Landslide disaster early warning system and method

Abstract

The invention discloses a landslide disaster early warning system and method, and belongs to the technical field of geological disaster monitoring, and the method comprises the steps: carrying out the omnibearing monitoring of a target region, and collecting monitoring data; the method comprises the following steps: preprocessing collected monitoring data to obtain monitoring processing data; analyzing the monitoring processing data, and performing feature extraction to obtain a target feature vector; constructing a landslide disaster risk assessment model by adopting logistic regression and a decision tree; acquiring historical monitoring data according to a large amount of historical data and landslide cases, and training the model by using the historical monitoring data; real-time monitoring data are collected and input into the landslide disaster risk assessment model, the landslide risk level of the target area at the current moment is judged, and a corresponding early warning prompt is given; the method comprehensively considers the linear and non-linear relationship of the data, can accurately evaluate the landslide risk grade of the target area, and gives out graded early warning prompts. Compared with an existing scheme, the method has the advantage that the monitoring data source, the data processing precision and the model prediction capability are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster monitoring, and in particular to a landslide disaster early warning system and method. Background Art

[0002] Landslide refers to a natural phenomenon in which soil or rock on a slope slides downward along a certain weak surface or weak zone, either as a whole or in a dispersed manner, under the action of gravity, due to the influence of factors such as river scouring, groundwater activity, rain soaking, earthquakes and artificial slope cutting.

[0003] Traditional landslide disaster monitoring and early warning mainly rely on manual inspections and simple monitoring equipment, which has many limitations. Manual inspections are difficult to achieve real-time and continuous monitoring of large target areas, and are prone to miss some potential danger signs; and simple monitoring equipment is insufficient in terms of data collection accuracy, comprehensiveness, and data processing capabilities, and cannot meet the requirements for refined early warning of landslide disasters;

[0004] This monitoring method has limitations and it is difficult to comprehensively and accurately warn of landslide disasters. Therefore, the present invention proposes a landslide disaster warning system and method to improve the warning capability of landslide disasters and effectively reduce the losses caused by landslide disasters. Summary of the invention

[0005] The purpose of the present invention is to provide a landslide disaster early warning system and method to solve the deficiencies in the background technology.

[0006] The purpose of the present invention can be achieved by the following technical solution: A landslide disaster early warning method comprises the following steps:

[0007] Step 1: Conduct all-round monitoring of the target area and collect monitoring data;

[0008] Step 2: pre-process the collected monitoring data to obtain monitoring processing data;

[0009] Step 3: Analyze the monitored data and extract features to obtain the target feature vector;

[0010] Step 4: Use logistic regression and decision tree to build a landslide disaster risk assessment model; obtain historical monitoring data based on a large amount of historical data and landslide cases, and use the historical monitoring data to train the model;

[0011] Step 5: Collect real-time monitoring data and input it into the landslide disaster risk assessment model to determine the landslide risk level of the target area at the current moment and give corresponding early warning prompts.

[0012] Preferably, in step 1, the monitoring data includes vibration wave data, vegetation physiological parameter data, environmental microclimate data and meteorological data;

[0013] Deploy highly sensitive vibration sensors at different depths in the target area to collect vibration wave data and record vibration frequency, amplitude and timestamp;

[0014] Use professional vegetation monitoring equipment to measure the photosynthesis efficiency and stomatal conductance of the slope vegetation, and record the corresponding time to form vegetation physiological parameter data;

[0015] The temperature data, humidity data and air pressure data are collected through the temperature, humidity sensor and air pressure sensor to obtain the environmental microclimate data;

[0016] By accessing the meteorological data interface, the meteorological data of the target area including precipitation data and wind speed data are obtained.

[0017] Preferably, in step 2, preprocessing the collected monitoring data includes:

[0018] Use filtering algorithm to remove noise interference from vibration wave data and obtain filtered vibration frequency and amplitude;

[0019] Identify and correct the vegetation physiological parameter data, calculate the mean and standard deviation of each type of vegetation physiological parameter, mark the data values ​​that exceed 3 times the standard deviation as outliers, remove outliers, and ensure data reliability;

[0020] The environmental microclimate data were smoothed by the sliding average method to remove the spike noise in the data;

[0021] All monitoring data will be cleaned and aligned by time to ensure that data of each dimension can be correlated and analyzed in the same time dimension.

