Landslide risk monitoring method and system based on multi-source data

Abstract

The invention discloses a landslide risk monitoring method and system based on multi-source data, and relates to the field of geological monitoring early warning, and the method comprises the steps: obtaining the environment monitoring data of a to-be-monitored slope area, and constructing a multi-dimensional feature matrix according to the environment monitoring data; performing risk prediction classification based on the multi-dimensional feature matrix according to a risk classifier to obtain a landslide risk level, and performing prediction based on the multi-dimensional feature matrix according to a neural network model to obtain a landslide displacement prediction quantity; calculating a landslide stability coefficient according to the environmental monitoring data based on a limit equilibrium method, calculating a rainfall induction threshold by combining the rainfall data and the soil humidity data, and obtaining a landslide comprehensive risk coefficient by integrating the landslide stability coefficient and the rainfall induction threshold; and determining a landslide risk monitoring result of the to-be-monitored slope area in combination with the landslide risk grade, the landslide displacement prediction amount and the landslide comprehensive risk coefficient. According to the application, the technical problem that the existing landslide risk monitoring accuracy is poor can be solved.

Description

Technical Field

[0001] This application relates to the field of geological monitoring and early warning, and in particular to a landslide risk monitoring method and system based on multi-source data. Background Technology

[0002] Landslides are natural phenomena caused by various natural or non-natural factors, such as geological conditions, rainfall, or earthquakes, where surface soil or rock masses slide down a slope. They not only damage the ecological environment but also cause severe property loss and casualties, posing a significant threat, especially to critical infrastructure such as transportation and buildings. To reduce the damage caused by landslides, high-precision monitoring and accurate early warning before landslides occur are crucial. Recent research on landslide risk monitoring has generally focused on improving the accuracy of various sensors.

[0003] Existing methods for landslide risk monitoring and early warning mainly focus on improving the accuracy of sensors that collect landslide measurement data, with less consideration given to improving the methods themselves. This has led to the following problems: (1) Limited data sources: Current landslide monitoring generally relies on strain monitoring and soil moisture, which is difficult to reflect the dynamic changes inside the slope and the surrounding environment, and is also prone to missing key factors, thus affecting the accuracy of landslide risk assessment; (2) Lack of intelligent analysis capabilities: Existing monitoring methods are mainly based on static thresholds for risk judgment, which is difficult to integrate the complex spatiotemporal coupling effects in the landslide process, resulting in serious lag in monitoring and early warning, and easy misjudgment and omission. Therefore, how to improve the accuracy of landslide risk monitoring is still an urgent problem to be solved by existing technologies. Summary of the Invention

[0004] This application provides a landslide risk monitoring method and system based on multi-source data to solve the technical problem of insufficient accuracy in existing landslide risk monitoring.

[0005] According to a first aspect of the embodiments of this application, a landslide risk monitoring method based on multi-source data is provided, comprising:

[0006] Environmental monitoring data of the slope area to be monitored is acquired, and a multidimensional feature matrix is ​​constructed based on the environmental monitoring data; wherein, the environmental monitoring data includes strain data, soil displacement data, rainfall data, temperature and humidity data, wind speed data, groundwater level data, and soil moisture data;

[0007] Based on the preset risk classifier, risk prediction and classification are performed using the multidimensional feature matrix to obtain the landslide risk level. Based on the preset neural network model, the predicted landslide displacement is then obtained.

[0008] Based on the limit equilibrium method, the landslide stability coefficient is calculated according to the environmental monitoring data. Combined with the rainfall data and the soil moisture data, the rainfall-induced threshold is calculated. Finally, the comprehensive landslide risk coefficient is obtained by combining the landslide stability coefficient and the rainfall-induced threshold.

[0009] By combining the landslide risk level, the predicted landslide displacement, and the comprehensive landslide risk coefficient, the landslide risk monitoring results for the slope area to be monitored are determined.

[0010] This application first acquires environmental monitoring data of the slope area to be monitored and constructs a multi-dimensional feature matrix, which can integrate data from different sources and of different types, providing a data foundation for subsequent data prediction. Compared with the single data source of existing technologies, multiple data sources improve the comprehensiveness of the data, thereby avoiding the adverse effects of a single data source on the accuracy of landslide risk assessment. Then, risk prediction classification and displacement prediction are performed based on a preset risk classifier and neural network model. The comprehensive landslide risk coefficient is obtained by calculating the landslide stability coefficient and rainfall-induced threshold based on the limit equilibrium method. Finally, the landslide risk monitoring results are comprehensively determined. Compared with existing technologies, by comprehensively utilizing different models for landslide risk assessment and combining different calculation indicators for auxiliary judgment, the application can avoid misjudgments and omissions caused by the static thresholds of existing technologies, thereby improving the accuracy of landslide risk monitoring.

