refer to events caused by a series of geological processes that pose a serious threat to human life, property, nature and ecology. Geological disasters including landslides, collapses, mudslides, etc. occur frequently in mountainous areas and are extremely destructive. Based on the understanding of the laws of past geological disasters, conducting geological disaster risk analysis can provide theoretical support and scientific basis for regional disaster prevention and mitigation. Geological disaster risk analysis is a quantitative analysis and analysis of the possibility of risk areas suffering from geological disasters of different intensities and the possible disaster losses caused. It includes three levels of analysis: susceptibility, hazard and vulnerability. Susceptibility refers to the possibility of geological disasters controlled by disaster-prone conditions in a certain area. Hazard refers to the possibility of a certain scale and type of geological disaster occurring in a certain period of time in a certain area under the influence of certain triggering factors. Vulnerability refers to the severity of damage that the disaster-bearing body may suffer from geological disasters. Hazard = susceptibility × inducing factors, risk = hazard × vulnerability.

[0003] Geological hazard risk analysis is about the coupling result of the selection of analysis factors, the selection of analysis samples, the selection of analysis models, and the selection of analysis scales. A large number of geological hazard risk analysis methods have made important progress, but the existing methods still have problems such as strong subjectivity and lack of accuracy. How to select appropriate methods in each link to objectively conduct geological hazard risk analysis, improve the accuracy of the results, and match the actual geological hazard risk situation is a challenging technical problem. Summary of the invention

[0004] The present invention provides a county-level geological disaster risk analysis method and system, the main purpose of which is to match the actual geological disaster risk situation by improving the accuracy of geological disaster analysis results.

[0005] To achieve the above purpose, the present invention provides a method for analyzing the risk of geological disasters in a county, comprising:

[0006] Identify the geological disaster area and non-geological disaster area of ​​the county, divide the slope units of the geological disaster area, screen the susceptibility analysis factors in the slope units, and select training sets and test sets from the geological disaster area and the non-geological disaster area based on the susceptibility analysis factors;

[0007] Establishing an initial analysis model of the training set, performing hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters, obtaining an optimized analysis model corresponding to the optimized hyperparameters, using the optimized analysis model to output susceptibility classification results of the training set and the test set, analyzing the classification accuracy of the susceptibility classification results, and determining the susceptibility analysis results of the slope unit according to the classification accuracy;

[0008] Calculating the hazard index of the slope unit, and dividing the hazard zones in the geological disaster area by using the hazard index;

[0009] Screening the vulnerability analysis factors in the slope unit, generating an index hierarchy of the vulnerability analysis factors, constructing a judgment matrix of the index hierarchy, and performing a hierarchical analysis on the vulnerability analysis factors using the judgment matrix to obtain a hierarchical analysis result;

[0010] Calculating the vulnerability index of the slope unit using the hierarchical analysis result, and determining the vulnerability analysis result of the geological disaster area according to the vulnerability index;

[0011] Based on the susceptibility analysis results, the hazard zoning and the vulnerability analysis results, a geological hazard risk analysis result of the slope unit is generated.

[0012] Optionally, the slope units for dividing the geological disaster area include:

[0013] Constructing a terrain surface model of the geological disaster area;

[0014] Filling depressions in the terrain surface model to obtain a depression-free model;

[0015] extracting water flow direction from the depression-free model;

[0016] Calculating the cumulative amount of runoff of the depression-free model according to the water flow direction;

[0017] Based on the accumulated flow, a river network map of the depression-free model is generated;

[0018] Identifying river connection information in the river network map;

[0019] Determine the water catchment basin and reverse water catchment basin of the depression-free model by using the river connection information;

[0020] Merging the watershed and the reverse watershed to obtain a slope unit model;

[0021] A slope unit corresponding to the slope unit model is queried in the geological disaster area.

[0022] Optionally, the screening of the susceptibility analysis factors in the slope unit includes:

[0023] dividing the grid cells in the slope cells;

[0024] Selecting topographic factors, geological condition factors, hydrological condition factors, engineering activity factors and vegetation normalization index in the grid cells;

[0025] Quantitatively grading the topographic and geomorphic factors, the geological condition factors, the hydrological condition factors, the engineering activity factors and the vegetation normalization index to obtain quantitative grading factors;

[0026] The Pearson coefficient of the quantitative grading factor is calculated using the following formula:

[0027]

[0028] Among them, ρ x,y represents the Pearson coefficient, cov(x,y) represents the covariance between the x quantitative grading factor and the y quantitative grading factor, σ x represents the standard deviation of the quantitative grading factor x, σ y Represents the standard deviation of the y quantitative grading factor, μ x represents the mean value of the quantization grading factor x, μ y represents the mean value of the y quantitative grading factor, E[(x-μ x )(y-μ y )] represents (x-μ x )(y-μ y ), x represents the quantitative grading factor, and y represents the quantitative grading factor other than x;

[0029] Eliminate the quantitative grading factors whose absolute values ​​of the Pearson coefficients are greater than the preset factor threshold to obtain the remaining grading factors;

[0030] Calculating the factor median of the remaining classification factors;

[0031] Generate a factor partition diagram of the slope unit using the factor median to obtain a susceptibility analysis factor in the slope unit;

[0032] Among them, the topographic factors include elevation, slope, slope aspect, curvature, surface roughness and terrain undulation index; the geological condition factors include stratigraphic lithology, engineering geological rock formations, slope structure and fault distance; the hydrological condition factors include terrain humidity index, water flow intensity index and water system distance; the engineering activity factors include medium and high risk slope cutting and building density and road distance.

