Reservoir area landslide susceptibility evaluation method considering reservoir water level change factor

Abstract

The invention provides a reservoir area landslide susceptibility evaluation method considering reservoir water level change factors. The method comprises the following steps: preparing data; constructing a model; introducing an adjustment coefficient; probability calculation and loss function definition; optimizing parameters; carrying out regularization processing; and outputting a result. The invention mainly considers the influence of reservoir water level change on reservoir area landslide susceptibility evaluation, provides a reservoir area landslide dynamic susceptibility evaluation method considering reservoir water level change, and mainly solves the following problems: a traditional landslide susceptibility evaluation technology is mainly oriented to administrative division, and the method mainly aims at evaluating the reservoir area landslide susceptibility, so that the reservoir area landslide dynamic susceptibility evaluation method can be used for evaluating the reservoir area landslide susceptibility. A river distance is introduced as an important influence factor of reservoir area landslide susceptibility evaluation.

Description

Technical Field

[0001] This invention relates to the field of landslide disaster early warning and machine learning, specifically to a method for evaluating the susceptibility of landslides in reservoir areas that takes into account reservoir water level changes. Background Technology

[0002] Hydropower, as an important clean energy source, plays a vital role in achieving energy conservation, emission reduction, and promoting low-carbon environmental protection. Southwest my country, located at the intersection of the Qinghai-Tibet Plateau, the Yunnan-Guizhou Plateau, and the Sichuan Basin, is rich in hydropower resources. Its undulating terrain, concentrated rainfall, and dense river network make it an ideal region for constructing water conservancy and hydropower projects.

[0003] However, during the operation of a hydropower station, the periodic fluctuations in the reservoir's water level can trigger a series of geological problems. As the impoundment process progresses, the weight of the reservoir slopes increases, reducing the shear strength of the soil and rock, making them highly susceptible to landslides. For example... Figure 1 As shown, the landslide poses a serious threat to the structural safety of water conservancy projects and the safety of life and property of surrounding residents.

[0004] Landslide susceptibility assessment is a technical means used to predict the probability of landslides occurring in a specific area. It can effectively delineate high-risk and low-risk areas, providing technical support for landslide prevention and mitigation. Most traditional landslide susceptibility assessment methods are based on static geological parameters, which are difficult to respond to the dynamic changes in landslide risk caused by water level changes during reservoir operation.

[0005] Therefore, there is an urgent need for an evaluation method that can dynamically reflect the impact of reservoir water level changes on landslide risk, so as to improve the real-time and scientific nature of water conservancy project risk management. Summary of the Invention

[0006] The main objective of this invention is to provide a method for evaluating the susceptibility of landslides in reservoir areas that takes into account changes in reservoir water level, thereby solving the problem in existing technologies that make it difficult to dynamically reflect the impact of changes in reservoir water level on landslide risk.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: The method steps are as follows: S1. Data preparation: Collect historical unstable slope data and multiple key factors affecting landslides in the reservoir area as input conditions for the model; S2. Model Construction: A landslide susceptibility assessment model is established based on the logistic regression algorithm, and high-risk landslide areas are identified by calculating the probability. S3. Adjustment coefficient introduction: Considering the amplification effect of the rise in reservoir water level on the "distance from the river" factor, an adjustment coefficient is set to correct the relevant model parameters; S4. Probability Calculation and Loss Function Definition: Construct a probability calculation formula for landslide occurrence, measure the error between the model's predicted value and the actual value through the loss function, and construct a cost function to evaluate the overall performance of the model. S5. Parameter Optimization: The gradient descent algorithm is used to iteratively optimize the model parameters and improve the model accuracy. S6. Regularization: Introduce a regularization term to suppress model overfitting and improve generalization ability and stability; S7. Results Output: The optimized model is used to accurately delineate high-risk landslide areas and complete the landslide susceptibility assessment.

[0008] Preferably, the collection of historical unstable slope data in the reservoir area in step S1 includes: We comprehensively employed InSAR deformation interpretation, satellite remote sensing, and field surveys to acquire historical unstable slope data in the reservoir area. Of this, 80% of the data was used for model training, and 20% was used for testing to verify the prediction accuracy of the logistic regression model.

[0009] Preferably, the key factors affecting landslides in step S1 include: Topographic factors: reservoir area elevation, slope angle distribution, slope aspect distribution, topographic plane curvature, topographic profile curvature; Surface properties: lithology distribution, vegetation cover; Distance-related factors: distance from river, distance from road, distance from fault; Meteorological conditions: Rainfall distribution; Among them, "distance from the river" is the core factor affected by changes in reservoir water level. All influencing factor data are classified using the natural breakpoint classification method to facilitate model analysis and calculation.

