Wireless star chain real-time early warning data processing method and system for mountain landslide

Abstract

The present application relates to the technical field of data processing, in particular to a wireless satellite chain real-time early warning data processing method and system for landslides. It includes the following steps: collecting data affecting the occurrence of landslides, using historical data to learn time series patterns, selecting key time series features from different time windows, and finally assigning weights to each key time series feature; according to the real-time data affecting the occurrence of landslides, combined with the selected key time series features and the weights of the key time series features, a prediction model is generated, and using fuzzy reasoning rules, the probability is combined with the key time series features to evaluate the risk level in real time. In the wireless satellite chain real-time early warning data processing method and system for landslides, the most valuable features for landslide prediction can be more effectively selected, the correlation between these features is strengthened by calculating the interaction information between them, thereby improving the accuracy of the landslide prediction model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a wireless satellite chain real-time early warning data processing method and system for landslides. BACKGROUND

[0002] In earthquake-prone areas and areas with concentrated rainfall, landslide disasters pose a serious threat to people's life and property safety. Extreme weather events caused by climate change, such as heavy rainfall and rainstorms, exacerbate soil moisture levels, leading to reduced stability of mountains. In earthquake-prone areas, the impact of seismic vibrations on soil pressure is also an important factor in landslides. Earthquakes not only directly trigger landslides, but also can change the existing geological balance at the critical moment of landslide occurrence, further increasing the risk of landslides.

[0003] Therefore, timely and accurate landslide early warning is crucial to reducing disaster losses and protecting people's lives and property. Landslide occurrence is usually a complex process influenced by multiple factors, including rainfall, soil moisture, seismic vibrations, etc., and the interaction of these factors can exacerbate the risk of landslides. Existing early warning systems mostly rely on single features or simple feature weighting, which has certain limitations in the face of complex landslide occurrence processes, and cannot fully capture the interaction between different features, thereby affecting the accuracy and response speed of early warning. Therefore, a wireless satellite chain real-time early warning data processing method and system for landslides are provided. SUMMARY

[0004] The purpose of the present application is to provide a wireless satellite chain real-time early warning data processing method and system for landslides to solve the problem that existing early warning systems mostly rely on single features or simple feature weighting, which has certain limitations in the face of complex landslide occurrence processes, and cannot fully capture the interaction between different features, thereby affecting the accuracy and response speed of early warning.

[0005] To achieve the above purpose, the present application provides a wireless satellite chain real-time early warning data processing method for landslides, comprising the following steps:

[0006] S1, collecting data affecting landslide occurrence through sensors arranged on the mountain, and transmitting the collected data affecting landslide occurrence to the ground base station through radio waves;

[0007] S2, the ground base station performs real-time preprocessing on the received data affecting landslide occurrence, and preliminarily judges whether each data affecting landslide occurrence exceeds its own safety threshold. If one data exceeds its own safety threshold, a first early warning signal is generated and transmitted to the early warning system through the wireless satellite chain, otherwise the prediction of landslide occurrence is performed;

[0008] S3. Use historical data on landslide occurrence stored at ground base stations to learn time-series patterns, select key time-series features from different time windows, and optimize the selection process by introducing the mutual influence of soil moisture and precipitation as well as the mutual influence of earthquake magnitude and soil pressure. Finally, assign weights to each key time-series feature.

[0009] S4. Based on real-time data affecting landslide occurrence, and combined with selected key time-series features and their weights, a prediction model is generated. This prediction model can predict the time window and probability of landslide occurrence, and uses fuzzy inference rules to combine the probability with key time-series features to assess the risk level in real time.

[0010] S5. Once the probability of a landslide occurring within the predicted landslide occurrence time window exceeds the safe probability value or the risk level exceeds the safe risk level, a second early warning signal is generated and transmitted to the early warning system via multiple channels using wireless starlink. The early warning system then issues warnings to the received first and second early warning signals through multiple channels.

