Avalanche hazard susceptibility assessment method and system based on GIS and machine learning

Abstract

The invention discloses a method and system for evaluating the susceptibility of avalanches based on GIS and machine learning, and belongs to the technical field of avalanche susceptibility evaluation. The method comprises the following processes: determining the influencing factors causing avalanches in a target area according to the terrain data, meteorological data and avalanche data of the target area; performing correlation detection on the influencing factors causing avalanches preliminarily selected to determine the evaluation factors; constructing a positive sample set and a negative sample set according to the evaluation factors and avalanche point data information of the target area, training a machine learning model according to the positive sample set and the negative sample set, and obtaining an avalanche susceptibility evaluation model after training; producing an avalanche susceptibility zoning map according to the prediction results of the avalanche susceptibility evaluation model, and determining the avalanche susceptibility of each area; the method can improve the rationality and scientificity of the avalanche susceptibility evaluation results, and provide theoretical support and method guidance for the avalanche disaster prevention and control system.

Description

Technical Field

[0001] The present invention relates to the technical field of avalanche disaster susceptibility assessment, and in particular to a method and system for avalanche disaster susceptibility assessment based on GIS and machine learning. Background Art

[0002] Avalanche is a sudden, destructive and unpredictable geological disaster that usually blocks roads and railways, causing serious disruptions to traffic. The powerful destructive force it brings is enough to destroy buildings, roads, bridges and other infrastructure, causing huge property losses. Mastering the distribution characteristics of avalanche disasters can improve the regional avalanche disaster prevention and mitigation capabilities. Avalanche susceptibility assessment is the basis for predicting the distribution characteristics of avalanche disasters, an important prerequisite for avalanche risk management, and an indispensable part of engineering design such as avalanche prevention, land use planning and road selection. At present, the disaster prevention and mitigation method is prevention-oriented, combined with prevention and control. Avalanche susceptibility assessment and zoning are effective means to prevent avalanche disasters.

[0003] Traditional avalanche susceptibility evaluation methods such as expert scoring method and analytic hierarchy process are highly subjective and require large amounts of computation. However, avalanche hazards are complex and uncertain, with many influencing factors, and there is a highly nonlinear relationship between the development of avalanches and influencing factors. Therefore, they are increasingly unable to meet the current requirements for refined and accurate avalanche susceptibility evaluation. With the rapid development of artificial intelligence, geospatial information systems and machine learning are widely used in geological disaster susceptibility evaluation. Machine learning algorithms based on decision trees, Bayesian, Logistic regression, support vector machines, multi-layer perceptrons, etc. have good performance in geological disaster susceptibility evaluation research, which can solve the nonlinear problem between avalanche hazards and influencing factors and improve the reliability of evaluation results. Machine learning has been applied in avalanche susceptibility evaluation research, but it also faces some problems: insufficient analysis of evaluation factors and arbitrary selection. Sample data is not accurate enough. Positive sample (avalanche point) data is usually obtained from drone remote sensing interpretation, field investigation, and historical avalanche data. The authenticity of negative sample (non-avalanche point) acquisition is insufficient, resulting in reduced accuracy of evaluation results. Based on the above problems, the present invention conducts in-depth research on a road avalanche disaster susceptibility assessment method based on GIS and machine learning. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a method and system for evaluating the susceptibility of avalanche disasters based on GIS and machine learning. The evaluation factor data and spatial distribution are analyzed and processed through the GIS platform. The use of machine learning can effectively solve the problem of strong subjectivity of traditional evaluation methods, providing theoretical support and methodological reference for avalanche disaster prevention and control work.

