Earthquake landslide disaster chain risk assessment method, medium and equipment

Abstract

The invention provides an earthquake landslide disaster chain risk assessment method, medium and equipment, and relates to the technical field of geological disaster chain risk assessment, and the method comprises the steps: obtaining earthquake-induced disaster chain data, and constructing an earthquake landslide disaster chain knowledge graph; according to the knowledge graph, acquiring disaster influence factors, determining association among the disaster influence factors, and determining key disaster influence factors with relatively high association degree; key disaster influence factors are used as Bayesian network nodes, node variable data are discretized, a maximum-minimum hill climbing algorithm is used for structural learning, and a directed acyclic graph structure of the Bayesian network is determined; performing Bayesian network parameter learning by using an expectation maximization algorithm to obtain a prior probability of Bayesian network nodes; if the prediction result of the model does not reach the expectation, the association between the nodes is adjusted, and if the prediction result reaches the expectation, the model is used for earthquake landslide disaster chain risk assessment. According to the scheme, the reliability of the prediction result can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster chain risk assessment, and in particular to a method, medium, and device for earthquake landslide disaster chain risk assessment. Background Art

[0002] Earthquake disasters have a huge impact. There are not only huge casualties and property losses caused by the earthquake itself, but also the impacts of landslides induced by earthquakes and their chain disasters cannot be ignored. Earthquake disasters induce landslides, and the landslides block rivers to form barrier lakes. The formed barrier dams are unstable. Once they burst, the floods caused by the discharge of lake water will undoubtedly bring serious harm to downstream residents and threaten the infrastructure along the line. The bursting of barrier dams has characteristics such as suddenness and difficulty in prediction and control. Therefore, it is very necessary to evaluate the stability of barrier dams. The risk assessment method for earthquake landslide geological disaster chains based on dynamic Bayesian networks has a significant effect on predicting landslide scale, flood disaster risk of barrier dams, and scientifically formulating emergency plans. The volume of the landslide is an evaluation index for the size of the landslide scale. The stability and instability of the barrier dam are used as evaluation indexes for the stability of the barrier dam, and the size of the storage capacity of the barrier lake is used as an evaluation index for the danger level of the barrier lake. In evaluating the danger of earthquake disaster chains, existing methods consider qualitative evaluation methods of the topography, slope, elevation difference of landslides, and the material composition of the barrier dam body. There are also many models that are in the primary stage of qualitative description and semi-quantitative statistics for the research of disaster chains. These methods cannot well describe the structure of landslide disaster chains and the possibility of disasters occurring.

[0003] The Bayesian network (BN) can combine probability methods with clear charts to better describe the causal relationship between variables, and thus more comprehensively conduct quantitative analysis on the structure and risk assessment of disaster chains. At the same time, the BN model can also provide a suitable framework for dealing with the uncertainty and complexity in the landslide disaster chain system, and is suitable for expressing and analyzing probabilistic and uncertain events. It can combine prior knowledge and data and is suitable for situations with limited data. The risk assessment method for earthquake landslide disaster chains based on the Bayesian network can solve the problem of strong subjectivity in risk assessment, comprehensively consider various factors such as the material composition of the barrier dam, the geomorphic parameters of the barrier lake, the height and length of the barrier lake dam, etc. Moreover, the constructed model will have continuously enhanced reliability as the data is continuously expanded. In addition, a node in the disaster chain often affects not only a specific node. For example, the height of the barrier dam affects the volume of the barrier dam and the storage capacity of the barrier lake. The influence between nodes can be inversely deduced and adjusted through the verified degree of fit, making the risk assessment more accurate. Summary of the Invention