[0033] Optionally, the establishing of the initial analysis model of the training set includes:

[0034] Randomly split the training set to obtain a split training set;

[0035] Calculate the training set similarity between each split training set;

[0036] Performing training set screening on the segmented training set according to the training set similarity to obtain a screening training set;

[0037] Randomly extract training samples from the screening training set;

[0038] After performing sample splitting on the training samples using the sample features of the training samples, a decision tree of the training set is determined;

[0039] The decision tree is used to establish an initial analysis model for the training set.

[0040] Optionally, the performing hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters includes:

[0041] Dividing the subspace to be optimized of the initial analysis model;

[0042] Uniformly sampling the hyperparameters to be optimized in the subspace to be optimized;

[0043] Analyze the objective function of the hyperparameter to be optimized;

[0044] Calculating the posterior distribution of the objective function;

[0045] According to the posterior distribution, the acquisition function value of the hyperparameter to be optimized is calculated using the following formula:

[0046] AC(z)=μ(z)+kσ(z)

[0047] Where AC(z) represents the acquisition function value, μ(z) represents the mean of the posterior distribution z, σ(z) represents the variance of the posterior distribution, and k represents the tuning parameter;

[0048] Based on the acquisition function value, identifying a target subspace in the subspace to be optimized;

[0049] The hyperparameters to be optimized in the target subspace are optimized by using the objective function corresponding to the target subspace to obtain optimized hyperparameters.

[0050] Optionally, analyzing the classification accuracy of the susceptibility classification result includes:

[0051] Constructing a confusion matrix of the susceptibility classification results;

[0052] Query the true positive rate and false positive rate of the susceptibility classification result from the confusion matrix;

[0053] Constructing a ROC curve of the true positive rate and the false positive rate;

[0054] Calculating the area under the ROC curve;

[0055] The classification accuracy of the susceptibility classification result is analyzed using the curve area.

[0056] Optionally, the calculating the danger index of the slope unit includes:

[0057] The instability probability of the slope unit is calculated using the following formula:

[0058]

[0059] Among them, P i represents the probability of instability of the i-th slope unit within a preset time period under rainfall conditions, L max / day It represents the maximum daily rainfall of the slope unit since the monitoring record, L i Indicates the annual average maximum daily rainfall under different rainfall conditions;

[0060] According to the instability probability, the hazard index of the slope unit is calculated using the following formula:

[0061]

[0062] Among them, H i represents the hazard index of the ith slope unit under rainfall conditions, P i represents the probability of instability of the i-th slope unit within a preset time period under rainfall conditions, Y i represents the susceptibility index of the i-th slope unit, Y max Represents the maximum susceptibility index of the slope unit.

[0063] Optionally, the screening of the vulnerability analysis factors in the slope unit includes:

[0064] Select the number of people, buildings, traffic roads, living facilities and important facilities in the slope unit;

[0065] Quantitatively grading the population, the number of buildings, the number of traffic roads, the number of living facilities and the number of important facilities to obtain a vulnerability analysis factor;

[0066] Among them, the number of living facilities includes the number of oil and gas pipelines, the number of water pipelines, the number of power transmission lines and the number of communication lines; the number of important facilities includes the number of schools, the number of hospitals, the number of nursing homes and the number of power stations.

[0067] Optionally, the calculating the vulnerability index of the slope unit by using the hierarchical analysis result includes:

[0068] Obtaining the weight values ​​of indicators in the hierarchical analysis results;

[0069] According to the indicator weight value, the vulnerability index of the slope unit is calculated using the following formula:

[0070]

[0071] Among them, V i represents the vulnerability index, i represents the serial number of the slope unit, ω j represents the index weight value of the jth vulnerability analysis factor of the i-th slope unit, y j represents the standardized value of the jth vulnerability analysis factor of the i-th slope unit, S i Represents the slope reduction coefficient of the i-th slope unit.

[0072] In order to solve the above problems, the present invention also provides a county geological disaster risk analysis system, the system comprising:

[0073] A training set selection module is used to identify the geological disaster area and the non-geological disaster area in the county, divide the slope units in the geological disaster area, screen the susceptibility analysis factors in the slope units, and select training sets and test sets from the geological disaster area and the non-geological disaster area based on the susceptibility analysis factors;

[0074] A susceptibility analysis module is used to establish an initial analysis model of the training set, perform hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters, obtain an optimized analysis model corresponding to the optimized hyperparameters, use the optimized analysis model to output susceptibility classification results of the training set and the test set, analyze the classification accuracy of the susceptibility classification results, and determine the susceptibility analysis results of the slope unit according to the classification accuracy;

[0075] A zoning division module, used for calculating the hazard index of the slope unit, and dividing the hazard zones in the geological disaster area by using the hazard index;

[0076] A hierarchical analysis module, used for screening the vulnerability analysis factors in the slope unit, generating an index hierarchy of the vulnerability analysis factors, constructing a judgment matrix of the index hierarchy, performing a hierarchical analysis on the vulnerability analysis factors using the judgment matrix, and obtaining a hierarchical analysis result;

[0077] A vulnerability analysis module, used to calculate the vulnerability index of the slope unit using the hierarchical analysis result, and determine the vulnerability analysis result of the geological disaster area according to the vulnerability index;

[0078] The risk analysis module is used to generate a geological hazard risk analysis result of the slope unit based on the susceptibility analysis result, the hazard zoning and the vulnerability analysis result.

