Digital modeling method for landslide surge disaster prediction

Abstract

The invention belongs to the technical field of surge disaster prediction, and particularly relates to a digital modeling method for landslide surge disaster prediction. Acquiring landslide surge test data by using a wave pool model test device, analyzing the sensitivity of surge influence parameters to surge characteristic parameters, and acquiring a landslide surge disaster prediction model under test conditions; substituting the actual surge influence parameters in the engineering case into the prediction model under the test condition to obtain predicted surge characteristic parameters of the engineering case; and the prediction precision of the prediction model is analyzed based on the predicted surge characteristic parameters and the actual surge characteristic parameters, network parameters of the prediction model under the test condition are adjusted based on a model optimization strategy of surge influence parameter sensitivity, the prediction model is continuously optimized, and the prediction model conforming to actual engineering is obtained. The defect of lack of engineering data is overcome, the prediction model is more fit with engineering practice, and the model prediction effect is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of surge disaster prediction, and in particular relates to a digital modeling method for landslide surge disaster prediction. Background Art

[0002] At present, the simulation and prediction methods for surge disasters caused by landslides mainly include numerical simulation methods based on physical processes and data-driven machine learning methods.

[0003] The numerical simulation method based on physical processes is to discretize the landslide and water body into a finite number of units, and simulate the movement of the landslide and the fluctuation of the water body by solving mechanical equations. For example, a two-dimensional or three-dimensional landslide-surge model is established in the relevant simulation software. Taking into account the material properties, initial conditions and boundary conditions of the landslide, the sliding process of the landslide under the action of gravity and the impact on the water body are simulated, thereby obtaining the generation and propagation process of the surge. This method takes into account the influence of multiple physical factors and can more accurately describe the complex mechanical behavior of the landslide and water body, but the calculation cost is high and the initial conditions and parameter settings of the model are strictly required.

[0004] The data-driven machine learning method is to build a multi-layer neural network and use a large amount of landslide surge case data for training. It learns the mapping relationship between input features such as the volume, slope, and speed of the landslide body and output results such as surge height and propagation speed. After training, surge disaster prediction is performed on new landslide situations. This method has strong nonlinear mapping capabilities and can handle complex input-output relationships, but it is highly dependent on data and requires a large amount of experimental or engineering data for training. The physical meaning of the model is unclear and its interpretability is poor.

[0005] For data-driven machine learning methods, if only experimental data is used to train the model, there are usually certain differences between the experimental device and the actual project. For example, in terms of still water depth, landslide size, and landslide impact velocity, the actual project is usually larger than the experimental device. Then the actual influencing parameters exceed the training range of the landslide surge disaster prediction model, which may cause the model prediction results to be inconsistent with the actual project; if only engineering data is used to train the model, but engineering data is often less, it may lead to poor model prediction accuracy and poor model prediction effect. Summary of the Invention

[0006] In order to solve the above technical problems, the present invention provides a digital modeling method for landslide surge disaster prediction.

[0007] To achieve the above-mentioned purpose of the invention, the technical solution adopted by the present invention is: a digital modeling method for predicting landslide surge disasters, which uses a wave pool model test device to obtain landslide surge test data, analyzes the sensitivity of surge influencing parameters to surge characteristic parameters, and uses the landslide surge test data to train a landslide surge disaster prediction model to obtain a landslide surge disaster prediction model under test conditions; substitutes the actual surge influencing parameters in the engineering case into the landslide surge disaster prediction model under test conditions to obtain the predicted surge characteristic parameters of the engineering case; analyzes the prediction accuracy of the landslide surge disaster prediction model based on the predicted surge characteristic parameters and the actual surge characteristic parameters, adjusts the network parameters of the landslide surge disaster prediction model under test conditions based on the model optimization strategy of the surge influencing parameters sensitivity, continuously optimizes the landslide surge disaster prediction model, and finally obtains a landslide surge disaster prediction model that conforms to actual engineering.