[0011] As a further improvement to this technical solution, in S1, the data affecting the occurrence of landslides include geological data, meteorological data, earthquake data, and topographic data; the geological data includes at least soil moisture and soil pressure; the meteorological data includes at least precipitation and temperature; the earthquake data includes at least earthquake magnitude; and the topographic data includes at least slope angle and aspect.

[0012] As a further improvement to this technical solution, in step S3, historical data on landslide occurrences stored at ground base stations are used for time-series pattern learning, and key time-series features are selected from different time windows, as follows:

[0013] Set time window Labels related to landslides ,in, This indicates that a landslide has occurred. This indicates that no landslide occurred, and the processed data affecting landslide occurrence in S2 are used to construct a feature set. Feature set Includes soil moisture, soil pressure, precipitation, temperature, earthquake magnitude, slope angle, and slope aspect;

[0014] Calculate the label of landslide occurrence With the Features Mutual information between them:

[0015] ;

[0016] in, is mutual information; is a time step is a label of landslide occurrence is a time step is the first feature is a dynamic weighting factor is a time step is a label of landslide occurrence is information entropy is a time step is the first feature is information entropy is a time step is a label of landslide occurrence is joint information entropy ; is a length of a time window

[0017] According to the calculated mutual information values, a number of features with high ranking mutual information values are selected as key time sequence features, and the number of key time sequence features are taken as inputs of the prediction model.

[0018] As a further improvement of the technical solution, in the S3, the mutual influence of soil moisture and precipitation and the mutual influence of earthquake magnitude and soil pressure are introduced in the process of selecting key time sequence features for optimization, and the optimization is specifically:

[0019] Joint information entropy of soil moisture and precipitation:

[0020] ;

[0021] wherein, is joint information entropy of soil moisture and precipitation at time is a soil moisture feature value at time is a precipitation feature value at time is a joint probability distribution of soil moisture and precipitation at time Mutual information of soil moisture and precipitation: ;

[0022] wherein, is mutual information between soil moisture and precipitation

[0023] is individual information entropy of soil moisture

[0024] ​​​​The individual information entropy of the precipitation; The joint information entropy of the soil moisture and the precipitation;

[0025] The joint information entropy of the seismic vibration and the soil pressure:

[0026]

[0027] The joint information entropy of the seismic magnitude and the soil pressure at time ; The characteristic value of the seismic magnitude at time ; The characteristic value of the soil pressure at time ; The joint probability distribution of the seismic magnitude and the soil pressure at time ; The mutual information of the seismic vibration and the soil pressure:

[0028]

[0029]

[0030] The mutual information between the seismic magnitude and the soil pressure; The individual information entropy of the seismic magnitude; The individual information entropy of the soil pressure; The joint information entropy of the seismic magnitude and the soil pressure; Key timing feature selection:

[0031]

[0032]

[0033] The optimized mutual information; The soil moisture characteristic value; The precipitation characteristic value; The seismic magnitude characteristic value; The soil pressure characteristic value; The mutual information between the label of landslide occurrence and the soil moisture and the precipitation; The mutual information between the label of landslide occurrence and the seismic magnitude and the soil pressure; The optimized mutual information value is introduced into a ranking list, and a number of features ranking at the top in the optimized ranking list are selected as key timing features, and the number of key timing features are taken as inputs of a prediction model.

[0034] As a further improvement of the technical solution, in S3, weights are assigned to each key timing feature, specifically:

[0035] ​​​​​​

[0036] ;

[0037] in, For the first Weights of key time-series features; Features and characteristics The interaction between them.

[0038] As a further improvement to this technical solution, in step S4, a prediction model is generated based on real-time data affecting landslide occurrence, combined with selected key time-series features and their weights. This prediction model can predict the time window and probability of landslide occurrence. The specific steps are as follows:

[0039] Prediction of landslide occurrence time window:

[0040] ;

[0041] in, This refers to the predicted time window for landslide occurrence; The number of key temporal features; ; Labels for landslides With the Features In time Mutual information on;

[0042] ;

[0043] ;

[0044] in, The weighted sum of all features; It is a nonlinear transformation function; For landslides within the predicted landslide occurrence time window The probability of occurrence; It is the sum of all key temporal features within the predicted landslide occurrence time window. A set of.