[0004] The purpose of the present invention is to propose a method for earthquake landslide disaster chain risk assessment to solve the problems of insufficient adaptability and stability of the microseismic detection data processing algorithm in the existing geological exploration field, including the following steps: S1. Obtain earthquake-induced disaster chain data and construct a knowledge graph of earthquake landslide disaster chains; S2. According to the knowledge graph, obtain disaster influencing factors, determine the associations between disaster influencing factors, and determine the key disaster influencing factors with the degree of association greater than the set threshold; S3. Use the key disaster influencing factors as Bayesian network nodes, discretize the node variable data, and perform structure learning (BNSL) on the discretized node variable data using the maximum-minimum hill-climbing algorithm (MMHC) to determine the directed acyclic graph structure of the Bayesian network; S4. Based on the directed acyclic graph (DAG) structure of the Bayesian network, use the expectation maximization algorithm (EM algorithm) for Bayesian network parameter learning to obtain the prior probabilities of the Bayesian network nodes; S5. Obtain the prediction results of the Bayesian network based on historical earthquake-induced disaster chain data. If the deviation between the prediction results and the actual results is greater than the preset value, adjust the associations between the nodes and return to step S3. If the deviation between the prediction results and the actual results is less than or equal to the preset value, use the Bayesian network disaster chain model for earthquake landslide disaster chain risk assessment.

[0005] Furthermore, the key disaster influencing factors include: earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, front-back edge height difference, rock strength, river, height of the barrier lake dam, width of the barrier lake dam, volume of the barrier lake dam, material of the barrier lake dam, valley shape, landslide volume, storage capacity of the barrier lake, and stability of the barrier lake dam.

[0006] Furthermore, divide the Bayesian network nodes into four levels according to the degree of association with each other. Among them, the first-level nodes include: earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, front-back edge height difference, rock strength, river, height of the barrier lake dam, width of the barrier lake dam, material of the barrier lake dam, valley shape; the second-level node includes: landslide volume; the third-level nodes include: storage capacity of the barrier lake, volume of the barrier lake dam; the fourth-level node includes: stability of the barrier lake dam.

[0007] Furthermore, The specific method for determining the node probability distribution is: Divide the states of the nodes and determine the value ranges of each state of the nodes; based on the earthquake-induced disaster chain data and the value ranges of each state of the nodes, determine the probability distribution of the nodes.

[0008] Furthermore, the joint distribution probability of the Bayesian network nodes is: ; Among them, represents the joint distribution probability of 16 Bayesian network nodes, represents the earthquake magnitude, represents the elevation, represents the vegetation coverage rate, represents the slope, represents the geological structure, represents the elevation difference between the front and rear edges, represents the rock strength, represents the river, represents the height of the barrier lake dam, represents the width of the barrier lake dam, represents the volume of the barrier lake dam, represents the material of the barrier lake dam body, represents the shape of the river valley, represents the volume of the landslide, represents the storage capacity of the barrier lake, represents the stability of the barrier lake dam, , , , , , , , , , , , respectively represent the probabilities of each node in a certain state; represents the probability of the volume of the landslide occurring in a certain state under the influence of a certain state based on the earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, elevation difference between the front and rear edges, and rock strength; represents the probability of the storage capacity of the barrier lake occurring in a certain state under the influence of a certain state based on the volume of the landslide, river, and height of the barrier lake dam; represents the probability of the volume of the barrier lake dam occurring in a certain state under the influence of a certain state based on the volume of the landslide, height of the barrier lake dam, and width of the barrier lake dam; represents the probability of the stability of the barrier lake dam occurring in a certain state under the influence of a certain state based on the storage capacity of the barrier lake, volume of the barrier lake dam, material of the barrier lake dam body, and shape of the river valley.

[0009] Furthermore, the expectation-maximization algorithm is used for Bayesian network parameter learning, specifically: Initialize the parameters as the states of the observed Bayesian network nodes: , where the node lacks state data, and the missing state data is the latent variable Z; E-step: Start iteration according to the following formula to calculate the expectation of the latent variable Z: ; where, represents the log-likelihood function, represents the joint distribution probability of under the parameters Z and ; represents the conditional probability distribution of the latent variable under the given observed data and the current parameter estimate Z ; represents the i -th iteration estimate; M-step: Update the parameter using the expectation of the latent variable calculated in the E-step: ; where represents the i -th iteration updated estimate; When the norm of is less than the set iteration convergence condition, stop the iteration and output the at this time as the network conditional probability distribution.

[0010] Furthermore, use the Brier score to test the prediction results of the Bayesian network disaster chain model, and the calculation is as follows: ; where BrierScore represents the Brier score, N represents the number of samples, represents the number of classes of the t -th task, represents the true label of the i -th sample belonging to the t -th class in the j -th task, represents the probability that the model predicts the i -th sample belongs to the t -th class in the j -th task.

[0011] The present invention also proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned seismic landslide disaster chain risk assessment method is implemented.

