Ice rock collapse-river plugging-dam break flood disaster chain quantitative risk assessment method and system

Abstract

The invention provides a quantitative risk assessment method and system for an ice rock collapse-river blockage-dam break flood disaster chain. The method comprises the following steps: acquiring an ice rock collapse risk grade index by adopting a BP neural network improved by an Adam optimizer; if the ice rock collapse occurs, correcting and calculating an ice rock collapse movement path based on a D8 algorithm in combination with a rock-soil friction coefficient, correcting the volume in real time through an erosion-accumulation dynamic balance model, and constructing a river blocking risk grade index in combination with the river entering volume and speed; the stability of the dam body is evaluated through a dynamic catchment area model and a Sigmoid function, and the dam break risk is calculated in combination with the ratio of the volume to the area; and if the dam break disaster does not occur in the barrier dam, integrating the ice rock collapse risk grade, the river blocking risk grade and the dam break risk grade index, and evaluating the comprehensive risk of the disaster chain. According to the method, disaster prediction of each stage is realized through neural network improvement, path optimization analysis and variance tracking change, and comprehensive risks are evaluated through multi-element coupling.

Description

Technical Field

[0001] The present application relates to the technical field of geological disaster risk assessment, and specifically to a quantitative risk assessment method and system for an ice and rock collapse-river blockage-dam breach flood disaster chain. Background Art

[0002] Due to global warming, ice-rich, steep slopes located at high altitudes of the plateau are frequently collapsing and failing, such as the Yigong landslide, the Sedongpugou glacial avalanche debris flow, and the Chamoli glacial rock avalanche. Unlike traditional landslides and avalanches, these plateau high-altitude slope collapses often manifest as a chain of hazard failures: glacial rock avalanches, river blockages, and dam-break floods. Due to the vertical cascade risk amplification effect of this hazard chain, such disasters pose a significant threat to the lives, property, and infrastructure of residents downstream. Therefore, for this glacial rock avalanche-river blockage-dam-break flood hazard chain, it is important to consider not only the hazards of the landslide source area but also the multi-hazard effects and consequences along the damage path. Currently, risk assessments for landslide and collapse hazard chains in alpine canyons are primarily based on the hazard of the collapse source area, without considering the cascading effects of dam-break floods and river blockages caused by glacial rock avalanches along the way. Summary of the Invention

[0003] The present application provides a quantitative risk assessment method and system for the ice and rock avalanche-river blockage-dam breach flood disaster chain, which can solve the technical problem that the risk assessment of the collapse and landslide disaster chain in high mountain canyon areas in the existing technology is mainly based on the danger of the collapse source area, and does not consider the cascade amplification effect of the risk of river blockage and dam breach flood risk caused by ice and rock avalanche destruction along the way.

[0004] In a first aspect, the present application provides a quantitative risk assessment method for an ice and rock avalanche-river blockage-dam breach flood disaster chain, comprising the following steps: A BP neural network improved by the Adam optimizer was used as input for regional digital elevation, meteorological data, and historical disaster data. The probability of ice and rock avalanche disasters was output by dynamically optimizing the network weights, and the ice and rock avalanche risk level index was obtained by grading. If an ice and rock avalanche occurs, the D8 algorithm is used to calculate the movement path of the ice and rock avalanche in combination with the correction of the rock-soil friction coefficient. The volume is corrected in real time through the erosion-deposition dynamic balance model, and the river blockage risk level index is constructed based on the volume and speed of the river. If a landslide dam blocks the river, the dam failure risk level index is calculated based on the dynamic growth model of the dam's catchment area and the Sigmoid function to characterize the relationship between the dam's stability and time. If the landslide dam does not fail, the ice and rock avalanche risk level, river blockage risk level and dam failure risk level indicators are integrated, and a multi-factor coupling weight allocation strategy is used to construct a cascade disaster vulnerability calculation formula, dynamically adjust the risk weights of each stage, and evaluate the comprehensive risk of the disaster chain.

[0005] Furthermore, the BP neural network improved by the Adam optimizer is input with regional digital elevation, meteorological data and historical disaster data, and the probability of ice and rock avalanche disaster is output by dynamically optimizing the network weights, and the ice and rock avalanche risk level index is obtained by grade classification. Specifically, the following steps are included: Construct BP neural network; Initialize the BP neural network weight parameters, use the Adam optimizer to dynamically adjust the learning rate of the BP neural network weight parameters, calculate the first-order moment estimate and second-order moment estimate of the gradient, and adaptively update the network weights in combination with the bias correction term to obtain the BP neural network improved by the Adam optimizer; A binary cross-entropy loss function is defined, and the weight gradients of each layer are calculated using the back-propagation algorithm. For the binary classification task of whether an ice and rock avalanche has occurred, the loss function is used to measure the error between the probability of an ice and rock avalanche disaster and the true label. The gradient of the hidden layer weight is calculated using the chain rule to obtain the gradient update value of the weights of each layer of the neural network. The standardized historical disaster data is input into the BP neural network improved by the Adam optimizer for training. The gradient update value of the weights of each layer of the neural network is used for iterative optimization to converge the binary cross entropy loss function and obtain the iteratively optimized BP neural network. The standardized real-time monitoring data is input into the iteratively optimized BP neural network, and the output probability of ice and rock avalanche disaster is classified into different levels to obtain the ice and rock avalanche risk level index.

[0006] Furthermore, the model architecture of the BP neural network includes an input layer, a hidden layer and an output layer. The input layer uses digital elevation data, rainfall data and temperature data as feature inputs, the hidden layer uses the ReLU activation function for nonlinear feature transformation, and the output layer maps the calculation results to the probability of disaster occurrence through the Sigmoid activation function.

[0007] Furthermore, if an ice and rock avalanche occurs, the movement path of the ice and rock avalanche is calculated based on the D8 algorithm combined with the correction of the rock-soil friction coefficient. The volume is corrected in real time through the erosion-deposition dynamic equilibrium model, and the river blocking risk level index is constructed based on the volume and speed of the river entering. The specific steps include: Step S21: Based on the coordinates of the ice and rock avalanche risk points, the elevation data of the corresponding grid cells in the digital elevation model are extracted, the slope between the central grid cell and the neighboring grid cells is calculated using the D8 algorithm, and the terrain correction is performed based on the friction coefficient of different rock and soil types to obtain the effective slope; The direction with the largest effective slope is selected as the movement direction of the ice-rock avalanche to obtain the movement path of the ice-rock avalanche. The volume of the ice-rock avalanche entering the river is calculated by combining the erosion and accumulation effects on the movement path and the principle of volume conservation. When the movement path reaches the river channel grid, the cumulative volume is calculated based on the angle between the ice and rock avalanche flow direction and the river channel direction, and the water entry velocity is calculated using the Manning formula; Based on the ratio of the volume of water entering the river, the water inflow speed and the river flow speed, combined with the geometric parameters of the river valley, the river blocking value is calculated and the river blocking value is segmented to obtain the river blocking risk level index.