[0045] As a further improvement to this technical solution, in step S4, fuzzy inference rules are used to combine probability with key temporal features to assess the risk level in real time, as follows:

[0046] S41. Standardize the key characteristics and prediction probabilities that affect landslides;

[0047] S42. Select key features and prediction probabilities related to landslide occurrence as input variables for fuzzy inference and transform them into fuzzy sets;

[0048] S43, form a fuzzy rule base to describe the fuzzy relationship between input variables;

[0049] S44, through the fuzzification of input data, the application of fuzzy rules for reasoning, get the intermediate results, and then through the defuzzification to convert the results into the output value of landslide risk;

[0050] S45, based on the output of fuzzy reasoning, evaluate the risk level of landslide occurrence.

[0051] As a further improvement of the technical solution, in the S5, the early warning system carries out instant early warning on the first early warning signal, carries out early warning of different degrees according to the risk level of the second early warning signal, and generates an emergency strategy.

[0052] As a further improvement of the technical solution, in the S5, the multi-channel includes mobile phone APP, short message, email and social platform.

[0053] On the other hand, the present application provides a wireless star chain real-time early warning data processing system for landslides, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to realize the wireless star chain real-time early warning data processing method for landslides.

[0054] Compared with the prior art, the present application has the following advantages:

[0055] 1. In the wireless star chain real-time early warning data processing method and system for landslides, the interaction effect between soil humidity and precipitation, earthquake vibration and soil pressure is introduced to optimize the process of time sequence feature selection. Through comprehensive evaluation of the relationship between multiple features, the most valuable features for landslide prediction can be effectively selected. The correlation between features is strengthened by calculating the interaction information between these features, thereby improving the accuracy of the landslide prediction model.

[0056] 2. In the wireless star chain real-time early warning data processing method and system for landslides, the response capability of the early warning system is effectively improved through a two-level early warning mechanism. When the first early warning signal exceeds the safety threshold, an alarm can be quickly sent to ensure timely handling of potential landslide risks. The generation of the second early warning signal is based on the prediction of landslide occurrence probability and risk level, providing a hierarchical and dynamically adjusted early warning mechanism that can take appropriate emergency response measures under different risk levels, improving the accuracy and emergency response efficiency of the system. This hierarchical early warning mechanism ensures comprehensive identification and effective response to landslide risks, minimizing potential disaster losses. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1The overall method flowchart of the present application. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0059] Embodiment 1: Please refer to Figure 1 As shown in the figure, the embodiment provides a wireless satellite real-time early warning data processing method for landslides, including the following steps:

[0060] S1, collecting data affecting landslide occurrence through sensors arranged on the mountain, and transmitting the collected data affecting landslide occurrence to the ground base station through radio waves;

[0061] In S1, the data affecting landslide occurrence includes geological data, meteorological data, seismic data and topographic data; the geological data at least includes soil moisture and soil pressure; the meteorological data at least includes precipitation and temperature; the seismic data at least includes earthquake magnitude; and the topographic data at least includes slope angle and slope direction;

[0062] S2, the ground base station pre-processes the received data affecting landslide occurrence in real time, and data cleaning is needed to solve the problems of noise interference or data missing in the transmission process. Interpolation algorithm is used to fill in missing values, and smoothing method is used to reduce noise influence; in order to avoid the influence between different data scales, normalization or standardization processing method can be used, common methods are maximum normalization or Z-score standardization; the time of data collected by different sensors may not be completely consistent, and time series alignment algorithm is needed to synchronize all data according to the unified timestamp, and it is preliminarily judged whether each data affecting landslide occurrence exceeds the respective safety threshold value, if one data exceeds the safety threshold value, the first early warning signal is generated and transmitted to the early warning system through the wireless satellite, otherwise the landslide occurrence is predicted;

