Landslide displacement prediction method for reservoir bank inspection

Abstract

The invention discloses a landslide displacement prediction method for reservoir bank inspection, and relates to the technical field of disaster prevention, and the method comprises the following steps: collecting original data of geological conditions, meteorological monitoring and bank landslide mass monitoring of reservoir bank inspection, and carrying out the preprocessing of the original data; extracting features related to the landslide deformation mode from the preprocessed original data, and integrating the features to obtain a feature set; and training the extracted features by adopting a support vector machine algorithm based on the related data of the feature set. According to the landslide displacement prediction method for reservoir bank inspection, by combining geological conditions, meteorological monitoring and reservoir bank landslide mass monitoring data, landslide deformation inducements can be analyzed more comprehensively, the influence of rainfall and earthquakes on landslides is considered, the possibility of landslide occurrence can be predicted more accurately, and the prediction accuracy is improved. And by setting a risk assessment threshold and assessing different landslide risk levels, early warning can be performed before the occurrence of the landslide, so that early recognition of the landslide risk is realized.

Description

Technical Field

[0001] The invention relates to the technical field of disaster prevention, and in particular to a landslide displacement prediction method for reservoir bank inspection. Background Art

[0002] Reservoir bank landslide is a common natural disaster that causes serious harm to human society and the natural environment. Landslides not only cause damage to reservoir dams, but may also affect the safety of life and property of surrounding residents. Landslide displacement prediction is a key link in the prevention and control of landslide disasters. By predicting the displacement of landslides, the possibility and degree of harm of landslides can be evaluated. In addition, landslide displacement prediction can also help relevant departments to issue early warnings and reduce the losses caused by landslide disasters. In recent years, with the rapid development of artificial intelligence technologies such as machine learning and deep learning, landslide displacement prediction methods have gradually developed towards the direction of intelligence and automation.

[0003] Due to the complexity of internal geological conditions and the interference of external meteorological factors, the deformation mode of reservoir bank landslide is difficult to determine accurately, which makes the prediction accuracy low. Therefore, how to combine the multi-source factors of geology and meteorology, comprehensively analyze the deformation causes of landslide, and predict the deformation mode of reservoir bank landslide is the problem we need to solve. To this end, a landslide displacement prediction method for reservoir bank inspection is proposed. Summary of the invention

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: A landslide displacement prediction method for reservoir bank inspection includes the following steps:

[0005] Step 1: Collect the original data of geological conditions, meteorological monitoring and landslide monitoring of the reservoir bank inspection, and pre-process the original data;

[0006] Step 2: extracting features related to the landslide deformation mode from the preprocessed raw data and integrating them to obtain a feature set;

[0007] Step 3: Based on the relevant data of the feature set, the support vector machine algorithm is used to train the extracted features, establish a landslide deformation pattern recognition model, and identify the landslide state of the reservoir bank;

[0008] Step 4: Combine geological, meteorological and reservoir bank landslide monitoring data to conduct a comprehensive analysis of the causes of landslide deformation, analyze the interaction and impact between different factors, and build a landslide risk assessment model to assess the risk of reservoir bank landslides;

[0009] Step 5: Based on the assessed risk of the reservoir bank landslide body, different risk levels are preset to classify the landslide risk of the reservoir bank landslide body;

[0010] Step 6: According to the landslide displacement prediction results and actual conditions, match the corresponding early warning mechanism for each risk level and formulate corresponding response measures.

[0011] Preferably, the collection and preprocessing process of the raw data includes:

[0012] The original data of geological conditions for reservoir bank inspections are obtained through geological survey reports, geological maps, and geological profiles, including information on the geological structure, rock and soil types, stratum distribution, faults, and folds of the reservoir bank;

[0013] The original meteorological monitoring data of reservoir bank inspections are obtained through meteorological monitoring stations, meteorological satellites, and meteorological radar equipment, including rainfall, temperature, wind speed, and wind direction meteorological information;

[0014] For the original data of the reservoir bank landslide monitoring, crack meters, GNSS monitors, and synthetic aperture radars are used to obtain data, including the displacement, deformation rate, crack width, and depth parameters of the landslide body;

[0015] Organize the collected data according to time series and monitoring points to ensure the integrity and consistency of the data, and classify and label data from different sources and types for subsequent processing;

[0016] Preprocess the classified and labeled data, including data cleaning and data conversion steps;

[0017] The pre-processed geological condition data, meteorological monitoring data and reservoir bank landslide monitoring data are integrated to form a complete data set for subsequent analysis and processing.

[0018] Preferably, the process of acquiring the feature set includes:

[0019] Extract geological characteristics from geological condition data, including the distribution of rock and soil types, the inclination angle of the strata, and the density of faults and folds; extract meteorological characteristics from meteorological monitoring data, including the total amount and intensity of rainfall, the number of consecutive dry or wet days, wind speed and direction, and the frequency of extreme meteorological events; extract landslide characteristics from landslide monitoring data, including displacement change rate, crack width and depth, crack expansion speed, and groundwater level fluctuation;

[0020] The extracted features are further processed, the rate of change of the features is calculated, and the extracted features are normalized to eliminate the impact of different dimensions and value ranges of different features on the model, so that they are on the same scale for easy comparison and analysis;

[0021] The extracted relevant feature data are integrated to obtain a feature set, each of which is associated with the landslide deformation pattern, ensuring that the features in the feature set cover different aspects and conditions that may cause landslides, which serves as the basis for subsequent landslide prediction and analysis.