[0008] Preferably, the method comprises the following steps:

[0009] S1. Determine surge influencing parameters and surge characteristic parameters, build a wave pool model test device, conduct experiments using the wave pool model test device, study surge characteristic parameters under different surge influencing parameters, and analyze the sensitivity of surge influencing parameters to surge characteristics;

[0010] S2. Using surge impact parameters as model input and surge characteristic parameters as model output, a landslide surge disaster prediction model is constructed using a neural network. The simulated landslide surge test data is used as a training set to train the landslide surge disaster prediction model, thereby obtaining a landslide surge disaster prediction model under test conditions.

[0011] S3. Collect engineering cases of landslide surges, substitute actual surge impact parameters of the engineering cases into the landslide surge disaster prediction model under experimental conditions, and obtain predicted surge characteristic parameters of the engineering cases;

[0012] S4. Draw a visual chart based on the predicted surge characteristic parameters and the actual surge characteristic parameters, determine the abnormal points predicted by the landslide surge disaster prediction model, analyze the surge impact parameters based on the abnormal points, and adjust the network parameters of the landslide surge disaster prediction model under experimental conditions based on the model optimization strategy of the surge impact parameter sensitivity. Continuously optimize the landslide surge disaster prediction model and finally obtain a landslide surge disaster prediction model that conforms to the actual project.

[0013] Preferably, the landslide surge disaster prediction model includes an input layer, a hidden layer, and an output layer, wherein the input layer includes 6 input neurons, the output layer includes 3 output neurons, and two hidden layers are set, the first hidden layer is set with 12 neurons, and the second hidden layer is set with 8 neurons; the surge impact parameters are used as model input, and the surge characteristic parameters are used as model output, the surge impact parameters include still water depth, landslide inclination, landslide size, landslide density, landslide impact velocity, and landslide incident angle, and the surge characteristic parameters include maximum standing wave height, propagation velocity, and maximum standing wave height propagation distance.

[0014] Preferably, step S1 includes the following process:

[0015] S11. Build a wave pool model test device;

[0016] S12. Determine surge impact parameters and surge characteristic parameters. The surge impact parameters include still water depth, landslide inclination, landslide size, landslide density, landslide impact velocity, and landslide incident angle. The surge characteristic parameters include maximum standing wave height, propagation velocity, and maximum standing wave height propagation distance.

[0017] S13. Each type of surge impact parameter is set to three levels: small, medium, and large. Combinations of surge impact parameters at each level are designed. Simulation tests are conducted on these combinations using a wave tank model test apparatus to obtain a data set P of experimental surge impact parameters and experimental surge characteristic parameters.

[0018] S14. Determine the contribution rate of the interaction between surge influencing parameters to the change of surge characteristic parameter results using the variance analysis method;

[0019] S15. Construct a sensitivity matrix of surge influence parameters and surge characteristic parameters, and for each surge characteristic parameter, sort the six surge influence parameters from large to small according to their contribution rates.

[0020] Preferably, in step S1, a multi-factor combination method is used to analyze the sensitivity of surge influencing parameters to surge characteristics. On the basis of full factorial design, an orthogonal design method is used to select n suitable and representative combinations for testing. The selected n combinations ensure that each level of each surge influencing parameter appears the same number of times in the combination and can balance the interactions between the surge influencing parameters.

[0021] Preferably, step S4 includes the following process:

[0022] S41, drawing a visualization chart based on the predicted surge characteristic parameters and the actual surge characteristic parameters, and determining the abnormal points predicted by the landslide surge disaster prediction model;

[0023] S42, traverse all abnormal points, and screen out extreme abnormal points based on whether the actual surge impact parameters exceed the model training range;

[0024] S43. According to whether the abnormal point is an extreme abnormal point, different optimization strategies are adopted to adjust the model network parameters to obtain the adjusted landslide surge disaster prediction model.

[0025] S44. Re-input the actual surge impact parameters in the engineering case into the adjusted landslide surge disaster prediction model to generate new predicted surge characteristic parameters, and compare the error changes of the original abnormal points until the requirements are met; at the same time, adjust the sensitivity of the surge impact parameters, and update the surge impact parameter sensitivity ranking and sensitivity matrix.