[0079] Compared with the problems described in the background technology, this method takes susceptibility-hazard-vulnerability evaluation as the idea, and based on random forest model, Bayesian optimization method, extreme rainfall condition hazard zoning, tomography analysis method and slope loss analysis, a random forest-extreme rainfall hypothesis method-analytic hierarchy process coupling model (RF-ARE-AHP) is constructed for the risk assessment of slope units at the county scale. The model is verified by the area under the ROC curve (AUC) value, confusion matrix parameters and field survey results, and a high-precision risk zoning map based on slope units is produced. The verification results show that the traditional method has made up for the strong subjectivity and lack of accuracy, and provides a reference for the risk assessment and disaster prevention and mitigation of geological hazards at the county scale. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 A schematic diagram of a process for analyzing the risk of geological disasters in a county according to an embodiment of the present invention;

[0081] Figure 2 A schematic diagram of a confusion matrix for implementing the county geological disaster risk analysis method provided by an embodiment of the present invention;

[0082] Figure 3 A schematic diagram of a module for implementing the county geological disaster risk analysis method provided in one embodiment of the present invention.

[0083] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0084] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0085] The embodiment of the present application provides a method for analyzing the risk of geological disasters in a county. The execution subject of the method for analyzing the risk of geological disasters in a county includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for analyzing the risk of geological disasters in a county can be executed by software or hardware installed in a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0086] Embodiment 1:

[0087] Reference Figure 1 FIG. 1 is a flow chart of a method for analyzing the risk of geological disasters in a county according to an embodiment of the present invention. In this embodiment, the method for analyzing the risk of geological disasters in a county includes:

[0088] S1. Identify the geological disaster areas and non-geological disaster areas in the county, divide the slope units of the geological disaster areas, screen the susceptibility analysis factors in the slope units, and select training sets and test sets from the geological disaster areas and the non-geological disaster areas based on the susceptibility analysis factors.

[0089] In the embodiment of the present invention, the geological disaster area refers to an area where a geological disaster has occurred, and the non-geological disaster area refers to an area where a geological disaster has not occurred.

[0090] Furthermore, the embodiment of the present invention divides the geological disaster area into slope units to conduct risk analysis on the geological source causing the geological disaster. The slope unit refers to a terrain unit defined by ridge lines and valley lines, which is related to the hydrological conditions and processes that shape the topography and is the basic unit for the development of geological disasters.

[0091] In one embodiment of the present invention, the slope units for dividing the geological disaster area include: constructing a terrain surface model of the geological disaster area; filling depressions in the terrain surface model to obtain a depression-free model; extracting a water flow direction from the depression-free model; calculating the cumulative flow of the depression-free model according to the water flow direction; generating a river network map of the depression-free model based on the cumulative flow; identifying river connection information in the river network map; determining the water catchment basin and reverse water catchment basin of the depression-free model using the river connection information; fusing the water catchment basin and the reverse water catchment basin to obtain a slope unit model; and querying the slope unit corresponding to the slope unit model in the geological disaster area.

[0092] The terrain surface model refers to the DEM model.

[0093] Optionally, the process of filling depressions in the terrain surface model is implemented by the "fill depression" operation of ArcGIS software. Furthermore, the process of extracting the water flow direction from the depression-free model is implemented by the "flow direction" operation of ArcGIS software. Furthermore, the process of calculating the cumulative runoff of the depression-free model based on the water flow direction refers to assigning water levels of different depths at each grid node of the DEM, and calculating the amount of water flowing through each node according to the regional terrain and water flow direction in accordance with the natural law that water flows from high to low, thereby obtaining the cumulative runoff of the area. Furthermore, the process of generating a river network map of the depression-free model based on the cumulative runoff refers to setting a threshold value for the cumulative runoff based on test results and auxiliary inspections such as terrain data, wherein different levels of valleys correspond to different thresholds, and valleys of the same level but in different areas have different thresholds, and valleys greater than the threshold value are set. The grid attribute of is assigned a value of 1, and the grid attribute that is less than or equal to the threshold is set to no data (null value), so that the grid file of the river network can be obtained, and then the river network map is generated through raster-vector conversion. The process of identifying the river connection information in the river network map refers to identifying the connection information between the nodes in the river network map, for example, each arc segment connects two nodes that serve as water outlets or confluence points, or connects the node that serves as the water outlet and the starting point of the river network, so that the starting point and the ending point of each river network arc segment can be obtained. The process of determining the watershed and the reverse watershed of the depression-free model using the river connection information refers to extracting the ridge line through the hydrological analysis of the positive DEM, and extracting the valley line through the hydrological analysis of the reverse DEM, and the ridge line and the valley line represent the watershed and the reverse watershed, and the process of fusing the watershed and the reverse watershed refers to using the positions of the ridge line and the valley line in the DEM as the slope unit model.

[0094] Furthermore, the embodiment of the present invention screens the susceptibility analysis factors in the slope unit to screen the factors that can be used for susceptibility risk analysis.