[0022] Preferably, the process of identifying the reservoir bank landslide state includes:

[0023] The feature set data is divided into a training set and a test set. The training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model. This ensures that the data in the training set and the test set are representative and can reflect the various modes of landslide deformation. The RBF kernel function of the support vector machine (SVM) algorithm is selected to process the nonlinear relationship between features and identify the landslide deformation mode.

[0024] The training set data is used to train the SVM model, and the landslide deformation pattern recognition model is constructed to obtain the classifier. The trained landslide deformation pattern recognition model is tested using the test set data to evaluate the recognition accuracy and generalization ability of the model. The landslide feature data to be identified is input into the trained landslide deformation pattern recognition model to obtain the predicted categories of landslide deformation, which are stable, slow deformation, and accelerated deformation.

[0025] Compare the overlap between the predicted landslide area and the actual landslide area, calculate the landslide identification matching index, and measure the matching degree between the model prediction results and the actual landslide state. The higher the matching index value, the higher the matching degree between the model prediction results and the actual landslide state.

[0026] According to the landslide identification matching index, the recognition matching degree of the reservoir bank landslide state is analyzed, and a matching threshold is set. The landslide identification matching index is compared with the set matching threshold to determine whether the landslide state is accurately identified.

[0027] Preferably, the calculation process of the landslide identification matching index includes:

[0028] Based on the historical data in the original data, the spatial data of the actual landslide area is obtained, the spatial data of the predicted landslide area predicted by the landslide deformation pattern recognition model is extracted, and the coordinate transformation is performed so that the spatial data sets of the actual landslide area and the predicted landslide area are in the same coordinate reference system to ensure the consistency of the comparison;

[0029] Use geographic information system software to calculate the area of ​​the actual landslide area and the predicted landslide area, and calculate the intersection area of ​​the actual landslide area and the predicted landslide area;

[0030] Based on the calculated actual landslide area and the predicted landslide area, the absolute value of the area difference between the actual landslide area and the predicted landslide area is calculated, and the landslide identification matching index is obtained through comprehensive analysis;

[0031] The calculated landslide identification matching index value is analyzed to determine the accuracy of the model prediction. If the landslide identification matching index value is close to 1, it means that the model prediction result matches the actual landslide state well. If the landslide identification matching index value is low, it means that the model prediction has a large deviation.

[0032] The calculation expression of the landslide identification matching index is:

[0033]

[0034] Where I is the landslide identification matching index, |A| is the actual landslide area, |B| is the predicted landslide area, |A∩B| is the intersection area of ​​the actual landslide area and the predicted landslide area, i.e. the overlapping part, λ is an adjustment parameter used to balance the influence of the area difference between the actual landslide area and the predicted landslide area, λ≥0, ||A|-|B|| is the absolute value of the area difference between the actual landslide area and the predicted landslide area, which is used to penalize the situation where the area difference between the predicted area and the actual area is large.

[0035] Preferably, the risk assessment process of the reservoir bank landslide body includes:

[0036] Integrate geological condition data, meteorological monitoring data and reservoir bank landslide monitoring data to analyze the causes of landslide deformation, which are rainfall and earthquake. Rainfall is one of the main factors causing landslides. Rainfall increases soil moisture content and reduces soil shear strength, thereby increasing the risk of landslides. Analyze the relationship between rainfall and landslides, especially the impact of heavy rain and continuous rainfall on landslides. The vibrations caused by earthquakes can cause landslides to lose balance, thereby triggering landslides. Analyze the impact of factors such as earthquake magnitude, focal depth, and seismic wave type on landslides;

[0037] Extract characteristic data related to landslide displacement, such as rainfall, earthquake data, surface displacement rate, deep displacement, and groundwater level change, from the pre-processed raw data. Use the extracted characteristic data as input and the landslide displacement risk probability as output. Combined with the neural network model, build a landslide risk assessment model. Verify the model by comparing the model prediction results with the actual observation results to verify the accuracy and reliability of the model. Based on the verified landslide risk assessment model, evaluate the risk of the landslide area.

[0038] Input the geological, meteorological and monitoring data characteristics of the landslide body into the landslide risk assessment model and output the landslide displacement risk probability value;

[0039] Based on the assessment results of the landslide risk assessment model, the displacement risk assessment coefficient of the landslide body is calculated in combination with the landslide identification matching index. According to the size of the displacement risk assessment coefficient, it is judged whether the landslide body is in a dangerous state and the possibility and potential impact of the landslide are determined.