[0010] Preferably, the model construction in step S2 specifically includes: Using collected historical unstable slope data and multiple key factors affecting landslides in the reservoir area as input conditions, the landslide susceptibility of the reservoir area is evaluated based on the logistic regression algorithm of machine learning, and high-risk areas for landslide disasters are delineated. The calculation model is shown in Equations (1) and (2): (1); (2); In the above formula, x1~x 11 These represent dummy variables, b and w, representing various factors influencing landslides. i Let g(z=1) represent the coefficients of the model, e be the natural logarithm, h(x) be the decision value, and x be the eigenvalue.

[0011] Preferably, the adjustment coefficient introduced in step S3 specifically includes: Considering the amplification effect of the distance from the river after the reservoir water level rises, an adjustment coefficient μ is introduced. The formula for calculating the adjustment coefficient is shown in equation (3): (3); In the formula, h1(m) represents the initial reservoir water level, h2(m) represents the reservoir water level after the water level change, and h0 represents the elevation of the bottom of the reservoir. Assuming the entire river channel is of uniform width, and factor x1 represents the distance of the influencing factor from the river, the expression for z after the reservoir water level changes is as follows: (4); In equation (4), μ adjusts the coefficient w1 to obtain a new coefficient w1.

[0012] Preferably, the probability calculation and loss function definition in step S4 specifically include: Probability calculation: (5); Where p is the probability of a random event occurring, and the range of p is [0, 1], h(x) is the decision value, and x is the eigenvalue; (6); (7); (8); (9); in Indicates predicted value , Represents the actual value; The loss function measures the error of a single-sample prediction. To measure the algorithm's performance on all training samples, a cost function is defined. This cost function is the sum of the loss functions for m samples, divided by m. (10); The specific solution process is as follows, and the likelihood function is: (11); (12); The cost function is: (13).

[0013] Preferably, the parameter optimization in step S5 specifically includes: The gradient descent solution process, where α is the learning rate: (14); (15); (16); ; Therefore, we can conclude that: (17); The parameter update formula for gradient descent is as follows: (18).

[0014] Preferably, in step S6, after solving for the gradient descent parameters, a regularization function is added to prevent overfitting. The objective function after adding the regularization function is as follows: (19); in For regularization terms, when When the value of starts to rise, the variance decreases.

[0015] This invention provides a method for evaluating the landslide susceptibility of reservoir areas considering reservoir water level changes. This invention primarily considers the impact of reservoir water level changes on the evaluation of landslide susceptibility in reservoir areas, and proposes a dynamic landslide susceptibility evaluation method that considers reservoir water level changes. This method can mainly solve the following problems: Traditional landslide susceptibility evaluation techniques are mainly geared towards administrative divisions, while this invention mainly evaluates the landslide susceptibility of reservoir areas by introducing "distance from the river" as an important influencing factor in the evaluation of landslide susceptibility in reservoir areas. Attached Figure Description

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a diagram of the existing reservoir and its bank slope structure. Figure 2 This is a landslide susceptibility assessment map of the reservoir area before the first stage of water impoundment; Figure 3 This is a landslide susceptibility assessment map of the reservoir area after the second phase of water impoundment; Figure 4 This is a diagram of the new unstable slope that appears after the second stage of water impoundment. Detailed Implementation

[0017] Example 1 A method for assessing the susceptibility of reservoir landslides considering reservoir water level changes is proposed. This invention considers the impact of reservoir water level changes on the susceptibility of reservoir landslides. Based on the logistic regression algorithm of machine learning, "distance from the river" is taken as an important influencing factor. An amplification factor is introduced to consider the amplification effect of the "distance from the river" factor after the reservoir water level rises, resulting in a dynamic susceptibility assessment method for reservoir landslides considering reservoir water level changes. This enables real-time dynamic updating of the susceptibility assessment results for hydropower station reservoir landslides.

[0018] Collection of datasets on unstable slopes in reservoir areas: A dataset of historical unstable slopes in the reservoir area was obtained by combining multiple methods, including InSAR deformation interpretation, satellite remote sensing, and field surveys. 80% of the unstable slope dataset was selected as the training set, and 20% as the test set for training and accuracy verification of a machine learning model using a logistic regression algorithm.

[0019] Collection of main influencing factors of landslides in reservoir area: This method considers 11 main influencing factors of reservoir area landslides, including reservoir elevation, reservoir slope angle distribution, reservoir slope aspect distribution, reservoir topographic plan curvature, reservoir topographic profile curvature, reservoir lithology distribution, reservoir vegetation cover, distance from river, distance from road, distance from fault, and rainfall distribution. Among these, "distance from river" is a factor directly related to reservoir water level changes. Data collection for these influencing factors included database platforms from NASA and the National Tibetan Plateau Scientific Data Center, as well as meteorological and rainfall stations in the reservoir area. The natural discontinuity method was used to classify the collected influencing factors.