[0012] The present invention also provides an electronic device, including a processor and a memory, where the processor is connected to the memory. The memory is configured to store a computer program, and the computer program includes computer-readable instructions. The processor is configured to call the computer-readable instructions to execute the above-described seismic landslide disaster chain risk assessment method.

[0013] The beneficial effects brought by the technical solution provided by the present invention are as follows: Based on the landslide disaster chain knowledge graph, the present invention constructs a Bayesian network to quantitatively assess the risk of the seismic-induced landslide disaster chain. The knowledge graph can continuously expand the case database. The dynamic Bayesian network can adjust the relationships between disaster nodes to optimize the relationships between various factors, improve the reliability of prediction results, suppress the occurrence of landslide disasters, and minimize disaster losses to the greatest extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a flowchart of the seismic landslide disaster chain risk assessment method according to an embodiment of the present invention; Figure 2 is the division and correlation relationship of each level node according to an embodiment of the present invention; Figure 3 is a schematic diagram of the Bayesian network model of four variables according to an embodiment of the present invention; Figure 4 is the Bayesian network diagram of the seismic landslide disaster chain according to an embodiment of the present invention; Figure 5 is the Bayesian network model with prior probabilities obtained through machine learning according to an embodiment of the present invention; Figure 6 is the landslide posterior probability model of the research area obtained according to an embodiment of the present invention; Figure 7 is a block diagram of an electronic device in an exemplary embodiment according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0016] The flowchart of the seismic landslide disaster chain risk assessment method according to the embodiment of the present invention is as Figure 1 , and specifically includes the following steps: S1. Obtain seismic-induced disaster chain data and construct a seismic landslide disaster chain knowledge graph.

[0017] Use AI tools to collect and screen literature related to earthquake-induced landslide disasters and their secondary disasters. The AI tools also extract relevant disaster data. Adopt bibliometric analysis methods, and based on the collected literature as data, analyze the development status and trends of the earthquake-induced landslide disaster chain, and generate a knowledge map of this related field through VOSviewer software.

[0018] The knowledge map of the earthquake landslide disaster chain takes "risk analysis" as the core and constructs a geological disaster risk assessment system, which includes four major modules: (1) Risk assessment: Quantify the disaster risk level through hazard assessment, stability analysis and numerical simulation; (2) Disaster chain logic: Reveal the chain evolution path of "earthquake → landslide → river blockage → barrier lake → dam break flood", and identify key nodes (such as landslide river blockage, dam break of the barrier dam); (3) Disaster types and mechanisms: Cover types such as landslides (high-speed long-distance landslides, rock landslides), debris flows, barrier lakes, etc., and analyze the dam break mechanism, flood evolution and landslide-debris flow movement characteristics; (4) Technical methods: Combine specific cases, use numerical simulation and model tests to reproduce the process of the disaster chain, and use the VOSviewer tool to mine the complex correlation network between elements. The map provides structured decision support for disaster early warning and prevention through the logic chain of "risk identification - mechanism analysis - impact prediction", especially strengthening the data-driven analysis of the key nodes of the chain disaster.

[0019] S2. According to the knowledge map, obtain the disaster influencing factors and determine the associations between the disaster influencing factors. Take the disaster influencing factors with the association degree greater than the set threshold as the key disaster influencing factors.

[0020] According to the disaster influencing factors shown in the knowledge map, expert experience and the mechanism of secondary disasters, clarify the association degree of the disaster influencing factors through the knowledge map, and determine 16 key disaster influencing factors with the association degree greater than the set threshold, including: earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, height difference between front and rear edges, rock strength, river, height of the barrier dam, width of the barrier dam, volume of the barrier dam, material of the barrier dam body, valley shape, landslide volume, storage capacity of the barrier lake, stability of the barrier dam. And determine the associations between the disaster influencing factors.

[0021] Divide each node into four levels according to the degree of association with each other. Among them, the first-level nodes include: earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, height difference between front and rear edges, rock strength, river, height of the barrier dam, width of the barrier dam, material of the barrier dam body, valley shape; the second-level node includes: landslide volume; the third-level nodes include: storage capacity of the barrier lake, volume of the barrier dam; the fourth-level node includes: stability of the barrier dam. The division and association relationship of each level of nodes are as Figure 2 shown. The high-level nodes are the results of the low-level nodes, and the low-level nodes are the causes of the high-level nodes.