[0008] Furthermore, based on the coordinates of the ice and rock avalanche risk points, the elevation data of the corresponding grid cells in the digital elevation model are extracted, the slope between the central grid cell and the neighboring grid cells is calculated using the D8 algorithm, and the terrain correction is performed in combination with the friction coefficient of different rock and soil types to obtain the effective slope. Specifically, the following steps are included: The slope between the central grid cell and the neighboring grid cells is calculated based on the elevation difference between the central grid cell and the adjacent 8 grid cells, combined with the grid resolution. The friction coefficient of different rock and soil types is introduced as a terrain correction factor to correct the slope and obtain the effective slope between the central grid cell and the neighboring grid cell.

[0009] Furthermore, if the landslide dam blocks the river, then according to the dynamic growth model of the water catchment area of ​​the landslide dam, a Sigmoid function is used to characterize the relationship between the stability of the dam and time, and a dam break risk level index is calculated by the ratio of volume to water catchment area, which specifically includes the following steps: Based on the obtained volume of ice and rock avalanches into the river and combined with the valley geometric parameters, the initial catchment area is calculated; A dynamic growth model of the water catchment area of ​​the landslide dam is established. The water level rise at any time is calculated based on the regional flow and the time of river blockage. The water catchment area at the corresponding time is then derived to obtain the dynamic water catchment area of ​​the landslide dam. Through the ratio of ice and rock avalanche volume to dynamic catchment area, the temporal characteristics of dam stability are quantified, and a dam failure risk level index based on Sigmoid function is constructed. The dam break risk level index is divided into dam break risk level.

[0010] Furthermore, if the landslide dam does not fail, the ice and rock avalanche risk level, river blockage risk level, and dam failure risk level indicators are integrated, and a multi-factor coupling weight allocation strategy is adopted to construct a cascade disaster vulnerability calculation formula, dynamically adjust the risk weights of each stage, and evaluate the comprehensive risk of the disaster chain. Specifically, the following steps are included: Normalizing the obtained ice-rock avalanche disaster hazard level index, river blocking disaster risk index, and dam break risk index to obtain a normalized ice-rock avalanche risk level index; Based on the normalized ice and rock avalanche risk level index, the activation state of the disaster assessment process is controlled by setting a two-level threshold and combining the activation function Re to obtain the hard gating signal of each disaster stage. The analytic hierarchy process is used to assign weights to the three types of disaster-prone objects and obtain the normalized vulnerability factor of the disaster-prone objects. The output results of the previous steps are integrated through a dynamic weight assessment model, in which the ice and rock avalanche risk level index, river blockage risk level index and dam break risk level index are coupled with the vulnerability factors of the disaster-bearing bodies in the corresponding stages and constrained by hard gating signals to obtain the disaster vulnerability level at each stage.

[0011] Furthermore, the output results of the previous steps are integrated through the dynamic weight assessment model, wherein the ice and rock avalanche risk level index, river blockage risk level index, and dam break risk level index are coupled with the vulnerability factors of the hazard-bearing bodies at the corresponding stages and are constrained by hard gating signals to obtain the disaster vulnerability level at each stage, as shown in the following formula:

[0012] In the formula, the subscript They are ice and rock avalanche, river blockage and dam failure; α i is the risk amplification factor of the cascade effect; I i0 is the normalized risk index, W i is the vulnerability factor, and Re is the activation function.

[0013] In a second aspect, the present application provides a quantitative risk assessment system for the ice and rock avalanche-river blockage-dam breach flood disaster chain, including: The module for obtaining the ice and rock avalanche risk level index is used to input the regional digital elevation, meteorological data, and historical disaster data into the BP neural network improved by the Adam optimizer. It outputs the probability of ice and rock avalanche disasters by dynamically optimizing the network weights, and then classifies them into different levels to obtain the ice and rock avalanche risk level index. The river blocking risk level indicator acquisition module is used to calculate the movement path of the ice and rock avalanche based on the D8 algorithm combined with the rock-soil friction coefficient correction if an ice and rock avalanche occurs. The volume is corrected in real time through the erosion-deposition dynamic balance model, and the river blocking risk level indicator is constructed based on the volume and velocity of the inflow into the river. The dam break risk level index acquisition module is used to calculate the dam break risk level index by the ratio of volume to catchment area based on the dynamic growth model of the dam's catchment area, using the Sigmoid function to characterize the relationship between the dam body stability and time if a landslide dam blocks the river. The module for obtaining the comprehensive risk of the disaster chain is communicated with the ice and rock avalanche risk level indicator acquisition module, the river blocking risk level indicator acquisition module and the dam break risk level indicator acquisition module. If no dam break disaster occurs in the landslide dam, the module is used to integrate the ice and rock avalanche risk level, the river blocking risk level and the dam break risk level indicators, adopt a multi-factor coupling weight allocation strategy to construct a cascade disaster vulnerability calculation formula, dynamically adjust the risk weights of each stage, and evaluate the comprehensive risk of the disaster chain.

[0014] Furthermore, the river blocking risk level indicator acquisition module includes: The effective slope acquisition unit is used to extract the elevation data of the corresponding grid cell in the digital elevation model based on the coordinates of the ice and rock avalanche risk point, calculate the slope between the central grid cell and the neighboring grid using the D8 algorithm, and perform terrain correction based on the friction coefficient of different rock and soil types to obtain the effective slope; a river-entry volume acquisition unit, in communication with the effective slope acquisition unit, configured to select a direction with the largest effective slope as the movement direction of the ice-rock avalanche to obtain a movement path of the ice-rock avalanche, and calculate the river-entry volume of the ice-rock avalanche by combining erosion and accumulation effects on the movement path and the principle of volume conservation; The water entry velocity acquisition unit is used to calculate the cumulative volume according to the angle between the ice and rock avalanche flow direction and the river channel direction when the movement path reaches the river channel grid, and calculate the water entry velocity using the Manning formula; The river blocking risk level index acquisition unit is communicatively connected to the effective slope acquisition unit, the river inflow volume acquisition unit and the water inflow speed acquisition unit, and is used to obtain the river blocking value based on the ratio of the river inflow volume, the water inflow speed and the river water flow rate, combined with the river valley geometric parameters, and to obtain the river blocking risk level index by performing segmented judgment on the river blocking value.