[0011] In some embodiments of this application, the step of acquiring environmental monitoring data of the slope area to be monitored and constructing a multidimensional feature matrix based on the environmental monitoring data specifically includes:

[0012] Based on multiple monitoring devices in the slope area to be monitored, the environmental monitoring data is acquired, and noise filtering and missing value completion are performed on the environmental monitoring data to obtain the first monitoring data;

[0013] Based on Kalman filtering, the data of various types in the first monitoring data are fused to obtain the second monitoring data, and the time series features of the data of various types in the second monitoring data are extracted based on wavelet transform to obtain multiple sets of landslide time series feature data.

[0014] The temporal and spatial alignment of each group of landslide time series feature data was performed to construct a multidimensional feature matrix.

[0015] This application acquires environmental monitoring data through multiple monitoring devices and performs noise filtering and missing value completion, which improves the effectiveness and accuracy of the data, thereby enhancing the accuracy of subsequent landslide risk monitoring. Furthermore, it fuses various types of data based on Kalman filtering, improving the comprehensiveness of the data. Finally, it extracts temporal features based on wavelet transform and constructs a multidimensional feature matrix through spatiotemporal alignment. Compared to existing technologies, this approach integrates temporal features into the data, improving data effectiveness and thus enhancing the accuracy of subsequent landslide risk monitoring.

[0016] In some embodiments of this application, the step of predicting the landslide displacement based on the multidimensional feature matrix according to a preset neural network model specifically includes:

[0017] Based on a pre-defined convolutional neural network, features are extracted from the soil displacement data in the multidimensional feature matrix to obtain soil displacement features.

[0018] Based on a preset long short-term memory network, the soil displacement characteristics are predicted to obtain the predicted landslide displacement.

[0019] This application uses a pre-defined convolutional neural network to extract features from soil displacement data, which can accurately obtain the displacement characteristics of the soil. Then, it uses a long short-term memory network to predict the landslide displacement, which can improve the accuracy of prediction based on the temporal features in the data, thereby improving the accuracy of subsequent landslide risk monitoring.

[0020] In some embodiments of this application, the calculation of the landslide stability coefficient based on the environmental monitoring data using the limit equilibrium method specifically includes:

[0021] Acquire slope measurement data for the slope area to be monitored; wherein, the slope measurement data includes measurement data of the slope sliding surface and the slope sliding body;

[0022] Based on the strain data in the environmental monitoring data and the slope sliding surface measurement data in the slope measurement data, the soil anti-sliding force is calculated;

[0023] Based on the groundwater level data in the environmental monitoring data and the slope sliding surface and slope sliding body measurement data in the slope measurement data, the soil sliding force is calculated.

[0024] Based on the limit equilibrium method, the landslide stability coefficient is obtained by combining the soil resistance force and the soil sliding force.

[0025] This application first obtains slope measurement data, then calculates soil anti-sliding force and soil sliding force by combining environmental monitoring data, and then obtains landslide stability coefficient based on the limit equilibrium method. It can accurately obtain the characterization index of landslide stability through the limit equilibrium method, thereby improving the accuracy of subsequent landslide risk monitoring based on the landslide stability coefficient.

[0026] In some embodiments of this application, calculating the rainfall-induced threshold by combining the rainfall data and the soil moisture data specifically includes:

[0027] Based on the rainfall data, determine the current and maximum rainfall, current and maximum rainfall intensity of the slope area to be monitored;

[0028] Based on the soil moisture data, determine the current soil moisture and maximum soil moisture of the slope area to be monitored;

[0029] The rainfall-induced threshold is calculated by weighting the current and maximum rainfall, current and maximum rainfall intensity, and current and maximum soil moisture in the slope area to be monitored.

[0030] This application first determines the relevant characterization quantities of rainfall amount, rainfall intensity and soil moisture based on rainfall data and soil moisture data, and then calculates the rainfall-induced threshold by weighting. This can accurately obtain the characterization index of rainfall-induced landslides among the landslide influencing factors, and improve the accuracy of subsequent landslide risk monitoring based on rainfall-induced thresholds.

[0031] According to a second aspect of the embodiments of this application, a landslide risk monitoring system based on multi-source data is provided, including a feature matrix construction module, a risk prediction and classification module, a risk coefficient calculation module, and a comprehensive risk monitoring module.

[0032] The feature matrix construction module is used to acquire environmental monitoring data of the slope area to be monitored, and to construct a multidimensional feature matrix based on the environmental monitoring data; wherein, the environmental monitoring data includes strain data, soil displacement data, rainfall data, temperature and humidity data, wind speed data, groundwater level data and soil moisture data;

[0033] The risk prediction and classification module is used to perform risk prediction and classification based on the multidimensional feature matrix according to the preset risk classifier to obtain the landslide risk level, and to predict the landslide displacement based on the multidimensional feature matrix according to the preset neural network model.

[0034] The risk coefficient calculation module is used to calculate the landslide stability coefficient based on the limit equilibrium method and the environmental monitoring data, calculate the rainfall-induced threshold by combining the rainfall data and the soil moisture data, and obtain the comprehensive landslide risk coefficient by combining the landslide stability coefficient and the rainfall-induced threshold.