[0005] The present invention is achieved through the following technical solutions:

[0006] In a first aspect, the present application provides a method for assessing the susceptibility of avalanche disasters based on GIS and machine learning, comprising the following steps:

[0007] Step 1: Determine the influencing factors that cause avalanches in the target area based on the terrain data, meteorological data and avalanche data of the target area;

[0008] Step 2: Conduct correlation test on the initially selected factors that may cause avalanches and determine the evaluation factors;

[0009] Step 3: construct a positive sample set and a negative sample set according to the evaluation factors and the avalanche point data information of the target area, train a machine learning model according to the positive sample set and the negative sample set, and obtain an avalanche susceptibility evaluation model after training;

[0010] Step 4: Create an avalanche susceptibility zoning map based on the prediction results of the avalanche susceptibility assessment model to determine the avalanche susceptibility of each area.

[0011] Preferably, the method for determining the impact factor in step 1 is as follows:

[0012] The influencing factors in terrain data are obtained by using digital elevation model calculation and analysis;

[0013] The interpolation method is used to determine the influencing factors of meteorological data and snow data.

[0014] Preferably, in step 2, a correlation analysis method is used to determine the correlation of the influencing factors, and the evaluation factors are determined according to the correlation of the influencing factors.

[0015] Preferably, determining the evaluation factor according to the correlation of the influencing factors includes:

[0016] The influencing factors with correlation greater than the threshold are taken as candidate evaluation factors;

[0017] The detection value of each candidate evaluation factor is determined, the non-linearly weakened candidate evaluation factors are eliminated according to the detection value, and the evaluation factor is determined according to the contribution rate of the remaining candidate evaluation factors.

[0018] Preferably, the correlation analysis method includes Pearson correlation coefficient method, Spearman rank correlation coefficient method or Kendall correlation coefficient method.

[0019] Preferably, the construction of the positive sample set and the negative sample set includes:

[0020] Construct a positive sample set based on the relevant data of the evaluation factors extracted from the avalanche point;

[0021] A buffer zone is set according to the development range of avalanche points. In the area outside the avalanche buffer zone, non-avalanche points with the same number as avalanche points are randomly selected. The relevant data in the evaluation factors are extracted from the non-avalanche points to construct a negative sample set.

[0022] Preferably, the machine learning model includes a support vector machine, a multilayer perceptron, a decision tree, a logistic regression model, and a random forest model;

[0023] Confusion matrix, precision, recall rate, and ROC curve were used to verify the accuracy of each machine learning model after training, and the optimal machine learning model was determined as the avalanche susceptibility evaluation model.

[0024] Preferably, the preparation of the avalanche susceptibility zoning map in step 4 includes:

[0025] According to the output results of the avalanche susceptibility assessment model, the evaluation factors are superimposed according to the weights to obtain an avalanche susceptibility assessment index map. The target area is divided into zones using the natural break point method to obtain an avalanche susceptibility zoning map. The avalanche susceptibility of each area is determined based on the number of grids, grid area and number of avalanches in each area.

[0026] In the second aspect, the present application provides an avalanche disaster susceptibility assessment system based on GIS and machine learning, including:

[0027] An influencing factor module is used to determine the influencing factors that cause avalanches in the target area based on the terrain data, meteorological data and avalanche data of the target area;

[0028] The factor analysis module is used to perform correlation test on the initially selected factors that affect avalanche occurrence and determine the evaluation factors;

[0029] A model training module is used to construct a positive sample set and a negative sample set according to the evaluation factors and the avalanche point data information of the target area, and train the machine learning model according to the positive sample set and the negative sample set to obtain an avalanche susceptibility evaluation model after training;

[0030] The evaluation module is used to prepare an avalanche susceptibility zoning map based on the prediction results of the avalanche susceptibility evaluation model and determine the avalanche susceptibility of each area.

[0031] In a second aspect, the present application provides an electronic device, including:

[0032] Memory for storing computer programs;

[0033] A processor is used to implement the steps of the avalanche disaster susceptibility assessment method based on GIS and machine learning when executing the computer program.

