Slope multi-parameter fusion monitoring and data analysis system

Abstract

The invention discloses a slope multi-parameter fusion monitoring and data analysis system, belongs to the technical field of slope monitoring and disaster prevention, and aims to accurately identify the slope instability risk. The system comprises a partition sensing module which divides a slope into a plurality of slope height segmentation units and collects multi-source monitoring data of each slope height segmentation unit; the data transmission module performs edge preprocessing on the multi-source monitoring data and then transmits the multi-source monitoring data to the cloud; the fusion calculation module performs anomaly elimination and feature extraction on the multi-source monitoring data subjected to edge end preprocessing, and then generates a regional instability index by combining a limit equilibrium method with a slope height correction factor; the prediction module divides a cooperative influence unit through a bormor index, and outputs a short-term overall instability probability through a long and short-term neural network; and the visual application module displays the data through a slope height layering GIS and carries out landslide risk judgment according to indexes and probabilities. According to the invention, refined monitoring and accurate short-term prediction are realized, and support is provided for landslide disaster prevention.

Description

Technical Field

[0001] This invention relates to the field of slope monitoring and disaster prevention technology, specifically to a multi-parameter fusion monitoring and data analysis system for slopes. Background Technology

[0002] In engineering fields such as transportation, water conservancy, mining, and building foundation pits, slope stability directly affects the safe operation of projects and the safety of surrounding personnel and property. Under the combined effects of their own weight, rainfall, earthquakes, and human engineering activities, slopes are prone to instability and landslides. Globally, over ten thousand traffic disruptions occur annually due to slope instability, and accidents in mining and water conservancy sectors cause substantial losses. Therefore, real-time and accurate monitoring and early warning of slopes has become a core requirement in the field of engineering safety. With technological advancements, slope monitoring has evolved from traditional manual inspections to automated and multi-parameter-based systems. Existing systems can integrate multiple sensors, such as GNSS displacement meters and rain gauges, to collect multi-source data. Some systems incorporate zonal monitoring and use the limit equilibrium method to calculate stability coefficients or achieve risk assessment through simple data thresholds, providing fundamental support for slope safety management.

[0003] However, existing slope monitoring systems still suffer from several technical bottlenecks, making it difficult to meet the needs of early identification, early warning, and early response. Existing zoning is mostly based on macro-geological or engineering functions, neglecting the crucial factor of slope height. This leads to insufficient monitoring resources in high-risk medium-to-high slope sections or data redundancy in low slope sections, resulting in unbalanced resource allocation. Multi-source data fusion remains at the data overlay level, failing to consider the amplifying effect of slope height on gravity when calculating stability coefficients, and lacking effective coupling of dynamic parameter correlations, leading to distorted instability indicators and frequent false alarms or missed alarms. Predictive models focus on long-term overall assessment, lacking short-term local prediction capabilities and failing to consider the risk transmission effect between adjacent units, resulting in delayed early warnings. Furthermore, risk assessment relies on a single threshold, lacking a dual-dimensional assessment combining indicators and probabilities. Visualization is mostly two-dimensional planar maps, not layered by slope height, making it difficult for managers to quickly locate high-risk sections and weakening decision support. These problems collectively restrict the accurate prevention and control of slope instability risks, urgently requiring targeted technological breakthroughs. Therefore, this paper proposes a monitoring and data analysis system capable of optimizing zoning based on slope height characteristics, deeply integrating multi-source data, and accurately achieving short-term collaborative prediction and visualized risk assessment. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-parameter fusion monitoring and data analysis system for slopes to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A multi-parameter fusion monitoring and data analysis system for slopes includes a zone sensing module, a data transmission module, a fusion calculation module, a prediction module, and a visualization application module. The partitioned sensing module divides the slope into n slope height segment units and collects multi-source monitoring data of each slope height segment unit through a multi-parameter sensor group deployed within the slope height segment unit. The data transmission module transmits the multi-source monitoring data of each slope height segment unit to the cloud in real time after preprocessing it at the edge terminal; The fusion computing module performs anomaly removal and feature extraction on the multi-source monitoring data received from the cloud and preprocessed at the edge. Then, it uses the limit balance method to calculate the stability coefficient of each slope height segment unit. Combined with the slope height correction factor and data-driven anomaly correction, it generates the regional instability index of each slope height segment unit. The prediction module is based on the regional instability index of each high-segment unit of the slope. It divides the slope into m co-influence units through the Moran index and combines the overall regional instability index and environmental parameters of each co-influence unit to output the short-term overall instability probability of each co-influence unit through a long short-term neural network. The visualization application module displays the short-term overall instability probability of each collaboratively affected unit and the regional instability index of each slope segment unit using a slope height-layered GIS map, and makes a landslide risk assessment based on the short-term overall instability probability and regional instability index.