[0015] The beneficial effects of the technical solutions provided in the embodiments of the present application include at least: The BP neural network, improved by the Adam optimizer, achieved accurate prediction of ice and rock avalanche disasters under complex terrain conditions. The neural network's adaptive learning mechanism effectively improved model training results, enabling the model to accurately capture the complex correlation characteristics between terrain, meteorology, and historical disaster data. The motion path calculation method based on the D8 algorithm combined with the correction of the rock and soil friction coefficient significantly improved the prediction accuracy under complex terrain conditions. The introduction of the erosion-deposition dynamic equilibrium model achieved real-time tracking of volume changes during ice and rock avalanche movement, providing dynamic data support for river blockage risk assessment. An innovative Sigmoid function characterization method is used to analyze dam break risk. By organically combining a dynamic catchment area model with dam stability analysis, a time-series assessment of dam break risk is achieved, which can accurately reflect the dynamic changes in dam stability. Through the dynamic weight allocation strategy and activation function control mechanism, an integrated assessment of the comprehensive risk of disaster chain risks is carried out, fully considering the vulnerability differences of different disaster-prone bodies, and providing comprehensive guidance for disaster risk management. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flowchart of the quantitative risk assessment method for the ice and rock avalanche-river blockage-dam breach flood disaster chain applied in this application; Figure 2 Another flowchart of the quantitative risk assessment method for the ice and rock avalanche-river blockage-dam breach flood disaster chain applied in this application; Figure 3 A top view and a three-dimensional view of a region of an embodiment of the present application; Figure 4 This is a statistical diagram of ice and rock avalanche hazard areas in an embodiment of the present application; Figure 5 This is a diagram showing the ice and rock avalanche danger zone according to an embodiment of the present application; Figure 6 The D8 algorithm of the embodiment of the present application calculates the ice and rock avalanche flow path diagram; Figure 7 This is a diagram of a calculation model for accumulation body deposition according to an embodiment of the present application; Figure 8 This is a diagram of a dam-break index calculation model according to an embodiment of the present application; Figure 9 This is a regional vulnerability index diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit the "first", "second" and "third" to different types.

[0019] In the description of the embodiments of the present application, the words "exemplary," "for example," or "for example" are used as examples, illustrations, or explanations. Any embodiment or design described in the embodiments of the present application as "exemplary," "for example," or "for example" should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0020] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can be A or B; “and / or” in the text is merely a description of the association relationship of associated objects, which means that there can be three relationships, for example, A and / or B can be: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” means two or more than two.

[0021] In some processes described in the embodiments of the present application, multiple operations or steps are included that appear in a specific order. However, it should be understood that these operations or steps may not be performed in the order in which they appear in the embodiments of the present application or may be performed in parallel. The sequence numbers of the operations are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be performed in sequence or in parallel, and these operations or steps may be combined.

[0022] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0023] like Figure 1-Figure 2 As shown, the present application provides a quantitative risk assessment method for the ice and rock avalanche-river blockage-dam breach flood disaster chain, including the following steps: Step S1: Using the BP neural network improved by the Adam optimizer, the regional digital elevation, meteorological data and historical disaster data are input, the probability of ice and rock avalanche disaster is output by dynamically optimizing the network weights, and the ice and rock avalanche risk level index is obtained by grade classification; Step S2: If an ice-rock avalanche occurs, the avalanche path is calculated based on the D8 algorithm combined with the correction of the rock-soil friction coefficient. The volume is corrected in real time using the erosion-deposition dynamic equilibrium model, and the river blockage risk level index is constructed based on the volume and velocity of the inflow into the river. Step S3: If a landslide dam blocks the river, the dam is subjected to a dynamic growth model of its catchment area, and the sigmoid function is used to characterize the relationship between the dam stability and time. The dam failure risk index is calculated using the ratio of volume to catchment area. Step S4: If the landslide dam does not fail, the ice and rock avalanche risk level, river blockage risk level, and dam failure risk level indicators are integrated, and a multi-factor coupling weight allocation strategy is used to construct a cascade disaster vulnerability calculation formula. The risk weights of each stage are dynamically adjusted to evaluate the comprehensive risk of the disaster chain.

[0024] This application uses a BP neural network improved by the Adam optimizer to achieve accurate prediction of ice and rock avalanche disasters under complex terrain conditions. The adaptive learning mechanism of the neural network effectively improves the model training effect, enabling the model to accurately capture the complex correlation characteristics between terrain, meteorology and historical disaster data. The motion path calculation method based on the D8 algorithm combined with the correction of the rock and soil friction coefficient significantly improved the prediction accuracy under complex terrain conditions. The introduction of the erosion-deposition dynamic equilibrium model achieved real-time tracking of volume changes during ice and rock avalanche movement, providing dynamic data support for river blockage risk assessment. An innovative Sigmoid function characterization method is used to analyze dam break risk. By organically combining a dynamic catchment area model with dam stability analysis, a time-series assessment of dam break risk is achieved, which can accurately reflect the dynamic changes in dam stability. Through the dynamic weight allocation strategy and activation function control mechanism, an integrated assessment of the comprehensive risk of disaster chain risks is carried out, fully considering the vulnerability differences of different disaster-prone bodies, and providing comprehensive guidance for disaster risk management.

[0025] In one embodiment, step S1: using a BP neural network improved by the Adam optimizer, inputting regional digital elevation data, meteorological data, and historical disaster data, outputting the probability of ice and rock avalanche disasters by dynamically optimizing network weights, and performing grade classification to obtain an ice and rock avalanche risk level index, specifically includes the following steps: Step S11: Constructing a BP neural network; the model architecture of the BP neural network includes an input layer, a hidden layer, and an output layer. The input layer uses digital elevation data, rainfall data, and temperature data as feature inputs. The hidden layer uses a ReLU activation function for nonlinear feature transformation. The output layer uses a Sigmoid activation function to map the calculation results to the probability of disaster occurrence. Step S12: Initialize the BP neural network weight parameters, use the Adam optimizer to dynamically adjust the learning rate of the BP neural network weight parameters, calculate the first-order moment estimate and second-order moment estimate of the gradient, and adaptively update the network weights in combination with the bias correction term to obtain the BP neural network improved by the Adam optimizer; Step S13: Define a binary cross entropy loss function, calculate the weight gradients of each layer through the back propagation algorithm, and use the loss function to measure the error between the probability of ice and rock avalanche disaster and the true label for the binary classification task of whether ice and rock avalanche has occurred. The hidden layer weight gradient is calculated by the chain rule to obtain the gradient update value of the weight of each layer of the neural network; Step S14: input the standardized historical disaster data into the BP neural network improved by the Adam optimizer for training, perform iterative optimization based on the gradient update value of the weights of each layer of the neural network to make the binary cross entropy loss function converge, and obtain the iteratively optimized BP neural network; Step S15: input the standardized real-time monitoring data into the iteratively optimized BP neural network, classify the output probability of ice and rock avalanche disaster occurrence, and obtain the ice and rock avalanche risk level index.

[0026] In one embodiment, the Adam optimizer is used in step S12 to dynamically adjust the learning rate of the BP neural network weight parameters, by calculating the first-order moment estimate and the second-order moment estimate of the gradient, and combining the deviation correction term to adjust the network weight θ t For adaptive update, the process is as follows: Adaptive estimation of the first-order moment (mean) of the gradient As shown in the following formula: Formula (1) Where, is the first-order moment estimate with an iteration step of t-1, is the decay rate of the first-order moment estimate, is the historical first-order moment estimate with the iteration step t-1, is the gradient of the loss function at iteration step t-1; Adaptive estimation of the second moment of the gradient (uncentered variance) As shown in the following formula: Formula (2) Where, is the second-order moment estimate with an iteration step of t, is the decay rate of the second-order moment estimate, is the historical second-order moment estimate with the iteration step t-1, is the gradient of the loss function at iteration step t-1.

