Geological disaster risk prediction method and device

Abstract

The embodiment of the application is suitable for the field of computer technology, and provides a geological disaster risk prediction method and device, the method comprising: collecting meteorological and hydrological data of each point in a target area at different time points, the meteorological and hydrological data comprising: rainfall, surface runoff, air temperature, rainfall intensity; collecting geographical information of the target area; determining point features corresponding to each point based on the meteorological and hydrological data, the point features being capable of representing feedback features of the points to meteorological and hydrological conditions and change features of the feedback features over time; constructing a surface water system topology structure of the target area based on the meteorological and hydrological data and the geographical information, the surface water system topology structure being used to represent the correlation between each point in the target area; and predicting geological disaster risks of each point in the target area according to the point features and the surface water system topology structure. Through the above method, the accuracy of geological disaster risk prediction can be improved.

Description

Technical Field

[0001] This application belongs to the field of computer technology, and in particular relates to a method and device for predicting geological disaster risks. Background Technology

[0002] Meteorological factors such as rainfall are important causes of geological disasters such as landslides and debris flows. If high-risk areas for geological disasters can be identified in advance, preventive measures can be taken to avoid losses caused by geological disasters.

[0003] Existing methods for predicting geological disaster risks typically rely on time-series data analysis and statistical perspectives, using single or limited meteorological and hydrological factor data. This makes it difficult to analyze the deep-seated correlation between geological disasters and meteorological factors across spatial and temporal dimensions. Consequently, the accuracy of existing prediction methods is not high. Summary of the Invention

[0004] In view of this, embodiments of this application provide a geological disaster risk prediction method and equipment to improve the accuracy of geological disaster risk prediction.

[0005] The first aspect of this application provides a method for predicting geological disaster risks, including:

[0006] Meteorological and hydrological data of various points in the target area at different time points are collected. The meteorological and hydrological data include at least one of the following: rainfall, surface runoff, temperature, and rainfall intensity.

[0007] Collect geographic information of the target area, including watershed information and elevation information;

[0008] Based on the meteorological and hydrological data, the location characteristics corresponding to each location are determined. The location characteristics can characterize the feedback characteristics of the location to meteorological and hydrological conditions and the changes of the feedback characteristics over time.

[0009] Based on the meteorological and hydrological data and the geographic information, a surface water system topology structure of the target area is constructed, which is used to characterize the correlation between various points within the target area;

[0010] Based on the location characteristics and the surface water system topology, geological hazard risk prediction is performed for each location within the target area.

[0011] In one possible implementation, determining the location characteristics of each location based on the meteorological and hydrological data includes:

[0012] Based on the meteorological and hydrological data of each point in the target area at different time points, a tensor is constructed using the surface runoff of each point as the filling element;

[0013] Tensor decomposition is performed on the tensor to obtain the time-varying and space-varying characteristics of each point.

[0014] Based on the time-varying characteristics and the spatial-varying characteristics, the point characteristics of each point are calculated.

[0015] In one possible implementation, the tensor decomposition to obtain the time-varying and space-varying features of each point includes:

[0016] Based on the objective function, the tensor is decomposed into multiple decomposition results, including time-varying vectors and space-varying vectors. The objective function is used to characterize the generalized distance between the time-varying vectors and the space-varying vectors.

[0017] The time-varying vector and the space-varying vector corresponding to the minimum value of the objective function are determined as the time-varying feature and the space-varying feature, respectively.

[0018] In one possible implementation, the objective function is:

[0019]

[0020] Where T is the tensor. For the reconstructed tensor, Let T be an element of tensor T. Let T be an element of tensor T. For tensor In the equation, α and β are the control parameters for the divergence. These are the types of meteorological and hydrological data, respectively. Let K be the value of the objective function, and K be the set of known data in the tensor T.

[0021] In one possible implementation, constructing the surface water system topology of the target area based on the meteorological and hydrological data and the geographic information includes:

[0022] Based on the surface runoff information within the target area, connect the various points within the target area that are located in the same water system according to the direction of water flow;

[0023] Based on the watershed information and elevation information within the target area, connect the various points within the target area in elevation order;

[0024] Based on the topology map composed of the connected points, a topology matrix is ​​constructed, which is used to characterize the topology of the surface water system.

[0025] In one possible implementation, the step of predicting geological hazard risks for each point within the target area based on the point characteristics and the surface water system topology includes:

[0026] Based on historical data, known high-risk locations among the stated locations were identified;

[0027] Based on the location characteristics and the surface water system topology, the locations within the target area are clustered to divide each location into at least one cluster.