[0022] Preferably, in step 3, the method for obtaining the target feature vector is as follows:

[0023] Traverse the monitoring and processing data to obtain the amplitude of the vibration wave data, and use the formula Calculate the vibration energy to reflect the internal activity state of the rock layer and soil in the target area; where E ij (t) is the vibration energy, t is the time point; k is a coefficient related to the sensor characteristics and the elastic modulus of the rock layer and soil, which is determined by historical calibration test data; A ij (t) is the amplitude; i = 1, 2, ..., n, represents the depth of the rock layer, j = 1, 2, ..., m, represents different plane points;

[0024] The vibration energy at different positions and depths is integrated to calculate the average vibration energy;

[0025] The calculation formula is:

[0026]

[0027] Where E(t) represents the average vibration energy.

[0028] Preferably, in step 3, it also includes: traversing the monitoring and processing data to obtain photosynthesis efficiency and stomatal conductance, and using the formula The vegetation stability coefficient is calculated to reflect the vegetation health index of the target area; where S(t) is the vegetation stability coefficient, t is the time point; η(t) is the vegetation photosynthesis efficiency; G(t) is the stomatal conductance; w1 and w2 are the weight coefficients of vegetation photosynthesis efficiency and stomatal conductance, respectively, which are set based on expert experience and historical data analysis; η max and G max They are the historical maximum values ​​of vegetation photosynthesis efficiency and stomatal conductance, respectively.

[0029] Preferably, in step 3, it also includes: traversing the monitoring and processing data to obtain environmental microclimate data, and using the formula Calculate the pressure change rate between adjacent time points to reflect the fluctuation of pressure in a short period of time; where ΔP(t) is the pressure change rate; P(t) is the pressure data at time point t; Δt is the time interval; P(t-Δt) represents the pressure data at the time interval before time point t;

[0030] The comprehensive influencing factor of temperature data and humidity data is calculated by the formula F(t)=w3×T(t)+w4×RH(t); where F(t) is the comprehensive influencing factor; T(t) is the temperature data; RH(t) is the humidity data; w3 and w4 are empirical coefficients, which are determined through statistical analysis based on the sensitivity of climate in different regions to landslides;

[0031] Traverse the monitoring and processing data to obtain meteorological data, and directly extract precipitation data and wind speed data;

[0032] The extracted average vibration energy, vegetation stability coefficient, air pressure change rate, comprehensive influencing factors, precipitation data, and wind speed data are integrated to obtain the target feature vector.

[0033] Preferably, in step 4, the logistic regression specifically includes:

[0034] The target feature vector in the historical monitoring data is extracted as the input variable, the target model parameters are obtained through the maximum likelihood estimation method, and the probability of landslide occurrence is calculated;

[0035] The probability of landslide occurrence is expressed as:

[0036]

[0037] Where X = [E, S, ΔP, F, R, V] is the input variable, E is the average vibration energy, S is the vegetation stability coefficient, ΔP is the air pressure change rate, F is the comprehensive influencing factor, R is the precipitation data, and V is the wind speed data; Y indicates whether the landslide occurs, Y = 1 indicates occurrence, and Y = 0 indicates no occurrence; β0, β1, β2, β3, β4, β5, β6 are all model parameters to be trained; P (Y = 1X) is the probability of landslide occurrence;

[0038] The maximum likelihood estimation method includes: constructing a maximum likelihood function, taking the logarithm of the maximum likelihood function to obtain a logarithmic likelihood function; using the gradient ascent method to find the parameter value that maximizes the logarithmic likelihood function to obtain the target model parameters.