[0027] The bias correction is shown below: Formula (3) Where, is the first-order moment estimate after the modified iteration step is t, is the revised second-order moment estimate with an iteration step of t; Neural network weight parameters with iteration step t The update is as follows: Formula (4) Where t is the iteration step, is the loss function The gradient of weight θ; α is the learning rate, β1, β2 are the decay rates, and ε is a small constant term; Among them, for the binary classification task of whether ice and rock avalanche occurs, the loss function is used to measure the probability of ice and rock avalanche disaster and the true label The error is as follows: Formula (5) First For example, the hidden layer weights The gradient of is calculated by the chain rule as follows: Formula (6) Where, For the The error of the layer, is the weight, For the Layer activation value, σ′ is the inverse of the ReLU activation function, which takes the value of 1 when Z is greater than 0, otherwise it is 0.

[0028] In one embodiment, the probability of occurrence of the ice and rock avalanche disaster is The calculation formula is as follows: Formula (7) Where b is the threshold of each layer of the network, L is the number of layers, and the probability of ice and rock avalanche disaster is The Sigmoid activation function can be expressed as: Formula (8) In one embodiment, the outputted probability of occurrence of ice and rock avalanche disaster is graded in the step S15 to obtain an ice and rock avalanche risk level index, which is specifically implemented as follows: Formula (9) Figure 6 (a) is a diagram illustrating the calculation of the slope between the central grid cell and the adjacent grid cells using the D8 algorithm; Figure 6 Middle (b) is the ice-rock avalanche flow path diagram calculated by D8 algorithm; In one embodiment, if Figure 6 As shown, step S2: if an ice and rock avalanche occurs, the movement path of the ice and rock avalanche is calculated based on the D8 algorithm combined with the correction of the rock-soil friction coefficient, the volume is corrected in real time through the erosion-deposition dynamic balance model, and the river blocking risk level index is constructed based on the volume and speed of the river entering. Specifically, the following steps are included: Step S21: Based on the coordinates of the ice and rock avalanche risk points, extract the elevation data of the corresponding grid cells in the digital elevation model. Use the D8 algorithm to calculate the slope between the central grid cell and the eight adjacent grid cells. Combined with the friction coefficient of different rock and soil types, perform terrain correction to obtain the effective slope. Specifically, the slope is calculated based on the elevation difference z(i, j) of the central grid cell (i, j) and the elevation difference Δz of the eight adjacent grid cells, combined with the grid resolution d, as shown in the following formula: Formula (10) Where, is the slope between the central grid cell and the neighboring grid cells in direction k, is the elevation of the neighborhood grid cell in direction k (direction k=1~8), is the Euclidean distance between the center and the neighboring grid cells in direction k.

[0029] By introducing the friction coefficient μ of different rock and soil types as the terrain correction factor, the slope is corrected and the effective slope is obtained as shown in the following formula: Formula (11) Where, is the effective slope from the central grid cell to the neighboring grid cells in direction k, and g is the acceleration of gravity at the foot of the slope.

[0030] Step S22: Select the direction with the largest effective slope as the movement direction of the ice and rock avalanche to obtain the movement path of the ice and rock avalanche, and calculate the volume of the ice and rock avalanche entering the river by combining the erosion and accumulation effects on the movement path and the principle of volume conservation; Formula (13) Where, V accum is the cumulative ice-rock avalanche volume of the nth grid unit, Vorigin is the volume of the source sliding area, V deposit is the deposition volume of the current grid unit, V erosion is the erosion volume of the current grid cell.

[0031] In formula (13), the deposition volume V deposit It can be calculated by the following formula: Formula (14) Where S dep is the critical sedimentation slope, S dep =tanφ, where φ is the internal friction angle of the slope and can be determined from survey data; ζ is related to the ice-rock avalanche movement speed of the current grid unit and takes a value between 0 and 1.

[0032] In formula (13), the erosion volume V erosion It can be calculated by the following formula: Formula (15) Where C erosion is the erosion coefficient, ρ is the density of the ice-rock avalanche mixture, v is the current ice-rock avalanche velocity, A n is the cross-sectional area of ​​the grid unit, and Δt is the time increment.

[0033] Step S23: When the motion path reaches the river grid, the cumulative volume is calculated based on the angle between the ice and rock avalanche flow direction and the river direction, and the water entry velocity is calculated using the Manning formula; the specific implementation is as follows: Based on the obtained ice-rock avalanche movement volume, the arrival channel grid unit (i river , j river ) into the river is: Formula (16) Where θ r It is the angle between the direction of the ice and rock avalanche and the direction of the river channel.

[0034] The velocity of ice and rock avalanches entering the river surface can be calculated using the Manning formula, as shown below: Formula (17) Where, is the velocity of the ice and rock avalanche entering the river surface, n v is the roughness coefficient that depends on the surface type, R is the hydraulic radius, which can be estimated as the depth of the ice and rock avalanche, and S L is the hydraulic slope.

[0035] Step S24: Based on the volume of water entering the river, the ratio of the water inflow velocity to the river flow velocity, and the geometric parameters of the river valley, the river blocking value is calculated and obtained, and the river blocking value is segmented to obtain the river blocking risk level index; the specific implementation is as follows: Generally, the shallower the river depth and the slower the river flow in the area where the ice and rock avalanche occurs, the larger the landslide volume and the faster the speed, and the more likely it is to cause a river blockage disaster. Therefore, the following normalized river blockage risk level index is defined: : Formula (18) Where A b 、B b 、C b is the empirical fitting coefficient, in this application, A b 、B b 、C b 0.38, 1.09 and 0.42 were used respectively; v L is the landslide velocity; v R is the river flow velocity; is the average depth of the river; B is the equivalent width of the valley, which can be calculated by the following formula: Formula (19) Where, φ R and φ L are the slope angles of the two banks of the U-shaped valley, b r The valley bottom is wide.

[0036] In one embodiment, step S3: if a landslide dam blocks the river, then based on the dynamic growth model of the landslide dam's catchment area, a Sigmoid function is used to characterize the relationship between the dam body's stability and time, and a dam break risk level index is calculated by the ratio of volume to catchment area. Specifically, the following steps are included: Step S31: Calculate the initial catchment area based on the obtained volume of ice and rock avalanches entering the river and the valley geometric parameters ; The specific implementation is: The dam failure risk is assessed based on the sediment volume. The dam failure risk is related to the catchment area of ​​the dam and the volume of ice and rock avalanches. The initial catchment area can be estimated by the following formula: Formula (20) Where, α r is the river channel inclination; Step S32: establishing a dynamic growth model for the catchment area of ​​the landslide dam, calculating the water level rise at any time based on the regional flow and the time of river blocking, and deriving the catchment area at the corresponding time to obtain the dynamic catchment area of ​​the landslide dam; If a landslide blocks the river, the catchment area A at any time t of the blockage c (t) That is, the dynamic catchment area can be: Formula (21) Where Δh is the height of the water level rising in the blocked river, B newis the equivalent width of the rising water surface height, B new is a function of Δh.