[0026] S45. Substitute the actual surge impact parameters in the engineering case into the adjusted landslide surge disaster prediction model to obtain new predicted surge characteristic parameters of the engineering case. Draw a visualization chart based on the new predicted surge characteristic parameters and the actual surge characteristic parameters. If abnormal points still appear, repeat steps S42-S44.

[0027] Preferably, in step S43, if the outlier is an extreme outlier and the extreme surge impact parameter is a high-sensitivity impact parameter corresponding to the corresponding surge characteristic parameter, then in the input layer, an extreme mapping neuron is added to the surge impact parameter, and the feature extraction of the extreme value is enhanced by a piecewise function; and / or, the weight initialization method of the surge impact parameter in the hidden layer is adjusted.

[0028] Preferably, in step S43, if the outlier is not an extreme outlier, for the highly sensitive influencing parameter of the outlier, if the model prediction deviation is inconsistent with the actual change trend of the parameter, a nonlinear transformation layer of the highly sensitive influencing parameter is added to the hidden layer to strengthen the nonlinear mapping; and / or, the input features of the highly sensitive influencing parameter are standardized.

[0029] Preferably, in step S43, if the outlier is not an extreme outlier and is caused by a low-sensitivity influencing parameter, L1 regularization is applied to the weight corresponding to the low-sensitivity influencing parameter to force the network to reduce its dependence on it; and / or, a perturbation sample of the low-sensitivity influencing parameter is added to the training data.

[0030] The present invention has the following beneficial effects:

[0031] A large amount of landslide surge test data is obtained using a wave pool model test device, and the sensitivity of various surge influencing parameters to various surge characteristic parameters is analyzed using the landslide surge test data; the experimental surge influencing parameters are used as model input, and the experimental surge characteristic influencing parameters are used as model output. The landslide surge test data is used to train a landslide surge disaster prediction model to obtain a landslide surge disaster prediction model under experimental conditions; a certain amount of engineering case data is obtained, and the actual surge influence parameters in the engineering case are substituted into the landslide surge disaster prediction model under experimental conditions to obtain the predicted surge characteristic parameters of the engineering case; the prediction accuracy of the landslide surge disaster prediction model is analyzed based on the predicted surge characteristic parameters and the actual surge characteristic parameters, and the network parameters of the landslide surge disaster prediction model under experimental conditions are adjusted based on the model optimization strategy of the surge influence parameters sensitivity, and the landslide surge disaster prediction model is continuously optimized to finally obtain a landslide surge disaster prediction model that conforms to actual engineering. The landslide surge disaster prediction model is trained using landslide surge test data, and then the prediction effect of the landslide surge disaster prediction model is verified and optimized using actual engineering data. This solves the problem of insufficient engineering data, makes the landslide surge disaster prediction model more in line with engineering practice, and improves the model prediction effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flowchart of the digital modeling method of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the accompanying drawings to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.

[0034] It should be noted that the surge influencing parameters studied in the digital modeling method for predicting landslide surge disasters of the present invention include still water depth, landslide inclination, landslide size, landslide density, landslide impact velocity, landslide incidence angle, etc., among which the landslide impact velocity is calculated from the landslide drop height and the time of falling into the water; the surge characteristic parameters studied include maximum standing wave height, propagation velocity, maximum standing wave height propagation distance (specifically referring to the distance between the location where the maximum standing wave height occurs and the landslide location), etc.

[0035] The invention concept of the digital modeling method disclosed in the present invention is as follows: a large amount of landslide surge test data is obtained using a wave pool model test device, and the sensitivity of various surge influencing parameters to various surge characteristic parameters is analyzed using the landslide surge test data; the test surge influencing parameters are used as model inputs, and the test surge characteristic influencing parameters are used as model outputs. The landslide surge test data is used to train a landslide surge disaster prediction model, thereby obtaining a landslide surge disaster prediction model under test conditions;

[0036] A certain amount of engineering case data is obtained, and the actual surge impact parameters in the engineering cases are substituted into the landslide surge disaster prediction model under experimental conditions to obtain the predicted surge characteristic parameters of the engineering cases; the prediction accuracy of the landslide surge disaster prediction model is analyzed based on the predicted surge characteristic parameters and the actual surge characteristic parameters, and the network parameters of the landslide surge disaster prediction model under experimental conditions are adjusted based on the model optimization strategy based on the surge impact parameter sensitivity, and the landslide surge disaster prediction model is continuously optimized to finally obtain a landslide surge disaster prediction model that conforms to actual engineering projects.