[0095] In one embodiment of the present invention, the screening of the susceptibility analysis factors in the slope unit includes: dividing the grid units in the slope unit; selecting the topographic factors, geological condition factors, hydrological condition factors, engineering activity factors and vegetation normalization index in the grid units; quantitatively grading the topographic factors, geological condition factors, hydrological condition factors, engineering activity factors and vegetation normalization index to obtain quantitative grading factors; and calculating the Pearson coefficient of the quantitative grading factor using the following formula:

[0096]

[0097] Among them, ρ x,yrepresents the Pearson coefficient, cov(x,y) represents the covariance between the x quantitative grading factor and the y quantitative grading factor, σ x represents the standard deviation of the quantitative grading factor x, σ y Represents the standard deviation of the y quantitative grading factor, μ x represents the mean value of the quantization grading factor x, μ y represents the mean value of the y quantitative grading factor, E[(x-μ x )(y-μ y )] represents (x-μ x )(y-μ y ), x represents the quantitative grading factor, and y represents the quantitative grading factor other than x;

[0098] Eliminate the quantitative grading factors whose absolute values ​​of the Pearson coefficient are greater than the preset factor threshold to obtain the remaining grading factors; calculate the factor median of the remaining grading factors; use the factor median to generate the factor partition map of the slope unit to obtain the susceptibility analysis factor in the slope unit; wherein the topographic and geomorphic factors include elevation, slope, slope aspect, curvature, surface roughness and terrain undulation index, the geological condition factors include stratum lithology, engineering geological rock formations, slope structure and fault distance, the hydrological condition factors include terrain humidity index, water flow intensity index and water system distance, and the engineering activity factors include medium and high risk slope cutting and building density and road distance.

[0099] Among them, the grid unit refers to the space divided into regular grids, each grid is called a unit, and the preset factor threshold refers to the threshold of the absolute value of the Pearson coefficient, such as 0.9, which aims to reduce data redundancy and improve the relative independence between susceptibility evaluation factors. The correlation between evaluation factors is evaluated by calculating the Pearson correlation coefficient, and factors with absolute values ​​of correlation coefficients greater than 0.5 are eliminated, thereby avoiding the problem of multicollinearity.

[0100] Optionally, the process of dividing the grid cells in the slope unit, for example, converts all susceptibility assessment factors into factors within a 10m×10m grid cell, the process of quantitatively grading the topographic factors, the geological condition factors, the hydrological condition factors, the engineering activity factors and the vegetation normalized index to obtain the quantitative grading factors refers to quantitatively grading multiple factors in the grid cell using the natural break point method, and the process of using the median of the factors to generate the factor zoning map of the slope unit refers to using the "Display Zoning Statistics as Table" tool in ArcGIS10.3 software to calculate the median of the susceptibility assessment factors in each slope unit, and finally generating a susceptibility assessment factor zoning map based on the slope unit, and using the susceptibility assessment factor zoning map as a susceptibility analysis factor.

[0101] Optionally, the process of selecting a training set and a test set from the geological disaster area and the non-geological disaster area based on the susceptibility analysis factor is, for example: in order to construct a binary classification model for machine learning, it is necessary to select geological disaster positive samples and non-geological disaster negative samples to train the random forest model, the slope unit where the geological disaster is located is used as the geological disaster positive sample, and the same number of non-geological disaster slope units are randomly selected as negative samples, and the sample set consisting of positive samples and negative samples is divided into a training set and a test set, wherein 70% of the samples are used to train the model and 30% of the samples are used to test the model. It should be noted that the training set and the test set are the data of the slope unit in the historical period, and in the subsequent process of determining the susceptibility analysis result of the slope unit, the susceptibility analysis result is determined based on the data of the slope unit in the current period.

[0102] S2. Establish an initial analysis model for the training set, perform hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters, obtain an optimized analysis model corresponding to the optimized hyperparameters, use the optimized analysis model to output susceptibility classification results of the training set and the test set, analyze the classification accuracy of the susceptibility classification results, and determine the susceptibility analysis results of the slope unit based on the classification accuracy.

[0103] The embodiment of the present invention establishes an initial analysis model of the training set to establish a random forest model using the training set.

[0104] In one embodiment of the present invention, the establishment of the initial analysis model of the training set includes: randomly splitting the training set to obtain split training sets; calculating the training set similarity between each split training set; screening the split training sets according to the training set similarity to obtain a screened training set; randomly extracting training samples from the screened training set; determining a decision tree of the training set after performing sample splitting on the training samples using the sample characteristics of the training samples; and establishing the initial analysis model of the training set using the decision tree.

[0105] Optionally, the process of randomly splitting the training set to obtain a split training set means that the training set for each decision tree is composed of n training samples randomly and with replacement from the original training set, and the training sets for each decision tree are different. The training set may also contain repeatedly selected training samples. Therefore, it is necessary to randomly split the training set to obtain a split training set, and it is also necessary to calculate the training set similarity to ensure that each split training set is different, so as to establish a random forest model for each split training set. Further, the process of randomly extracting training samples from the filtered training set means that if there are M features in the filtered training set, then when generating each node, m features (m < M) are randomly and without repetition selected from the M features, and the samples are divided using the m features to find the best feature to split and generate a node. Further, after splitting the training samples using the sample features of the training samples, the process of determining the decision tree of the training set means repeating the foregoing steps K times, that is, obtaining K decision trees, and finally integrating the classification results of the K decision trees (that is, the susceptibility classification results corresponding to each training set) to obtain an initial analysis model.

[0106] Further, in the embodiment of the present invention, the initial analysis model is optimized for hyperparameters to optimize the parameters of the established random forest model and improve the output accuracy of the random forest model.

[0107] In an embodiment of the present invention, optimizing the hyperparameters of the initial analysis model to obtain optimized hyperparameters includes: dividing the subspace to be optimized of the initial analysis model; uniformly sampling the hyperparameters to be optimized in the subspace to be optimized; analyzing the objective function of the hyperparameters to be optimized; calculating the posterior distribution of the objective function; and calculating the acquisition function value of the hyperparameters to be optimized according to the posterior distribution using the following formula:

[0108] AC(z) = μ(z) + kσ(z)

[0109] where AC(z) represents the acquisition function value, μ(z) represents the mean of the posterior distribution z, σ(z) represents the variance of the posterior distribution, and k represents the tuning parameter;

[0110] Based on the acquisition function value, identifying the target subspace in the subspace to be optimized; and optimizing the hyperparameters to be optimized in the target subspace through the objective function corresponding to the target subspace to obtain optimized hyperparameters.