[0063] S3, using the historical data affecting landslide occurrence stored in the ground base station to learn time sequence pattern, selecting key time sequence features from different time windows, and introducing the mutual influence of soil moisture and precipitation and the mutual influence of earthquake magnitude and soil pressure in the process of selecting key time sequence features for optimization, and finally assigning weights to each key time sequence feature;

[0064] In S3, historical data on landslide occurrences stored at ground base stations are used for time-series pattern learning. Key time-series features are selected from different time windows, as follows:

[0065] Set time window Labels related to landslides ,in, This indicates that a landslide has occurred. This indicates that no landslide occurred, and the processed data affecting landslide occurrence in S2 are used to construct a feature set. Feature set Includes soil moisture, soil pressure, precipitation, temperature, earthquake magnitude, slope angle, and slope aspect;

[0066] Calculate the labels for landslide occurrence With the Features Mutual information between them:

[0067] ;

[0068] in, For mutual information; For time steps The label indicating the occurrence of the landslide; For time steps Upper One feature; The dynamic weighting factor reflects the relative importance of characteristics at different time steps for landslide prediction. It can be a function that decays over time, such as exponential decay. Or, the weights can be adaptively adjusted based on the specific scenario; For time steps Labels of landslides Information entropy; For time steps Upper Features Information entropy; For time steps The label of the landslide occurred and the first The joint information entropy of the features; ; The length of the time window;

[0069] The calculated mutual information values ​​are sorted, and the top-ranked features are selected as key time-series features. These key time-series features are then used as inputs to the prediction model.

[0070] Traditional mutual information calculation usually assumes that the relationship between each time series feature and the target label (landslide occurrence) is static, i.e. ignores the temporal dependency between features. To consider the dynamic changes of time series features and their mutual influence, we introduce a time series based dynamic mutual information adjustment mechanism. By introducing a time weighting factor, the influence of features on the target event at different time steps is dynamically adjusted, enhancing the accuracy of time series feature selection.

[0071] By introducing a dynamic weighting factor, the formula can capture the dynamic relationship between features and landslide occurrence at different time steps. For example, when a landslide is about to occur, recent meteorological, seismic and other features may have a greater impact on the landslide, while long-term data may contribute less. Therefore, by adjusting the weighting factor , we can make the model pay more attention to features that are closer in time to the occurrence of the landslide, thereby improving the accuracy of the prediction.

[0072] ;

[0073] ;

[0074] ;

[0075] wherein, is the marginal distribution of ; is the marginal distribution of ; is the joint distribution of X and ;

[0076] In S3, the mutual influence of soil moisture and precipitation and the mutual influence of earthquake magnitude and soil pressure are introduced in the process of selecting key time series features for optimization. After optimization, the specific is:

[0077] Joint information entropy of soil moisture and precipitation:

[0078] ;

[0079] wherein, is the joint information entropy of soil moisture and precipitation at time , reflecting the joint uncertainty of the two features; is the soil moisture feature value at time ; is the precipitation feature value at time ; is the joint probability distribution of soil moisture and precipitation at time ;

[0080] This formula represents the mutual information between soil moisture and precipitation, reflecting the interdependence between these two features. The greater the mutual information, the stronger the relationship between these two features, and the smaller the weaker the relationship between them.