[0020] Machine learning algorithm model: Using the historical unstable slope data and 11 influencing factors collected above as input conditions, the landslide susceptibility of the reservoir area is evaluated based on the logistic regression algorithm of machine learning, and high-risk areas for landslide disasters are delineated. The calculation model is shown in Equations (1) and (2): (1); (2); In the above formula, x1~x 11 These represent dummy variables, b and w, representing various factors influencing landslides. i Let g(z=1) represent the coefficients of the model, e be the natural logarithm, h(x) be the decision value, and x be the eigenvalue.

[0021] Considering the amplification effect of "distance from the river" after the reservoir water level rises, an adjustment coefficient μ is introduced. The formula for calculating the adjustment coefficient is shown in equation (3): (3); In the formula, h1(m) represents the initial reservoir water level, h2(m) represents the reservoir water level after the water level change, and h0 represents the elevation of the bottom of the reservoir. Assuming the entire river channel is of uniform width, and factor x1 represents the distance of the influencing factor from the river, the expression for z after the reservoir water level changes is as follows: (4); In equation (4), μ adjusts the coefficient w1 to obtain a new coefficient w1.

[0022] The probability solution, loss function, gradient descent and model parameter update process are shown in equations (4) to (19).

[0023] (5); Where p is the probability of a random event occurring, and the range of p is [0, 1], h(x) is the decision value, and x is the eigenvalue; (6); (7); (8); (9); in Indicates predicted value , Represents the actual value. Since the loss function measures the error of a single-sample prediction, to measure the algorithm's performance on all training samples, a cost function needs to be defined. The cost function is the sum of the loss functions for m samples, divided by m. (10); The specific solution process is as follows, and the likelihood function is: (11); (12); The cost function is: (13).

[0024] The solution process of gradient descent, where Learning rate: (14); (15); (16); ; Therefore, we can conclude that: (17); The parameter update formula for gradient descent is as follows: (18).

[0025] After solving for the gradient descent parameters, a regularization function is added to prevent overfitting.

[0026] The objective function after adding the regularization function is as follows: (19); in For regularization terms, when As the value of increases, the model variance decreases.

[0027] Example 2 like Figures 2-4 As shown in Example 1, a reservoir area is analyzed as an implementation case. The first phase of water impoundment in the reservoir area was from December 3rd to December 5th, 2020, during which the water level rose from 2608 m to 2681 m. The second phase of water impoundment was from May 15th to August 24th, 2021, during which the water level rose from 2681 m to 2790 m. Using this reservoir area as a research case, the proposed technology was employed to collect historical unstable slope datasets and 11 influencing factors from the Lianghekou Reservoir area. Machine learning calculations were performed to obtain the landslide susceptibility assessment results of the reservoir area before the first phase of water impoundment and after the second phase of water impoundment. Figure 2 and Figure 3 As shown in Table 1, the model accuracy verification results indicate that the landslide susceptibility level significantly increases with the rise in reservoir water level. After the second phase of water impoundment, 18 new unstable slopes were created, such as... Figure 4 As shown, Figure 2 The results could only predict 16 unstable slopes; predictions for the 17th and 18th unstable slopes failed. The results were adjusted using different coefficients. Figure 3 The results showed that the 17th and 18th unstable slopes after the second stage of water impoundment could be successfully predicted.

[0028] Table 1: Model Accuracy Evaluation

[0029] It can be understood that this method uses "distance from the river" as the core evaluation indicator and dynamically adjusts its weight by adjusting the coefficient to reflect the amplification effect of rising reservoir water level on landslide risk, thereby achieving dynamic assessment of landslide susceptibility and thus realizing the purpose of dynamic landslide susceptibility assessment in the reservoir area.

[0030] An adjustment coefficient (Formula 3) was proposed to quantitatively describe the impact of reservoir water level changes on landslide susceptibility in the reservoir area. Distance from the river is considered an important influencing factor in the assessment of landslide susceptibility in the reservoir area. After the reservoir is filled and the water level changes, the importance of the "distance from the river" factor is adjusted in real time by adjusting the coefficients. The updating of the model coefficients relies on the iterative optimization of the gradient descent algorithm and also combines the adjustment coefficients to dynamically correct specific influencing factors (such as distance from the river) (Formulas 4 to 19). This achieves the goal of dynamic assessment of landslide susceptibility in the reservoir area, realizes real-time updates of future landslide susceptibility assessment results, and avoids the workload of secondary surveys and secondary collection of datasets.