[0111] Among them, the subspace to be optimized refers to a continuous forest area randomly selected in the initial analysis model and containing N hyperparameters, and the objective function refers to the loss value corresponding to the hyperparameters to be optimized.

[0112] Optionally, the process of calculating the posterior distribution of the objective function refers to estimating the posterior distribution of the objective function of each sampled hyperparameter combination using a Gaussian process. Furthermore, the process of identifying the target subspace in the subspace to be optimized based on the acquisition function value refers to using the acquisition function value to identify the hyperparameter forest area that may optimize the objective function. The process of performing hyperparameter optimization on the hyperparameters to be optimized in the target subspace through the objective function corresponding to the target subspace to obtain the optimized hyperparameters refers to optimizing the objective function corresponding to the target subspace using a Bayesian optimization algorithm.

[0113] Furthermore, in an embodiment of the present invention, the susceptibility classification result refers to the susceptibility evaluation result of the optimization analysis model on the susceptibility analysis factors in the training set and the test set.

[0114] Furthermore, the embodiment of the present invention analyzes the classification accuracy of the susceptibility classification results to evaluate the classification accuracy of the optimization analysis model, and demonstrates the classification ability of the model by plotting the true positive rate (TPR) and the false positive rate (FPR) under different thresholds. The AUC (Area Under the Curve), that is, the area under the ROC curve, is a quantitative indicator for measuring the performance of the random forest model. The closer the AUC value is to 1, the better the classification performance of the random forest model. Since the ROC curve is insensitive to changes in the distribution of positive and negative samples, it can provide a relatively comprehensive evaluation of the model performance.

[0115] In one embodiment of the present invention, the analysis of the classification accuracy of the susceptibility classification results includes: constructing a confusion matrix of the susceptibility classification results; querying the true positive rate and the false positive rate of the susceptibility classification results from the confusion matrix; constructing a ROC curve of the true positive rate and the false positive rate; calculating the curve area of ​​the ROC curve; and using the curve area to analyze the classification accuracy of the susceptibility classification results.

[0116] See also Figure 2 FIG. 1 is a schematic diagram of a confusion matrix for implementing the county geological disaster risk analysis method provided by an embodiment of the present invention. Figure 2 In the formula, true positive rate refers to the positive sample sensitivity TPR, false positive rate refers to the positive sample specificity FPR, true negative rate refers to the negative sample sensitivity TNR, false negative rate refers to the negative sample specificity FNR, P is the number of medium and high prone slope units obtained from the field survey, N is the number of low prone slope units obtained from the field survey, true positive (TP) and true negative (TN) are the number of correctly classified positive and negative samples, respectively, false positive (FP) and false negative (FN) are the number of incorrectly classified positive and negative samples, respectively, and true positive rate TPR represents the proportion of correctly classified positive samples.

[0117] Optionally, the process of determining the susceptibility analysis result of the slope unit based on the classification accuracy means continuing to optimize the random forest model when the classification accuracy is insufficient, and using the optimized random forest model to evaluate the susceptibility of the slope unit in the current time period when the classification accuracy is sufficient.

[0118] S3. Calculate the hazard index of the slope unit, and divide the hazard zones in the geological disaster area using the hazard index.

[0119] The embodiment of the present invention calculates the hazard index of the slope unit to evaluate the hazard in the geological disaster area through the hazard index, adopts a geological disaster hazard evaluation method based on the extreme rainfall hypothesis method, fully considers the historical fact that extreme rainfall in the area induces geological disasters, combines extreme rainfall records with dynamic rainfall intensity, highlights the relationship between geological disaster hazard evaluation and the probability of inducing factors, and reflects the dynamic evaluation process of geological disaster hazard.

[0120] In one embodiment of the present invention, the calculating of the hazard index of the slope unit includes: calculating the instability probability of the slope unit using the following formula:

[0121]

[0122] Among them, P i represents the probability of instability of the i-th slope unit within a preset time period under rainfall conditions, L max / day It represents the maximum daily rainfall of the slope unit since the monitoring record, L i Indicates the annual average maximum daily rainfall under different rainfall conditions;

[0123] According to the instability probability, the hazard index of the slope unit is calculated using the following formula:

[0124]

[0125] Among them, H i represents the hazard index of the ith slope unit under rainfall conditions, P i represents the probability of instability of the i-th slope unit within a preset time period under rainfall conditions, Y i represents the susceptibility index of the i-th slope unit, Y max Represents the maximum susceptibility index of the slope unit.

[0126] Optionally, the process of dividing the geological disaster area into hazard zones using the hazard index refers to calculating the hazard index of each slope unit to obtain the hazard zones under different rainfall conditions.

[0127] S4. Screening the vulnerability analysis factors in the slope unit, generating an index hierarchy of the vulnerability analysis factors, constructing a judgment matrix of the index hierarchy, and performing a hierarchical analysis on the vulnerability analysis factors using the judgment matrix to obtain a hierarchical analysis result.

[0128] The embodiment of the present invention selects factors capable of evaluating the vulnerability of the slope unit by screening the vulnerability analysis factors in the slope unit.