[0006] Preferably, the method for dividing the slope height into segments is as follows: based on the vertical height gradient from the top to the bottom of the slope, the slope is divided into n segments with a segment interval of 5-10m. If the slope experiences more than 2 landslides within 3 years, the segment interval is shortened to 3-5m. The multi-source monitoring data includes deformation monitoring data, stress monitoring data, and environmental monitoring data. The deformation monitoring data includes GNSS displacement collected by a GNSS displacement meter, tilt data collected by a tilt sensor, and crack width collected by a crack gauge. The stress monitoring data includes earth pressure data collected by an earth pressure cell and anchor cable load collected by an anchor cable dynamometer. The environmental monitoring data includes rainfall collected by a rain gauge, groundwater level collected by a groundwater level gauge, and soil moisture content collected by a soil moisture sensor.

[0007] Preferably, the edge preprocessing specifically includes: for the multi-source monitoring data collected by each slope height segment unit, firstly, local caching is performed to prevent the loss of multi-source monitoring data due to disconnection; then, for high-frequency interference signals mixed in the multi-source monitoring data, a Kalman filter algorithm is used to remove high-frequency noise, thereby improving the effectiveness of the multi-source monitoring data; subsequently, time synchronization of the multi-source monitoring data of all slope height segment units is achieved through BeiDou satellite time synchronization technology, ensuring that the time base of the multi-source monitoring data of each slope height segment unit is consistent; finally, the LZ4 compression algorithm is used to compress the multi-source monitoring data of each slope height segment unit after BeiDou time synchronization, reducing the bandwidth occupation for subsequent transmission to the cloud; the real-time transmission specifically includes: according to the network coverage conditions at the slope site, one of 5G network, 4G network, or LoRa wireless communication technology is selected for transmitting the multi-source monitoring data after the above edge preprocessing; and before transmitting the multi-source monitoring data after edge preprocessing, each edge preprocessed multi-source monitoring data is appended with the slope height segment unit identifier of its corresponding slope height segment unit and the timestamp after BeiDou synchronization.

[0008] Preferably, the anomaly removal involves: preprocessing the deformation detection data and stress detection data at the edge end using... Quantile method is used to remove anomalous mutation points, and environmental data preprocessed at the edge is then processed using... The criteria remove abnormal fluctuation points, and the missing values ​​of the removed abnormal mutation points and abnormal fluctuation points are filled by linear interpolation. The feature extraction specifically involves extracting deformation features, stress features, and environmental features from the multi-source monitoring data after anomaly removal. The deformation features include GNSS displacement rate, tilt angle change rate, and crack width propagation rate. The stress features include peak earth pressure and anchor cable load variance. The environmental features include daily rainfall intensity, daily groundwater level rise slope, and daily average increase in soil moisture content. The peak earth pressure is the maximum value of the earth pressure data within 24 hours of the day, and the daily rainfall intensity is 1 / 24th of the daily rainfall.