[0022] S3. Use the key disaster influencing factors as the nodes of the Bayesian network, discretize the node variable data, and use the max-min hill climbing algorithm to perform structure learning on the discretized node variable data to determine the directed acyclic graph structure of the Bayesian network.

[0023] Each node in the Bayesian network has different states, and each state corresponds to a value range. The states and thresholds of the Bayesian network nodes refer to Table 1.

[0024] Table 1

[0025] Based on the data obtained in S1, obtain the probability distribution of the nodes. The joint distribution of these 16 variables can be factorized in 16! ways.

[0026] , where, represents the joint distribution probability of the Bayesian network nodes, represents the joint distribution probability of n Bayesian network nodes, , represents the parent node of, represents under the condition of , the probability of occurrence.

[0027] Each factorization can represent a Bayesian network model. Construct the directed graph of the network by creating a node for each factor in the distribution (label each node with the name of the variable before the condition), and draw directed arcs between them, always from the variable on the right side of the condition to the variable on the left side. For example, in a simple four-node model, earthquake (E), landslide (L), debris flow (D), barrier lake (B), the joint probability distribution of these four variables can be obtained in 4! ways. For one of the factorizations: , the schematic diagram of the Bayesian network model of the four variables in the embodiment of the present invention refers to Figure 3 .

[0028] The present invention infers the topological structure of the Bayesian network from the data and obtains one of the 16! factorizations: , where, represents the joint distribution probability of 16 Bayesian network nodes, represents the earthquake magnitude, represents the elevation, represents the vegetation coverage rate, represents the slope, represents the geological structure, Indicates the elevation difference between the front and rear edges, Indicates the rock strength, Indicates a river, Indicates the height of the barrier dam, Indicates the width of the barrier dam, Indicates the volume of the barrier dam, Indicates the material of the barrier dam body, Indicates the shape of the river valley, Indicates the volume of the landslide, Indicates the storage capacity of the barrier lake, Indicates the stability of the barrier dam, 、 、 、 、 、 、 、 、 、 、 、 respectively represent the probabilities of each node in a certain state; Indicates the probability of the volume of the landslide occurring in a certain state under the influence of a certain state based on the earthquake magnitude, elevation, vegetation coverage, slope, geological structure, elevation difference between the front and rear edges, and rock strength; Indicates the probability of the storage capacity of the barrier lake occurring in a certain state under the influence of a certain state based on the volume of the landslide, river, and height of the barrier dam; Indicates the probability of the volume of the barrier dam occurring in a certain state under the influence of a certain state based on the volume of the landslide, height of the barrier dam, and width of the barrier dam; Indicates the probability of the stability of the barrier dam occurring in a certain state under the influence of a certain state based on the storage capacity of the barrier lake, volume of the barrier dam, material of the barrier dam body, and shape of the river valley.

[0029] The Bayesian network diagram of the earthquake landslide disaster chain in the embodiment of the present invention is as shown in Figure 4 shown.

[0030] S4. Based on the directed acyclic graph structure of the Bayesian network, use the Expectation-Maximization algorithm (EM algorithm) to perform Bayesian network parameter learning to obtain the prior probabilities of the Bayesian network nodes. In the embodiment of the present invention, specifically: Initialize the model parameters For the states of the observed Bayesian network nodes: , where the node lacks state data, is incomplete data, and the missing state data is the hidden variable Z, and Z together are called complete data.

[0031] ; Step E: Calculate the expectation (posterior probability) of the latent variable Z; Start iteration, Denoted as the i estimated value of the th iteration, calculate the function: , where, represents the log-likelihood function, represents the joint distribution probability of the latent variable Z and the variable under the parameter , represents the conditional probability distribution of the latent variable data Z given the observed data and the current parameter estimate . represents the i th iteration of the estimated value (current parameter estimate).

[0032] Step M: Update the parameter using the latent variable expectation calculated in Step E: , where, represents the i updated estimated value after the

[0033] When the norm of is less than the set iteration convergence condition, stop the iteration and output the at this time as the network conditional probability distribution and as the prior probability of the Bayesian network node.