[0129] In one embodiment of the present invention, the screening of the vulnerability analysis factor in the slope unit includes: selecting the number of population, the number of buildings, the number of traffic roads, the number of living facilities and the number of important facilities in the slope unit; quantitatively grading the number of population, the number of buildings, the number of traffic roads, the number of living facilities and the number of important facilities to obtain a vulnerability analysis factor; wherein the number of living facilities includes the number of oil and gas pipelines, the number of water pipelines, the number of power transmission lines and the number of communication lines, and the number of important facilities includes the number of schools, the number of hospitals, the number of nursing homes and the number of power stations.

[0130] Optionally, the process of quantitatively grading the population, the number of buildings, the number of traffic roads, the number of living facilities and the number of important facilities to obtain the vulnerability analysis factor is carried out by quantitatively grading the vulnerability assessment factor based on the grid unit using the natural break point method.

[0131] Optionally, the process of generating the indicator hierarchy of the vulnerability analysis factor refers to: decomposing decision-related elements into multiple levels such as goals, criteria, plans, etc., the target layer is the vulnerability assessment result, population vulnerability, material vulnerability and important facility vulnerability are the criterion layer, and the five evaluation factors of population, buildings, transportation roads, living facilities and important facilities are the plan layer. Furthermore, the process of constructing the judgment matrix of the indicator hierarchy refers to determining the weight of each criterion layer to the target layer through subjective mutual pairwise comparison between each vulnerability analysis factor based on analysis and research. Furthermore, the process of using the judgment matrix to perform hierarchical analysis on the vulnerability analysis factor to obtain the hierarchical analysis result is, for example: using the hierarchical analysis method to calculate the relative importance of each evaluation factor by combining qualitative and quantitative methods, thereby calculating the weight value of the vulnerability analysis factor.

[0132] S5. Calculate the vulnerability index of the slope unit using the hierarchical analysis result, and determine the vulnerability analysis result of the geological disaster area according to the vulnerability index.

[0133] The embodiment of the present invention calculates the vulnerability index of the slope unit by using the hierarchical analysis result, so as to calculate the vulnerability index of the slope unit by weighted summation. When evaluating the vulnerability of a large-scale region, the terrain is generally undulating, and a large number of population buildings are gathered in the county towns and township settlements with flat terrain. These areas are often classified as high vulnerability or extremely high vulnerability areas in the vulnerability evaluation. However, the risk evaluation method based on the traditional formula "risk (R) = hazard (H) × vulnerability (V)" will inevitably lead to some flat areas that are not actually threatened by geological disasters being classified as medium and high risk areas. Therefore, the vulnerability evaluation is performed according to the loss coefficient.

[0134] In one embodiment of the present invention, the step of calculating the vulnerability index of the slope unit using the hierarchical analysis result includes: obtaining an indicator weight value in the hierarchical analysis result; and calculating the vulnerability index of the slope unit using the following formula according to the indicator weight value:

[0135]

[0136] Among them, V i represents the vulnerability index, i represents the serial number of the slope unit, ω j represents the index weight value of the jth vulnerability analysis factor of the i-th slope unit, y j represents the standardized value of the jth vulnerability analysis factor of the i-th slope unit, S i Represents the slope reduction coefficient of the i-th slope unit.

[0137] The vulnerability analysis result is the vulnerability index.

[0138] It should be noted that when the slope of the slope unit is less than 5°, there is no risk of instability in the slope unit, and the slope reduction coefficient is 0; when the slope is between 5° and 15°, only the buildings near the toe of the slope are at risk of geological disasters, and the slope reduction coefficient is 0.1; and when the slope is greater than 35°, all threatened objects in the entire slope unit face the risk of geological disasters, and the slope reduction coefficient is 1. The slope reduction coefficient is determined by the expert experience method, so that the object of vulnerability assessment is focused from the entire domain to the area actually threatened by geological disasters.

[0139] S6. Generate a geological hazard risk analysis result of the slope unit based on the susceptibility analysis result, the hazard zoning and the vulnerability analysis result.

[0140] Optionally, the process of generating the geological hazard risk analysis results of the slope unit based on the susceptibility analysis results, the hazardous zoning and the vulnerability analysis results refers to: through the evaluation of the susceptibility, hazardousness and vulnerability of geological hazards in the evaluation area, after standardizing the evaluation results, based on the risk assessment model "risk (R) = hazardousness (H) × vulnerability (V)" and the susceptibility analysis results, the geological hazard risk zones for different rainfall conditions are divided by the natural breakpoint method, that is, divided into low risk areas, medium risk areas, high risk areas and extremely high risk areas for geological hazards.

[0141] Compared with the problems described in the background technology, this method takes susceptibility-hazard-vulnerability evaluation as the idea, and based on random forest model, Bayesian optimization method, extreme rainfall condition hazard zoning, tomography analysis method and slope loss analysis, a random forest-extreme rainfall hypothesis method-analytic hierarchy process coupling model (RF-ARE-AHP) is constructed for the risk assessment of slope units at the county scale. The model is verified by the area under the ROC curve (AUC) value, confusion matrix parameters and field survey results, and a high-precision risk zoning map based on slope units is produced. The verification results show that the traditional method has made up for the strong subjectivity and lack of accuracy, and provides a reference for the risk assessment and disaster prevention and mitigation of geological hazards at the county scale.

[0142] Embodiment 2:

[0143] like Figure 3 The figure shows a functional module diagram of a county geological disaster risk analysis system of the present invention.