[0028] Based on the clusters and the known high-risk locations, predict whether each location is a predicted high-risk location.

[0029] In one possible implementation, the step of clustering the points within the target area based on the point characteristics and the surface water system topology to divide the points into at least one cluster includes:

[0030] Based on the location characteristics and the surface water system topology, construct an attribute adjacency matrix;

[0031] The surface water system topology and the attribute adjacency matrix are fused to obtain a fusion matrix;

[0032] Calculate the Laplacian matrix of the fusion matrix;

[0033] Based on the Laplacian matrix, determine the characteristic matrix;

[0034] Multiple clustering methods are used to cluster the feature matrix corresponding to each point, resulting in multiple clustering results;

[0035] Based on multiple clustering results, each point is divided into at least one cluster.

[0036] In one possible implementation, the surface water system topology and the attribute adjacency matrix are fused to obtain a fused matrix in the following manner:

[0037] A = λG + (1 - λ)E

[0038] Where A is the fusion matrix, G is the surface water system topology, E is the attribute adjacency matrix, and λ is the weight of the surface water system topology.

[0039] In one possible implementation, the characteristic matrix is ​​constructed based on the Laplacian matrix, including:

[0040] The Laplacian matrix is ​​decomposed into multiple eigenvalues, each of which has a corresponding eigenvector.

[0041] A predetermined number of target feature values ​​are determined from the multiple feature values ​​in ascending order of numerical value;

[0042] The feature matrix is ​​formed based on the feature vectors corresponding to the multiple target feature values.

[0043] In one possible implementation, predicting whether each location is a predicted high-risk location based on the cluster and the known high-risk locations includes:

[0044] Calculate the cluster purity, normalization information, and RAND index for each of the clusters;

[0045] Based on the cluster purity, the normalization information, and the RAND index, the category of each cluster is determined, and the category is used to characterize the similarity between the cluster and the known high-risk location.

[0046] Based on the category, points with a similarity higher than a preset value to the known high-risk locations are identified as predicted high-risk points.

[0047] A second aspect of this application provides a geological disaster risk prediction device, comprising:

[0048] The meteorological and hydrological data acquisition module is used to collect meteorological and hydrological data of various points in the target area at different time points. The meteorological and hydrological data includes at least one of the following: rainfall, surface runoff, temperature, and rainfall intensity.

[0049] A geographic information acquisition module is used to collect geographic information of the target area, including watershed information and elevation information.

[0050] The location feature determination module is used to determine the location features corresponding to each location based on the meteorological and hydrological data. The location features can characterize the feedback features of the location to meteorological and hydrological conditions and the change features of the feedback features over time.

[0051] The surface water system topology determination module is used to construct the surface water system topology of the target area based on the meteorological and hydrological data and the geographic information. The surface water system topology is used to characterize the correlation between various points in the target area.

[0052] The risk prediction module is used to predict the geological disaster risk of each point in the target area based on the point characteristics and the surface water system topology.

[0053] A third aspect of this application provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in the first aspect above.

[0054] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect above.

[0055] A fifth aspect of this application provides a computer program product that, when run on a computer device, causes the computer device to perform the method described in the first aspect.

[0056] Compared with the prior art, the embodiments of this application have the following advantages:

[0057] When performing geological hazard risk prediction based on the method in this application, multiple points can be identified for the target area to be predicted. Meteorological and hydrological data of each point in the target area at different times, along with the geographical information of the target area, are collected. Based on the meteorological and hydrological data, the spatiotemporal variation characteristics of the meteorological and hydrological data at each point are determined. Based on the meteorological and hydrological data and geographical information, a surface water system topology structure of the target area is constructed, which is used to characterize the relationships between various points within the target area. Based on the spatiotemporal variation characteristics of the points and the surface water system topology structure, geological hazard risk prediction is performed for each point within the target area. The method in this application, when performing geological hazard risk prediction, can improve the prediction accuracy by basing geological risk prediction on the temporal and spatial characteristics of meteorological and hydrological data at different points within the target area, as well as the geographical relationships between these points. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0059] Figure 1 This is a schematic flowchart illustrating the steps of a geological disaster risk prediction method provided in an embodiment of this application;

[0060] Figure 2 This is a schematic diagram of tensor construction provided in an embodiment of this application;

[0061] Figure 3 This is a schematic diagram of a topological structure construction provided in an embodiment of this application;

[0062] Figure 4 This is a schematic diagram of a geological disaster risk prediction device provided in an embodiment of this application;