[0039] Preferably, in step 4, the decision tree specifically includes: taking the historical monitoring data as training samples and the training samples as root nodes, calculating the information gain of each feature in the first feature vector for whether the landslide occurs, selecting the feature with the largest information gain as the best feature, and performing node splitting;

[0040] For each child node after splitting, repeat the above steps and continue to select the best features for splitting until the stop condition is met, and obtain the landslide risk category label to which the sample belongs, including low risk label, medium risk label, and high risk label;

[0041] The stopping condition is to set the maximum tree depth and the leaf sample number threshold. When the number of leaf node samples is less than the leaf sample number threshold or the tree depth is greater than or equal to the maximum tree depth, the splitting is stopped.

[0042] Preferably, in step 5, the method for determining the landslide risk level of the target area at the current moment is as follows:

[0043] If the calculated landslide occurrence probability P is less than 0.3 and the decision tree predicts a low risk label, the landslide risk level of the target area at the current moment is comprehensively determined to be low risk, and no warning prompt is triggered;

[0044] If the calculated landslide occurrence probability P ≥ 0.7 and the decision tree predicts a high-risk label, the landslide risk level of the target area at the current moment is comprehensively determined to be high risk, and a red warning is issued to take prompt response measures;

[0045] In addition, if the landslide risk level in the target area is comprehensively determined to be medium risk at the current moment, an orange warning will be issued and preventive preparations will be made.

[0046] In order to solve the problem, the present invention also discloses a landslide disaster early warning system, which specifically includes:

[0047] The monitoring and collection module conducts all-round monitoring of the target area and collects monitoring data;

[0048] A preprocessing module preprocesses the collected monitoring data to obtain monitoring processing data;

[0049] The feature extraction module analyzes the monitoring and processing data and extracts features to obtain the target feature vector;

[0050] The fusion model building module uses logistic regression and decision trees to build a landslide disaster risk assessment model; historical monitoring data is obtained based on a large amount of historical data and landslide cases, and the model is trained using historical monitoring data;

[0051] The risk assessment and early warning module collects real-time monitoring data and inputs it into the landslide disaster risk assessment model to determine the landslide risk level in the target area at the current moment and give corresponding early warning prompts.

[0052] Compared with the existing solutions, the present invention has the following beneficial effects:

[0053] By conducting all-round monitoring of the target area and collecting monitoring data, as well as preprocessing the collected data, the present invention can more comprehensively and accurately grasp the relevant situation of the target area, ensure the quality of the data used for subsequent analysis, and lay a good foundation for accurate early warning.

[0054] The present invention further analyzes the monitoring and processing data and extracts the target feature vector, which helps to focus on key feature information, better grasp the key factors related to landslide disasters, and facilitate more targeted risk assessment.

[0055] The present invention adopts logistic regression and decision tree to construct a landslide disaster risk assessment model, and combines a large amount of historical data and landslide cases to train the model, so that the model can fully consider a variety of actual situations, has strong adaptability and scientificity, and effectively improves the accuracy of landslide disaster risk assessment.

[0056] The present invention utilizes a trained model, collects real-time monitoring data and inputs it into the model to determine the landslide risk level of the target area at the current moment, and gives corresponding early warning prompts, which can respond to landslide disaster risks in a timely and accurate manner, help to take corresponding preventive measures in advance, and minimize the losses caused by landslide disasters to the safety of life and property. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The present invention will be further described below in conjunction with the accompanying drawings.

[0058] Figure 1 The present invention provides a flowchart of a landslide disaster early warning method.

[0059] Figure 2This is a module block diagram of a landslide disaster early warning system proposed by the present invention. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0061] Embodiment 1, as Figure 1 As shown, a landslide disaster early warning method comprises the following steps:

[0062] Step 1: Conduct all-round monitoring of the target area and collect monitoring data; the monitoring data includes vibration wave data, vegetation physiological parameter data, environmental microclimate data and meteorological data; the target area is an area prone to landslide signs or is determined to have a high potential risk of landslide based on geological surveys;

[0063] Furthermore, highly sensitive vibration sensors are deployed at different depths in the target area to collect vibration wave data and record vibration frequency, amplitude and timestamp;

[0064] Use professional vegetation monitoring equipment to measure the photosynthesis efficiency and stomatal conductance of the slope vegetation, and record the corresponding time to form vegetation physiological parameter data;