[0035] The comprehensive risk monitoring module is used to determine the landslide risk monitoring results of the slope area to be monitored by combining the landslide risk level, the predicted landslide displacement, and the comprehensive landslide risk coefficient.

[0036] In some embodiments of this application, the feature matrix construction module includes a monitoring data acquisition unit, a data processing and extraction unit, and a feature matrix construction unit;

[0037] The monitoring data acquisition unit is used to acquire the environmental monitoring data from multiple monitoring devices in the slope area to be monitored, and to perform noise filtering and missing value completion on the environmental monitoring data to obtain the first monitoring data.

[0038] The data processing and extraction unit is used to fuse various types of data in the first monitoring data based on Kalman filtering to obtain the second monitoring data, and to extract time-series features of various types of data in the second monitoring data based on wavelet transform to obtain multiple sets of landslide time-series feature data.

[0039] The feature matrix construction unit is used to perform spatiotemporal alignment of each group of landslide time-series feature data to construct a multidimensional feature matrix.

[0040] In some embodiments of this application, the risk prediction and classification module includes a displacement feature extraction unit and a displacement feature prediction unit;

[0041] The displacement feature extraction unit is used to extract features from the soil displacement data in the multidimensional feature matrix based on a preset convolutional neural network to obtain soil displacement features.

[0042] The displacement feature prediction unit is used to predict the soil displacement features based on a preset long short-term memory network to obtain the predicted landslide displacement.

[0043] In some embodiments of this application, the risk coefficient calculation module includes a measurement data acquisition unit, an anti-slip force calculation unit, a sliding force calculation unit, and a stability coefficient calculation unit;

[0044] The measurement data acquisition unit is used to acquire slope measurement data of the slope area to be monitored; wherein, the slope measurement data includes measurement data of the slope sliding surface and the slope sliding body;

[0045] The anti-sliding force calculation unit is used to calculate the soil anti-sliding force based on the strain data in the environmental monitoring data and the slope sliding surface measurement data in the slope measurement data.

[0046] The sliding force calculation unit is used to calculate the soil sliding force based on the groundwater level data in the environmental monitoring data and the measurement data of the slope sliding surface and slope sliding body in the slope measurement data.

[0047] The stability coefficient calculation unit is used to obtain the landslide stability coefficient based on the limit equilibrium method, combining the soil anti-sliding force and the soil sliding force.

[0048] In some embodiments of this application, the risk coefficient calculation module includes a rainfall data determination unit, a soil moisture data determination unit, and a rainfall-induced threshold calculation unit;

[0049] The rainfall data determination unit is used to determine the current rainfall and maximum rainfall, current rainfall intensity and maximum rainfall intensity of the slope area to be monitored based on the rainfall data.

[0050] The soil moisture data determination unit is used to determine the current soil moisture and maximum soil moisture of the slope area to be monitored based on the soil moisture data.

[0051] The rainfall-induced threshold calculation unit is used to calculate the rainfall-induced threshold by combining the current and maximum rainfall, current and maximum rainfall intensity, and current and maximum soil moisture of the slope area to be monitored.

[0052] This application first acquires environmental monitoring data of the slope area to be monitored and constructs a multi-dimensional feature matrix, which can integrate data from different sources and of different types, providing a data foundation for subsequent data prediction. Compared with the single data source of existing technologies, multiple data sources improve the comprehensiveness of the data, thereby avoiding the adverse effects of a single data source on the accuracy of landslide risk assessment. Then, risk prediction classification and displacement prediction are performed based on a preset risk classifier and neural network model. The comprehensive landslide risk coefficient is obtained by calculating the landslide stability coefficient and rainfall-induced threshold based on the limit equilibrium method. Finally, the landslide risk monitoring results are comprehensively determined. Compared with existing technologies, by comprehensively utilizing different models for landslide risk assessment and combining different calculation indicators for auxiliary judgment, the application can avoid misjudgments and omissions caused by the static thresholds of existing technologies, thereby improving the accuracy of landslide risk monitoring. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating a landslide risk monitoring method based on multi-source data, as shown in some embodiments of this application.

[0054] Figure 2 This is a module structure diagram of a landslide risk monitoring system based on multi-source data, as shown in some embodiments of this application. Detailed Implementation

[0055] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below in conjunction with the accompanying drawings are exemplary and are only used to explain some embodiments of this application, and should not be construed as limiting the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments shown in this application without inventive effort are within the protection scope of this application.

[0056] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, unless otherwise explicitly specified, "a plurality of" or "several" means two or more.