[0034] Compared with the prior art, the present invention has the following beneficial technical effects:

[0035] The present application provides a method for evaluating the susceptibility of avalanche disasters based on GIS and machine learning. First, by comprehensively analyzing the terrain data, meteorological data and historical avalanche data of the target area, the key influencing factors that may cause avalanches are scientifically and preliminarily selected. Subsequently, advanced statistical methods and geographic detector technology are used to conduct in-depth detection and analysis of single and double factors, effectively eliminating redundant or weakly correlated influencing factors, ensuring the high correlation and representativeness of the final evaluation factors, and improving the accuracy and pertinence of the evaluation model; in view of the problems of inaccurate sample data and insufficient authenticity of negative samples in traditional methods, the present invention innovatively proposes a strategy for setting an avalanche point buffer zone, and combines GIS technology to accurately erase the avalanche directly affected area, thereby ensuring the purity of negative samples (non-avalanche points) while also ensuring the representativeness and authenticity of their spatial distribution. The application of this method significantly improves the quality of sample data, provides more reliable learning materials for machine learning models, and thus enhances the accuracy and credibility of evaluation results. In addition, the spatial analysis capabilities of GIS are closely combined with the powerful data processing and pattern recognition capabilities of machine learning. By training and optimizing machine learning models (such as decision trees, random forests, support vector machines, etc.), high-precision predictions of avalanche susceptibility are achieved. This interdisciplinary technology integration not only solves the complex nonlinear relationship between avalanche hazards and influencing factors, but also greatly improves the automation and efficiency of evaluation; based on the trained avalanche susceptibility evaluation model, the present invention can generate a detailed avalanche susceptibility zoning map, which intuitively displays the avalanche risk level in different regions. This method can provide theoretical support and methodological guidance for the avalanche disaster prevention and control system.

[0036] This application also proposes a GIS and machine learning-based avalanche susceptibility assessment system, an electronic device and a computer storage medium, which have all the advantages of the above-mentioned GIS and machine learning-based avalanche susceptibility assessment method. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 It is a flow chart of the avalanche disaster susceptibility assessment method based on GIS and machine learning of the present invention;

[0039] Figure 2 Avalanche point distribution diagram of the present invention;

[0040] Figure 3 This is a schematic diagram of the correlation of the impact factors of the present invention;

[0041] Figure 4 This is a schematic diagram of the double-factor detection results of the geographic detector of the present invention;

[0042] Figure 5 It is the non-avalanche point distribution diagram of the present invention;

[0043] Figure 6 This is a zoning diagram of the avalanche susceptibility evaluation results of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0045] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0046] like Figure 1 As shown in the figure, a method for evaluating the susceptibility of avalanche disasters based on GIS and machine learning specifically includes the following steps:

[0047] Step 1: Obtain the location information of avalanche point data in the study area, as well as the terrain data, meteorological data and avalanche data in the study area.

[0048] During the specific implementation, the coordinates of the avalanche points in the study area are obtained by means of UAV remote sensing interpretation, on-site surveys and investigations, historical avalanche data, etc., and a location data table of the avalanche points is established. At the same time, the terrain data, meteorological data and snow accumulation data of the study area are obtained.

[0049] Step 2: Set a unified coordinate system and grid pixel size for the study area;

[0050] In the specific implementation, the spatial resolution requirements and actual needs of the avalanche hazard research area, as well as the calculation and evaluation analysis of the influencing factors of avalanches are weighed. The coordinate system is changed to GCS2000 or WGS1984, and the raster data is unified into Am*Am pixels using the resampling tool; this is conducive to the subsequent raster calculation and evaluation analysis;

[0051] Step 3: Based on the data information of the avalanche point and combined with the terrain data, meteorological data, snow data and other data of the study area, preliminarily select several influencing factors that cause avalanches.