[0065] The temperature data, humidity data and air pressure data are collected through the temperature, humidity sensor and air pressure sensor to obtain the environmental microclimate data;

[0066] By accessing the meteorological data interface, the meteorological data of the target area, including precipitation data and wind speed data, is obtained;

[0067] Step 2: pre-process the collected monitoring data to obtain monitoring processing data;

[0068] Furthermore, a filtering algorithm is used to remove noise interference from the vibration wave data, such as a Butterworth low-pass filter, to obtain the filtered vibration frequency and amplitude;

[0069] Identify and correct the vegetation physiological parameter data, calculate the mean and standard deviation of each type of vegetation physiological parameter, mark the data values ​​that exceed 3 times the standard deviation as outliers, remove outliers, and ensure data reliability;

[0070] The environmental microclimate data are smoothed by the sliding average method to remove abnormal conditions such as spike noise in the data;

[0071] All monitoring data are cleaned and aligned by time to ensure that data of each dimension can be correlated and analyzed in the same time dimension;

[0072] Step 3: Analyze the monitored data and extract features to obtain the target feature vector;

[0073] Furthermore, the amplitude of the vibration wave data is obtained by traversing the monitoring and processing data, and the formula Calculate the vibration energy to reflect the internal activity state of the rock layer and soil in the target area; where E ij (t) is the vibration energy, t is the time point; k is a coefficient related to the sensor characteristics and the elastic modulus of the rock layer and soil, which is determined by historical calibration test data; A ij (t) is the amplitude; i = 1, 2, ..., n, represents the depth of the rock layer, j = 1, 2, ..., m, represents different plane points;

[0074] The vibration energy at different positions and depths is integrated to calculate the average vibration energy;

[0075] The calculation formula is:

[0076]

[0077] Where, E(t) represents the average vibration energy;

[0078] Further, the photosynthesis efficiency and stomatal conductance are obtained by traversing the monitoring and processing data, and the formula The vegetation stability coefficient is calculated to reflect the vegetation health index of the target area; where S(t) is the vegetation stability coefficient, t is the time point; η(t) is the vegetation photosynthesis efficiency; G(t) is the stomatal conductance; w1 and w2 are the weight coefficients of vegetation photosynthesis efficiency and stomatal conductance, respectively, which are set based on expert experience and historical data analysis; η max and G max They are the historical maximum values ​​of vegetation photosynthesis efficiency and stomatal conductance, representing the relative strength of the current vegetation root system's effect on soil stabilization;

[0079] Furthermore, the environmental microclimate data is obtained by traversing the monitoring and processing data, and the formula Calculate the pressure change rate between adjacent time points to reflect the fluctuation of pressure in a short period of time; where ΔP(t) is the pressure change rate; P(t) is the pressure data at time point t; Δt is the time interval; P(t-Δt) represents the pressure data at the time interval before time point t;

[0080] The comprehensive influencing factor of temperature data and humidity data is calculated by the formula F(t)=w3×T(t)+w4×RH(t); where F(t) is the comprehensive influencing factor; T(t) is the temperature data; RH(t) is the humidity data; w3 and w4 are empirical coefficients, which are determined through statistical analysis based on the sensitivity of climate in different regions to landslides;

[0081] Furthermore, the meteorological data is obtained by traversing the monitoring and processing data, and precipitation data and wind speed data are directly extracted;

[0082] The extracted average vibration energy, vegetation stability coefficient, air pressure change rate, comprehensive influencing factors, precipitation data, and wind speed data are integrated to obtain the target feature vector.