[0063] Figure 5 This is a schematic diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0064] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0065] Geological disasters have complex causes, with meteorological factors such as rainfall being a significant contributor to landslides and debris flows. Existing methods often rely on time-series data analysis and statistics, using single or limited meteorological and hydrological data, such as rainfall and soil moisture content, for geological disaster early warning studies. These methods fail to comprehensively organize and deeply mine the rich meteorological and hydrological data and historical geological disaster data, making it difficult to analyze the deep-seated correlations between geological disasters and meteorological factors across time and space. Therefore, this application designs a suitable data structure to accommodate multi-source, time-varying meteorological and hydrological data, maximizing the presentation of data correlations. Simultaneously, a tensor representation method based on generalized distance is used to extract time-varying attributes from different locations and embed these attributes into the calculated location topology graph to construct an attribute graph. Graph clustering is then used to identify similar nodes. Finally, combining historical geological disaster data and known location risk assessment data, the geological disaster risk at each location is analyzed.

[0066] The technical solution of this application will be described below through specific embodiments.

[0067] Reference Figure 1 The diagram illustrates a step-by-step flowchart of a geological disaster risk prediction method provided in an embodiment of this application, which may specifically include the following steps:

[0068] S101, collect meteorological and hydrological data of various points in the target area at different time points, wherein the meteorological and hydrological data includes at least one of the following: rainfall, surface runoff, temperature, and rainfall intensity.

[0069] The execution subject of this embodiment can be a computer device, and this embodiment does not limit the specific type of computer device.

[0070] The target area mentioned above can be the entire area to be monitored, which can be divided into multiple area blocks. Each area block can be represented by a point. These points can be used to refer to different area blocks within the target area.

[0071] For example, the target area can be divided into i*j blocks by matrix partitioning, and the location of each block can be represented by the two values ​​i and j.

[0072] The meteorological and hydrological data at the aforementioned locations can be represented by the average value of the entire regional block, or by the meteorological and hydrological data of the central coordinate point within the entire regional block; no limitation is made here. The meteorological and hydrological data can be obtained from publicly available meteorological data, reports, weather forecasts, and other information; no limitation is made here either.

[0073] S102, Collect geographic information of the target area, including watershed information and elevation information.

[0074] A watershed is a mountain range or high ground that separates adjacent watersheds. Elevation can characterize the height of each point. The above geographical information can be used to characterize the geographical relationships between various points in a target area, such as elevation trends, river system connections, and watershed trends. For example, points may be located on the same river system and connected sequentially according to the direction of river flow. Points may be located on the same watershed, and points may be located at the same elevation.

[0075] S103, Based on the meteorological and hydrological data, determine the location characteristics corresponding to each location, wherein the location characteristics can characterize the feedback characteristics of the location to meteorological and hydrological conditions and the change characteristics of the feedback characteristics over time.

[0076] Computer equipment can construct tensors based on meteorological and hydrological data of various points in the target area at different time points, using the surface runoff of each point as the filling element; perform tensor decomposition on the tensors to obtain the time-varying and spatial-varying characteristics of each point; and calculate the point characteristics of each point based on the time-varying and spatial-varying characteristics.

[0077] In tensor decomposition, the computer device can perform tensor decomposition based on the objective function to obtain various decomposition results, including time-varying vectors and space-varying vectors. The objective function is used to characterize the generalized distance between the time-varying vectors and the space-varying vectors. The time-varying vectors and space-varying vectors corresponding to the minimum value of the objective function are determined as time-varying features and space-varying features.

[0078] In one possible implementation, the target area N Point coordinate information n i，j ,in i =1, 2, 3, ...., j =1, 2, 3, ..., data time range P Hourly rainfall at the locationr ( n i，j ，t ), Surface runoff at specific locations (time-of-use). s ( n i，j ，t ), point-of-sale temperature w ( n i，j ，t ), hourly rainfall intensity rs ( n i，j ，t ),in t =1, 2, 3, ... P Surface water system topology, contour lines, basin and watershed information; site risk assessment data. d ( n i，j ，t ).

[0079] When constructing tensors, the point coordinates can be vectorized, that is... n i，j Mapped to Determine the range of hourly rainfall. r min , r max ], Time-of-day temperature range at the location [ w min , w max ], hourly rainfall intensity [ rs min , rs max Set the value range segmentation coefficients respectively. rn , wn and rsn .