[0040] Preferably, the calculation process of the displacement risk assessment coefficient includes:

[0041] Extract normalized rainfall, earthquake magnitude, surface displacement rate and groundwater level change characteristic data;

[0042] The rainfall risk function is calculated by the rainfall and the baseline value of the rainfall, and the impact of the rainfall risk on the landslide is analyzed. The earthquake risk function is calculated by using the earthquake magnitude, the minimum value and the maximum value of the earthquake magnitude, and the impact of the earthquake risk on the landslide is analyzed. The expression of the rainfall risk function is: C r is the rainfall risk function, P r is the rainfall, P r，base is the baseline value of rainfall, indicating the minimum rainfall that may cause landslides, and is a threshold set based on similar cases; the expression of the earthquake risk function is: C m is the earthquake risk function, M s is the earthquake magnitude, min(M s ) and max(M s ) are the minimum and maximum values ​​of earthquake magnitude, respectively, indicating the magnitude range that may trigger landslides, determined based on similar case analysis;

[0043] The surface displacement risk function is calculated using the surface displacement rate, the surface displacement rate benchmark value and the maximum observed value of the surface displacement rate to clarify the risk impact of landslides caused by surface displacement. The deep displacement, the deep displacement benchmark value and the maximum observed value of the deep displacement are used to calculate the deep displacement risk function to analyze the risk impact of landslides caused by deep displacement. The expression of the surface displacement risk function is: C d is the surface displacement risk function, D s is the surface displacement rate, D s，base is the baseline value of the surface displacement rate, indicating a safety threshold. When this value is exceeded, the landslide risk increases. max(D s ) is the maximum observed value of the surface displacement rate; the expression of the deep displacement risk function is: C b is the risk function of deep displacement, D b is the deep displacement, D b，base is the reference value of deep displacement, max(D b ) is the maximum observed value of deep displacement;

[0044] The groundwater level risk function is calculated by the groundwater level change, the base value of the groundwater level change and the quantitative scale of the groundwater level change, and the risk impact of the groundwater level change on the landslide is analyzed. The expression of the groundwater level risk function is: C g is the groundwater level risk function, G w is the groundwater level change, G w，base is the base value of groundwater level change, indicating the critical change of groundwater level, G w，s It is a quantitative scale of groundwater level change and is used to adjust the sensitivity of the exponential function;

[0045] Comprehensively analyze the impact of rainfall risk, earthquake risk, surface displacement risk, deep displacement risk and groundwater level risk on landslides, calculate the displacement risk assessment coefficient R, and analyze the landslide risk level of the reservoir bank.

[0046] Preferably, the calculation expression of the displacement risk assessment coefficient is:

[0047]

[0048] Among them, R is the displacement risk assessment coefficient, C r is the rainfall risk function, C m is the earthquake risk function, C d is the surface displacement risk function, C b is the risk function of deep displacement, C g is the groundwater level risk function, and the value of R ranges from 0 to 1, where 0 represents no risk and 1 represents the highest risk.

[0049] Preferably, the risk level classification process includes:

[0050] According to the severity of landslide disasters, different risk levels of landslide disasters are preset, namely low risk level, medium risk level, high risk level and severe risk level;

[0051] According to the size of the displacement risk assessment coefficient, a matching risk assessment threshold is set for each landslide risk level;

[0052] Based on the assessment results of the landslide risk assessment model, it is determined whether the landslide body is in a dangerous state, and the risk of the landslide area is classified by the size of the displacement risk assessment coefficient to determine the risk level of the reservoir bank landslide body;

[0053] Obtain the value of the displacement risk assessment coefficient and the preset evaluation threshold of the risk level to match the four risk levels, which are:

[0054] Low risk level: 0≤R <R1;

[0055] Medium risk level: R1≤R <R2;

[0056] High risk level: R2≤R <R3;

[0057] Severe risk level: R3≤R<1;

[0058] Among them, R is the displacement risk assessment coefficient, R1 is the lower threshold corresponding to the medium risk level and the upper threshold corresponding to the low risk level, R2 is the lower threshold corresponding to the high risk level and the upper threshold corresponding to the medium risk level, R3 is the lower threshold corresponding to the severe risk level and the upper threshold corresponding to the high risk level, R1=0.25, R2=0.5, R3=0.75.

[0059] Preferably, the process of formulating the early warning mechanism and response measures includes:

[0060] Based on historical landslide data and similar case analysis, a landslide early warning system is established to ensure the stability and reliability of the early warning system, which can warn of displacement changes of landslide bodies. When the risk of landslide bodies reaches or exceeds the risk assessment threshold, the early warning mechanism is triggered and early warning information is issued to relevant departments and personnel through SMS and phone calls so that they can take countermeasures.

[0061] According to the displacement prediction results and risk level of the landslide body, formulate corresponding emergency response plans and take corresponding prevention and control measures. For landslide bodies with low risk levels, strengthen monitoring and early warning to timely discover potential risks. For landslide bodies with medium risk levels, take engineering measures, such as reinforcing slopes and building retaining walls. For landslide bodies with high and severe risk levels, the emergency response plan should be immediately activated to evacuate personnel, transfer materials, and take emergency engineering measures for control;

[0062] During the implementation of early warning and prevention measures, the landslide situation will be continuously monitored, the effectiveness of prevention measures will be evaluated, and the response strategy will be adjusted according to actual conditions.

[0063] The present invention provides a landslide displacement prediction method for reservoir bank inspection. It has the following beneficial effects:

[0064] 1. The landslide displacement prediction method of the reservoir bank inspection can more comprehensively analyze the causes of landslide deformation by combining geological conditions, meteorological monitoring and reservoir bank landslide monitoring data, and can more accurately predict the possibility of landslide occurrence by considering the impact of rainfall and earthquakes on landslides. By setting risk assessment thresholds and evaluating different landslide risk levels, early warning can be given before a landslide occurs, thereby achieving early identification of landslide risks.