[0034] The Bayesian network model with prior probability obtained by the present invention through machine learning is as shown in Figure 5 . Figure 5 In it, "√" indicates that the state (probability distribution) of the node is the predicted result.

[0035] S4. Obtain the prediction result of the Bayesian network disaster chain model based on the historical earthquake-induced disaster chain data. If the deviation between the prediction result and the actual result is greater than the preset value, adjust the association between the nodes, return to S3. If the deviation between the prediction result and the actual result is less than or equal to the preset value, use the Bayesian network disaster chain model for earthquake landslide disaster chain risk assessment.

[0036] In the embodiments of the present invention, the parameters of the research area are as follows: the height of the barrier dam is extremely high; the width of the barrier dam is wide; the volume of the barrier dam is huge; the rock strength is high-strength rock; there is a river; the material of the barrier dam body is block stone type; the magnitude of the earthquake is strong; the vegetation coverage rate is relatively high; the slope is steep; the elevation is medium altitude; the height difference between the front and rear edges is high; the geological structure is unstable; the river valley is between U and V.

[0037] The Brier score is used to test the prediction results of the Bayesian network disaster chain model, and the calculation is as follows: , where BrierScore represents the Brier score, N represents the number of samples, represents the number of categories of the t-th task, represents the true label that the i-th sample belongs to the j-th category in the t-th task, represents the probability that the model predicts that the i-th sample belongs to the j-th category in the t-th task.

[0038] The posterior probability model of landslide in the research area is obtained as Figure 6 , Figure 6 in which " " indicates that the state of this node is the result of observation, and "√" indicates that the state of this node is the result of prediction. The probability table of the research area is shown in Table 2. Taking the maximum posterior probability as the predicted value, comparing the predicted value and the actual result, the prediction comparison is shown in Table 3.

[0039] Table 2

[0040] Table 3

[0041] In order to fully quantify the state of each node, the "extra-large" category of landslide volume is incorporated into the "large" category, and the "huge" of the reservoir capacity of the barrier lake is incorporated into the "large". Then, from the landslide samples in Table 2, Task 1 is to predict the landslide volume, the true label is Category 3 (large), the model prediction probability is [0.34, 0.14, 0.52], and BrierScore1 = 0.3656; Task 2 is to predict the reservoir capacity of the barrier lake, the true label is Category 3 (large), the model prediction probability is [0.17, 0.14, 0.69], and BrierScore2 = 0.1446; Task 3 is to predict the stability of the barrier dam, the true label is Category 2 (unstable), the model prediction probability is [0.33, 0.67], and BrierScore3 = 0.2178; then the average Brier score of the landslide is 0.2427, which is a very good prediction.

[0042] In an exemplary embodiment, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned seismic landslide disaster chain risk assessment method is implemented.

[0043] Please refer to Figure 7 , in an exemplary embodiment, an electronic device is further provided, which includes at least one processor, at least one memory, and at least one communication bus.

[0044] Among them, a computer program is stored on the memory. The computer program includes computer-readable instructions. The processor calls the computer-readable instructions stored in the memory through the communication bus to execute the above-mentioned seismic landslide disaster chain risk assessment method.

[0045] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for risk assessment of earthquake landslide disaster chains, characterized in that, It includes the following steps: S1. Obtain earthquake-induced disaster chain data and construct a knowledge graph of earthquake landslide disaster chains; S2. According to the knowledge graph, obtain disaster influencing factors, determine the associations between the disaster influencing factors, and determine the key disaster influencing factors with the association degree greater than the set threshold; S3. Use the key disaster influencing factors as Bayesian network nodes, discretize the node variable data, and perform structure learning on the discretized node variable data using the max-min hill climbing algorithm to determine the directed acyclic graph structure of the Bayesian network; S4. Based on the directed acyclic graph structure of the Bayesian network, use the expectation maximization algorithm for Bayesian network parameter learning to obtain the prior probabilities of the Bayesian network nodes; S5. Obtain the prediction results of the Bayesian network based on historical earthquake-induced disaster chain data. If the deviation between the prediction results and the actual results is greater than the preset value, adjust the associations between the nodes and return to step S3. If the deviation is less than or equal to the preset value, use the Bayesian network disaster chain model for earthquake landslide disaster chain risk assessment.