[0081] Mutual information of soil moisture and precipitation:

[0082] ;

[0083] where, is the mutual information between soil moisture and precipitation, reflecting the interdependence between soil moisture and precipitation, the greater the stronger the relationship between them, and the smaller the weaker the relationship between them; is the individual information entropy of soil moisture, describing the randomness or uncertainty of the soil moisture feature itself; is the individual information entropy of precipitation, describing the randomness or uncertainty of the precipitation feature itself; is the joint information entropy of soil moisture and precipitation, describing the randomness or uncertainty of the combination of these two features;

[0084] Joint information entropy of seismic vibration and soil pressure:

[0085] ;

[0086] where, is the joint information entropy of earthquake magnitude and soil pressure at time ; is the characteristic value of earthquake magnitude at time ; is the characteristic value of soil pressure at time ; is the joint probability distribution of earthquake magnitude and soil pressure at time ;

[0087] This formula represents the mutual information between earthquake magnitude and soil pressure, reflecting the interdependence between these two features. The greater the value of mutual information, the stronger the relationship between earthquake magnitude and soil pressure, and the higher the predictive value.

[0088] Mutual information of seismic vibration and soil pressure:

[0089] ;

[0090] where, is the mutual information between earthquake magnitude and soil pressure; is the individual information entropy of earthquake magnitude, describing the randomness or uncertainty of the earthquake magnitude feature itself; H (S) is the individual information entropy of soil pressure, describing the randomness or uncertainty of soil pressure characteristics itself; H (S, M) is the joint information entropy of earthquake magnitude and soil pressure, describing the randomness or uncertainty of the combination of these two characteristics;

[0091] Key timing feature selection:

[0092] ;

[0093] where, is the optimized mutual information; is the soil moisture feature value; is the precipitation feature value; is the earthquake magnitude feature value; is the soil pressure feature value; is the mutual information between the landslide occurrence label and soil moisture, precipitation; is the mutual information between the landslide occurrence label and earthquake magnitude, soil pressure;

[0094] The formula represents the optimized mutual information, which is used to select key timing features. The optimized mutual information considers the mutual information relationship between landslide occurrence and each key feature, and combines the mutual information between soil moisture and precipitation, and between earthquake magnitude and soil pressure. Through comprehensive evaluation of the relationship between multiple features, the most valuable features for landslide prediction can be effectively selected, thereby improving the accuracy of the prediction model. Finally, the optimized mutual information value is used for feature ranking, and the most important features are selected as the input of the prediction model to improve the prediction performance of the model.

[0095] The optimized mutual information value is introduced into the ranking list, and the top-ranked features in the optimized ranking list are selected as key timing features, and the several key timing features are used as the input of the prediction model. The role of these formulas is to evaluate the importance of each feature and its combination in landslide prediction by calculating the joint information entropy and mutual information between different features. Through this optimization process, the features that have the greatest impact on the prediction result can be selected, and the accuracy of landslide occurrence probability and time window prediction can be further improved.

[0096] In S3, each key timing feature is assigned a weight, specifically:

[0097] ;

[0098] where, is the weight of the th key timing feature; is the feature and feature interactions between features are quantified by computing the mutual information between features, which reflects the joint influence of different features on the time of landslide occurrence. For example, the interaction between soil moisture and precipitation, or the interaction between seismic vibration and soil pressure. Features with stronger interactions are likely to be more important for predicting landslide occurrence;

[0099] By this formula, we can calculate the weight of each feature according to its independent relationship with landslide occurrence and the interaction between features . This ensures that in the subsequent prediction model, each key time series feature is weighted according to its actual influence on landslide prediction. Specifically, features with greater mutual information and stronger interactions will be given higher weights, ensuring that these features play a more important role in the prediction process, thereby improving the accuracy and robustness of the prediction model.

[0100] S4, according to the real-time data affecting landslide occurrence, combined with the selected key time series features and the weights of the key time series features to generate a prediction model, the prediction model can predict the time window of landslide occurrence and the probability of occurrence, and use fuzzy reasoning rules to combine the probability with the key time series features to assess the risk level in real time;

[0101] In S4, according to the real-time data affecting landslide occurrence, combined with the selected key time series features and the weights of the key time series features to generate a prediction model, the prediction model can predict the time window of landslide occurrence and the probability of occurrence, the specific steps are as follows:

[0102] Prediction of landslide occurrence time window:

[0103] ;

[0104] Where, is the predicted landslide occurrence time window; is the number of key time series features; ; is the label of landslide occurrence and the first feature at time , this mutual information also contains the interaction between soil moisture and precipitation and the interaction between seismic vibration and soil pressure on landslide occurrence; by weighting the mutual information of these features, we can select the time point that best reflects landslide occurrence , i.e. the predicted landslide occurrence time window. This helps to accurately locate the possible time of landslide occurrence under dynamic environmental conditions.