[0065] 2. The landslide displacement prediction method of the reservoir bank inspection can timely discover the displacement anomalies and potential risks of the landslide body through monitoring the landslide body, ensure the safety and stability of the reservoir bank, and avoid the damage of landslide disasters to the facilities around the reservoir. During the monitoring process, the displacement, deformation, rainfall and other parameters of the landslide body are monitored and recorded in real time. Through data analysis, the displacement trend changes of the landslide body can be discovered, which provides a strong guarantee for the safe operation of the reservoir and thus improves the prevention and control efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A method flow chart of a landslide displacement prediction method for reservoir bank inspection according to the present invention;

[0067] Figure 2 A flow chart of the reservoir bank landslide state identification of the present invention;

[0068] Figure 3 The figure is a calculation flow chart of the landslide identification matching index of the present invention;

[0069] Figure 4 This is a flow chart for assessing the risk of reservoir bank landslides according to the present invention. DETAILED DESCRIPTION

[0070] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0071] The first embodiment, as Figures 1 to 3 As shown, the present invention provides a technical solution: a landslide displacement prediction method for reservoir bank inspection, comprising the following steps:

[0072] Step 1: Collect the original data of geological conditions, meteorological monitoring and landslide monitoring of the reservoir bank, and pre-process the original data. The original data of geological conditions of the reservoir bank inspection are obtained through geological survey reports, geological maps, and geological profiles, including the geological structure, rock and soil type, stratum distribution, fault and fold geological structure information of the reservoir bank. The original data of meteorological monitoring of the reservoir bank inspection are obtained through meteorological monitoring stations, meteorological satellites, and meteorological radar equipment, including rainfall, temperature, wind speed, and wind direction meteorological information. For the original data of landslide monitoring of the reservoir bank, crack meters, GNSS monitors, and synthetic aperture radars are used to obtain data, including the displacement, deformation rate, crack width, and depth parameters of the landslide body. The collected data are sorted according to time series and monitoring points to ensure the integrity and consistency of the data, and data from different sources and types are classified and marked for subsequent use. Processing: preprocessing the classified and marked data, including data cleaning and data conversion steps. Data cleaning is used to remove abnormal values, check abnormal or unreasonable values ​​in the data, such as data beyond the normal range, and remove or correct them. For missing data, interpolation, regression and other methods are used to fill in the missing data according to the distribution law and correlation of the data, and the data is smoothed to reduce the impact of noise and fluctuations. Data conversion includes data format conversion and data scale conversion. Data format conversion converts data from different sources and formats into a unified format for subsequent processing and analysis. Data scale conversion converts the scale of the data, such as converting displacement data into relative displacement or absolute displacement, so as to better reflect the changes in the landslide body. The preprocessed geological condition data, meteorological monitoring data and reservoir bank landslide body monitoring data are integrated to form a complete data set for subsequent analysis and processing;

[0073] Step 2: Extract features related to the landslide deformation mode from the preprocessed raw data and integrate them to obtain a feature set. Extract geological features from geological condition data, including the distribution of rock and soil types, the inclination angle of the strata, and the density of faults and folds. Extract meteorological features from meteorological monitoring data, including the total amount and intensity of rainfall, the number of consecutive dry or wet days, wind speed and direction, and the frequency of extreme meteorological events. Extract landslide body features from landslide body monitoring data, including displacement change rate, crack width and depth, crack expansion rate, and groundwater level fluctuation. Further process the extracted features, calculate the change rate of the features, and normalize the extracted features to eliminate the impact of different dimensions and value ranges of different features on the model, so that they are on the same scale for easy comparison and analysis. Integrate the extracted relevant feature data to obtain a feature set, in which each feature is associated with the landslide deformation mode, ensuring that the features in the feature set cover different aspects and conditions that may cause landslides, as the basis for subsequent landslide prediction and analysis;

[0074] Step 3: Based on the relevant data of the feature set, the support vector machine algorithm is used to train the extracted features, and a landslide deformation pattern recognition model is established to identify the landslide state on the reservoir bank. The feature set data is divided into a training set and a test set. The training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model. It is ensured that the data in the training set and the test set are representative and can reflect various modes of landslide deformation. The RBF kernel function of the support vector machine (SVM) algorithm is selected to process the nonlinear relationship between the features and identify the landslide deformation mode. The training set data is used to train the SVM model, and the landslide deformation pattern recognition model is constructed to obtain a classifier. The test set data is used to classify the trained landslide deformation pattern recognition model. The model was tested to evaluate the recognition accuracy and generalization ability of the model. The feature data of the landslide to be identified was input into the trained landslide deformation pattern recognition model to obtain the predicted categories of landslide deformation, namely stable, slow deformation and accelerated deformation. The overlap between the predicted landslide area and the actual landslide area was compared, and the landslide identification matching index was calculated to measure the matching degree between the model prediction result and the actual landslide state. The higher the matching index value, the higher the matching degree between the model prediction result and the actual landslide state. According to the landslide identification matching index, the recognition matching degree of the reservoir bank landslide state was analyzed, and a matching threshold was set. The landslide identification matching index was compared with the set matching threshold to determine whether the landslide state was accurately identified.