[0037] The landslide surge disaster prediction model is trained using landslide surge test data, and then the prediction effect of the landslide surge disaster prediction model is verified and optimized using actual engineering data. This solves the problem of insufficient engineering data, makes the landslide surge disaster prediction model more in line with engineering practice, and improves the model prediction effect.

[0038] like Figure 1 As shown, the present invention discloses a digital modeling method for landslide surge disaster prediction, which includes the following specific steps:

[0039] S1. Determine the surge influencing parameters and surge characteristic parameters. The surge influencing parameters include still water depth, landslide inclination, landslide size, landslide density, landslide impact velocity, and landslide incidence angle. The surge characteristic parameters include maximum standing wave height, propagation velocity, and maximum standing wave height propagation distance. Build a wave pool model test device, conduct experiments using the wave pool model test device, and construct a data set P of experimental surge influencing parameters and experimental surge characteristic parameters. Study the surge characteristic parameters under different surge influencing parameters, and analyze the sensitivity of the surge influencing parameters to the surge characteristic parameters.

[0040] S2. Using surge impact parameters as model input and surge characteristic parameters as model output, a landslide surge disaster prediction model is constructed using a neural network. The simulated landslide surge test data is used as a training set to train the landslide surge disaster prediction model, and a landslide surge disaster prediction model under experimental conditions is obtained.

[0041] S3. Collect as many engineering cases of landslide surge as possible and construct a data set P of actual surge impact parameters and actual surge characteristic parameters. zThe actual surge impact parameters of each engineering case are substituted into the landslide surge disaster prediction model under experimental conditions to obtain the predicted surge characteristic parameters of the engineering case, and a data set P of actual surge impact parameters-predicted surge characteristic parameters is constructed. c .

[0042] Specifically, P z =(p z1 ,p z2 …p zi …p zm ), p zi =(X zi ,Y zi ), i=1,2…m, m is the number of engineering cases; X zi =(x z1 ,x z2 ,x z3 ,x z4 ,x z5 ,x z6 ), X zi is the ith actual surge impact parameter group, x z1 is the actual still water depth, x z2 is the actual landslide inclination, x z3 is the actual landslide size, x z4 is the actual landslide density, x z5 is the actual landslide impact velocity, x z6 is the actual landslide incident angle; Y zi =(y z1 ,y z2 ,y z3 ), Y zi is the i-th predicted surge characteristic parameter group, y z1 To predict the maximum standing wave height, y z2 To predict the propagation speed, y z3 To predict the maximum standing wave height propagation distance.

[0043] Specifically, P c =(p c1 ,p c2 …p ci …p cm ), p ci =(X zi ,Y ci ), i=1,2…m, m is the number of engineering cases; X zi =(x z1 ,x z2 ,x z3 ,x z4 ,x z5 ,x z6 ), X ziis the ith actual surge impact parameter group, x z1 is the actual still water depth, x z2 is the actual landslide inclination, x z3 is the actual landslide size, x z4 is the actual landslide density, x z5 is the actual landslide impact velocity, x z6 is the actual landslide incident angle; Y ci =(y c1 ,y c2 ,y c3 ), Y ci is the i-th predicted surge characteristic parameter group, y c1 To predict the maximum standing wave height, y c2 To predict the propagation speed, y c3 To predict the maximum standing wave height propagation distance.

[0044] S4. Draw a visualization chart based on the predicted and actual surge characteristic parameters to identify outliers in the landslide surge disaster prediction model. Analyze the surge influencing parameters based on the outliers. Adjust the network parameters of the landslide surge disaster prediction model under experimental conditions using a model optimization strategy based on the sensitivity of the surge influencing parameters. Continuously optimize the landslide surge disaster prediction model until the error between the predicted and actual surge characteristic parameters falls below a set error threshold, ultimately obtaining a landslide surge disaster prediction model that meets the requirements of the actual project. Outliers refer to situations where the error between the predicted and actual surge characteristic parameters is greater than or equal to a set error threshold. The set error threshold can be set to 5-10%.