[0105] Selecting the feature with the maximum weighted mutual information i.e., select time points such that the prediction of landslide occurrence is strongest.

[0106] ;

[0107] ;

[0108] where, is the weighted sum of all features, representing the risk of landslide occurrence; is a nonlinear transformation function (such as the output of a deep neural network) that captures complex relationships between features; is the probability of landslide occurrence within the predicted landslide occurrence time window ; is the set of all key time-series features within the predicted landslide occurrence time window ;

[0109] Through these three steps, the model can effectively predict the time window of landslide occurrence and calculate the probability of landslide occurrence. The first step selects the time points that best predict landslide occurrence through weighted mutual information, the second step calculates the landslide risk through the nonlinear transformation of weighted features, and the final step calculates the probability of landslide occurrence through logistic regression. These steps form a complete prediction process for real-time assessment of landslide risk and provision of accurate early warning information.

[0110] In S4, the probability is combined with the key time-series features using fuzzy inference rules to assess the risk level in real time, as follows:

[0111] Since landslide occurrence is usually influenced by multiple uncertain factors, the fuzzy logic model can handle these uncertainties. The fuzzy logic model converts different risk factors (such as rainfall, soil moisture, etc.) into "likelihood" values of landslide occurrence through fuzzy processing. Using fuzzy set theory to process input data (such as precipitation, soil moisture, etc.), the risk level of landslide occurrence is obtained through fuzzy reasoning.

[0112] S41, standardize the key features affecting landslide and the prediction probability, so that different types of data can be unified in scale, facilitating the subsequent fuzzy reasoning process;

[0113] S42, select key features related to landslide occurrence (such as soil moisture, precipitation, etc.) and prediction probability as input variables for fuzzy reasoning, and convert them into fuzzy sets such as "low", "medium", "high";

[0114] S43, According to the experience of experts in the field or data analysis, a fuzzy rule base is developed to describe the fuzzy relationship between input variables, such as soil moisture and precipitation, and the probability of landslide occurrence; Common fuzzy rules are as follows:

[0115] Rule 1: If soil moisture is high and precipitation is large, the probability of landslide occurrence is high;

[0116] Rule 2: If the predicted probability is medium and the seismic vibration is strong, the probability of landslide occurrence is high;

[0117] Rule 3: If the predicted probability is low, and soil moisture and precipitation are both low, the probability of landslide occurrence is low;

[0118] Rule 4: If all features are low, the probability of landslide occurrence is low.

[0119] S44, By fuzzifying the input data, applying fuzzy rules for reasoning, and converting the results to output values of landslide risk through defuzzification;

[0120] S45, Based on the output of fuzzy reasoning, the risk level of landslide occurrence is evaluated;

[0121] S5, Once the probability of landslide occurrence in the predicted landslide occurrence time window exceeds the safe probability value or the risk level exceeds the safe risk level, a second warning signal is generated, and the second warning signal is transmitted to the warning system through multiple channels using wireless star chain, and the warning system receives the first warning signal and the second warning signal through multiple channels for warning; Multiple channels include mobile phone APP, SMS, email and social platform.

[0122] In S5, the warning system immediately warns the first warning signal, and different degrees of warning are performed according to the risk level of the second warning signal, and an emergency strategy is generated.