[0075] Furthermore, the calculation process of the landslide identification matching index includes:

[0076] Based on the historical data in the original data, the spatial data of the actual landslide area is obtained, the spatial data of the predicted landslide area predicted by the landslide deformation pattern recognition model is extracted, and the coordinate transformation is performed so that the spatial data sets of the actual landslide area and the predicted landslide area are in the same coordinate reference system to ensure the consistency of the comparison;

[0077] Use geographic information system software to calculate the area of ​​the actual landslide area and the predicted landslide area, and calculate the intersection area of ​​the actual landslide area and the predicted landslide area;

[0078] Based on the calculated actual landslide area and the predicted landslide area, the absolute value of the area difference between the actual landslide area and the predicted landslide area is calculated, and the landslide identification matching index is obtained through comprehensive analysis;

[0079] The calculated landslide identification matching index value is analyzed to determine the accuracy of the model prediction. If the landslide identification matching index value is close to 1, it means that the model prediction result matches the actual landslide state well. If the landslide identification matching index value is low, it means that the model prediction has a large deviation.

[0080] The calculation expression of landslide identification matching index is:

[0081]

[0082] Wherein, I is the landslide identification matching index, |A| is the actual landslide area, |B| is the predicted landslide area, |A∩B| is the intersection area of ​​the actual landslide area and the predicted landslide area, i.e., the overlapping part, λ is an adjustment parameter used to balance the influence of the difference between the actual landslide area and the predicted landslide area, λ≥0, ||A|-|B|| is the absolute value of the difference between the actual landslide area and the predicted landslide area, used to punish the situation where the area difference between the predicted area and the actual area is large, the value range of I is limited to between 0 and 1, when I=1, it means that the predicted landslide area is completely consistent with the actual landslide area, when I=0, it means that the predicted landslide area does not overlap with the actual landslide area, the higher the value, the higher the matching degree between the model prediction result and the actual landslide state, when the predicted area completely contains the actual landslide area, and there is no additional predicted area, I reaches the maximum value of 1, if the predicted area includes the area outside the actual landslide area, or misses part of the actual landslide area, the value of I will decrease;

[0083] Step 4: Combine geological, meteorological and reservoir bank landslide monitoring data to conduct a comprehensive analysis of the causes of landslide deformation, analyze the interaction and impact between different factors, and build a landslide risk assessment model to assess the risk of reservoir bank landslides;

[0084] Step 5: Based on the assessed risk of the reservoir bank landslide body, different risk levels are preset to classify the landslide risk of the reservoir bank landslide body;

[0085] Step 6: According to the landslide displacement prediction results and actual conditions, match the corresponding early warning mechanism for each risk level and formulate corresponding response measures.

[0086] The second embodiment is based on the first embodiment. Figure 4 As shown in the figure, the assessment process of reservoir bank landslide risk includes:

[0087] Integrate geological condition data, meteorological monitoring data and reservoir bank landslide monitoring data, analyze the causes of landslide deformation, which are rainfall and earthquake. Rainfall is one of the main factors causing landslides. Rainfall will increase soil moisture content and reduce soil shear strength, thereby increasing the risk of landslides. Analyze the relationship between rainfall and landslide occurrence, especially the impact of heavy rain and continuous rainfall on landslides. The vibration caused by earthquakes can cause landslides to lose balance, thereby triggering landslides. Analyze the impact of factors such as earthquake magnitude, focal depth, and seismic wave type on landslides. Extract characteristic data related to landslide displacement, such as rainfall, earthquake data, surface displacement rate, deep displacement, and groundwater level change, from the pre-processed raw data. Use extraction The characteristic data is used as input, and the landslide displacement risk probability is used as output. A landslide risk assessment model is constructed in combination with a neural network model. The model is verified by comparing the model prediction results with the actual observation results to verify the accuracy and reliability of the model. Based on the verified landslide risk assessment model, the risk of the landslide area is assessed. The geological, meteorological and monitoring data characteristics of the landslide body are input into the landslide risk assessment model, and the landslide displacement risk probability value is output. Based on the assessment results of the landslide risk assessment model, the displacement risk assessment coefficient of the landslide body is calculated in combination with the landslide identification matching index. According to the size of the displacement risk assessment coefficient, it is judged whether the landslide body is in a dangerous state and the possibility and potential impact of the landslide are determined.

[0088] Furthermore, the calculation process of the displacement risk assessment coefficient includes:

[0089] Extract the normalized rainfall, earthquake magnitude, surface displacement rate and groundwater level change characteristic data, calculate the rainfall risk function through the rainfall and the baseline value of rainfall, analyze the rainfall risk impact of landslide, and use the earthquake magnitude, the minimum and maximum values ​​of the earthquake magnitude to calculate the earthquake risk function, and analyze the earthquake risk impact of landslide. The expression of the rainfall risk function is: C r is the rainfall risk function, P r is the rainfall, P r，base is the baseline value of rainfall, indicating the minimum rainfall that may cause landslides, and is a threshold set based on similar cases; the expression of the earthquake risk function is: C m is the earthquake risk function, M s is the earthquake magnitude, min(M s ) and max(M sare the minimum and maximum values ​​of earthquake magnitude, respectively, indicating the magnitude range that may cause landslides. Based on similar case analysis, the surface displacement rate, the surface displacement rate benchmark value and the maximum observed value of the surface displacement rate are used to calculate the surface displacement risk function, clarify the risk impact of surface displacement causing landslides, and the deep displacement, the deep displacement benchmark value and the maximum observed value of the deep displacement are used to calculate the deep displacement risk function, and analyze the risk impact of deep displacement causing landslides. The expression of the surface displacement risk function is: C d is the surface displacement risk function, D s is the surface displacement rate, D s，base is the baseline value of the surface displacement rate, indicating a safety threshold. When this value is exceeded, the landslide risk increases. max(D s ) is the maximum observed value of the surface displacement rate; the expression of the deep displacement risk function is: C b is the risk function of deep displacement, D b is the deep displacement, D b，base is the reference value of deep displacement, max(D b ) is the maximum observed value of deep displacement. The groundwater level risk function is calculated through groundwater level change, the reference value of groundwater level change and the quantitative scale of groundwater level change to analyze the impact of groundwater level change on landslide risk. The expression of groundwater level risk function is: C g is the groundwater level risk function, G w is the groundwater level change, G w，base is the base value of groundwater level change, indicating the critical change of groundwater level, G w，s It is a quantitative scale for groundwater level changes, used to adjust the sensitivity of the exponential function, comprehensively analyze the impact of rainfall risk, earthquake risk, surface displacement risk, deep displacement risk and groundwater level risk on landslides, calculate the displacement risk assessment coefficient R, and analyze the landslide risk level of the reservoir bank;

[0090] Furthermore, the calculation expression of the displacement risk assessment coefficient is:

[0091]

[0092] Among them, R is the displacement risk assessment coefficient, C r is the rainfall risk function, C m is the earthquake risk function, C d is the surface displacement risk function, C b is the risk function of deep displacement, C gis the groundwater level risk function, and the value of R ranges from 0 to 1, where 0 indicates no risk and 1 indicates the highest risk. When rainfall, earthquake magnitude, surface displacement rate, and groundwater level change all exceed the benchmark values, the R value increases, indicating an increase in landslide risk. When these factors are below the benchmark values, the R value decreases, indicating a decrease in landslide risk.

[0093] The risk classification process includes:

[0094] According to the severity of landslide disasters, different risk levels of landslide disasters are preset, namely low risk level, medium risk level, high risk level and severe risk level. For low risk level, the possibility of landslide is small. Even if it occurs, the loss caused is small and the impact on society and environment is limited. For medium risk level, the possibility of landslide is moderate. If it occurs, it may cause certain losses and have certain impact on society and environment. For high risk level, the possibility of landslide is high. If it occurs, it may cause serious losses and have a greater impact on society and environment. For severe risk level, the possibility of landslide is extremely high. If it occurs, it may cause catastrophic losses and have extremely serious impact on society and environment. According to the size of displacement risk assessment coefficient, matching risk assessment thresholds are set for each landslide risk level. Based on the assessment results of landslide risk assessment model, it is judged whether the landslide body is in a dangerous state. The size of displacement risk assessment coefficient is used to classify the risk of landslide area and determine the risk level of reservoir bank landslide body.

[0095] Obtain the value of the displacement risk assessment coefficient and the preset evaluation threshold of the risk level to match the four risk levels, which are:

[0096] Low risk level: 0≤R <R1;

[0097] Medium risk level: R1≤R <R2;

[0098] High risk level: R2≤R <R3;

[0099] Severe risk level: R3≤R<1;

[0100] Among them, R is the displacement risk assessment coefficient, R1 is the lower threshold corresponding to the medium risk level and the upper threshold corresponding to the low risk level, R2 is the lower threshold corresponding to the high risk level and the upper threshold corresponding to the medium risk level, R3 is the lower threshold corresponding to the severe risk level and the upper threshold corresponding to the high risk level, R1 = 0.25, R2 = 0.5, R3 = 0.75;

[0101] The process of developing early warning mechanisms and response measures includes:

[0102] Based on historical landslide data and similar case analysis, a landslide early warning system is established to ensure the stability and reliability of the early warning system, which can warn of displacement changes of landslide bodies. When the risk of landslide bodies reaches or exceeds the risk assessment threshold, the early warning mechanism is triggered, and early warning information is issued to relevant departments and personnel through multiple channels such as text messages and telephone calls so that they can take countermeasures. According to the displacement prediction results and risk level of the landslide body, a corresponding emergency response plan is formulated and corresponding prevention and control measures are taken. For landslide bodies with low risk levels, monitoring and early warning are strengthened to timely discover potential risks. For landslide bodies with medium risk levels, engineering measures are taken, such as reinforcing slopes and building retaining walls. For landslide bodies with high and severe risk levels, the emergency response plan should be immediately activated to evacuate personnel, transfer materials, and take emergency engineering measures for control. During the implementation of early warning and prevention measures, the landslide situation is continuously monitored, the effectiveness of prevention measures is evaluated, and the response strategy is adjusted according to actual conditions.