[0052] Select influencing factors such as elevation, slope, aspect, terrain undulation, surface cut, ground roughness, elevation variation coefficient, etc. from the terrain data;

[0053] Select influencing factors such as average temperature during the snow accumulation period, average rainfall, and maximum wind speed from the meteorological data;

[0054] Select influencing factors such as average snowfall during the snow accumulation period, maximum snow depth, and average annual number of snowfall days from the snow accumulation data;

[0055] Select vegetation coverage index, faults, water system and other influencing factors from other data, and select at least one of the above influencing factors;

[0056] The influencing factors of terrain data are calculated and analyzed using the digital elevation model (DEM) of the study area. The influencing factors of meteorological data and snow data are analyzed by interpolation. The influencing factors of other data are further analyzed based on data websites or on-site surveys.

[0057] R=H max -H min , D=H mean -H min 、V=H std ÷H mean ,

[0058] Among them, R represents the terrain undulation, S represents the ground roughness, D represents the surface cut degree, V represents the elevation variation coefficient, A represents the slope, H represents the max Indicates the highest altitude point, H min Indicates the lowest altitude point, H mean Indicates the average altitude point, H std represents the standard deviation of altitude;

[0059] Step 4: Conduct correlation test on the initially selected factors that cause avalanche to determine the evaluation factors; the specific implementation includes the following process:

[0060] S4.1. Use the Pearson correlation coefficient method to calculate the correlation coefficient p of each initially selected influencing factor, and use the factor with greater correlation as the candidate evaluation factor

[0061] The expression of the correlation coefficient p is as follows:

[0062]

[0063] Among them, COV(x1,x2) represents the covariance of the initial selected influencing factors, It represents the standard deviation of the initially selected impact factors. When p≤0.3, the impact factors are irrelevant. When 0.3≤p<0.5, the impact factors have a low correlation. When 0.5≤p<0.8, the impact factors have a high correlation. When p≥0.8, the impact factors are highly correlated. The impact factors with greater correlation are selected as the alternative evaluation factor pairs.

[0064] S4.2. Use geographic detectors to calculate the detection values ​​q of the alternative evaluation factors, and eliminate the alternative evaluation factors that are non-linearly weakened.

[0065] The expression of the detection value q is as follows:

[0066] SST=Nσ 2

[0067] Where i represents the number of categories i = 1, 2, ... n, N i represents the number of units in category i, represents the variance of the dependent variable Y value of category i, N represents the total number of units in the study area, σ 2 represents the variance of the dependent variable Y value in the study area, SSW represents the sum of n types of variance, and SST represents the total variance value in the study area. Interaction detection is used to explore the effect of the interaction between the two influencing factors X1 and X2. q(X1∩X2)<Min(q(X1),q(X2)) represents nonlinear attenuation, Min(q(X1),q(X2))<q(X1∩X2)<Max(q(X1),q(X2)) represents single factor nonlinear attenuation, q(X1∩X2)>Max(q(X1),q(X2)) represents double factor enhancement, q(X1∩X2)=q(X1)+q(X2) represents mutual independence, and q(X1∩X2)>q(X1)+q(X2) represents nonlinear enhancement. The alternative evaluation factor pairs with nonlinear attenuation are eliminated.

[0068] S4.3. Set a contribution rate threshold, eliminate the candidate evaluation factors obtained in S4.2 whose contribution rates are lower than the threshold, and use the remaining candidate evaluation factors as evaluation factors.

[0069] Step 5: construct a positive sample set and a negative sample set according to the evaluation factors and avalanche points, construct a data sample set according to the positive sample set and the negative sample set, and train an avalanche susceptibility evaluation model according to the data sample set.

[0070] In the specific implementation, the positive sample set is constructed based on the relevant data of the evaluation factors extracted from the avalanche points; a buffer zone is set according to the development range of the avalanche points, and in the area outside the avalanche buffer zone, non-avalanche points with the same number as the avalanche points are randomly selected, and the negative sample set is constructed based on the relevant data of the evaluation factors extracted from the non-avalanche points.