[0037] According to the river channel inclination α r As well as the regional flow Q and the time t of river blocking, the water level rise Δh in the blocked river in formula (21) can be estimated by solving the following equation: Formula (22) Step S33: quantify the temporal variation of dam stability through the ratio of ice and rock avalanche volume to dynamic catchment area, and construct a dam failure risk level index based on the Sigmoid function. The specific implementation is as follows: Once the catchment area A of the dam is obtained c (t) The total volume of the landslide dam and the stability of the dam can be used to analyze the potential losses caused by the discharge of high-energy floods to downstream residents, towns, and water lifeline projects, and the dam failure risk level index I k To express: Formula (23) Where, I k It is expressed in the form of a Sigmoid function and is determined by the ratio of the accumulation volume to the time-dependent catchment area. Step S34: Classify the dam break risk level indicators into dam break risk levels, specifically as follows: When I k When the value is >0.75, the landslide dam cannot be formed (it will break at the beginning) or the landslide dam will become unstable and fail; When I k When ∈[0.56,0.75], the landslide dam is unstable and has a high risk of dam failure; When I k When ∈[0.25,0.56], the landslide dam is relatively unstable and has the risk of dam failure; When I k When it is <0.25, the landslide dam is relatively stable and there is no risk of dam failure.

[0038] In one embodiment, if a landslide dam fails, the dam failure flood hazard is calculated.

[0039] In one embodiment, step S4: if the landslide dam does not fail, then the ice and rock avalanche risk level, river blockage risk level, and dam failure risk level indicators are integrated, a multi-factor coupling weight allocation strategy is adopted to construct a cascade disaster vulnerability calculation formula, and the risk weights of each stage are dynamically adjusted to assess the comprehensive risk of the disaster chain. Specifically, the following steps are included: Step S41: normalizing the obtained ice-rock avalanche hazard level index, river blocking hazard risk index, and dam break risk index to obtain a normalized ice-rock avalanche risk level index; Step S42: Based on the normalized ice and rock avalanche risk level index, a two-level threshold is set to control the activation state of the disaster assessment process and combined with the activation function Re to obtain a hard gating signal for each disaster stage; Step S43: Using the analytic hierarchy process to assign weights to the three types of disaster-prone objects, and obtaining the normalized vulnerability factor of the disaster-prone objects, which is specifically implemented as follows: Based on the risk indicators of ice-rock avalanche, river blockage, and dam failure obtained in all the above steps, the vulnerability level W of ice-rock avalanche cascade disaster is divided. The calculation formula is as follows: Formula (24) In the formula, the subscript They are ice and rock avalanche, river blockage and dam failure; α i is the risk amplification factor of the cascade effect; I i0 is the normalized risk index, W i is the vulnerability factor, Re is the activation function, and the activation state of the disaster assessment process is controlled by setting a two-level threshold and combining the activation function Re to obtain the hard gating signal of each disaster stage as shown in the following formula: Formula (25) Formula (26) The activation function Re is a hard gate, which significantly controls the nodes where each disaster occurs. If no ice and rock collapse geological disaster occurs, the subsequent two cascade disasters will not be induced.

[0040] Category 1: Vulnerability factor W in formula (24) i , which is expressed as follows: Formula (27) In the formula, the subscript They are ice and rock avalanche, river blockage, and dam failure; P is the personnel risk, which is determined by population density × evacuation difficulty coefficient; H is housing property, which is the housing value per unit area × building disaster resistance coefficient; A is the area affected by river blockage backwater, which is determined by the proportion of the potential flooded area upstream and determined by numerical inversion; E is the economic loss, which comprehensively reflects direct economic losses and indirect economic losses, and the calculation expression is as follows: Formula (28) In formula (28), direct economic loss E direct It can be determined by the following formula: Formula (29) Where N i is the area of ​​the i-th type of assets, V i is the unit asset value, D i is the damage rate, the value is between [0,1], Ccritical for loss of critical infrastructure; In formula (28), indirect economic loss E indirect It can be determined by the following formula: Formula (30) Where, T is the number of days of production interruption caused by the disaster, G d is the average daily GDP of the region, R is the emergency response and recovery cost, and λ is the industry impact coefficient, which takes a value of [1, 2], with larger values ​​being used in areas sensitive to agriculture and tourism; In formula (28), the total regional economic volume E regional Used to normalize economic losses, it can be determined by the following formula: Formula (31) Category 2: Risk amplification factors of ice and rock avalanches, river blockages, and dam breaches that reflect cascading effects α l , α b , α k are the risk amplification factors of ice and rock avalanche, river blockage, and dam failure, and the default value is 1 / 3, i.e., α l + α b + α k = 1. If the risk at a certain stage is emphasized, the parameters can be adjusted, for example, the amplification factor of ice and rock avalanche disaster can be increased to 0.4.

[0041] Category 3: Ice and rock avalanche, river blockage, and dam breach risk indicators I l0 , I b0 , I k0 are the normalized risk indicators of ice and rock avalanche, river blockage, and dam failure, respectively. The above expressions are all normalized to the interval [0,1] using the maximum and minimum values, where 0 represents no risk and 1 represents the highest risk.

[0042] Category 4: Weight coefficient w lp + w lh + w le = 1, w b = 1, w kp + w kh + w ke = 1, the specific numerical distribution can be determined by the hierarchical analysis method. In this application, personnel risk is the most important influencing factor, so the personnel risk weight is 0.5.

[0043] The vulnerability level W is divided into 0-0.2 low vulnerability, 0.2-0.4 medium vulnerability, 0.4-0.6 high vulnerability, and 0.6-1 extremely high vulnerability.

[0044] Step S44: The output results of the previous steps are integrated through a dynamic weight assessment model, wherein the ice and rock avalanche risk level index, river blocking risk level index and dam break risk level index are coupled with the vulnerability factors of the disaster-bearing bodies at the corresponding stages and are constrained by hard gating signals to obtain the disaster vulnerability levels of each stage.

[0045] In a specific embodiment, the present application provides a quantitative risk assessment method for the ice and rock avalanche-river blocking-dam breach flood disaster chain, including ice and rock avalanche risk level indicators, river blocking risk level indicators, dam breach risk level indicators, and regional vulnerability, which are specifically implemented as follows: (1) Ice and rock avalanche risk index The model's effectiveness was verified using a mountainous area in a river basin in western China as an example. This region boasts a vast ice and snow belt, severe mountain weathering, and large temperature swings between day and night. Its unique climatic conditions make it a high-risk area for geological disasters, particularly as global warming significantly increases the risk of ice and rock avalanches. This makes it a good test area for validating the developed model. Therefore, it is feasible to use this region as a case study for model validation. The detailed steps are as follows: Figure 3 (a) is a 2D topographic map drawn for visualization using Python program; Figure 3 (b) is a 3D topographic map drawn using Python program for visualization; First, based on the data model of the mountain area and the steps designed in this application, a Python program is used to draw visual 2D and 3D topographic maps, such as Figure 3 As shown in the figure, the area is 2350 m × 2350 m, with an elevation range of 654 m to 1678 m. This two-dimensional data is stored in the data.xlsx file, with the table name "Terrain." Combined with rainfall data, temperature data, and historical landslide occurrence data from the local meteorological department, the data is stored in the data.xlsx file, respectively named "Precipitation," "Temperature," and "Landslides." Note that the size of these data must be consistent. For example, if there is no rainfall, the data can be replaced with 0.