[0083] Step 4: Use logistic regression and decision tree to build a landslide disaster risk assessment model; obtain historical monitoring data based on a large amount of historical data and landslide cases, and use the historical monitoring data to train the model;

[0084] In some preferred embodiments, the logistic regression specifically includes:

[0085] The target feature vector in the historical monitoring data is extracted as the input variable, the target model parameters are obtained through the maximum likelihood estimation method, and the probability of landslide occurrence is calculated;

[0086] The probability of landslide occurrence is expressed as:

[0087]

[0088] Where X = [E, S, ΔP, F, R, V] is the input variable, E is the average vibration energy, S is the vegetation stability coefficient, ΔP is the air pressure change rate, F is the comprehensive influencing factor, R is the precipitation data, and V is the wind speed data; Y indicates whether the landslide occurs, Y = 1 indicates occurrence, and Y = 0 indicates no occurrence; β0, β1, β2, β3, β4, β5, β6 are all model parameters to be trained; P (Y = 1X) is the probability of landslide occurrence;

[0089] Furthermore, the maximum likelihood estimation method includes:

[0090] Construct the maximum likelihood function;

[0091] The calculation formula is:

[0092]

[0093] In the formula, P(Y=1X n ) represents the probability of landslide occurrence of the nth sample; P(Y=0X n ) represents the probability of no landslide in the nth sample; Y n Indicates whether the landslide of the nth sample occurs, Y n =1 means it happens, Y n =0 means it has not happened; L(β) is the maximum likelihood function; β = {β0, β1, β2, β3, β4, β5, β6} is the model parameter;

[0094] Taking the logarithm of the maximum likelihood function, we obtain the log-likelihood function;

[0095] The gradient ascent method is used to find the parameter value that maximizes the log-likelihood function and obtain the target model parameters;

[0096] In some preferred embodiments, the decision tree specifically includes: using historical monitoring data as training samples and the training samples as root nodes, calculating the information gain of each feature in the first feature vector for whether a landslide occurs, selecting the feature with the largest information gain as the best feature, and performing node splitting;

[0097] For each child node after splitting, repeat the above steps and continue to select the best features for splitting until the stop condition is met, and obtain the landslide risk category label to which the sample belongs, including low risk label, medium risk label, and high risk label;

[0098] The stopping condition is to set the maximum tree depth and the leaf sample number threshold. When the number of leaf node samples is less than the leaf sample number threshold or the tree depth is greater than or equal to the maximum tree depth, the splitting is stopped.

[0099] Step 5: Collect real-time monitoring data and input it into the landslide disaster risk assessment model to determine the landslide risk level of the target area at the current moment and give corresponding early warning prompts; among which, the landslide risk level includes low risk, medium risk and high risk;

[0100] If the calculated landslide occurrence probability P is less than 0.3 and the decision tree predicts a low-risk label, the landslide risk level of the target area at the current moment is comprehensively determined to be low risk, and the daily monitoring frequency and status are maintained without triggering an early warning;

[0101] If the calculated landslide probability P ≥ 0.7 and the decision tree predicts a high-risk label, the landslide risk level of the target area at the current moment is comprehensively determined to be high risk, and a red warning is issued so that relevant departments and personnel in the target area can quickly take countermeasures, such as organizing personnel evacuation and setting up warning signs, to reduce the losses that may be caused by landslide disasters;

[0102] In addition, if the landslide risk level of the target area is determined to be medium at the current moment, an orange warning will be issued, monitoring efforts will be appropriately strengthened, and residents in the surrounding area will be notified to be vigilant and make certain preventive preparations;

[0103] In the steps of the present invention, by integrating the prediction results of logistic regression and decision tree, the linear and nonlinear relationships of the data are comprehensively considered, and the landslide risk of the target area at the current moment can be evaluated timely and relatively accurately, thereby improving the accuracy of early warning.

[0104] Embodiment 2, as Figure 2As shown, a landslide disaster early warning system comprises:

[0105] The monitoring and collection module conducts all-round monitoring of the target area and collects monitoring data;

[0106] A preprocessing module preprocesses the collected monitoring data to obtain monitoring processing data;

[0107] The feature extraction module analyzes the monitoring and processing data and extracts features to obtain the target feature vector;

[0108] The fusion model building module uses logistic regression and decision trees to build a landslide disaster risk assessment model; historical monitoring data is obtained based on a large amount of historical data and landslide cases, and the model is trained using historical monitoring data;

[0109] The risk assessment and early warning module collects real-time monitoring data and inputs it into the landslide disaster risk assessment model to determine the landslide risk level in the target area at the current moment and give corresponding early warning prompts.