[0057] Existing methods for landslide risk monitoring and early warning suffer from the following problems: Current landslide monitoring generally relies on strain monitoring and soil moisture, which struggles to reflect dynamic changes within the slope and its surrounding environment, and is prone to overlooking key factors, thus affecting the accuracy of landslide risk assessment. Furthermore, existing monitoring methods primarily rely on static thresholds for risk determination, failing to incorporate the complex spatiotemporal coupling effects of the landslide process, resulting in significant monitoring and early warning delays and a high risk of misjudgment and missed detections. Therefore, improving the accuracy of landslide risk monitoring remains a pressing issue that needs to be addressed with current technology.

[0058] Based on the above technical background, please refer to Figure 1 This application provides a landslide risk monitoring method based on multi-source data, including steps S101 to S104, each step of which is as follows:

[0059] Step S101: Obtain environmental monitoring data of the slope area to be monitored, and construct a multidimensional feature matrix based on the environmental monitoring data; wherein, the environmental monitoring data includes strain data, soil displacement data, rainfall data, temperature and humidity data, wind speed data, groundwater level data and soil moisture data.

[0060] In some embodiments of this application, the soil displacement data includes first soil displacement measurement data and second soil displacement measurement data; wherein, the first soil displacement measurement data is obtained by measuring a first soil displacement sensor, and the second soil displacement measurement data is obtained by measuring a second soil displacement sensor; the measurement accuracy of the second soil displacement sensor is much greater than the measurement accuracy of the first soil displacement sensor. For example, the ratio of the measurement accuracy of the second soil displacement sensor to the measurement accuracy of the first soil displacement sensor is greater than 10, that is, when the measurement accuracy of the first soil displacement sensor is 10 mm, the measurement accuracy of the second soil displacement sensor is 1 mm.

[0061] In some embodiments of this application, the step of acquiring environmental monitoring data of the slope area to be monitored and constructing a multidimensional feature matrix based on the environmental monitoring data specifically includes:

[0062] Based on multiple monitoring devices in the slope area to be monitored, the environmental monitoring data is acquired, and noise filtering and missing value completion are performed on the environmental monitoring data to obtain the first monitoring data;

[0063] Based on Kalman filtering, the data of various types in the first monitoring data are fused to obtain the second monitoring data, and the time series features of the data of various types in the second monitoring data are extracted based on wavelet transform to obtain multiple sets of landslide time series feature data.

[0064] The temporal and spatial alignment of each group of landslide time series feature data was performed to construct a multidimensional feature matrix.

[0065] In some embodiments of this application, when obtaining second monitoring data based on first monitoring data and then obtaining landslide time-series characteristic data, wavelet packet decomposition and time series similarity analysis are performed on the second soil landslide time-series characteristic data obtained from the second soil displacement measurement data in the first monitoring data to obtain second soil landslide abrupt change characteristic data; the second soil landslide abrupt change characteristic data is used as the landslide time-series characteristic data obtained from the second soil displacement measurement data.

[0066] This application acquires environmental monitoring data through multiple monitoring devices and performs noise filtering and missing value completion, which improves the effectiveness and accuracy of the data, thereby enhancing the accuracy of subsequent landslide risk monitoring. Furthermore, it fuses various types of data based on Kalman filtering, improving the comprehensiveness of the data. Finally, it extracts temporal features based on wavelet transform and constructs a multidimensional feature matrix through spatiotemporal alignment. Compared to existing technologies, this approach integrates temporal features into the data, improving data effectiveness and thus enhancing the accuracy of subsequent landslide risk monitoring.

[0067] Step S102: Based on the preset risk classifier, perform risk prediction and classification based on the multidimensional feature matrix to obtain the landslide risk level, and based on the preset neural network model, predict the landslide displacement based on the multidimensional feature matrix.

[0068] In some embodiments of this application, the risk classifier is specifically a classification model built based on machine learning algorithms, including but not limited to random forest models and support vector machine models, with the preferred embodiment being the support vector machine model.

[0069] Specifically, landslide risk levels include high risk, medium risk, and low risk, determined by the risk probability output by the risk classifier; more specifically, when the risk probability is greater than 75%, it is determined to be high risk, when the risk probability is less than 20%, it is determined to be low risk, and the remaining probabilities are determined to be medium risk.

[0070] In some embodiments of this application, the step of predicting the landslide displacement based on the multidimensional feature matrix according to a preset neural network model specifically includes:

[0071] Based on a pre-defined convolutional neural network, features are extracted from the soil displacement data in the multidimensional feature matrix to obtain soil displacement features.

[0072] Based on a preset long short-term memory network, the soil displacement characteristics are predicted to obtain the predicted landslide displacement.

[0073] In some embodiments of this application, the convolutional neural network includes, but is not limited to, CNN and its improved models, with CNN being the preferred embodiment; the long short-term memory network includes, but is not limited to, LSTM and Bi-LSTM and their improved models, with LSTM being the preferred embodiment.

[0074] In some embodiments of this application, when making predictions using a preset long short-term memory network, the soil displacement characteristics used for prediction are obtained based on second soil displacement measurement data.

[0075] In some embodiments of this application, the predicted landslide displacement is a predicted cumulative displacement over a future preset time period, in mm / 6h.