[0144] The county geological disaster risk analysis system 300 described in the present invention can be installed in an electronic device. According to the functions implemented, the county geological disaster risk analysis system can include a training set selection module 301, a susceptibility analysis module 302, a partitioning module 303, a hierarchical analysis module 304, a vulnerability analysis module 305 and a risk analysis module 306. The module described in the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0145] In the embodiment of the present invention, the functions of each module / unit are as follows:

[0146] The training set selection module 301 is used to identify the geological disaster area and the non-geological disaster area in the county, divide the slope units in the geological disaster area, screen the susceptibility analysis factors in the slope units, and select the training set and the test set from the geological disaster area and the non-geological disaster area based on the susceptibility analysis factors;

[0147] The susceptibility analysis module 302 is used to establish an initial analysis model of the training set, perform hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters, obtain an optimized analysis model corresponding to the optimized hyperparameters, use the optimized analysis model to output susceptibility classification results of the training set and the test set, analyze the classification accuracy of the susceptibility classification results, and determine the susceptibility analysis results of the slope unit according to the classification accuracy;

[0148] The zoning division module 303 is used to calculate the hazard index of the slope unit and divide the hazard zones in the geological disaster area using the hazard index;

[0149] The hierarchical analysis module 304 is used to screen the vulnerability analysis factors in the slope unit, generate an index hierarchy of the vulnerability analysis factors, construct a judgment matrix of the index hierarchy, and perform a hierarchical analysis on the vulnerability analysis factors using the judgment matrix to obtain a hierarchical analysis result;

[0150] The vulnerability analysis module 305 is used to calculate the vulnerability index of the slope unit using the hierarchical analysis result, and determine the vulnerability analysis result of the geological disaster area according to the vulnerability index;

[0151] The risk analysis module 306 is used to generate a geological hazard risk analysis result of the slope unit based on the susceptibility analysis result, the hazard zoning and the vulnerability analysis result.

[0152] In detail, each module in the county geological disaster risk analysis system 300 in the embodiment of the present invention is used in the same manner as above. Figure 1 The technical means are the same as the county geological disaster risk analysis method described in, and can produce the same technical effects, so I will not go into details here.

[0153] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for analyzing the risk of geological disasters in a county, characterized in that: The method comprises: Identify the geological disaster area and non-geological disaster area of ​​the county, divide the slope units of the geological disaster area, screen the susceptibility analysis factors in the slope units, and select training sets and test sets from the geological disaster area and the non-geological disaster area based on the susceptibility analysis factors; Establishing an initial analysis model of the training set, performing hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters, obtaining an optimized analysis model corresponding to the optimized hyperparameters, using the optimized analysis model to output susceptibility classification results of the training set and the test set, analyzing the classification accuracy of the susceptibility classification results, and determining the susceptibility analysis results of the slope unit according to the classification accuracy; Calculating the hazard index of the slope unit, and dividing the hazard zones in the geological disaster area by using the hazard index; Screening the vulnerability analysis factors in the slope unit, generating an index hierarchy of the vulnerability analysis factors, constructing a judgment matrix of the index hierarchy, and performing a hierarchical analysis on the vulnerability analysis factors using the judgment matrix to obtain a hierarchical analysis result; Calculating the vulnerability index of the slope unit using the hierarchical analysis result, and determining the vulnerability analysis result of the geological disaster area according to the vulnerability index; Based on the susceptibility analysis results, the hazard zoning and the vulnerability analysis results, a geological hazard risk analysis result of the slope unit is generated.

2. The method for analyzing county geological disaster risks according to claim 1, characterized in that: The slope unit for dividing the geological disaster area includes: Constructing a terrain surface model of the geological disaster area; Filling depressions in the terrain surface model to obtain a depression-free model; extracting water flow direction from the depression-free model; Calculating the cumulative amount of runoff of the depression-free model according to the water flow direction; Based on the accumulated flow, a river network map of the depression-free model is generated; Identifying river connection information in the river network map; Determine the water catchment basin and reverse water catchment basin of the depression-free model by using the river connection information; Merging the watershed and the reverse watershed to obtain a slope unit model; A slope unit corresponding to the slope unit model is queried in the geological disaster area.

3. The method for analyzing the risk of geological disasters in a county according to claim 1, characterized in that: The screening of the susceptibility analysis factors in the slope unit includes: dividing the grid cells in the slope cells; Selecting topographic factors, geological condition factors, hydrological condition factors, engineering activity factors and vegetation normalization index in the grid cells; Quantitatively grading the topographic and geomorphic factors, the geological condition factors, the hydrological condition factors, the engineering activity factors and the vegetation normalization index to obtain quantitative grading factors; The Pearson coefficient of the quantitative grading factor is calculated using the following formula: Among them, ρ x,y represents the Pearson coefficient, cov(x,y) represents the covariance between the x quantitative grading factor and the y quantitative grading factor, σ x represents the standard deviation of the quantitative grading factor x, σ y Represents the standard deviation of the y quantitative grading factor, μ x represents the mean value of the quantization grading factor x, μ y represents the mean value of the y quantitative grading factor, E[(x-μ x )(y-μ y )] represents (x-μ x )(y-μ y ), x represents the quantitative grading factor, and y represents the quantitative grading factor other than x; Eliminate the quantitative grading factors whose absolute values ​​of the Pearson coefficients are greater than the preset factor threshold to obtain the remaining grading factors; Calculating the factor median of the remaining classification factors; Generate a factor partition diagram of the slope unit using the factor median to obtain a susceptibility analysis factor in the slope unit; Among them, the topographic factors include elevation, slope, slope aspect, curvature, surface roughness and terrain undulation index; the geological condition factors include stratigraphic lithology, engineering geological rock formations, slope structure and fault distance; the hydrological condition factors include terrain humidity index, water flow intensity index and water system distance; the engineering activity factors include medium and high risk slope cutting and building density and road distance.