[0123] The setting of S2 step effectively improves the response ability of the warning system through two-level warning mechanism. When the first warning signal exceeds the safety threshold, it can quickly issue an alarm to ensure timely handling of potential landslide risks. The generation of the second warning signal is based on the prediction of landslide occurrence probability and risk level, providing a hierarchical and dynamic adjustment of the warning mechanism, which can take appropriate emergency response measures under different risk levels, improving the accuracy and emergency response efficiency of the system. This hierarchical warning mechanism ensures comprehensive identification and effective response to landslide risks, minimizing potential disaster losses.

[0124] Embodiment 2: The embodiment provides a wireless star chain real-time early warning data processing system for mountain landslide, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the wireless star chain real-time early warning data processing method for mountain landslide.

[0125] The basic principles, main features and advantages of the present application are shown and described above. Those skilled in the art should understand that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application.

Claims

1. A wireless star-chain real-time early warning data processing method for landslides, characterized in that, The method comprises the following steps: S1, collecting data affecting landslide occurrence through sensors arranged on the mountain, and transmitting the collected data affecting landslide occurrence to the ground base station through radio waves; S2, the ground base station pre-processes the received data affecting landslide occurrence in real time, and preliminarily judges whether each data affecting landslide occurrence exceeds the respective safety threshold value, if one data exceeds the safety threshold value, a first warning signal is generated and transmitted to the warning system through the wireless star chain, otherwise, the landslide occurrence is predicted; S3, time sequence pattern learning is performed on the historical data affecting landslide occurrence stored in the ground base station, key time sequence features are selected from different time windows, and the mutual influence of soil humidity and precipitation and the mutual influence of earthquake magnitude and soil pressure are introduced in the process of selecting the key time sequence features for optimization, and finally each key time sequence feature is assigned a weight; Key time sequence feature selection: ； wherein, is the optimized mutual information; is the soil moisture characteristic value; is the precipitation characteristic value; is the earthquake magnitude characteristic value; is the soil pressure characteristic value; is the mutual information between the label of landslide occurrence and the soil moisture and the precipitation; is the mutual information between the label of landslide occurrence and the earthquake magnitude and the soil pressure, are mutual information, is the length of the time window, is the dynamic reinforcement factor; The optimized mutual information value is introduced into the ranking list, and a number of features ranking at the top of the optimized ranking list are selected as key time sequence features, and the number of key time sequence features are used as inputs of the prediction model; Each key time sequence feature is assigned a weight, specifically: ； in, For the first Weights of key time-series features; Features and characteristics The interaction between them For mutual information; S4, according to the real-time data affecting landslide occurrence, in combination with the selected key time sequence features and the weights of the key time sequence features, a prediction model is generated, which can predict the time window of landslide occurrence and the probability of occurrence, and using fuzzy reasoning rules, the probability is combined with the key time sequence features to evaluate the risk level in real time; S5, once the probability of landslide occurrence in the predicted time window exceeds the safety probability value or the risk level exceeds the safety risk level, a second warning signal is generated, and the second warning signal is transmitted to the warning system through multiple channels using the wireless star chain, and the warning system receives the first warning signal and the second warning signal through multiple channels to generate a warning.

2. The wireless star-chain real-time early warning data processing method for mountain landslide according to claim 1, characterized in that: In the S1, the data affecting landslide occurrence includes geological data, meteorological data, seismic data and topographic data; the geological data at least includes soil humidity and soil pressure; the meteorological data at least includes precipitation and temperature; the seismic data at least includes earthquake magnitude; and the topographic data at least includes slope angle and slope direction.

3. The wireless star-chain real-time early warning data processing method for mountain landslide according to claim 2, characterized in that: In the S3, the historical data affecting landslide occurrence stored in the ground base station is used for time sequence pattern learning, and key time sequence features are selected from different time windows, specifically as follows: Setting a time window and landslide occurrence wherein, indicates landslide occurrence, indicates no landslide occurrence, and the processed data in S2 that affect landslide occurrence are combined into a feature set ; feature set includes soil moisture, soil pressure, precipitation, temperature, earthquake magnitude, slope angle, and slope direction; Labeling the occurrence of landslides Mutual information between the first feature ​ ； wherein, is the mutual information; is the time step is the label of landslide occurrence; is the time step is the first feature; is the dynamic weighting factor; is the time step is the information entropy of the label of landslide occurrence ; is the time step is the first feature ; is the time step is the joint information entropy of the label of landslide occurrence and the first feature; ; is the length of the time window; According to the calculated mutual information value, a number of features ranking at the top of the mutual information value are selected as key time sequence features, and the number of key time sequence features are used as inputs of the prediction model.