[0080] Data fusion is performed using a tensor data structure. The model is defined as follows: location, time, hourly rainfall, hourly temperature, and hourly rainfall intensity. Meteorological and hydrological factors require initial segmentation based on their value range coefficients. rn , wn and rsn The process involves segmenting the data, and finally, using the time-varying surface runoff at each location as the value, it is placed into the corresponding meteorological and hydrological factor segment space to construct a 5th-order tensor. ,in I 1=| N |, I 2= P , I3=( r max -r min ) / rn , I 4=( w max - w min ) / wn , I 5=( rs max -rs min ) / rsn .

[0081] Figure 2 This is an example of constructing a third-order tensor using location, time, and hourly rainfall as models and surface runoff as values. The gray area represents the surface runoff reading, and the light-colored area is empty.

[0082] Based on CP decomposition, a low-rank feature space is constructed. Considering the rich physical meaning and statistical distribution of geological disaster big data, the following was introduced during the modeling process: α - β Divergence allows us to construct an objective function of the following form:

[0083]

[0084] Where T is a tensor. For the reconstructed tensor, Let T be an element of tensor T. Let T be an element of tensor T. For tensor In the equation, α and β are the control parameters for the divergence. These are the types of meteorological and hydrological data. Let K be the value of the objective function, and K be the set of known data in the tensor T.

[0085] in, α and β yes α - β The control parameters for divergence. By changing the control parameters, the objective function of the low-rank representation learning model for geological disaster big data can exhibit different data representation capabilities. When α=β=1, the objective function is transformed into a function based on Euclidean distance. Optimizing the objective function can obtain the feature vectors corresponding to the land parcels. and the feature vector Y corresponding to each time. t By taking the outer product of specific X and Y values, the time-varying point features can be obtained. .

[0086] S104. Based on the meteorological and hydrological data and the geographic information, construct the surface water system topology of the target area. The surface water system topology is used to characterize the correlation between various points in the target area.

[0087] The causes of geological hazards are complex, and the interactions between different locations cannot be ignored. Therefore, this application, based on the fusion of geological hazard big data at various locations, will further utilize geographic information data to construct a relationship map between locations. For example, surface runoff can reflect the spatial relationship between locations to a certain extent, and the relationship between locations can be abstracted and simplified into the relationship of water body activity between locations.

[0088] In one possible implementation, the computer device can connect various points within the target area based on surface runoff information; connect various points within the target area based on watershed and elevation information; and construct a topology matrix based on the connection topology map formed by the various points. This topology matrix is ​​used to characterize the surface water system topology. Specifically, during the connection process, points can be connected according to the direction of water flow; points located within the same watershed can be connected; points located at the same elevation can be connected; and points located on the same contour line can be connected.

[0089] Figure 3 This is a schematic diagram of a topological structure construction provided in an embodiment of this application; as shown... Figure 3 As shown, the target area is first divided to obtain a set of points. For example... Figure 3 As shown in (a), the target area can be divided into 9 blocks, each corresponding to a location. Then, by filling depressions in the target area using geographic information software, the water system of the target area is obtained, and the corresponding blocks are connected based on the water system. For example... Figure 3 As shown in (b), points within the same water system can be connected according to the direction of water flow; furthermore, connections between all points are completed based on information such as elevation and watershed. Figure 3 As shown in (c), points on the same watershed can be connected, as can points on the same contour line. After completing the connections, a topological map of the points can be obtained, i.e. Figure 3 The final connection diagram is shown in (c) above.

[0090] The above method allows us to obtain the topological structure between points in the target area. This topological relationship is a directed graph, which can include directions such as water flow direction, altitude, and watershed orientation. Therefore, the topological graph reflects the relationships between points, and the topological structure G can be output based on the graph. The topological structure G can be represented using a matrix.

[0091] S105, Based on the location characteristics and the surface water system topology, perform geological disaster risk prediction for each location within the target area.

[0092] Computer equipment can identify known high-risk locations based on historical data. For example, locations where major geological disasters have occurred within a preset time period can be identified as known high-risk locations.

[0093] Based on location characteristics and surface water system topology, points within the target area are clustered to divide each point into at least one cluster. In one possible implementation, during clustering, a computer device constructs an attribute adjacency matrix based on location characteristics and surface water system topology; the surface water system topology and attribute adjacency matrix are fused to obtain a fusion matrix; the Laplace matrix of the fusion matrix is ​​calculated; a feature matrix is ​​determined based on the Laplace matrix; multiple clustering methods are used to cluster the feature matrices corresponding to each point, resulting in multiple clustering results; and based on these multiple clustering results, each point is divided into at least one cluster.