2. The risk assessment method for earthquake landslide disaster chain according to claim 1, wherein The key disaster influencing factors include: earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, elevation difference between the front and rear edges, rock strength, river, height of the barrier dam, width of the barrier dam, volume of the barrier dam, material of the barrier dam, valley shape, landslide volume, storage capacity of the barrier lake, and stability of the barrier dam.

3. A method for risk assessment of earthquake landslide disaster chain according to claim 2, characterized in that, The Bayesian network nodes are divided into four levels according to the degree of association with each other. Among them, the first-level nodes include: earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, elevation difference between the front and rear edges, rock strength, river, height of the barrier dam, width of the barrier dam, material of the barrier dam, valley shape; the second-level node includes: landslide volume; the third-level nodes include: storage capacity of the barrier lake, volume of the barrier dam; the fourth-level node includes: stability of the barrier dam.

4. A method for earthquake landslide disaster chain risk assessment according to claim 1, wherein The specific method for determining the node probability distribution is: Divide the states of the nodes and determine the value ranges of the states of each node; through the earthquake-induced disaster chain data and the value ranges of the states of each node, obtain the probability distribution of the nodes.

5. A method for risk assessment of earthquake landslide disaster chain according to claim 2, characterized in that, The joint distribution probability of the Bayesian network nodes is: ； Among them, represents the joint distribution probability of 16 Bayesian network nodes, represents the earthquake magnitude, represents the elevation, represents the vegetation coverage rate, represents the slope, represents the geological structure, represents the elevation difference between the front and rear edges, represents the rock strength, represents the river, represents the height of the barrier lake dam, represents the width of the barrier lake dam, represents the volume of the barrier lake dam, represents the material of the barrier lake dam body, represents the valley shape, represents the volume of the landslide, represents the storage capacity of the barrier lake, represents the stability of the barrier lake dam, , , , , , , , , , , , respectively represent the probabilities of each node in a certain state; represents the probability of the occurrence of the landslide volume in a certain state under the influence of a certain state of the earthquake magnitude, elevation, vegetation coverage rate, slope, geological structure, elevation difference between the front and rear edges, and rock strength; represents the probability of the occurrence of the storage capacity of the barrier lake in a certain state under the influence of a certain state of the landslide volume, river, and height of the barrier lake dam; represents the probability of the occurrence of the volume of the barrier lake dam in a certain state under the influence of a certain state of the landslide volume, height of the barrier lake dam, and width of the barrier lake dam; represents the probability of the occurrence of the stability of the barrier lake dam in a certain state under the influence of a certain state of the storage capacity of the barrier lake, volume of the barrier lake dam, material of the barrier lake dam body, and valley shape.

6. The risk assessment method for earthquake landslide disaster chain according to claim 5, characterized in that Using the expectation maximization algorithm for Bayesian network parameter learning, specifically: Initialization parameters For the state of the observed Bayesian network nodes: ， Among them, the node lacks state data, and the missing state data is the latent variable Z; E step: Start iteration according to the following formula to calculate the expectation of the latent variable Z: ； Among them, represents the log-likelihood function, represents the joint distribution probability of Z and under the parameter ; represents the conditional probability distribution of the latent variable Z under the given observed data and the current parameter estimate ; represents the i -th iteration of the estimated value; M step: Update parameters using the expectations of the latent variables computed in the E step : ； Among them, represents the i estimated value after the (+1)-th round of iterative update; When the norm is less than the set iteration convergence condition, stop the iteration and output the current as the network conditional probability distribution.

7. A method for risk assessment of earthquake landslide disaster chain according to claim 1, characterized in that, Use the Brier score to test the prediction results of the Bayesian network disaster chain model, and calculate as follows: ； Among them, BrierScore represents Brier a fraction, N represents the number of samples, represents the t number of categories of the th i task, t represents the true label of the j th sample belonging to the th category in the i th task, t represents the probability that the model predicts the j th sample belonging to the th category in the th task.

8. A computer-readable storage medium storing a computer program, characterized in that: The computer program, when executed by a processor, implements the method according to any one of claims 1-7.

9. An electronic device, characterized in that, It includes a processor and a memory, and the processor is connected to the memory. Among them, the memory is used to store a computer program, the computer program includes computer-readable instructions, and the processor is configured to call the computer-readable instructions to execute the method according to any one of claims 1-7.

Images