[0046] The model developed in this application automatically calculates the slope of the relevant area without requiring additional data preparation. Therefore, as long as the provided DEM is sufficiently accurate, high-precision slope data can be automatically obtained as a landslide-inducing factor. The two-dimensional "Slope" data calculated by the program is combined with rainfall data ("Precipitation"), temperature data ("Temperature"), and historical landslide occurrences ("Landslides") obtained from the local meteorological department, and stored in the file data.xlsx. A BP neural network model is used to establish a correspondence between topographic, rainfall, and temperature characteristics and historical landslide data ("Landslides"). Data preprocessing, including missing value filling, normalization, and feature engineering, is performed to eliminate dimensional differences and extract valid features. Subsequently, a BP neural network model is constructed. The input layer receives features such as topography, rainfall, and temperature. The hidden layer learns the complex relationships between these features through multiple layers of nonlinear transformations. The output layer uses a sigmoid activation function to output a probability value (ranging from 0 to 1) for landslide occurrence. According to the predicted probability value, landslide risks are divided into five levels: the probability of occurrence is 0-0.2 as risk level 1 (very low risk), 0.2-0.4 as risk level 2 (low risk), 0.4-0.6 as risk level 3 (medium risk), 0.6-0.8 as risk level 4 (high risk), and 0.8-1.0 as risk level 5 (very high risk).

[0047] The trained model and trained normalizer are stored as "scaler.pkl" and "BP neural network_model.h5" respectively. The historical rainfall and temperature data of other months are processed by the data normalizer and input into the trained model to obtain the results. This application has standardized the program. You only need to provide a file named Prediction.xlsx and include relevant data in the tables named "Terrain", "Precipitation", and "Temperature". The following is the corresponding prediction result. Figure 4 The display shows the prediction results statistics for a total of 55,225 grid areas in the region. Figure 5 The following figure shows a high-risk visualization map of the area (red dots indicate high-risk landslide areas). As can be seen from the figure below, the program developed by this application can intelligently predict the areas and risk levels of future landslides based on historical data, and can effectively warn of possible landslide geological disasters.

[0048] (2) Calculation of river blocking risk level indicators According to the river blocking risk level index I bThe calculation principle is: the shallower the river water depth and the slower the river flow in the area where the ice and rock avalanche occurs, the larger the landslide volume and the faster the speed, the more likely it is to cause a river blockage disaster. b Index calculation, Table 1 lists the river blocking values ​​X of the three groups of landslides b Calculation results.

[0049] Table 1 Calculation results of river blocking values ​​for three groups of landslides

[0050] The above calculation results are consistent with the actual prediction results of landslide and river blocking disasters. Therefore, the river blocking risk level index I b The calculation is reasonable.

[0051] (3) Calculation of dam break risk level index Still taking Baige landslide as an example, according to the theoretical calculation model proposed in this application, Figure 7 As shown in the figure, the relevant parameters of the Baige landslide are as follows: the flow rate is 1400m^3 / s, the river channel slope is 3%, and the I calculated based on the above formulas (20)-(23) is k The changes during the 50 hours of blocking the river are as follows: Figure 8 As shown. According to I k The relationship with the risk level, that is, when I k ∈[0.56,0.75], the risk of dam failure is high; when I k When ∈[0.25,0.56], the landslide dam is relatively unstable and there is a risk of dam failure. It can be judged that after blocking the river for about 1.7 hours, there is a high risk of dam failure, and after blocking the river for about 48 hours, there is an extremely high risk of dam failure.

[0052] (4) Regional vulnerability calculation Taking the Baige landslide as an example, the design parameters of the area are given below to test whether the vulnerability index can reflect the vulnerability index of the ice-rock avalanche hazard chain and whether it is reasonable: Vulnerability factors: P: Maximum population density is 1000 / km 2 ,The evacuation difficulty coefficient is believed to be positively correlated with the population density of the disaster area; H: It is believed to be positively correlated with the population density of the disaster area, that is, the value of housing per unit area increases with population growth; E: Direct economic loss 10 5 (10,000 yuan), indirect economic losses 5 (10,000 yuan), total economic GDP10 7 (Ten thousand yuan). The calculation result is as follows Figure 9 shown.

[0053] Assuming that the regional economic losses remain unchanged, it can be seen from the figure that as the population density continues to approach the maximum population density, the vulnerability caused by ice and rock avalanche disasters shows an exponential growth, which is consistent with the conventional perception that "the more people affected, the more serious the disaster". Therefore, the vulnerability index designed in this application is reasonable. At the same time, according to the difference in risk indicators, when ice and rock avalanches are at a low probability of occurrence, even if the population density is close to the maximum, the vulnerability index designed in this application will not give an early warning. Once the probability of ice and rock avalanches increases to a medium probability, the medium vulnerability index will increase rapidly, which proves that the design of the vulnerability coefficient is reasonable.

[0054] In a second aspect, the present application provides a quantitative risk assessment system for the ice and rock avalanche-river blockage-dam breach flood disaster chain, including: The module for obtaining the ice and rock avalanche risk level index is used to input the regional digital elevation, meteorological data, and historical disaster data into the BP neural network improved by the Adam optimizer. It outputs the probability of ice and rock avalanche disasters by dynamically optimizing the network weights, and then classifies them into different levels to obtain the ice and rock avalanche risk level index. The river blocking risk level indicator acquisition module is used to calculate the movement path of the ice and rock avalanche based on the D8 algorithm combined with the rock-soil friction coefficient correction if an ice and rock avalanche occurs. The volume is corrected in real time through the erosion-deposition dynamic balance model, and the river blocking risk level indicator is constructed based on the volume and velocity of the inflow into the river. The dam break risk level index acquisition module is used to calculate the dam break risk level index by the ratio of volume to catchment area based on the dynamic growth model of the dam's catchment area, using the Sigmoid function to characterize the relationship between the dam body stability and time if a landslide dam blocks the river. The module for obtaining the comprehensive risk of the disaster chain is communicated with the ice and rock avalanche risk level indicator acquisition module, the river blocking risk level indicator acquisition module and the dam break risk level indicator acquisition module. If no dam break disaster occurs in the landslide dam, the module is used to integrate the ice and rock avalanche risk level, the river blocking risk level and the dam break risk level indicators, adopt a multi-factor coupling weight allocation strategy to construct a cascade disaster vulnerability calculation formula, dynamically adjust the risk weights of each stage, and evaluate the comprehensive risk of the disaster chain.