4. The wireless star-chain real-time early warning data processing method for mountain landslide according to claim 3, characterized in that: In the S3, the joint information entropy of soil humidity and precipitation: ； wherein H (S, R) is the joint information entropy of soil humidity and precipitation at time t; S (t) is the soil humidity characteristic value at time t; R (t) is the precipitation characteristic value at time t; P (S, R | t) is the joint probability distribution of soil humidity and precipitation at time t. The mutual information of soil humidity and precipitation: ； wherein, H(S) is the mutual information between soil moisture and precipitation; H(S) is the mutual information between soil moisture and precipitation; H(S) is the mutual information between soil moisture and precipitation; H(S) is the mutual information between soil moisture and precipitation; The joint information entropy of earthquake vibration and soil pressure: ； wherein, is the joint information entropy of the earthquake magnitude and the soil pressure at time ; is the characteristic value of the earthquake magnitude at time ; is the characteristic value of the soil pressure at time ; is the joint probability distribution of the earthquake magnitude and the soil pressure at time ; The mutual information of earthquake vibration and soil pressure: ； wherein, M is the mutual information between the earthquake magnitude and the soil pressure; H(M) is the individual information entropy of the earthquake magnitude; H(P) is the individual information entropy of the soil pressure; H(M,P) is the joint information entropy of the earthquake magnitude and the soil pressure.

5. The wireless star-chain real-time early warning data processing method for mountain landslide according to claim 4, characterized in that: In the S4, according to the real-time data affecting landslide occurrence, in combination with the selected key time sequence features and the weights of the key time sequence features, a prediction model is generated, which can predict the time window of landslide occurrence and the probability of occurrence, and the specific steps are as follows: Prediction of landslide occurrence time window: ； wherein, is the predicted landslide occurrence time window; is the number of key timing features; ; is the label of landslide occurrence with the first feature in time on mutual information; ； ； wherein, is a weighted sum of all features; is a non-linear transformation function; is the probability of a landslide occurring in the predicted landslide occurrence time window for the landslide.

6. The wireless star-chain real-time early warning data processing method for mountain landslide according to claim 5, characterized in that: In S4, the probability is combined with the key timing characteristics using fuzzy inference rules to evaluate the risk level in real time, specifically as follows: S41, standardizing the key features and prediction probability affecting the landslide; S42, selecting the key features and prediction probability related to the occurrence of landslide as the input variables of fuzzy inference, and converting them into fuzzy sets; S43, formulating a fuzzy rule base to describe the fuzzy relationship between input variables; S44, obtaining the intermediate result by fuzzifying the input data, applying fuzzy rules for inference, and converting the result to the output value of landslide risk by defuzzification; S45, based on the output of fuzzy inference, evaluating the risk level of landslide occurrence.

7. The wireless star-chain real-time early warning data processing method for mountain landslide according to claim 6, characterized in that: In S5, the early warning system immediately warns the first warning signal, and different degrees of warning are carried out according to the risk level of the second warning signal, and an emergency strategy is generated.

8. The wireless star-chain real-time early warning data processing method for mountain landslide according to claim 7, characterized in that: In S5, the multiple channels include mobile APP, SMS, email and social platform.

9. Wireless star-chain real-time early warning data processing system for landslides, comprising a memory, a processor and a computer program stored in said memory and executable on said processor, characterized in that: The processor executes a computer program to implement the steps of the wireless star chain real-time early warning data processing method for landslides as claimed in any one of claims 1-8.

Images