[0110] In the several embodiments provided by the present invention, it should be understood that the disclosed system can be implemented in other ways. For example, the above-described embodiments of the invention are only illustrative, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.

[0111] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0112] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0113] It is obvious to a person skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the essential characteristics of the present invention.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A landslide disaster early warning method, characterized in that: The following steps are involved: Step 1: Conduct all-round monitoring of the target area and collect monitoring data; Step 2: pre-process the collected monitoring data to obtain monitoring processing data; Step 3: Analyze the monitored data and extract features to obtain the target feature vector; Step 4: Use logistic regression and decision tree to build a landslide disaster risk assessment model; obtain historical monitoring data based on a large amount of historical data and landslide cases, and use the historical monitoring data to train the model; Step 5: Collect real-time monitoring data and input it into the landslide disaster risk assessment model to determine the landslide risk level of the target area at the current moment and give corresponding early warning prompts.

2. A landslide disaster early warning method according to claim 1, characterized in that: In step 1, the monitoring data includes vibration wave data, vegetation physiological parameter data, environmental microclimate data and meteorological data; Deploy highly sensitive vibration sensors at different depths in the target area to collect vibration wave data and record vibration frequency, amplitude and timestamp; Use professional vegetation monitoring equipment to measure the photosynthesis efficiency and stomatal conductance of the slope vegetation, and record the corresponding time to form vegetation physiological parameter data; The temperature data, humidity data and air pressure data are collected through the temperature, humidity sensor and air pressure sensor to obtain the environmental microclimate data; By accessing the meteorological data interface, the meteorological data of the target area including precipitation data and wind speed data are obtained.

3. A landslide disaster early warning method according to claim 2, characterized in that: In step 2, preprocessing the collected monitoring data includes: Use filtering algorithm to remove noise interference from vibration wave data and obtain filtered vibration frequency and amplitude; Identify and correct the vegetation physiological parameter data, calculate the mean and standard deviation of each type of vegetation physiological parameter, mark the data values ​​that exceed 3 times the standard deviation as outliers, remove outliers, and ensure data reliability; The environmental microclimate data were smoothed by the sliding average method to remove the spike noise in the data; All monitoring data will be cleaned and aligned by time to ensure that data of each dimension can be correlated and analyzed in the same time dimension.

4. A landslide disaster early warning method according to claim 3, characterized in that: In step 3, the method for obtaining the target feature vector is as follows: Traverse the monitoring and processing data to obtain the amplitude of the vibration wave data, and use the formula Calculate the vibration energy to reflect the internal activity state of the rock layer and soil in the target area; where E ij (t) is the vibration energy, t is the time point; k is a coefficient related to the sensor characteristics and the elastic modulus of the rock layer and soil, which is determined by historical calibration test data; A ij (t) is the amplitude; i = 1, 2, ..., n, represents the depth of the rock layer, j = 1, 2, ..., m, represents different plane points; The vibration energy at different positions and depths is integrated to calculate the average vibration energy; The calculation formula is: Where E(t) represents the average vibration energy.

5. A landslide disaster early warning method according to claim 4, characterized in that: In step 3, it also includes: traversing the monitoring and processing data to obtain photosynthesis efficiency and stomatal conductance, and using the formula The vegetation stability coefficient is calculated to reflect the vegetation health index of the target area; where S(t) is the vegetation stability coefficient, t is the time point; η(t) is the vegetation photosynthesis efficiency; G(t) is the stomatal conductance; w1 and w2 are the weight coefficients of vegetation photosynthesis efficiency and stomatal conductance, respectively, which are set based on expert experience and historical data analysis; η max and G max They are the historical maximum values ​​of vegetation photosynthesis efficiency and stomatal conductance, respectively.