[0055] In one embodiment, the river blocking risk level indicator acquisition module includes: The effective slope acquisition unit is used to extract the elevation data of the corresponding grid cell in the digital elevation model based on the coordinates of the ice and rock avalanche risk point, calculate the slope between the central grid cell and the neighboring grid using the D8 algorithm, and perform terrain correction based on the friction coefficient of different rock and soil types to obtain the effective slope; a river-entry volume acquisition unit, in communication with the effective slope acquisition unit, configured to select a direction with the largest effective slope as the movement direction of the ice-rock avalanche to obtain a movement path of the ice-rock avalanche, and calculate the river-entry volume of the ice-rock avalanche by combining erosion and accumulation effects on the movement path and the principle of volume conservation; The water entry velocity acquisition unit is used to calculate the cumulative volume according to the angle between the ice and rock avalanche flow direction and the river channel direction when the movement path reaches the river channel grid, and calculate the water entry velocity using the Manning formula; The river blocking risk level index acquisition unit is communicatively connected to the effective slope acquisition unit, the river inflow volume acquisition unit and the water inflow speed acquisition unit, and is used to obtain the river blocking value based on the ratio of the river inflow volume, the water inflow speed and the river water flow rate, combined with the river valley geometric parameters, and to obtain the river blocking risk level index by performing segmented judgment on the river blocking value.

[0056] Among them, the functional implementation of each module in the above-mentioned ice and rock avalanche-river blocking-dam break flood disaster chain quantitative risk assessment system corresponds to the various steps in the above-mentioned ice and rock avalanche-river blocking-dam break flood disaster chain quantitative risk assessment method embodiment, and its functions and implementation processes will not be repeated here one by one.

[0057] On the third aspect, an embodiment of the present application provides a quantitative risk assessment device for the ice and rock avalanche-river blocking-dam breach flood disaster chain. The quantitative risk assessment device for the ice and rock avalanche-river blocking-dam breach flood disaster chain can be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0058] Communication interfaces include input / output (I / O), physical, and logical interfaces, which interconnect components within the device and other devices (such as other computing devices or user devices) involved in the quantitative risk assessment of the ice and rock avalanche-river blockage-dam breach flood hazard chain. Physical interfaces can include Ethernet, fiber optic, or ATM interfaces, while user devices can include displays and keyboards.

[0059] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0060] The processor can be a general-purpose processor that can invoke a program stored in memory for the quantitative risk assessment of the ice and rock avalanche-river blocking-dam-break flood disaster chain and execute the quantitative risk assessment method for the ice and rock avalanche-river blocking-dam-break flood disaster chain provided in the embodiments of the present application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the program is invoked can be referenced from the various embodiments of the quantitative risk assessment method for the ice and rock avalanche-river blocking-dam-break flood disaster chain provided in the present application and will not be further described here.

[0061] In a fourth aspect, an embodiment of the present application also provides a readable storage medium.

[0062] The readable storage medium of the present application stores a quantitative risk assessment program for the ice and rock collapse-river blocking-dam break flood disaster chain, wherein when the quantitative risk assessment program for the ice and rock collapse-river blocking-dam break flood disaster chain is executed by a processor, the steps of the quantitative risk assessment method for the ice and rock collapse-river blocking-dam break flood disaster chain as described above are implemented.

[0063] Among them, the method implemented when the quantitative risk assessment program of the ice and rock avalanche-river blocking-dam break flood disaster chain is executed can refer to the various embodiments of the quantitative risk assessment method of the ice and rock avalanche-river blocking-dam break flood disaster chain of this application, and will not be repeated here.

[0064] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0065] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of this application.

[0066] The above are only preferred embodiments of the present application and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the description and drawings of this application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A quantitative risk assessment method for the ice and rock collapse-river blockage-dam breach flood disaster chain, characterized by: The following steps are involved: A BP neural network improved by the Adam optimizer was used as input for regional digital elevation, meteorological data, and historical disaster data. The probability of ice and rock avalanche disasters was output by dynamically optimizing the network weights, and the ice and rock avalanche risk level index was obtained by grading. If an ice and rock avalanche occurs, the D8 algorithm is used to calculate the movement path of the ice and rock avalanche in combination with the correction of the rock-soil friction coefficient. The volume is corrected in real time through the erosion-deposition dynamic balance model, and the river blockage risk level index is constructed based on the volume and speed of the river. If a landslide dam blocks the river, the dam failure risk level index is calculated based on the dynamic growth model of the dam's catchment area and the Sigmoid function to characterize the relationship between the dam's stability and time. If the landslide dam does not fail, the ice and rock avalanche risk level, river blockage risk level and dam failure risk level indicators are integrated, and a multi-factor coupling weight allocation strategy is used to construct a cascade disaster vulnerability calculation formula, dynamically adjust the risk weights of each stage, and evaluate the comprehensive risk of the disaster chain.

2. The quantitative risk assessment method for the ice and rock collapse-river blockage-dam breach flood disaster chain according to claim 1, characterized in that: The BP neural network improved by the Adam optimizer is input with regional digital elevation, meteorological data and historical disaster data, and the probability of ice and rock avalanche disaster is output by dynamically optimizing the network weights. The ice and rock avalanche risk level index is obtained by grading. Specifically, the following steps are included: Construct BP neural network; Initialize the BP neural network weight parameters, use the Adam optimizer to dynamically adjust the learning rate of the BP neural network weight parameters, calculate the first-order moment estimate and second-order moment estimate of the gradient, and adaptively update the network weights in combination with the bias correction term to obtain the BP neural network improved by the Adam optimizer; A binary cross-entropy loss function is defined, and the weight gradients of each layer are calculated using the back-propagation algorithm. For the binary classification task of whether an ice and rock avalanche has occurred, the loss function is used to measure the error between the probability of an ice and rock avalanche disaster and the true label. The gradient of the hidden layer weight is calculated using the chain rule to obtain the gradient update value of the weights of each layer of the neural network. The standardized historical disaster data is input into the BP neural network improved by the Adam optimizer for training. The gradient update value of the weights of each layer of the neural network is used for iterative optimization to converge the binary cross entropy loss function and obtain the iteratively optimized BP neural network. The standardized real-time monitoring data is input into the iteratively optimized BP neural network, and the output probability of ice and rock avalanche disaster is classified into different levels to obtain the ice and rock avalanche risk level index.

3. The quantitative risk assessment method for the ice and rock collapse-river blockage-dam breach flood disaster chain according to claim 2 is characterized in that: The model architecture of the BP neural network includes an input layer, a hidden layer and an output layer. The input layer uses digital elevation data, rainfall data and temperature data as feature inputs. The hidden layer uses the ReLU activation function for nonlinear feature transformation. The output layer uses the Sigmoid activation function to map the calculation results into the probability of disaster occurrence.

4. The quantitative risk assessment method for the ice and rock collapse-river blockage-dam breach flood disaster chain according to claim 1, characterized in that: If an ice and rock avalanche occurs, the movement path of the ice and rock avalanche is calculated based on the D8 algorithm combined with the correction of the rock-soil friction coefficient. The volume is corrected in real time through the erosion-deposition dynamic balance model. The river blocking risk level index is constructed based on the volume and speed of the river entering the river. The specific steps include: Based on the coordinates of the ice and rock avalanche risk points, the elevation data of the corresponding grid cells in the digital elevation model are extracted. The slope between the central grid cell and the neighboring grid cells is calculated using the D8 algorithm. The terrain is corrected by combining the friction coefficient of different rock and soil types to obtain the effective slope. The direction with the largest effective slope is selected as the movement direction of the ice-rock avalanche to obtain the movement path of the ice-rock avalanche. The volume of the ice-rock avalanche entering the river is calculated by combining the erosion and accumulation effects on the movement path and the principle of volume conservation. When the movement path reaches the river channel grid, the cumulative volume is calculated based on the angle between the ice and rock avalanche flow direction and the river channel direction, and the water entry velocity is calculated using the Manning formula; Based on the ratio of the volume of water entering the river, the water inflow speed and the river flow speed, combined with the geometric parameters of the river valley, the river blocking value is calculated and the river blocking value is segmented to obtain the river blocking risk level index.