[0071] The data sample set of the model is composed of the positive sample set and the negative sample set, and the training set and the test set are randomly divided according to the ratio of 7:3 on the data sample set. The selected training set is used as the input data of the machine learning model, and the machine learning model is trained to obtain the avalanche susceptibility evaluation model after training;

[0072] Machine learning models include support vector machines, multi-layer perceptrons, decision trees, logistic regression models, and random forest models;

[0073] A test set was used to test the accuracy of each avalanche susceptibility assessment model. The confusion matrix, precision, recall rate, ROC curve and other methods were used to verify the differences and applicability of different machine learning-based models in the avalanche susceptibility assessment research. Different research areas have different factors affecting avalanche development and different constructed evaluation index systems, which leads to different avalanche susceptibility assessment models constructed in different research areas. The optimal avalanche susceptibility assessment model was determined based on the actual situation of the study area and the accuracy test of the model.

[0074] Step 6: Prepare an avalanche susceptibility zoning map and conduct an evaluation and analysis of avalanche susceptibility.

[0075] In the specific implementation, the optimal avalanche susceptibility evaluation model is determined according to step 5, and the model outputs the evaluation factor weights. The evaluation factor grid data are superimposed according to the weights to obtain an avalanche susceptibility evaluation index map. The natural break point method is used to divide the study area into four areas: low susceptibility area, medium susceptibility area, high susceptibility area, and extremely high susceptibility area. Finally, an avalanche susceptibility zoning map is obtained. The number of grids, grid area, and number of avalanches in each area are counted to calculate the proportion, avalanche ratio, and avalanche density of different prone areas. According to the proportion, avalanche ratio, and avalanche density of prone areas, different prone areas are analyzed and compared.

[0076] This avalanche disaster susceptibility assessment method based on GIS and machine learning obtains avalanche point data information in the study area through UAV remote sensing interpretation, on-site surveys and investigations, historical avalanche data, etc., extracts avalanche influencing factors according to the conditions of the study area, and uses correlation detection and geographic detectors to screen out evaluation factors that meet the requirements, constructs an avalanche evaluation factor system, sets a buffer zone for the positive sample set, and selects negative sample sets on this basis. By constructing a scientific and reasonable avalanche evaluation factor system and strengthening the sample set, an avalanche susceptibility assessment model is constructed in combination with machine learning, thereby conducting avalanche susceptibility evaluation and analysis. This method has strong objectivity and rationality, and the reliability of the avalanche susceptibility assessment results is high, which can provide theoretical support and methodological reference for avalanche disaster prevention and mitigation.

[0077] Example 1

[0078] Taking a proposed expressway as an example, the avalanche disaster susceptibility assessment method based on GIS and machine learning of the present invention is further described, including the following process:

[0079] Step 1: Take a part of a planned highway as the study area of ​​the embodiment, and use UAV remote sensing interpretation, on-site survey, historical avalanche data, etc. to obtain 62 avalanche points in this study area, such as Figure 2 As shown;

[0080] Step 2: Based on the terrain data, meteorological data, snow data and other data of the study area, 17 influencing factors were initially selected, including elevation, slope, slope aspect, terrain undulation, ground roughness, surface cut degree, elevation variation coefficient, plane curvature, profile curvature, average temperature from December to March during the snow accumulation period, average precipitation, maximum precipitation, average air pressure, average wind speed, maximum wind speed, maximum snow depth and average snow depth. According to the actual situation of the study area, the pixel size of each influencing factor was set to 12.5m*12.5m, and the coordinate system was set to WGS-1984-UTM-Zone-46N.