[0031] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0032] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products of embodiments. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, produce a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0033] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0034] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0035] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, characterized by: The steps are as follows: S1. Data preparation: Collect historical unstable slope data and multiple key factors affecting landslides in the reservoir area as input conditions for the model; S2. Model Construction: A landslide susceptibility assessment model is established based on the logistic regression algorithm, and high-risk landslide areas are identified by calculating the probability. S3. Adjustment coefficient introduction: Considering the amplification effect of the rise in reservoir water level on the "distance from the river" factor, an adjustment coefficient is set to correct the relevant model parameters; S4. Probability Calculation and Loss Function Definition: Construct a probability calculation formula for landslide occurrence, measure the error between the model's predicted value and the actual value through the loss function, and construct a cost function to evaluate the overall performance of the model. S5. Parameter Optimization: The gradient descent algorithm is used to iteratively optimize the model parameters and improve the model accuracy. S6. Regularization: Introduce a regularization term to suppress model overfitting and improve generalization ability and stability; S7. Results Output: The optimized model is used to accurately delineate high-risk landslide areas and complete the landslide susceptibility assessment.

2. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 1, is characterized by: The collection of historical unstable slope data in the reservoir area in step S1 includes: We comprehensively employed InSAR deformation interpretation, satellite remote sensing, and field surveys to acquire historical unstable slope data in the reservoir area. Of this, 80% of the data was used for model training, and 20% was used for testing to verify the prediction accuracy of the logistic regression model.

3. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 1, is characterized by: The key factors affecting landslides in step S1 include: Topographic factors: reservoir area elevation, slope angle distribution, slope aspect distribution, topographic plane curvature, topographic profile curvature; Surface properties: lithology distribution, vegetation cover; Distance-related factors: distance from river, distance from road, distance from fault; Meteorological conditions: Rainfall distribution; Among them, the distance from the river is the core factor affected by the reservoir water level change. All influencing factor data are classified using the natural breakpoint classification method to facilitate model analysis and calculation.

4. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 1, is characterized in that: The model construction in step S2 specifically includes: Using collected historical unstable slope data and multiple key factors affecting landslides in the reservoir area as input conditions, the landslide susceptibility of the reservoir area is evaluated based on the logistic regression algorithm of machine learning, and high-risk areas for landslide disasters are delineated. The calculation model is shown in Equations (1) and (2): （1）； （2）； In the above formula, x1~x 11 These represent dummy variables, b and w, representing various factors influencing landslides. i Let g(z=1) represent the coefficients of the model, e be the natural logarithm, h(x) be the decision value, and x be the eigenvalue.

5. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 1, is characterized in that: The adjustment coefficients introduced in step S3 specifically include: Considering the amplification effect of the distance from the river after the reservoir water level rises, an adjustment coefficient μ is introduced. The formula for calculating the adjustment coefficient is shown in equation (3): （3）； In the formula, h1(m) represents the initial reservoir water level, h2(m) represents the reservoir water level after the water level change, and h0 represents the elevation of the bottom of the reservoir.

6. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 5, is characterized in that: Assuming the entire river channel is of uniform width, and factor x1 represents the distance of the influencing factor from the river, the expression for z after the reservoir water level changes is as follows: （4）； In equation (4), μ adjusts the coefficient w1 to obtain a new coefficient w1.

7. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 1, is characterized in that: Step S4, specifically the probability calculation and loss function definition, includes: Probability calculation: （5）； Where p is the probability of a random event occurring, and the range of p is [0, 1], h(x) is the decision value, and x is the eigenvalue; （6）； （7）； （8）； （9）； in Indicates predicted value , Represents the actual value.

8. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 7, is characterized by: considering... The loss function measures the error of a single-sample prediction. To measure the algorithm's performance on all training samples, a cost function is defined. The cost function is the sum of the loss functions for m samples, divided by m. （10）； The specific solution process is as follows, and the likelihood function is: （11）； （12）； The cost function is: （13）。 9. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 1, is characterized in that: Step S5 Parameter optimization specifically includes: The solution process of gradient descent, where Learning rate: （14）； （15）； （16）； ； Therefore, we can conclude that: （17）； The parameter update formula for gradient descent is as follows: （18）。 10. The method for evaluating the susceptibility of landslides in reservoir areas considering reservoir water level changes, as described in claim 1, is characterized in that: Step S6 involves adding a regularization function after solving for the gradient descent parameters to prevent overfitting. The objective function after adding the regularization function is as follows: （19）； in For regularization terms, when When the value of starts to rise, the variance decreases.

Images