[0094] For example, based on features Z and topology G, an attribute adjacency matrix E is constructed, where E(h, o) in the attribute adjacency matrix E represents the attribute similarity. Features Z can be the aforementioned point features. The process of determining the location features can be described as above. E(h, o) in the attribute adjacency matrix E can represent the attribute similarity between location h and location o, where attribute similarity can be represented by cosine similarity. Location h and location o can be represented by i*j. Then, the topological structure G and the attribute adjacency matrix E can be fused to obtain the fusion matrix A, A = λG + (1-λ)E, where λ is the weight. Based on the fusion matrix A, the fusion Laplace matrix L = DA can be calculated, where D is the degree matrix of A. The eigenvectors corresponding to the first n smallest eigenvalues ​​of L are taken to form the feature matrix U. Each row of U is normalized, and then various clustering methods, such as DBSCAN and K-means clustering (where the value of k is consistent with the number of categories in the location risk assessment data), are used to evaluate the optimal result combination through "compactness within clusters" and "separation between clusters".

[0095] Based on clusters and labeled high-risk locations, predict whether each location is a predicted high-risk location.

[0096] Computer equipment can calculate the cluster purity, normalized mutual information (NMI), and RAND index (RI) of each cluster, sum them, assess label consistency, determine the category of each cluster, and further identify high-risk locations. Specifically, computer equipment can determine the category of each cluster based on cluster purity, normalized mutual information, and RAND index; based on the category, it can determine whether the points in the cluster are predicted high-risk locations.

[0097] Cluster purity characterizes the uniformity of categories within each cluster, representing the concentration of samples within each cluster belonging to the same reference category. For example, the total number of nodes in each cluster and the number of nodes belonging to the high-risk category can be counted. The high-risk purity of a single cluster is obtained by the ratio of the number of high-risk nodes to the total number of nodes in that cluster. Then, a weighted average of all single-cluster purities is calculated using the number of nodes in each cluster as a weight, ultimately yielding the overall cluster high-risk purity. A cluster purity value closer to 1 indicates a higher concentration of high-risk nodes within the cluster.

[0098] Normalized information can characterize the degree of information matching between clustering results and reference categories. For example, the entropy of the clustering results and the entropy of the high-risk labels can be calculated separately. The entropy value reflects the uncertainty of the corresponding classification system. Then, the mutual information between the two can be calculated by statistically analyzing the distribution of points in each cluster and each risk label. The mutual information quantifies the closeness of the association between the two classification systems. Finally, the mutual information is divided by the maximum value of the clustering result entropy and the high-risk label entropy to complete the normalization process, ensuring that the index value is constrained between 0 and 1. The closer the normalized value is to 1, the higher the information overlap between the clustering structure and the high-risk distribution.

[0099] The RAND index can characterize the overall consistency between clustering results and reference categories. For example, the total number of sample pairs formed by all points can be counted, and then consistent in-group sample pairs and consistent out-of-group sample pairs can be distinguished. Consistent in-group sample pairs refer to point combinations that belong to both the same cluster and the same risk category, while consistent out-of-group sample pairs refer to point combinations that belong to neither the same cluster nor the same risk category. Finally, the RAND index is obtained by dividing the total number of consistent in-group and consistent out-of-group sample pairs by the total number of all sample pairs. The RAND index ranges from 0 to 1, with a value closer to 1 indicating a better overall match between the clustering results and the high-risk label.

[0100] When determining the similarity category of clusters based on the three indicators mentioned above, indicator weights can be set. For example, cluster purity directly reflects the concentration of high-risk points within a single cluster and can be given the highest weight; normalized mutual information can supplement the reflection of the inter-cluster distinction between the cluster structure and the high-risk distribution; and the RAND index ensures the global matching between the cluster and the high-risk label. Both can be given medium weights. The three indicators for each cluster are normalized to ensure they fall within a uniform range of 0 to 1. If an indicator meets this range requirement, it is used directly. Then, the normalized indicators are weighted and summed according to the set weights to obtain the comprehensive similarity score for each cluster. A threshold is set based on the comprehensive similarity score to divide the clusters into different similarity categories. The higher the score of the cluster, the higher the similarity to known high-risk locations. This category division clearly defines the degree of risk association between each cluster.