6. A landslide disaster early warning method according to claim 5, characterized in that: In step 3, it also includes: traversing the monitoring and processing data to obtain environmental microclimate data, and using the formula Calculate the pressure change rate between adjacent time points to reflect the fluctuation of pressure in a short period of time; where ΔP(t) is the pressure change rate; P(t) is the pressure data at time point t; Δt is the time interval; P(t-Δt) represents the pressure data at the time interval before time point t; The comprehensive influencing factor of temperature data and humidity data is calculated by the formula F(t)=w3×T(t)+w4×RH(t); where F(t) is the comprehensive influencing factor; T(t) is the temperature data; RH(t) is the humidity data; w3 and w4 are empirical coefficients, which are determined through statistical analysis based on the sensitivity of climate in different regions to landslides; Traverse the monitoring and processing data to obtain meteorological data, and directly extract precipitation data and wind speed data; The extracted average vibration energy, vegetation stability coefficient, air pressure change rate, comprehensive influencing factors, precipitation data, and wind speed data are integrated to obtain the target feature vector.

7. A landslide disaster early warning method according to claim 6, characterized in that: In step 4, the logistic regression specifically includes: The target feature vector in the historical monitoring data is extracted as the input variable, the target model parameters are obtained through the maximum likelihood estimation method, and the probability of landslide occurrence is calculated; The probability of landslide occurrence is expressed as: Where X = [E, S, ΔP, F, R, V] is the input variable, E is the average vibration energy, S is the vegetation stability coefficient, ΔP is the air pressure change rate, F is the comprehensive influencing factor, R is the precipitation data, and V is the wind speed data; Y indicates whether the landslide occurs, Y = 1 indicates occurrence, and Y = 0 indicates no occurrence; β0, β1, β2, β3, β4, β5, β6 are all model parameters to be trained; P (Y = 1X) is the probability of landslide occurrence; The maximum likelihood estimation method includes: constructing a maximum likelihood function, taking the logarithm of the maximum likelihood function to obtain a logarithmic likelihood function; using the gradient ascent method to find the parameter value that maximizes the logarithmic likelihood function to obtain the target model parameters.

8. A landslide disaster early warning method according to claim 7, characterized in that: In step 4, the decision tree specifically includes: taking the historical monitoring data as training samples and the training samples as root nodes, calculating the information gain of each feature in the first feature vector for whether the landslide occurs, selecting the feature with the largest information gain as the best feature, and performing node splitting; For each child node after splitting, repeat the above steps and continue to select the best features for splitting until the stop condition is met, and obtain the landslide risk category label to which the sample belongs, including low risk label, medium risk label, and high risk label; The stopping condition is to set the maximum tree depth and the leaf sample number threshold. When the number of leaf node samples is less than the leaf sample number threshold or the tree depth is greater than or equal to the maximum tree depth, the splitting is stopped.

9. A landslide disaster early warning method according to claim 8, characterized in that: In step 5, the method for determining the landslide risk level of the target area at the current moment is as follows: If the calculated landslide occurrence probability P is less than 0.3 and the decision tree predicts a low risk label, the landslide risk level of the target area at the current moment is comprehensively determined to be low risk, and no warning prompt is triggered; If the calculated landslide occurrence probability P ≥ 0.7 and the decision tree predicts a high-risk label, the landslide risk level of the target area at the current moment is comprehensively determined to be high risk, and a red warning is issued to take prompt response measures; In addition, if the landslide risk level in the target area is comprehensively determined to be medium risk at the current moment, an orange warning will be issued and preventive preparations will be made.

10. A landslide disaster early warning method, using a landslide disaster early warning system as claimed in any one of claims 1 to 9, characterized in that: include: The monitoring and collection module conducts all-round monitoring of the target area and collects monitoring data; A preprocessing module preprocesses the collected monitoring data to obtain monitoring processing data; The feature extraction module analyzes the monitoring and processing data and extracts features to obtain the target feature vector; The fusion model building module uses logistic regression and decision trees to build a landslide disaster risk assessment model; historical monitoring data is obtained based on a large amount of historical data and landslide cases, and the model is trained using historical monitoring data; The risk assessment and early warning module collects real-time monitoring data and inputs it into the landslide disaster risk assessment model to determine the landslide risk level in the target area at the current moment and give corresponding early warning prompts.

Images