[0076] This application uses a pre-defined convolutional neural network to extract features from soil displacement data, which can accurately obtain the displacement characteristics of the soil. Then, it uses a long short-term memory network to predict the landslide displacement, which can improve the accuracy of prediction based on the temporal features in the data, thereby improving the accuracy of subsequent landslide risk monitoring.

[0077] Step S103: Based on the limit equilibrium method, calculate the landslide stability coefficient according to the environmental monitoring data, calculate the rainfall-induced threshold by combining the rainfall data and the soil moisture data, and obtain the comprehensive landslide risk coefficient by combining the landslide stability coefficient and the rainfall-induced threshold.

[0078] In some embodiments of this application, the calculation of the landslide stability coefficient based on the environmental monitoring data using the limit equilibrium method specifically includes:

[0079] Acquire slope measurement data for the slope area to be monitored; wherein, the slope measurement data includes measurement data of the slope sliding surface and the slope sliding body;

[0080] Based on the strain data in the environmental monitoring data and the slope sliding surface measurement data in the slope measurement data, the soil anti-sliding force is calculated;

[0081] Based on the groundwater level data in the environmental monitoring data and the slope sliding surface and slope sliding body measurement data in the slope measurement data, the soil sliding force is calculated.

[0082] Based on the limit equilibrium method, the landslide stability coefficient is obtained by combining the soil resistance and the soil sliding force.

[0083] In some embodiments of this application, the landslide stability coefficient is specifically:

[0084]

[0085] Wherein, soil anti-sliding force R=c·L+σtan(φ)·L, c is soil cohesion, σ is normal stress on the slope sliding surface, φ is internal friction angle of the slope, and L is length of the slope sliding surface; soil sliding force S=W·sin(θ)+μ·L, W is weight of the slope sliding body, θ is inclination angle of the slope sliding surface, and μ is groundwater pressure on the slope sliding surface.

[0086] This application first obtains slope measurement data, then calculates soil anti-sliding force and soil sliding force by combining environmental monitoring data, and then obtains landslide stability coefficient based on the limit equilibrium method. It can accurately obtain the characterization index of landslide stability through the limit equilibrium method, thereby improving the accuracy of subsequent landslide risk monitoring based on the landslide stability coefficient.

[0087] In some embodiments of this application, calculating the rainfall-induced threshold by combining the rainfall data and the soil moisture data specifically includes:

[0088] Based on the rainfall data, determine the current and maximum rainfall, current and maximum rainfall intensity of the slope area to be monitored;

[0089] Based on the soil moisture data, determine the current soil moisture and maximum soil moisture of the slope area to be monitored;

[0090] The rainfall-induced threshold is calculated by weighting the current and maximum rainfall, current and maximum rainfall intensity, and current and maximum soil moisture in the slope area to be monitored.

[0091] In some embodiments of this application, the rainfall-induced threshold is specifically:

[0092]

[0093] Wherein, R, R max Let I be the current rainfall and the maximum rainfall. max Let H be the current rainfall intensity and the maximum rainfall intensity. max α represents the current soil moisture and the maximum soil moisture, and β and γ are weighting coefficients.

[0094] This application first determines the relevant characterization quantities of rainfall amount, rainfall intensity and soil moisture based on rainfall data and soil moisture data, and then calculates the rainfall-induced threshold by weighting. This can accurately obtain the characterization index of rainfall-induced landslides among the landslide influencing factors, and improve the accuracy of subsequent landslide risk monitoring based on rainfall-induced thresholds.

[0095] In some embodiments of this application, the comprehensive landslide risk coefficient is specifically:

[0096] P 综合 =δ·F s +(1-δ)·P c ;

[0097] Where δ represents the overall risk weight.

[0098] Step S104: Combine the landslide risk level, the predicted landslide displacement, and the comprehensive landslide risk coefficient to determine the landslide risk monitoring results for the slope area to be monitored.

[0099] In some embodiments of this application, the step of determining the landslide risk monitoring results of the slope area to be monitored by combining the landslide risk level, the predicted landslide displacement, and the comprehensive landslide risk coefficient specifically includes:

[0100] When the landslide risk level is high, the landslide risk monitoring result is determined to trigger an early warning and an early warning notice is issued;

[0101] When the landslide risk level is not high risk and the predicted landslide displacement is greater than the preset displacement threshold, the landslide risk monitoring result is determined to trigger an early warning and an early warning prompt is issued.

[0102] When the landslide risk level is not high risk, the predicted landslide displacement is not greater than the preset displacement threshold, and the comprehensive landslide risk coefficient is greater than the preset critical risk coefficient, the landslide risk monitoring result is determined to trigger an early warning and an early warning prompt is issued.

[0103] Otherwise, the landslide risk monitoring result is determined to be not triggering an early warning.

[0104] Specifically, the preset displacement threshold and the preset critical risk coefficient can be determined by historical monitoring data or by the implementation experience of those skilled in the art. The preferred preset displacement threshold is 10mm / 6h.