[0101] High-risk locations are predicted based on the similarity categories of clusters. A preset similarity threshold is set, which can be adjusted according to the geographical characteristics of the target area and actual business needs to ensure the flexibility and adaptability of the screening criteria. Clusters with a comprehensive similarity score higher than the preset threshold are identified as high-similarity clusters. These clusters perform well in terms of the concentration of high-risk locations, the degree of information matching with the high-risk distribution, and overall consistency, indicating that they have a strong similarity to known high-risk locations. All locations within high-similarity clusters are then identified as predicted high-risk locations.

[0102] In one possible implementation, based on historical meteorological and hydrological data, known high-risk locations can be identified from multiple locations in the target area. After determining the clusters, locations within the same cluster as the known high-risk locations can be predicted as high-risk locations.

[0103] In one possible implementation, known high-risk locations can have associated high-risk information; for example, known high-risk locations could be high-rainfall hazard areas, landslide hazard areas, debris flow hazard areas, etc. Locations within the same cluster can be predicted to have the same geological risk as known high-risk locations.

[0104] The method in this application can use a high-order tensor structure to accommodate various meteorological and hydrological data, achieving a unified representation of different types of data; it uses a tensor decomposition model based on a generalized distance metric to achieve efficient and accurate acquisition of the time-varying characteristics of different plots, i.e., different locations; and it constructs node topological relationships based on surface spatial information to achieve a structured representation of the time-varying hydrological attributes of plots, which is conducive to exploring the disaster evolution characteristics of inter-plot linkages and achieving accurate identification of high-risk points.

[0105] The method described in this application can be applied to multiple fields. In the field of urban planning and construction, the technology of this application can be used to conduct geological hazard risk assessments of construction areas during the urban planning and construction process. Based on the assessment results, urban layouts can be rationally planned to avoid large-scale construction in areas with high geological hazard risks, or corresponding preventative measures can be taken to ensure the safety of urban construction.

[0106] The method described in this application can be applied to the field of transportation engineering. For the construction and operation of transportation projects such as highways, railways, and bridges, the technology of this application can be used to assess the geological hazard risks along the route. During the site selection stage, high-risk areas for geological hazards can be avoided; during the construction and operation of the project, geological hazard risks can be monitored in real time, and protective measures can be taken in a timely manner to ensure the safety of transportation facilities.

[0107] In the planning, design, and construction of water conservancy projects, the impact of geological hazards on the projects needs to be considered. The technology in this application can be used to assess the geological hazard risks surrounding water conservancy projects, providing a basis for the rational design and construction of the projects. Furthermore, during the operation of the projects, it can also be used to monitor geological hazard risks, ensuring the safe operation of the water conservancy projects.

[0108] The method described in this application can be applied to the insurance industry. Insurance companies can use the technology presented in this application to assess geological disaster risks within their coverage areas, thereby more accurately determining insurance rates and terms. For areas and customers with high geological disaster risks, insurance rates can be appropriately increased or other risk management measures can be taken to reduce the insurance company's payout risk.

[0109] It should be noted that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0110] Reference Figure 4 The diagram illustrates a geological disaster risk prediction device provided in an embodiment of this application. Specifically, it may include a meteorological and hydrological data acquisition module 41, a geographic information acquisition module 42, a location feature determination module 43, a surface water system topology determination module 44, and a risk prediction module 45, wherein:

[0111] The meteorological and hydrological data acquisition module 41 is used to collect meteorological and hydrological data of various points in the target area at different time points. The meteorological and hydrological data includes at least one of the following: rainfall, surface runoff, temperature, and rainfall intensity.

[0112] Geographic information acquisition module 42 is used to acquire geographic information of the target area, including watershed information and elevation information;

[0113] The point feature determination module 43 is used to determine the point features corresponding to each point based on the meteorological and hydrological data. The point features can characterize the spatiotemporal variation characteristics of the meteorological and hydrological data of the point, and the point features can characterize the feedback characteristics of the point to meteorological and hydrological conditions and the change characteristics of the feedback characteristics over time.

[0114] The surface water system topology determination module 44 is used to construct the surface water system topology of the target area based on the meteorological and hydrological data and the geographic information. The surface water system topology is used to characterize the correlation between various points in the target area.

[0115] The risk prediction module 45 is used to predict the geological disaster risk of each point in the target area based on the point characteristics and the surface water system topology.

[0116] In one possible implementation, determining the location characteristics of each location based on the meteorological and hydrological data includes:

[0117] Based on the meteorological and hydrological data of each point in the target area at different time points, a tensor is constructed using the surface runoff of each point as the filling element;

[0118] Tensor decomposition is performed on the tensor to obtain the time-varying and space-varying characteristics of each point.