4. The method for analyzing county geological disaster risks according to claim 1, characterized in that: The step of establishing the initial analysis model of the training set comprises: Randomly splitting the training set to obtain a split training set; Calculate the training set similarity between each split training set; Performing training set screening on the segmented training set according to the training set similarity to obtain a screening training set; Randomly extract training samples from the screening training set; After performing sample splitting on the training samples using the sample features of the training samples, a decision tree of the training set is determined; The decision tree is used to establish an initial analysis model for the training set.

5. The method for analyzing county geological disaster risks according to claim 1, characterized in that: The step of performing hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters includes: Dividing the subspace to be optimized of the initial analysis model; Uniformly sampling the hyperparameters to be optimized in the subspace to be optimized; Analyze the objective function of the hyperparameter to be optimized; Calculating the posterior distribution of the objective function; According to the posterior distribution, the acquisition function value of the hyperparameter to be optimized is calculated using the following formula: AC(z)=μ(z)+kσ(z) Where AC(z) represents the acquisition function value, μ(z) represents the mean of the posterior distribution z, σ(z) represents the variance of the posterior distribution, and k represents the tuning parameter; Based on the acquisition function value, identifying a target subspace in the subspace to be optimized; The hyperparameters to be optimized in the target subspace are optimized by using the objective function corresponding to the target subspace to obtain optimized hyperparameters.

6. The method for analyzing the risk of geological disasters in a county according to claim 1, characterized in that: The analyzing the classification accuracy of the susceptibility classification result includes: Constructing a confusion matrix of the susceptibility classification results; Query the true positive rate and the false positive rate of the susceptibility classification result from the confusion matrix; Constructing a ROC curve of the true positive rate and the false positive rate; Calculating the area under the ROC curve; The classification accuracy of the susceptibility classification result is analyzed using the curve area.

7. The method for analyzing county geological disaster risks according to claim 1, characterized in that: The calculating the danger index of the slope unit includes: The instability probability of the slope unit is calculated using the following formula: Among them, P i represents the probability of instability of the i-th slope unit within a preset time period under rainfall conditions, L max / day It indicates the maximum daily rainfall of the slope unit since the monitoring record, L i Indicates the annual average maximum daily rainfall under different rainfall conditions; According to the instability probability, the hazard index of the slope unit is calculated using the following formula: Among them, H i represents the hazard index of the ith slope unit under rainfall conditions, P i represents the probability of instability of the i-th slope unit within a preset time period under rainfall conditions, Y i represents the susceptibility index of the i-th slope unit, Y max Represents the maximum susceptibility index of the slope unit.

8. The method for analyzing county geological disaster risks according to claim 1, characterized in that: The screening of the vulnerability analysis factor in the slope unit includes: Select the number of people, buildings, traffic roads, living facilities and important facilities in the slope unit; Quantitatively grading the population, the number of buildings, the number of traffic roads, the number of living facilities and the number of important facilities to obtain a vulnerability analysis factor; Among them, the number of living facilities includes the number of oil and gas pipelines, the number of water pipelines, the number of power transmission lines and the number of communication lines; the number of important facilities includes the number of schools, the number of hospitals, the number of nursing homes and the number of power stations.

9. The method for analyzing county geological disaster risks according to claim 1, characterized in that: The calculating the vulnerability index of the slope unit by using the hierarchical analysis result includes: Obtaining the weight values ​​of indicators in the hierarchical analysis results; According to the indicator weight value, the vulnerability index of the slope unit is calculated using the following formula: Among them, V i represents the vulnerability index, i represents the serial number of the slope unit, ω j represents the index weight value of the jth vulnerability analysis factor of the i-th slope unit, y j represents the standardized value of the jth vulnerability analysis factor of the i-th slope unit, S i Represents the slope reduction coefficient of the i-th slope unit.

10. A county geological disaster risk analysis system, characterized in that: The system comprises: A training set selection module is used to identify the geological disaster area and the non-geological disaster area in the county, divide the slope units in the geological disaster area, screen the susceptibility analysis factors in the slope units, and select training sets and test sets from the geological disaster area and the non-geological disaster area based on the susceptibility analysis factors; A susceptibility analysis module is used to establish an initial analysis model of the training set, perform hyperparameter optimization on the initial analysis model to obtain optimized hyperparameters, obtain an optimized analysis model corresponding to the optimized hyperparameters, use the optimized analysis model to output susceptibility classification results of the training set and the test set, analyze the classification accuracy of the susceptibility classification results, and determine the susceptibility analysis results of the slope unit according to the classification accuracy; A zoning division module, used for calculating the hazard index of the slope unit, and dividing the hazard zones in the geological disaster area by using the hazard index; A hierarchical analysis module, used for screening the vulnerability analysis factors in the slope unit, generating an index hierarchy of the vulnerability analysis factors, constructing a judgment matrix of the index hierarchy, performing a hierarchical analysis on the vulnerability analysis factors using the judgment matrix, and obtaining a hierarchical analysis result; A vulnerability analysis module, used to calculate the vulnerability index of the slope unit using the hierarchical analysis result, and determine the vulnerability analysis result of the geological disaster area according to the vulnerability index; The risk analysis module is used to generate a geological hazard risk analysis result of the slope unit based on the susceptibility analysis result, the hazard zoning and the vulnerability analysis result.

Images