[0105] This application first acquires environmental monitoring data of the slope area to be monitored and constructs a multi-dimensional feature matrix, which can integrate data from different sources and of different types, providing a data foundation for subsequent data prediction. Compared with the single data source of existing technologies, multiple data sources improve the comprehensiveness of the data, thereby avoiding the adverse effects of a single data source on the accuracy of landslide risk assessment. Then, risk prediction classification and displacement prediction are performed based on a preset risk classifier and neural network model. The comprehensive landslide risk coefficient is obtained by calculating the landslide stability coefficient and rainfall-induced threshold based on the limit equilibrium method. Finally, the landslide risk monitoring results are comprehensively determined. Compared with existing technologies, by comprehensively utilizing different models for landslide risk assessment and combining different calculation indicators for auxiliary judgment, the application can avoid misjudgments and omissions caused by the static thresholds of existing technologies, thereby improving the accuracy of landslide risk monitoring.

[0106] For a method corresponding to the one described above, please refer to [link to relevant documentation]. Figure 2 The present application provides a landslide risk monitoring system based on multi-source data, including a feature matrix construction module 210, a risk prediction and classification module 220, a risk coefficient calculation module 230, and a comprehensive risk monitoring module 240.

[0107] The feature matrix construction module 210 is used to acquire environmental monitoring data of the slope area to be monitored, and construct a multidimensional feature matrix based on the environmental monitoring data; wherein, the environmental monitoring data includes strain data, soil displacement data, rainfall data, temperature and humidity data, wind speed data, groundwater level data and soil moisture data;

[0108] The risk prediction and classification module 220 is used to perform risk prediction and classification based on the multidimensional feature matrix according to the preset risk classifier to obtain the landslide risk level, and to predict the landslide displacement based on the multidimensional feature matrix according to the preset neural network model.

[0109] The risk coefficient calculation module 230 is used to calculate the landslide stability coefficient based on the limit equilibrium method and the environmental monitoring data, calculate the rainfall-induced threshold by combining the rainfall data and the soil moisture data, and obtain the comprehensive landslide risk coefficient by combining the landslide stability coefficient and the rainfall-induced threshold.

[0110] The comprehensive risk monitoring module 240 is used to determine the landslide risk monitoring results of the slope area to be monitored by combining the landslide risk level, the predicted landslide displacement, and the comprehensive landslide risk coefficient.

[0111] In some embodiments of this application, the feature matrix construction module 210 includes a monitoring data acquisition unit, a data processing and extraction unit, and a feature matrix construction unit;

[0112] The monitoring data acquisition unit is used to acquire the environmental monitoring data from multiple monitoring devices in the slope area to be monitored, and to perform noise filtering and missing value completion on the environmental monitoring data to obtain the first monitoring data.

[0113] The data processing and extraction unit is used to fuse various types of data in the first monitoring data based on Kalman filtering to obtain the second monitoring data, and to extract time-series features of various types of data in the second monitoring data based on wavelet transform to obtain multiple sets of landslide time-series feature data.

[0114] The feature matrix construction unit is used to perform spatiotemporal alignment of each group of landslide time-series feature data to construct a multidimensional feature matrix.

[0115] In some embodiments of this application, the risk prediction and classification module 220 includes a displacement feature extraction unit and a displacement feature prediction unit;

[0116] The displacement feature extraction unit is used to extract features from the soil displacement data in the multidimensional feature matrix based on a preset convolutional neural network to obtain soil displacement features.

[0117] The displacement feature prediction unit is used to predict the soil displacement features based on a preset long short-term memory network to obtain the predicted landslide displacement.

[0118] In some embodiments of this application, the risk coefficient calculation module 230 includes a measurement data acquisition unit, an anti-slip force calculation unit, a sliding force calculation unit, and a stability coefficient calculation unit;

[0119] The measurement data acquisition unit is used to acquire slope measurement data of the slope area to be monitored; wherein, the slope measurement data includes measurement data of the slope sliding surface and the slope sliding body;

[0120] The anti-sliding force calculation unit is used to calculate the soil anti-sliding force based on the strain data in the environmental monitoring data and the slope sliding surface measurement data in the slope measurement data.

[0121] The sliding force calculation unit is used to calculate the soil sliding force based on the groundwater level data in the environmental monitoring data and the measurement data of the slope sliding surface and slope sliding body in the slope measurement data.

[0122] The stability coefficient calculation unit is used to obtain the landslide stability coefficient based on the limit equilibrium method, combining the soil anti-sliding force and the soil sliding force.

[0123] In some embodiments of this application, the risk coefficient calculation module 230 includes a rainfall data determination unit, a soil moisture data determination unit, and a rainfall-induced threshold calculation unit;

[0124] The rainfall data determination unit is used to determine the current rainfall and maximum rainfall, current rainfall intensity and maximum rainfall intensity of the slope area to be monitored based on the rainfall data.