[0045] In a further embodiment, step S1 includes the following steps:

[0046] S11. Build a wave pool model test device: Use water from the Three Gorges Reservoir as the water for the wave pool model test device, and use the slopes surrounding the Three Gorges Reservoir as the slopes for the wave pool model test device. Set up multiple high-speed cameras beside the pool to simultaneously film the waves caused by the slope sliding. Add time codes or optical synchronization signals to the captured videos to ensure time alignment of multi-camera data. The high-speed cameras perform multi-angle synchronous shooting at a sampling frequency of ≥1000 frames per second, which can accurately record the parameters affecting the surge and the parameters of the surge characteristics. The shutter speed of the high-speed cameras is ≤1 / 10000s to avoid motion blur. Adjust the aperture and ISO according to the light source intensity to ensure a clear image.

[0047] S12. Determine surge impact parameters and surge characteristic parameters. The surge impact parameters include still water depth, landslide inclination, landslide size, landslide density, landslide impact velocity, and landslide incident angle. The surge characteristic parameters include maximum standing wave height, propagation velocity, and maximum standing wave height propagation distance.

[0048] S13. For each surge impact parameter, set three levels: small (low), medium, and large (high). Design combinations of the surge impact parameters at each level. Conduct simulation tests on these combinations using a wave tank model test apparatus to obtain a dataset P of experimental surge impact parameters and experimental surge characteristic parameters. It should be noted that each level corresponds to a specific range of values, and the specific range is determined based on previous research experience.

[0049] Specifically, P=(p1, p2…p i …p n ), p i =(X i ,Y i ), i = 1, 2…n, n is the number of experimental landslide surge data sets; X i =(x1,x2,x3,x4,x5,x6), X i is the i-th test surge impact parameter group, x1 is the test still water depth, x2 is the test landslide inclination, x3 is the test landslide size, x4 is the test landslide density, x5 is the test landslide impact velocity, and x6 is the test landslide incident angle; Y i =(y1,y2,y3), Y i is the ith test surge characteristic parameter group, y1 is the test maximum standing wave height, y2 is the test propagation speed, and y3 is the test maximum standing wave height propagation distance.

[0050] S14. Use the variance analysis method to determine the contribution rate of the surge influencing parameters and the interactions between them to the changes in the surge characteristic parameters. By calculating statistical quantities such as the variance ratio (F value), determine whether each surge influencing parameter and the interaction has a significant impact on the changes in the surge characteristic parameters. For example, if the F value of a surge influencing parameter is large and the corresponding P value is less than the set significance level (such as 0.05), it means that the surge influencing parameter has a significant impact on the surge characteristic parameters. Its contribution rate can be obtained by calculating the ratio of the sum of squares of the surge influencing parameter to the total sum of squares.

[0051] S15. Based on the analysis results of step S14, a sensitivity matrix of surge impact parameters-surge characteristic parameters is constructed, and for each surge characteristic parameter, the six surge impact parameters are sorted from large to small according to their contribution rate to obtain a sensitivity data set R.

[0052] Specifically, R=(R1, R2, R3), R1 is the surge impact parameter sensitivity set of the maximum standing wave height, R2 is the surge impact parameter sensitivity set of the propagation speed, and R3 is the surge impact parameter sensitivity set of the maximum standing wave height propagation distance.

[0053] The surge impact parameters of each type of surge characteristic parameter are divided into high-sensitivity impact parameters and low-sensitivity impact parameters according to the sensitivity. Among them, the high and low sensitivity impact parameters can be specifically set according to needs. For example, the surge impact parameters with a sensitivity greater than 0.5 are defined as high-sensitivity impact parameters, and the surge impact parameters with a sensitivity less than or equal to 0.5 are defined as low-sensitivity impact parameters; or the number of surge impact parameters in this application is 6, and the top three in the R1, R2, and R3 sets are defined as high-sensitivity impact parameters, and the last three are defined as low-sensitivity impact parameters. Of course, other definition methods are also possible.