[0103] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A landslide displacement prediction method for reservoir bank inspection, characterized in that: The following steps are involved: Step 1: Collect the original data of geological conditions, meteorological monitoring and landslide monitoring of the reservoir bank inspection, and pre-process the original data; Step 2: extracting features related to the landslide deformation mode from the preprocessed raw data and integrating them to obtain a feature set; Step 3: Based on the relevant data of the feature set, the support vector machine algorithm is used to train the extracted features, establish a landslide deformation pattern recognition model, and identify the landslide state of the reservoir bank; Step 4: Combine geological, meteorological and reservoir bank landslide monitoring data to conduct a comprehensive analysis of the causes of landslide deformation, analyze the interaction and impact between different factors, and build a landslide risk assessment model to assess the risk of reservoir bank landslides; Step 5: Based on the assessed risk of the reservoir bank landslide body, different risk levels are preset to classify the landslide risk of the reservoir bank landslide body; Step 6: According to the landslide displacement prediction results and actual conditions, match the corresponding early warning mechanism for each risk level and formulate corresponding response measures.

2. The landslide displacement prediction method for reservoir bank inspection according to claim 1 is characterized by: The collection and preprocessing process of the raw data includes: The original data of geological conditions for reservoir bank inspections are obtained through geological survey reports, geological maps, and geological profiles, including information on the geological structure, rock and soil types, stratum distribution, faults, and folds of the reservoir bank; The original meteorological monitoring data of reservoir bank inspections are obtained through meteorological monitoring stations, meteorological satellites, and meteorological radar equipment, including rainfall, temperature, wind speed, and wind direction meteorological information; For the original data of the reservoir bank landslide monitoring, crack meters, GNSS monitors, and synthetic aperture radars are used to obtain data, including the displacement, deformation rate, crack width, and depth parameters of the landslide body; Organize the collected data according to time series and monitoring points, and classify and label data from different sources and types; Preprocess the classified and labeled data, including data cleaning and data conversion steps; The preprocessed geological condition data, meteorological monitoring data and reservoir bank landslide monitoring data are integrated to form a complete data set.

3. The landslide displacement prediction method for reservoir bank inspection according to claim 2 is characterized by: The process of acquiring the feature set includes: Extract geological characteristics from geological condition data, including the distribution of rock and soil types, the inclination angle of the strata, and the density of faults and folds; extract meteorological characteristics from meteorological monitoring data, including the total amount and intensity of rainfall, the number of consecutive dry or wet days, wind speed and direction, and the frequency of extreme meteorological events; extract landslide characteristics from landslide monitoring data, including displacement change rate, crack width and depth, crack expansion speed, and groundwater level fluctuation; The extracted features are further processed to calculate the rate of change of the features and normalize the extracted features to eliminate the dimensional effects of different features so that they are on the same scale; The extracted relevant feature data are integrated to obtain a feature set, each of which is associated with the landslide deformation mode.

4. The landslide displacement prediction method for reservoir bank inspection according to claim 3 is characterized by: The process of identifying the state of the reservoir bank landslide includes: The feature set data is divided into a training set and a test set. The training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model. The RBF kernel function of the support vector machine algorithm is selected to process the nonlinear relationship between the features and identify the landslide deformation mode. The training set data is used to train the SVM model, and the landslide deformation pattern recognition model is constructed to obtain the classifier. The trained landslide deformation pattern recognition model is tested using the test set data. The landslide feature data to be identified is input into the trained landslide deformation pattern recognition model to obtain the predicted categories of landslide deformation, which are stable, slow deformation, and accelerated deformation. Compare the overlap between the predicted landslide area and the actual landslide area, calculate the landslide identification matching index, and measure the matching degree between the model prediction results and the actual landslide status; According to the landslide identification matching index, the recognition matching degree of the reservoir bank landslide state is analyzed, and a matching threshold is set. The landslide identification matching index is compared with the set matching threshold to determine whether the landslide state is accurately identified.

5. The landslide displacement prediction method for reservoir bank inspection according to claim 4 is characterized by: The calculation process of the landslide identification matching index includes: Based on the historical data in the original data, the spatial data of the actual landslide area is obtained, the spatial data of the predicted landslide area predicted by the landslide deformation pattern recognition model is extracted, and the coordinate transformation is performed so that the spatial data sets of the actual landslide area and the predicted landslide area are in the same coordinate reference system; Use geographic information system software to calculate the area of ​​the actual landslide area and the predicted landslide area, and calculate the intersection area of ​​the actual landslide area and the predicted landslide area; Based on the calculated actual landslide area and the predicted landslide area, the absolute value of the area difference between the actual landslide area and the predicted landslide area is calculated, and the landslide identification matching index is obtained through comprehensive analysis; Analyze the calculated landslide identification matching index value to determine the accuracy of the model prediction; The calculation expression of the landslide identification matching index is: Where I is the landslide identification matching index, |A| is the actual landslide area, |B| is the predicted landslide area, |A∩B| is the intersection area of ​​the actual landslide area and the predicted landslide area, that is, the overlapping part, λ is the adjustment parameter, λ≥0, and ||A|-|B|| is the absolute value of the difference between the actual landslide area and the predicted landslide area.