[0081] Step 3: Use the digital elevation model (DEM) of the study area to calculate and analyze the influencing factors of terrain conditions. The influencing factors under meteorological data and snow data are obtained by interpolation analysis. The ground roughness, surface cut degree, and elevation variation coefficient are calculated and analyzed using a raster calculator. The formula is R = H max -H min , D=H mean -H min 、V=H std ÷H mean , R represents terrain undulation, S represents ground roughness, D represents surface cut degree, V represents elevation variation coefficient, A represents slope, H represents max Indicates the highest altitude point, H min Indicates the lowest altitude point, H mean Indicates the average altitude point, H std It indicates the standard deviation of altitude, elevation, slope, aspect, plan curvature, and profile curvature. The average temperature, average precipitation, maximum precipitation, average air pressure, average wind speed, maximum wind speed, maximum snow depth, and average snow depth during the snow season from December to March are interpolated and analyzed using the inverse distance weighted method;

[0082] Table 1 Classification and grading of impact factors

[0083]

[0084]

[0085]

[0086]

[0087] Step 4: Use the Pearson correlation coefficient method to obtain the correlation coefficient p of the initially selected influencing factors, expressed as COV(x1,x2) represents the covariance of the initial selected influencing factors. It indicates that the standard deviation of the initial selection of impact factors is not relevant when p≤0.3, the impact factors are less relevant when 0.3≤p<0.5, the impact factors are more relevant when 0.5≤p<0.8, and the impact factors are highly relevant when p≥0.8. After removing the highly relevant impact factors, the other impact factors are used as alternative evaluation factors. The correlation of the avalanche impact factors is as follows: Figure 3 As shown in Table 2, elevation, terrain undulation, average precipitation, maximum wind speed, and maximum snow depth are eliminated; the alternative evaluation factors are detected by single factor detection and double factor interactive detection using geographic detectors, and the q value is obtained, q slope = 0.55, q slope aspect = 0.46, q ground roughness = 0.15, q surface cut degree = 0.12, q elevation variation coefficient = 0.26, q plane curvature = 0.15, q profile curvature = 0.23, q average temperature = 0.10, q maximum precipitation = 0.07, q average air pressure = 0.13, q average wind speed = 0.43, q average snow depth = 0.77, and the double factor interactive detection is as follows Figure 4 As shown, avalanches under the double-factor interaction of geographic detectors are double-factor enhanced or nonlinearly enhanced. Since the maximum rainfall has a weak contribution in this study area, the maximum rainfall is eliminated;

[0088] Finally, an avalanche evaluation factor system was constructed by slope, aspect, ground roughness, surface cut, elevation variation coefficient, plane curvature, profile curvature, average temperature from December to March during the snow accumulation period, average air pressure, average wind speed, and average snow depth;

[0089] Step 5: The 62 avalanche points in the study area are the positive sample set. According to the development range of the avalanche points, a 50m buffer zone is set, the area where the avalanche buffer zone occurs is deleted, and 62 non-avalanche points with the same number as the avalanche points are randomly selected in other areas as the negative sample set, such as Figure 5 As shown in the figure, the positive sample set and the negative sample set constitute the sample set, and the training set and the test set are randomly divided in the sample set according to the ratio of 7:3. The selected sample set is used as the input data of machine learning and is imported into the support vector machine, multi-layer perceptron, decision tree, logistic regression, random forest and other algorithms respectively to build an avalanche susceptibility evaluation model based on machine learning;

[0090] Step 6: Perform accuracy test on the sample set, verify the support vector machine, multi-layer perceptron, decision tree, logistic regression, and random forest avalanche susceptibility evaluation model through ROC curve, and conclude that the avalanche susceptibility evaluation model based on support vector machine is more suitable for the area, and draw an avalanche susceptibility zoning map;

[0091] The avalanche susceptibility zoning map uses the natural break point method to divide the study area into low susceptibility area, medium susceptibility area, high susceptibility area, and extremely high susceptibility area. Figure 6 As shown in Table 2, the avalanche ratio and avalanche density are calculated based on the number of grids and avalanche points in each prone area. As shown in Table 2, with the increase of avalanche susceptibility level, the corresponding avalanche ratio and avalanche density increase.