5. The quantitative risk assessment method for the ice and rock collapse-river blockage-dam breach flood disaster chain according to claim 4, characterized in that: Based on the coordinates of the ice and rock avalanche risk points, the elevation data of the corresponding grid cells in the digital elevation model are extracted, the slope between the central grid cell and the neighboring grid cells is calculated using the D8 algorithm, and the terrain correction is performed based on the friction coefficient of different rock and soil types to obtain the effective slope. Specifically, the following steps are included: The slope between the central grid cell and the neighboring grid cells is calculated based on the elevation difference between the central grid cell and the adjacent 8 grid cells, combined with the grid resolution. The friction coefficient of different rock and soil types is introduced as a terrain correction factor to correct the slope and obtain the effective slope between the central grid cell and the neighboring grid cell.

6. The quantitative risk assessment method for the ice and rock avalanche-river blockage-dam breach flood disaster chain according to claim 1, characterized in that: If a landslide dam blocks the river, the dam failure risk level index is calculated by using the ratio of volume to catchment area, and the following steps are specifically included: Based on the obtained volume of ice and rock avalanches into the river and combined with the valley geometric parameters, the initial catchment area is calculated; A dynamic growth model of the water catchment area of ​​the landslide dam is established. The water level rise at any time is calculated based on the regional flow and the time of river blockage. The water catchment area at the corresponding time is then derived to obtain the dynamic water catchment area of ​​the landslide dam. Through the ratio of ice and rock avalanche volume to dynamic catchment area, the temporal characteristics of dam stability are quantified, and a dam failure risk level index based on Sigmoid function is constructed. The dam break risk level index is divided into dam break risk level.

7. The quantitative risk assessment method for the ice and rock avalanche-river blockage-dam breach flood disaster chain according to claim 1, characterized in that: If the barrier dam does not fail, the ice and rock avalanche risk level, river blockage risk level, and dam failure risk level indicators are integrated, and a multi-factor coupling weight allocation strategy is adopted to construct a cascade disaster vulnerability calculation formula, dynamically adjust the risk weights of each stage, and evaluate the comprehensive risk of the disaster chain. Specifically, the following steps are included: Normalizing the obtained ice-rock avalanche disaster hazard level index, river blocking disaster risk index, and dam break risk index to obtain a normalized ice-rock avalanche risk level index; Based on the normalized ice and rock avalanche risk level index, the activation state of the disaster assessment process is controlled by setting a two-level threshold and combining the activation function Re to obtain the hard gating signal of each disaster stage. The analytic hierarchy process is used to assign weights to the three types of disaster-prone objects and obtain the normalized vulnerability factor of the disaster-prone objects. The output results of the previous steps are integrated through a dynamic weight assessment model, in which the ice and rock avalanche risk level index, river blockage risk level index and dam break risk level index are coupled with the vulnerability factors of the disaster-bearing bodies in the corresponding stages and constrained by hard gating signals to obtain the disaster vulnerability level at each stage.

8. The quantitative risk assessment method for the ice and rock collapse-river blockage-dam breach flood disaster chain according to claim 7, characterized in that: The output results of the previous steps are integrated through the dynamic weight assessment model. The ice and rock avalanche risk level index, river blockage risk level index, and dam break risk level index are coupled with the vulnerability factors of the hazard-bearing bodies at the corresponding stages and are constrained by hard gating signals to obtain the disaster vulnerability level at each stage, as shown in the following formula: Formula (24) In the formula, the subscript They are ice and rock avalanche, river blockage and dam failure; α i is the risk amplification factor of the cascade effect; I i0 is the normalized risk index, W i is the vulnerability factor, and Re is the activation function.

9. A quantitative risk assessment system for the ice and rock collapse-river blockage-dam breach flood disaster chain, characterized by: include: The module for obtaining the ice and rock avalanche risk level index is used to input the regional digital elevation, meteorological data, and historical disaster data into the BP neural network improved by the Adam optimizer. It outputs the probability of ice and rock avalanche disasters by dynamically optimizing the network weights, and then classifies them into different levels to obtain the ice and rock avalanche risk level index. The river blocking risk level indicator acquisition module is used to calculate the movement path of the ice and rock avalanche based on the D8 algorithm combined with the rock-soil friction coefficient correction if an ice and rock avalanche occurs. The volume is corrected in real time through the erosion-deposition dynamic balance model, and the river blocking risk level indicator is constructed based on the volume and velocity of the inflow into the river. The dam break risk level index acquisition module is used to calculate the dam break risk level index by the ratio of volume to catchment area based on the dynamic growth model of the dam's catchment area, using the Sigmoid function to characterize the relationship between the dam body stability and time if a landslide dam blocks the river. The module for obtaining the comprehensive risk of the disaster chain is communicated with the ice and rock avalanche risk level indicator acquisition module, the river blocking risk level indicator acquisition module and the dam break risk level indicator acquisition module. If no dam break disaster occurs in the landslide dam, the module is used to integrate the ice and rock avalanche risk level, the river blocking risk level and the dam break risk level indicators, adopt a multi-factor coupling weight allocation strategy to construct a cascade disaster vulnerability calculation formula, dynamically adjust the risk weights of each stage, and evaluate the comprehensive risk of the disaster chain.

10. The quantitative risk assessment system for the ice and rock collapse-river blockage-dam breach flood disaster chain according to claim 9, characterized in that: The river blocking risk level indicator acquisition module includes: The effective slope acquisition unit is used to extract the elevation data of the corresponding grid cell in the digital elevation model based on the coordinates of the ice and rock avalanche risk point, calculate the slope between the central grid cell and the neighboring grid using the D8 algorithm, and perform terrain correction based on the friction coefficient of different rock and soil types to obtain the effective slope; a river-entry volume acquisition unit, in communication with the effective slope acquisition unit, configured to select a direction with the largest effective slope as the movement direction of the ice-rock avalanche to obtain a movement path of the ice-rock avalanche, and calculate the river-entry volume of the ice-rock avalanche by combining erosion and accumulation effects on the movement path and the principle of volume conservation; The water entry velocity acquisition unit is used to calculate the cumulative volume according to the angle between the ice and rock avalanche flow direction and the river channel direction when the movement path reaches the river channel grid, and calculate the water entry velocity using the Manning formula; The river blocking risk level index acquisition unit is communicatively connected to the effective slope acquisition unit, the river inflow volume acquisition unit and the water inflow speed acquisition unit, and is used to obtain the river blocking value based on the ratio of the river inflow volume, the water inflow speed and the river water flow rate, combined with the river valley geometric parameters, and to obtain the river blocking risk level index by performing segmented judgment on the river blocking value.

Images