[0054] In a preferred embodiment, since the interaction between surge influencing parameters jointly affects the changes in surge characteristic parameters, a multi-factor combination method is used to analyze the sensitivity of surge influencing parameters to surge characteristics. Specifically, considering all combinations of all levels of all surge influencing parameters, there are 6 surge influencing parameters, each of which has 3 levels, so the full factor design has 3 6 = 729 combinations. If all of these combinations are tested, the workload will be enormous. In a preferred embodiment, based on the full factorial design, an orthogonal design method is used to select n appropriate and representative combinations for testing. The n combinations selected ensure that each level of each surge impact parameter appears the same number of times in the combination, and the interactions between the surge impact parameters can be well balanced.

[0055] In a further implementation method, in step S2, the experimental landslide surge disaster prediction model includes an input layer, a hidden layer, and an output layer, wherein the input layer includes 6 input neurons, the output layer includes 3 output neurons, and two hidden layers are set, the first hidden layer is set with 12 neurons, and the second hidden layer is set with 8 neurons.

[0056] In a further embodiment, step S4 includes the following steps:

[0057] S41. Draw a visualization chart based on the predicted surge characteristic parameters and the actual surge characteristic parameters, determine the abnormal points predicted by the landslide surge disaster prediction model, determine the highly sensitive influencing parameters of the surge characteristic parameters corresponding to the abnormal points, and give priority to analyzing and adjusting the model network parameters of these highly sensitive influencing parameters. Draw different charts based on different surge characteristic parameters, such as the maximum standing wave height chart, the propagation speed chart, and the maximum standing wave height propagation distance chart. The visualization chart can be a line chart or a bar chart that draws the actual value and the predicted value for each surge characteristic parameter; with the actual value as the horizontal axis and the predicted value as the vertical axis, each point represents an engineering case, and the deviation value is calculated for each point, and the deviation size is marked with a line segment perpendicular to the horizontal axis; of course, other forms of charts are also possible.

[0058] S42. Traverse all outliers and screen out extreme outliers based on whether the actual surge impact parameters exceed the model training range. An extreme outlier refers to a point where, when the actual surge impact parameters of the engineering case corresponding to the outlier are compared with the distribution of the test surge impact parameters in the test surge impact parameter-test surge characteristic parameter data set P, the actual surge impact parameters are greater than the maximum value of the test surge impact parameters, or the actual surge impact parameters are less than the minimum value of the test surge impact parameters. In other words, determining whether an outlier is an extreme outlier is actually determining whether the actual surge impact parameters exceed the training range of the test surge impact parameters used to train the landslide surge disaster prediction model. Surge impact parameters that exceed the training range are defined as extreme surge impact parameters.

[0059] S43. The strategy for adjusting the model network parameters based on sensitivity is as follows:

[0060] If the outlier is an extreme outlier, and the extreme surge impact parameter is a highly sensitive impact parameter corresponding to the corresponding surge characteristic parameter, then in the input layer, an extreme mapping neuron is added to the surge impact parameter, and the feature extraction of extreme values ​​is enhanced through piecewise functions to avoid simple truncation or linear mapping of extreme values, thereby improving the model's nonlinear response to out-of-range inputs; and / or, the weight initialization method of the surge impact parameter in the hidden layer is adjusted, such as using Xavier initialization instead of default initialization to avoid gradient disappearance. The verification indicator mainly depends on the reduction in prediction error under extreme working conditions. For example, if the relative error is reduced from 50% to within 15%, the optimization is considered complete.