[0092] The method first preliminarily selects influencing factors suitable for the study area, performs correlation tests on the influencing factors, uses geographic detectors to perform single-factor and double-factor detection, removes influencing factors that do not meet the requirements, and determines the final evaluation factors; sets a buffer zone for avalanche points, erases avalanche areas, selects non-avalanche points, and constructs a more accurate negative sample set. On this basis, a reliable avalanche susceptibility evaluation model is constructed, which not only objectively analyzes avalanche influencing factors, but also improves the rationality and scientificity of avalanche susceptibility evaluation results. The invention can provide theoretical support and methodological guidance for the avalanche disaster prevention and control system. It also provides a scientific basis for the prevention, early warning and emergency response of avalanche disasters, and helps to implement more accurate and effective disaster prevention and mitigation measures.

[0093] Based on the above-mentioned avalanche disaster susceptibility assessment method based on GIS and machine learning, the present application also provides an avalanche disaster susceptibility assessment system based on GIS and machine learning, including:

[0094] An influencing factor module is used to determine the influencing factors that cause avalanches in the target area based on the terrain data, meteorological data and avalanche data of the target area;

[0095] The factor analysis module is used to perform correlation test on the initially selected factors that affect avalanche occurrence and determine the evaluation factors;

[0096] A model training module is used to construct a positive sample set and a negative sample set according to the evaluation factors and the avalanche point data information of the target area, and train the machine learning model according to the positive sample set and the negative sample set to obtain an avalanche susceptibility evaluation model after training;

[0097] The evaluation module is used to prepare an avalanche susceptibility zoning map based on the prediction results of the avalanche susceptibility evaluation model and determine the avalanche susceptibility of each area.

[0098] It should be noted that in the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of each module is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules can be combined or integrated into another device, or some features can be ignored or not executed. The module described as a separate component may or may not be physically separated. The component displayed as a module may be a physical unit or multiple physical units, that is, it may be located in one place, or it may be distributed in multiple different places. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0099] In addition, each module in each embodiment of the present invention may be integrated into a processing unit, each module may exist physically separately, or two or more modules may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0100] An electronic device provided in an embodiment of the present application includes a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the steps of the avalanche hazard susceptibility assessment method based on GIS and machine learning as described in any of the above embodiments are implemented.

[0101] Another electronic device provided in the embodiment of the present application may also include: an input port connected to the processor, used to transmit multimodal data collected by an external acquisition device to the processor; and a display unit connected to the processor, used to display the processing results of the processor to the outside world; a communication module connected to the processor, used to realize communication between the electronic device and the outside world. The display unit can be a display panel, a laser scanning display, etc.; the communication mode adopted by the communication module includes but is not limited to mobile high-definition link technology (HML), universal serial bus (USB), high-definition multimedia interface (HDMI), wireless connection (including wireless fidelity technology (WiFi), Bluetooth communication technology, low-power Bluetooth communication technology, and communication technology based on IEEE802.11s).

[0102] An embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of the avalanche hazard susceptibility assessment method based on GIS and machine learning as described in any of the above embodiments are implemented.

[0103] For the description of the relevant parts of the avalanche susceptibility assessment system based on GIS and machine learning, the electronic device, and the computer-readable storage medium provided in the embodiments of the present application, please refer to the detailed description of the corresponding parts of the avalanche susceptibility assessment method based on GIS and machine learning provided in the embodiments of the present application, which will not be repeated here. In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive elaboration.

[0104] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for assessing the susceptibility of avalanche disasters based on GIS and machine learning, characterized in that: The following steps are involved: Step 1: Determine the influencing factors that cause avalanches in the target area based on the terrain data, meteorological data and avalanche data of the target area; Step 2: Conduct correlation test on the initially selected factors that may cause avalanche occurrence and determine the evaluation factors; Step 3: construct a positive sample set and a negative sample set according to the evaluation factors and the avalanche point data information of the target area, train a machine learning model according to the positive sample set and the negative sample set, and obtain an avalanche susceptibility evaluation model after training; Step 4: Create an avalanche susceptibility zoning map based on the prediction results of the avalanche susceptibility assessment model to determine the avalanche susceptibility of each area.