6. The landslide displacement prediction method for reservoir bank inspection according to claim 5 is characterized by: The assessment process of the reservoir bank landslide risk includes: Integrate geological condition data, meteorological monitoring data and reservoir bank landslide monitoring data to analyze the causes of landslide deformation; Extract characteristic data related to landslide displacement, such as rainfall, earthquake data, surface displacement rate, deep displacement, and groundwater level change, from the pre-processed raw data. Use the extracted characteristic data as input and the landslide displacement risk probability as output. Combined with the neural network model, build a landslide risk assessment model. Verify the model by comparing the model prediction results with the actual observation results. Based on the verified landslide risk assessment model, evaluate the risk of the landslide area. Input the geological, meteorological and monitoring data characteristics of the landslide body into the landslide risk assessment model and output the landslide displacement risk probability value; Based on the assessment results of the landslide risk assessment model and combined with the landslide identification matching index, the displacement risk assessment coefficient of the landslide body is calculated. According to the size of the displacement risk assessment coefficient, it is judged whether the landslide body is in a dangerous state.

7. The method for predicting landslide displacement during reservoir bank inspection according to claim 6, characterized in that: The calculation process of the displacement risk assessment coefficient includes: Extract normalized rainfall, earthquake magnitude, surface displacement rate and groundwater level change characteristic data; The rainfall risk function is calculated by the rainfall and the baseline value of the rainfall, and the impact of the rainfall risk on the landslide is analyzed. The earthquake risk function is calculated by using the earthquake magnitude, the minimum value and the maximum value of the earthquake magnitude, and the impact of the earthquake risk on the landslide is analyzed. The expression of the rainfall risk function is: C r is the rainfall risk function, P r is the rainfall, P r，base is the baseline value of rainfall; the expression of the earthquake risk function is: C m is the earthquake risk function, M s is the earthquake magnitude, min(M s ) and max(M s ) are the minimum and maximum values ​​of earthquake magnitude, indicating the magnitude range that may trigger landslides; The surface displacement risk function is calculated using the surface displacement rate, the surface displacement rate benchmark value and the maximum observed value of the surface displacement rate to clarify the risk impact of landslides caused by surface displacement. The deep displacement, the deep displacement benchmark value and the maximum observed value of the deep displacement are used to calculate the deep displacement risk function to analyze the risk impact of landslides caused by deep displacement. The expression of the surface displacement risk function is: C d is the surface displacement risk function, D s is the surface displacement rate, D s，base is the reference value of the surface displacement rate, max(D s ) is the maximum observed value of the surface displacement rate; the expression of the deep displacement risk function is: C b is the risk function of deep displacement, D b is the deep displacement, D b，base is the reference value of deep displacement, max(D b ) is the maximum observed value of deep displacement; The groundwater level risk function is calculated by the groundwater level change, the base value of the groundwater level change and the quantitative scale of the groundwater level change, and the risk impact of the groundwater level change on the landslide is analyzed. The expression of the groundwater level risk function is: C g is the groundwater level risk function, G w is the groundwater level change, G w，base is the base value of groundwater level change, indicating the critical change of groundwater level, G w，s It is a quantitative measure of groundwater level change; Comprehensively analyze the impact of rainfall risk, earthquake risk, surface displacement risk, deep displacement risk and groundwater level risk on landslides, calculate the displacement risk assessment coefficient R, and analyze the landslide risk level of the reservoir bank.

8. The method for predicting landslide displacement during reservoir bank inspection according to claim 7, characterized in that: The calculation expression of the displacement risk assessment coefficient is: Among them, R is the displacement risk assessment coefficient, C r is the rainfall risk function, C m is the earthquake risk function, C d is the surface displacement risk function, C b is the risk function of deep displacement, C g is the groundwater level risk function, and the value of R ranges from 0 to 1, where 0 represents no risk and 1 represents the highest risk.

9. The landslide displacement prediction method for reservoir bank inspection according to claim 8 is characterized by: The risk level classification process includes: According to the severity of landslide disasters, different risk levels of landslide disasters are preset, namely low risk level, medium risk level, high risk level and severe risk level; According to the size of the displacement risk assessment coefficient, a matching risk assessment threshold is set for each landslide risk level; Based on the assessment results of the landslide risk assessment model, it is determined whether the landslide body is in a dangerous state, and the risk of the landslide area is classified by the size of the displacement risk assessment coefficient to determine the risk level of the reservoir bank landslide body; Obtain the value of the displacement risk assessment coefficient and the preset evaluation threshold of the risk level to match the four risk levels, which are: Low risk level: 0≤R <R1; Medium risk level: R1≤R <R2; High risk level: R2≤R <R3; Severe risk level: R3≤R<1; Among them, R is the displacement risk assessment coefficient, R1 is the lower threshold corresponding to the medium risk level and the upper threshold corresponding to the low risk level, R2 is the lower threshold corresponding to the high risk level and the upper threshold corresponding to the medium risk level, R3 is the lower threshold corresponding to the severe risk level and the upper threshold corresponding to the high risk level, R1=0.25, R2=0.5, R3=0.

75.

10. A landslide displacement prediction method for reservoir bank inspection according to claim 9, characterized in that: The process of developing the early warning mechanism and response measures includes: A landslide early warning system is established based on historical landslide data and similar case analysis. When the risk of a landslide reaches or exceeds the risk assessment threshold, the early warning mechanism is triggered and early warning information is issued to relevant departments and personnel through SMS and phone calls. According to the displacement prediction results and risk level of the landslide body, formulate corresponding emergency response plans and take corresponding prevention and control measures; During the implementation of early warning and prevention measures, the landslide situation will be continuously monitored, the effectiveness of prevention measures will be evaluated, and the response strategy will be adjusted according to actual conditions.

Images