[0119] Based on the time-varying characteristics and the spatial-varying characteristics, the point characteristics of each point are calculated.

[0120] In one possible implementation, the tensor decomposition to obtain the time-varying and space-varying features of each point includes:

[0121] Based on the objective function, the tensor is decomposed into multiple decomposition results, including time-varying vectors and space-varying vectors. The objective function is used to characterize the generalized distance between the time-varying vectors and the space-varying vectors.

[0122] The time-varying vector and the space-varying vector corresponding to the minimum value of the objective function are determined as the time-varying feature and the space-varying feature, respectively.

[0123] In one possible implementation, the objective function is:

[0124]

[0125] Where T is the tensor. For the reconstructed tensor, Let T be an element of tensor T. Let T be an element of tensor T. For tensor In the equation, α and β are the control parameters for the divergence. These are the types of meteorological and hydrological data, respectively. Let K be the value of the objective function, and K be the set of known data in the tensor T.

[0126] In one possible implementation, constructing the surface water system topology of the target area based on the meteorological and hydrological data and the geographic information includes:

[0127] Based on the surface runoff information within the target area, connect the various points within the target area that are located in the same water system according to the direction of water flow;

[0128] Based on the watershed information and elevation information within the target area, connect the various points within the target area in elevation order;

[0129] Based on the topology map composed of the connected points, a topology matrix is ​​constructed, which is used to characterize the topology of the surface water system.

[0130] In one possible implementation, the step of predicting geological hazard risks for each point within the target area based on the point characteristics and the surface water system topology includes:

[0131] Based on historical data, known high-risk locations among the stated locations were identified;

[0132] Based on the location characteristics and the surface water system topology, the locations within the target area are clustered to divide each location into at least one cluster.

[0133] Based on the clusters and the known high-risk locations, predict whether each location is a predicted high-risk location.

[0134] In one possible implementation, the step of clustering the points within the target area based on the point characteristics and the surface water system topology to divide the points into at least one cluster includes:

[0135] Based on the location characteristics and the surface water system topology, construct an attribute adjacency matrix;

[0136] The surface water system topology and the attribute adjacency matrix are fused to obtain a fusion matrix;

[0137] Calculate the Laplacian matrix of the fusion matrix;

[0138] Based on the Laplacian matrix, determine the characteristic matrix;

[0139] Multiple clustering methods are used to cluster the feature matrix corresponding to each point, resulting in multiple clustering results;

[0140] Based on multiple clustering results, each point is divided into at least one cluster.

[0141] In one possible implementation, the surface water system topology and the attribute adjacency matrix are fused to obtain a fusion matrix in the following manner:

[0142] A = λG + (1 - λ)E

[0143] Where A is the fusion matrix, G is the surface water system topology, E is the attribute adjacency matrix, and λ is the weight of the surface water system topology.

[0144] In one possible implementation, the characteristic matrix is ​​constructed based on the Laplacian matrix, including:

[0145] The Laplacian matrix is ​​decomposed into multiple eigenvalues, each of which has a corresponding eigenvector.

[0146] A predetermined number of target feature values ​​are determined from the multiple feature values ​​in ascending order of numerical value;

[0147] The feature matrix is ​​formed based on the feature vectors corresponding to the multiple target feature values.

[0148] In one possible implementation, predicting whether each location is a predicted high-risk location based on the cluster and the known high-risk locations includes:

[0149] Calculate the cluster purity, normalization information, and RAND index for each of the clusters;

[0150] Based on the cluster purity, the normalization information, and the RAND index, the category of each cluster is determined, and the category is used to characterize the similarity between the cluster and the known high-risk location.

[0151] Based on the category, points with a similarity higher than a preset value to the known high-risk locations are identified as predicted high-risk points.

[0152] As the apparatus embodiments are basically similar to the method embodiments, they are described in a relatively simple manner. For relevant details, please refer to the description in the method embodiment section.

[0153] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Figure 5As shown, the computer device 5 of this embodiment includes: at least one processor 50 ( Figure 5 (Only one is shown in the diagram), memory 51, and computer program 52 stored in said memory 51 and executable on said at least one processor 50, wherein said processor 50 executes said computer program 52 to implement the steps in any of the above method embodiments.