[0061] If the outlier is not an extreme outlier, for the highly sensitive parameter of the outlier, if the model prediction deviation does not match the actual change trend of the parameter (for example, the predicted value decreases when the actual parameter increases), then a nonlinear transformation layer for the highly sensitive parameter is added to the hidden layer, such as first passing it through the ReLU activation function and then crossing it with other parameters to strengthen the nonlinear mapping; and / or, the input features of the highly sensitive parameter are standardized, such as Z-score standardization, to avoid imbalance in weight updates due to a large value range;

[0062] If the outlier is not an extreme outlier and is caused by a low-sensitivity parameter, the cause of the anomaly may be model overfitting. Apply L1 regularization to the weight corresponding to the low-sensitivity parameter to force the network to reduce its dependence on it; and / or, add perturbation samples of the low-sensitivity parameter to the training data to increase model robustness.

[0063] S44. Re-input the actual surge impact parameters in the engineering case into the adjusted landslide surge disaster prediction model to generate new predicted surge characteristic parameters, and compare the error changes of the original abnormal points until the requirements are met; at the same time, adjust the sensitivity of the surge impact parameters, and update the surge impact parameter sensitivity ranking and sensitivity matrix.

[0064] S45. Substitute the actual surge impact parameters in the engineering case into the adjusted landslide surge disaster prediction model to obtain new predicted surge characteristic parameters of the engineering case. Draw a visualization chart based on the new predicted surge characteristic parameters and the actual surge characteristic parameters. If abnormal points still appear, repeat steps S42-S44.

[0065] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various deformations, modifications, and substitutions made to the technical solutions of the present invention by ordinary technicians in this field should fall within the scope of protection determined by the claims of the present invention.

Claims

1. A digital modeling method for landslide surge disaster prediction, characterized by: Landslide surge test data are obtained using a wave pool model test device, and the sensitivity of surge influencing parameters to surge characteristic parameters is analyzed. The landslide surge disaster prediction model is trained using the landslide surge test data to obtain a landslide surge disaster prediction model under test conditions. The actual surge influencing parameters in the engineering case are substituted into the landslide surge disaster prediction model under test conditions to obtain the predicted surge characteristic parameters of the engineering case. The prediction accuracy of the landslide surge disaster prediction model is analyzed based on the predicted surge characteristic parameters and the actual surge characteristic parameters. The network parameters of the landslide surge disaster prediction model under test conditions are adjusted based on the model optimization strategy based on the sensitivity of surge influencing parameters. The landslide surge disaster prediction model is continuously optimized to finally obtain a landslide surge disaster prediction model that conforms to actual engineering.

2. The digital modeling method for landslide surge disaster prediction according to claim 1, characterized in that: The steps include: S1. Determine surge influencing parameters and surge characteristic parameters, build a wave pool model test device, conduct experiments using the wave pool model test device, study surge characteristic parameters under different surge influencing parameters, and analyze the sensitivity of surge influencing parameters to surge characteristics; S2. Using surge impact parameters as model input and surge characteristic parameters as model output, a landslide surge disaster prediction model is constructed using a neural network. The simulated landslide surge test data is used as a training set to train the landslide surge disaster prediction model, thereby obtaining a landslide surge disaster prediction model under test conditions. S3. Collect engineering cases of landslide surges, substitute actual surge impact parameters of the engineering cases into the landslide surge disaster prediction model under experimental conditions, and obtain predicted surge characteristic parameters of the engineering cases; S4. Draw a visual chart based on the predicted surge characteristic parameters and the actual surge characteristic parameters, determine the abnormal points predicted by the landslide surge disaster prediction model, analyze the surge impact parameters based on the abnormal points, and adjust the network parameters of the landslide surge disaster prediction model under experimental conditions based on the model optimization strategy of the surge impact parameter sensitivity. Continuously optimize the landslide surge disaster prediction model and finally obtain a landslide surge disaster prediction model that conforms to the actual project.

3. The digital modeling method for landslide surge disaster prediction according to claim 2, characterized in that: The landslide surge disaster prediction model includes an input layer, a hidden layer, and an output layer, wherein the input layer includes 6 input neurons, the output layer includes 3 output neurons, and two hidden layers are set, the first hidden layer is set with 12 neurons, and the second hidden layer is set with 8 neurons; the surge impact parameters are used as model inputs, and the surge characteristic parameters are used as model outputs, the surge impact parameters include still water depth, landslide inclination, landslide size, landslide density, landslide impact velocity, and landslide incident angle, and the surge characteristic parameters include maximum standing wave height, propagation velocity, and maximum standing wave height propagation distance.