[0125] The soil moisture data determination unit is used to determine the current soil moisture and maximum soil moisture of the slope area to be monitored based on the soil moisture data.

[0126] The rainfall-induced threshold calculation unit is used to calculate the rainfall-induced threshold by combining the current and maximum rainfall, current and maximum rainfall intensity, and current and maximum soil moisture of the slope area to be monitored.

[0127] This application first acquires environmental monitoring data of the slope area to be monitored and constructs a multi-dimensional feature matrix, which can integrate data from different sources and of different types, providing a data foundation for subsequent data prediction. Compared with the single data source of existing technologies, multiple data sources improve the comprehensiveness of the data, thereby avoiding the adverse effects of a single data source on the accuracy of landslide risk assessment. Then, risk prediction classification and displacement prediction are performed based on a preset risk classifier and neural network model. The comprehensive landslide risk coefficient is obtained by calculating the landslide stability coefficient and rainfall-induced threshold based on the limit equilibrium method. Finally, the landslide risk monitoring results are comprehensively determined. Compared with existing technologies, by comprehensively utilizing different models for landslide risk assessment and combining different calculation indicators for auxiliary judgment, the application can avoid misjudgments and omissions caused by the static thresholds of existing technologies, thereby improving the accuracy of landslide risk monitoring.

[0128] It should be understood that the system provided in this application corresponds to the aforementioned method. The landslide risk monitoring system based on multi-source data provided in this application can realize the landslide risk monitoring method based on multi-source data provided in any of the embodiments of this application.

[0129] Adaptively, embodiments of this application also provide a computer device and a computer-readable storage medium.

[0130] The computer device includes: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor;

[0131] The processor executes the computer program to implement a landslide risk monitoring method based on multi-source data, as described in this application.

[0132] The computer-readable storage medium stores multiple instructions, which are adapted for loading by a processor to execute a landslide risk monitoring method based on multi-source data according to this application.

[0133] The above description represents some embodiments of this application, providing a further detailed explanation of the purpose, technical solution, and beneficial effects of this application. It should be understood that the above-described embodiments of this application should not be construed as limiting this application. In particular, any changes, modifications, equivalent substitutions, and variations made by those skilled in the art within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A landslide risk monitoring method based on multi-source data, characterized in that, include: Environmental monitoring data of the slope area to be monitored is acquired, and a multidimensional feature matrix is ​​constructed based on the environmental monitoring data; wherein, the environmental monitoring data includes strain data, soil displacement data, rainfall data, temperature and humidity data, wind speed data, groundwater level data, and soil moisture data; Based on the preset risk classifier, risk prediction and classification are performed using the multidimensional feature matrix to obtain the landslide risk level. Based on the preset neural network model, the predicted landslide displacement is then obtained. Based on the limit equilibrium method, the landslide stability coefficient is calculated according to the environmental monitoring data. Combined with the rainfall data and the soil moisture data, the rainfall-induced threshold is calculated. Finally, the comprehensive landslide risk coefficient is obtained by combining the landslide stability coefficient and the rainfall-induced threshold. By combining the landslide risk level, the predicted landslide displacement, and the comprehensive landslide risk coefficient, the landslide risk monitoring results for the slope area to be monitored are determined.

2. The landslide risk monitoring method based on multi-source data according to claim 1, characterized in that, The process of acquiring environmental monitoring data for the slope area to be monitored and constructing a multidimensional feature matrix based on the environmental monitoring data specifically includes: Based on multiple monitoring devices in the slope area to be monitored, the environmental monitoring data is acquired, and noise filtering and missing value completion are performed on the environmental monitoring data to obtain the first monitoring data; Based on Kalman filtering, the data of various types in the first monitoring data are fused to obtain the second monitoring data, and the time series features of the data of various types in the second monitoring data are extracted based on wavelet transform to obtain multiple sets of landslide time series feature data. The temporal and spatial alignment of each group of landslide time series feature data was performed to construct a multidimensional feature matrix.

3. The landslide risk monitoring method based on multi-source data according to claim 1, characterized in that, The step of predicting the landslide displacement based on the multidimensional feature matrix according to the preset neural network model specifically includes: Based on a pre-defined convolutional neural network, features are extracted from the soil displacement data in the multidimensional feature matrix to obtain soil displacement features. Based on a preset long short-term memory network, the soil displacement characteristics are predicted to obtain the predicted landslide displacement.

4. The landslide risk monitoring method based on multi-source data according to claim 1, characterized in that, The calculation of the landslide stability coefficient based on the limit equilibrium method and the environmental monitoring data specifically includes: Acquire slope measurement data for the slope area to be monitored; wherein, the slope measurement data includes measurement data of the slope sliding surface and the slope sliding body; Based on the strain data in the environmental monitoring data and the slope sliding surface measurement data in the slope measurement data, the soil anti-sliding force is calculated; Based on the groundwater level data in the environmental monitoring data and the slope sliding surface and slope sliding body measurement data in the slope measurement data, the soil sliding force is calculated. Based on the limit equilibrium method, the landslide stability coefficient is obtained by combining the soil resistance force and the soil sliding force.