[0154] The computer device 5 may be a desktop computer, laptop, handheld computer, or cloud computing device, etc. This computer device may include, but is not limited to, a processor 50 and a memory 51. Those skilled in the art will understand that... Figure 5 The computer device 5 is merely an example and does not constitute a limitation on the computer device 5. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0155] The processor 50 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0156] In some embodiments, the memory 51 may be an internal storage unit of the computer device 5, such as a hard disk or memory of the computer device 5. In other embodiments, the memory 51 may be an external storage device of the computer device 5, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 5. Furthermore, the memory 51 may include both internal and external storage units of the computer device 5. The memory 51 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory 51 can also be used to temporarily store data that has been output or will be output.

[0157] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0158] This application provides a computer program product that, when run on a computer device, enables the computer device to perform the steps described in the above-described method embodiments.

[0159] The embodiments described above are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for predicting geological disaster risks, characterized in that, include: Meteorological and hydrological data of various points in the target area at different time points are collected. The meteorological and hydrological data include at least one of the following: rainfall, surface runoff, temperature, and rainfall intensity. Collect geographic information of the target area, including watershed information and elevation information; Based on the meteorological and hydrological data of each point in the target area at different time points, a tensor is constructed using the surface runoff of each point as the filling element; Tensor decomposition is performed on the tensor to obtain the time-varying and space-varying characteristics of each point. Based on the time-varying characteristics and the spatial-varying characteristics, the location characteristics of each location are calculated. The location characteristics can characterize the feedback characteristics of the location to meteorological and hydrological conditions and the changes of the feedback characteristics over time. Based on the meteorological and hydrological data and the geographic information, a surface water system topology structure of the target area is constructed, which is used to characterize the correlation between various points within the target area; Based on historical data, known high-risk locations among the stated locations were identified; Based on the location characteristics and the surface water system topology, the locations within the target area are clustered to divide each location into at least one cluster. Based on the clusters and the known high-risk locations, predict whether each location is a predicted high-risk location.

2. The method as described in claim 1, characterized in that, The tensor decomposition process, which yields the time-varying and space-varying characteristics of each point, includes: Based on the objective function, the tensor is decomposed into multiple decomposition results, including time-varying vectors and space-varying vectors. The objective function is used to characterize the generalized distance between the time-varying vectors and the space-varying vectors. The time-varying vector and the space-varying vector corresponding to the minimum value of the objective function are determined as the time-varying feature and the space-varying feature, respectively.

3. The method as described in claim 2, characterized in that, The objective function is: Where T is the tensor. For the reconstructed tensor, Let T be an element of tensor T. Let T be an element of tensor T. For tensor In the equation, α and β are the control parameters for the divergence. These are the types of meteorological and hydrological data, respectively. Let K be the value of the objective function, and K be the set of known data in the tensor T.

4. The method as described in claim 1, characterized in that, The construction of the surface water system topology of the target area based on the meteorological and hydrological data and the geographic information includes: Based on the surface runoff information within the target area, connect the various points within the target area that are located in the same water system according to the direction of water flow; Based on the watershed information and elevation information within the target area, connect the various points within the target area in elevation order; Based on the topology map composed of the connected points, a topology matrix is ​​constructed, which is used to characterize the topology of the surface water system.

5. The method as described in claim 1, characterized in that, The step of clustering points within the target area based on the point characteristics and the surface water system topology to divide the points into at least one cluster includes: Based on the location characteristics and the surface water system topology, construct an attribute adjacency matrix; The surface water system topology and the attribute adjacency matrix are fused to obtain a fusion matrix; Calculate the Laplacian matrix of the fusion matrix; Based on the Laplacian matrix, determine the characteristic matrix; Multiple clustering methods are used to cluster the feature matrix corresponding to each point, resulting in multiple clustering results; Based on multiple clustering results, each point is divided into at least one cluster.

6. The method as described in claim 5, characterized in that, Based on the Laplace matrix, the characteristic matrix is ​​determined, including: The Laplacian matrix is ​​decomposed into multiple eigenvalues, each of which has a corresponding eigenvector. A predetermined number of target feature values ​​are determined from the multiple feature values ​​in ascending order of numerical value; The feature matrix is ​​formed based on the feature vectors corresponding to the multiple target feature values.

7. The method according to any one of claims 5-6, characterized in that, The step of predicting whether each location is a predicted high-risk location based on the cluster and the known high-risk locations includes: Calculate the cluster purity, normalization information, and RAND index for each of the clusters; Based on the cluster purity, the normalization information, and the RAND index, the category of each cluster is determined, and the category is used to characterize the similarity between the cluster and the known high-risk location. Based on the category, points with a similarity higher than a preset value to the known high-risk locations are identified as predicted high-risk points.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-7.

Images