4. The digital modeling method for landslide surge disaster prediction according to claim 3 is characterized by: The step S1 includes the following process: S11. Build a wave pool model test device; S12. Determine surge impact parameters and surge characteristic parameters. The surge impact parameters include still water depth, landslide inclination, landslide size, landslide density, landslide impact velocity, and landslide incident angle. The surge characteristic parameters include maximum standing wave height, propagation velocity, and maximum standing wave height propagation distance. S13. Each type of surge impact parameter is set to three levels: small, medium, and large. Combinations of surge impact parameters at each level are designed. Simulation tests are conducted on these combinations using a wave tank model test apparatus to obtain a data set P of experimental surge impact parameters and experimental surge characteristic parameters. S14. Determine the contribution rate of the interaction between surge influencing parameters to the change of surge characteristic parameter results using the variance analysis method; S15. Construct a sensitivity matrix of surge influence parameters and surge characteristic parameters, and for each surge characteristic parameter, sort the six surge influence parameters from large to small according to their contribution rates.

5. The digital modeling method for landslide surge disaster prediction according to claim 4, characterized in that: In step S1, a multi-factor combination method is used to analyze the sensitivity of surge influencing parameters to surge characteristics. On the basis of full factorial design, an orthogonal design method is used to select n suitable and representative combinations for testing. The n combinations are selected so that each level of each surge influencing parameter appears the same number of times in the combination and the interactions between the surge influencing parameters can be balanced.

6. The digital modeling method for landslide surge disaster prediction according to claim 3, characterized in that: The step S4 includes the following process: S41, drawing a visualization chart based on the predicted surge characteristic parameters and the actual surge characteristic parameters, and determining the abnormal points predicted by the landslide surge disaster prediction model; S42, traverse all abnormal points, and screen out extreme abnormal points based on whether the actual surge impact parameters exceed the model training range; S43. According to whether the abnormal point is an extreme abnormal point, different optimization strategies are adopted to adjust the model network parameters to obtain the adjusted landslide surge disaster prediction model. S44. Re-input the actual surge impact parameters in the engineering case into the adjusted landslide surge disaster prediction model to generate new predicted surge characteristic parameters, and compare the error changes of the original abnormal points until the requirements are met; at the same time, adjust the sensitivity of the surge impact parameters, and update the surge impact parameter sensitivity ranking and sensitivity matrix. S45. Substitute the actual surge impact parameters in the engineering case into the adjusted landslide surge disaster prediction model to obtain new predicted surge characteristic parameters of the engineering case. Draw a visualization chart based on the new predicted surge characteristic parameters and the actual surge characteristic parameters. If abnormal points still appear, repeat steps S42-S44.

7. The digital modeling method for landslide surge disaster prediction according to claim 6, characterized in that: In step S43, if the outlier is an extreme outlier, and the extreme surge impact parameter is a high-sensitivity impact parameter corresponding to the corresponding surge characteristic parameter, then in the input layer, an extreme mapping neuron is added for the surge impact parameter, and feature extraction of extreme values ​​is enhanced by a piecewise function; And / or, adjust the weight initialization method of the surge impact parameter in the hidden layer.

8. The digital modeling method for landslide surge disaster prediction according to claim 6, characterized in that: In step S43, if the outlier is not an extreme outlier, for the highly sensitive influencing parameter of the outlier, if the model prediction deviation does not match the actual change trend of the parameter, a nonlinear transformation layer of the highly sensitive influencing parameter is added to the hidden layer to strengthen the nonlinear mapping; and / or, the input features of the highly sensitive influencing parameter are standardized.

9. The digital modeling method for landslide surge disaster prediction according to claim 6, characterized in that: In step S43, if the outlier is not an extreme outlier and is caused by a low-sensitivity influencing parameter, L1 regularization is applied to the weight corresponding to the low-sensitivity influencing parameter to force the network to reduce its dependence on it; and / or, a perturbation sample of the low-sensitivity influencing parameter is added to the training data.

Images