5. The landslide risk monitoring method based on multi-source data according to claim 1, characterized in that, The calculation of the rainfall-induced threshold by combining the rainfall data and the soil moisture data specifically includes: Based on the rainfall data, determine the current and maximum rainfall, current and maximum rainfall intensity of the slope area to be monitored; Based on the soil moisture data, determine the current soil moisture and maximum soil moisture of the slope area to be monitored; The rainfall-induced threshold is calculated by weighting the current and maximum rainfall, current and maximum rainfall intensity, and current and maximum soil moisture in the slope area to be monitored.

6. A landslide risk monitoring system based on multi-source data, characterized in that, It includes a feature matrix construction module, a risk prediction and classification module, a risk coefficient calculation module, and a comprehensive risk monitoring module; The feature matrix construction module is used to acquire environmental monitoring data of the slope area to be monitored, and to construct a multidimensional feature matrix based on the environmental monitoring data; wherein, the environmental monitoring data includes strain data, soil displacement data, rainfall data, temperature and humidity data, wind speed data, groundwater level data and soil moisture data; The risk prediction and classification module is used to perform risk prediction and classification based on the multidimensional feature matrix according to the preset risk classifier to obtain the landslide risk level, and to predict the landslide displacement based on the multidimensional feature matrix according to the preset neural network model. The risk coefficient calculation module is used to calculate the landslide stability coefficient based on the limit equilibrium method and the environmental monitoring data, calculate the rainfall-induced threshold by combining the rainfall data and the soil moisture data, and obtain the comprehensive landslide risk coefficient by combining the landslide stability coefficient and the rainfall-induced threshold. The comprehensive risk monitoring module is used to determine the landslide risk monitoring results of the slope area to be monitored by combining the landslide risk level, the predicted landslide displacement, and the comprehensive landslide risk coefficient.

7. A landslide risk monitoring system based on multi-source data according to claim 6, characterized in that, The feature matrix construction module includes a monitoring data acquisition unit, a data processing and extraction unit, and a feature matrix construction unit. The monitoring data acquisition unit is used to acquire the environmental monitoring data from multiple monitoring devices in the slope area to be monitored, and to perform noise filtering and missing value completion on the environmental monitoring data to obtain the first monitoring data. The data processing and extraction unit is used to fuse various types of data in the first monitoring data based on Kalman filtering to obtain the second monitoring data, and to extract time-series features of various types of data in the second monitoring data based on wavelet transform to obtain multiple sets of landslide time-series feature data. The feature matrix construction unit is used to perform spatiotemporal alignment of each group of landslide time-series feature data to construct a multidimensional feature matrix.

8. A landslide risk monitoring system based on multi-source data according to claim 6, characterized in that, The risk prediction and classification module includes a displacement feature extraction unit and a displacement feature prediction unit; The displacement feature extraction unit is used to extract features from the soil displacement data in the multidimensional feature matrix based on a preset convolutional neural network to obtain soil displacement features. The displacement feature prediction unit is used to predict the soil displacement features based on a preset long short-term memory network to obtain the predicted landslide displacement.

9. A landslide risk monitoring system based on multi-source data according to claim 6, characterized in that, The risk coefficient calculation module includes a measurement data acquisition unit, an anti-slip force calculation unit, a sliding force calculation unit, and a stability coefficient calculation unit; The measurement data acquisition unit is used to acquire slope measurement data of the slope area to be monitored; wherein, the slope measurement data includes measurement data of the slope sliding surface and the slope sliding body; The anti-sliding force calculation unit is used to calculate the soil anti-sliding force based on the strain data in the environmental monitoring data and the slope sliding surface measurement data in the slope measurement data. The sliding force calculation unit is used to calculate the soil sliding force based on the groundwater level data in the environmental monitoring data and the measurement data of the slope sliding surface and slope sliding body in the slope measurement data. The stability coefficient calculation unit is used to obtain the landslide stability coefficient based on the limit equilibrium method, combining the soil anti-sliding force and the soil sliding force.

10. A landslide risk monitoring system based on multi-source data according to claim 6, characterized in that, The risk coefficient calculation module includes a rainfall data determination unit, a soil moisture data determination unit, and a rainfall-induced threshold calculation unit. The rainfall data determination unit is used to determine the current rainfall and maximum rainfall, current rainfall intensity and maximum rainfall intensity of the slope area to be monitored based on the rainfall data. The soil moisture data determination unit is used to determine the current soil moisture and maximum soil moisture of the slope area to be monitored based on the soil moisture data. The rainfall-induced threshold calculation unit is used to calculate the rainfall-induced threshold by combining the current and maximum rainfall, current and maximum rainfall intensity, and current and maximum soil moisture of the slope area to be monitored.

Images