2. The method for assessing the susceptibility of avalanche disasters based on GIS and machine learning according to claim 1, characterized in that: The method for determining the impact factor in step 1 is as follows: The influencing factors in terrain data are obtained by using digital elevation model calculation and analysis; The interpolation method is used to determine the influencing factors of meteorological data and snow data.

3. The method for assessing the susceptibility of avalanche disasters based on GIS and machine learning according to claim 1, characterized in that: In step 2, the correlation analysis method is used to determine the correlation of the impact factors, and the evaluation factors are determined based on the correlation of the impact factors.

4. The method for assessing the susceptibility of avalanche disasters based on GIS and machine learning according to claim 3, characterized in that: Determining the evaluation factor according to the correlation of the impact factor includes: The influencing factors with correlation greater than the threshold are taken as candidate evaluation factors; The detection value of each candidate evaluation factor is determined, the non-linearly weakened candidate evaluation factors are eliminated according to the detection value, and the evaluation factor is determined according to the contribution rate of the remaining candidate evaluation factors.

5. The method for assessing the susceptibility of avalanche disasters based on GIS and machine learning according to claim 3, characterized in that: The correlation analysis method includes Pearson correlation coefficient method, Spearman rank correlation coefficient method or Kendall correlation coefficient method.

6. The method for assessing the susceptibility of avalanche disasters based on GIS and machine learning according to claim 1, characterized in that: The construction of the positive sample set and the negative sample set includes: Construct a positive sample set based on the relevant data of the evaluation factors extracted from the avalanche point; A buffer zone is set according to the development range of avalanche points. In the area outside the avalanche buffer zone, non-avalanche points with the same number as avalanche points are randomly selected. The relevant data in the evaluation factors are extracted from the non-avalanche points to construct a negative sample set.

7. The method for assessing the susceptibility of avalanche disasters based on GIS and machine learning according to claim 1, characterized in that: The machine learning models include support vector machines, multi-layer perceptrons, decision trees, logistic regression models, and random forest models; Confusion matrix, precision, recall rate, and ROC curve were used to verify the accuracy of each machine learning model after training, and the optimal machine learning model was determined as the avalanche susceptibility evaluation model.

8. The method for assessing the susceptibility of avalanche disasters based on GIS and machine learning according to claim 1, characterized in that: The preparation of the avalanche susceptibility zoning map in step 4 includes: According to the output results of the avalanche susceptibility assessment model, the evaluation factors are superimposed according to the weights to obtain an avalanche susceptibility assessment index map. The target area is divided into zones using the natural break point method to obtain an avalanche susceptibility zoning map. The avalanche susceptibility of each area is determined based on the number of grids, grid area and number of avalanches in each area.

9. A GIS and machine learning-based avalanche disaster susceptibility assessment system, characterized in that: include: An influencing factor module is used to determine the influencing factors that cause avalanches in the target area based on the terrain data, meteorological data and avalanche data of the target area; The factor analysis module is used to perform correlation test on the initially selected factors that affect avalanche occurrence and determine the evaluation factors; A model training module is used to construct a positive sample set and a negative sample set according to the evaluation factors and the avalanche point data information of the target area, and train the machine learning model according to the positive sample set and the negative sample set to obtain an avalanche susceptibility evaluation model after training; The evaluation module is used to prepare an avalanche susceptibility zoning map based on the prediction results of the avalanche susceptibility evaluation model and determine the avalanche susceptibility of each area.

10. An electronic device, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of the avalanche hazard susceptibility assessment method based on GIS and machine learning as described in any one of claims 1 to 8 when executing the computer program.

Images