[0009] Preferably, the method for obtaining the regional instability index specifically includes the following steps: S31. Calculation of Stability Coefficient and Slope Height Correction Factor S311. Geometric features and hydrological conditions of each high-segment unit of the slope are obtained in advance through geological surveys, and mechanical properties of each high-segment unit of the slope are obtained in advance through indoor geotechnical tests. Among them, geometric features include the unit soil weight, the unit bottom inclination angle and the unit width, hydrological conditions are the unit pore water pressure, and mechanical properties include the soil internal friction angle and soil cohesion. S312. The stability coefficient of each slope height segment element is calculated using the Bishop method in the limit equilibrium method. The Bishop method is as follows: ; in, It is the stability coefficient. It is the weight of a single soil unit. It is the inclination angle of the unit's bottom surface. It is the unit pore water pressure. It is the unit width. It is the internal friction angle of the soil. It is the soil cohesion; S313. Substitute the average slope height of each slope height segment unit and the overall average slope height of the slope into the slope height correction factor formula to calculate the slope height correction factor for each slope height segment unit. The average slope height is the vertical height difference between the top of the slope and the toe of the slope height segment unit, and the overall average slope height is the average of the average slope heights of all slope height segment units. The slope height correction factor formula is: ; in, It is the slope height correction factor. It is the average slope height. It is the overall average slope height; S32. Input the deformation characteristics, stress characteristics, and environmental characteristics of each high-segment unit of the slope into the pre-trained random forest regression model for processing, and output the data-driven anomaly correction value of each high-segment unit of the slope. The pre-trained random forest regression model is obtained by training the random forest regression model with no less than 1,000 sets of historical landslide event samples of the same type of slope as labels. The historical landslide event samples are the deformation characteristics, stress characteristics, and environmental characteristics of the slope that occurred one week before the landslide or the deformation characteristics, stress characteristics, and environmental characteristics of any one week of the slope that has not experienced a landslide. S33. Substitute the stability coefficient, slope height correction factor and data-driven anomaly correction value of each slope height segment unit into the regional instability index formula to calculate the regional instability index of each slope height segment unit. The formula for the regional instability index is: ; in, It is a regional instability indicator, and the value range of the regional instability indicator is 0-10. It is a data-driven anomaly correction value. It is the design safety factor, and the design safety factor is taken according to the "Technical Specification for Building Slope Engineering" GB51210-2016, which is generally 1.3.

[0010] Preferably, the specific method for dividing the collaborative influence unit is as follows: First, the geographic coordinates of each slope height segment unit are obtained through oblique photography by UAV or total station surveying. These geographic coordinates include X and Y plane coordinates and the slope height Z value. Then, the Delaunay triangulation algorithm is used to process the geographic coordinates of all slope height segment units to construct a spatial adjacency graph. Each node in this spatial adjacency graph corresponds to a slope height segment unit, and the lines between nodes represent that two corresponding slope height segment units have a spatial adjacency relationship. Based on this spatial adjacency graph, all adjacent slope height segment units of each slope height segment unit are determined. Based on this, the Moran index of each slope height segment unit and each adjacent slope height segment unit is calculated using the Moran index formula. Slope height segment units with Moran index > 0.7 are selected and merged into a co-influence unit until all slope height segment units are determined. At this point, the slope is re-divided into m co-influence units, where m ≤ n. The formula for the Moran index is: ; Where n is the total number of slope height segmentation units, and They are the first The slope height segment unit and the adjacent first Regional instability indicators for high-slope segmented units This represents the average regional instability index for all slope height segmentation units. , It is the first The set of all adjacent slope height segmented units of a given slope height segmented unit. It is the Moran Index.

[0011] Preferably, the overall regional instability index is the weighted average of the regional instability indices of each slope height segment unit within the collaborative influence unit, and the weight of each slope height segment unit is the proportion of the unit area of ​​each slope height segment unit to the unit area of ​​the collaborative influence unit; the environmental parameters include the highest groundwater level within the unit, the average rainfall intensity within the unit, and the average soil moisture content within the unit. The highest groundwater level within the unit is the maximum value among the groundwater levels monitored by groundwater level gauges in each slope height segment unit within the collaborative influence unit. The average rainfall intensity within the unit is the arithmetic mean of the rainfall collected by rain gauges in all slope height segment units within the collaborative influence unit, divided by 24. The average soil moisture content within the unit is the arithmetic mean of the soil moisture content collected by soil moisture sensors in all slope height segment units within the collaborative influence unit.

[0012] Preferably, the probability value of the short-term overall instability is in the range of 0-1. The short-term overall instability probability of the co-influencing unit is obtained by inputting the overall regional instability index of the co-influencing unit and the corresponding environmental parameters into a Long Short-Term Memory (LSTM) neural network for the next 0-24 hours. The closer the probability value of the short-term overall instability is to 1, the higher the short-term instability risk of the co-influencing unit. The LSTM is a pre-trained LSTM, and its specific calculation logic for the short-term overall instability probability is as follows: The overall regional instability index of the co-influencing unit The highest groundwater level within the unit Unit average rainfall intensity Average soil moisture content per unit A total of 4 feature parameters are used as the input vector. The input vector is processed by a hidden layer consisting of two LSTM layers. The first LSTM layer has 64 units, an activation function of tanh, and a dropout rate of 0.2. The second LSTM layer has 32 units, an activation function of tanh, and a dropout rate of 0.2. A fully connected layer maps the hidden layer output to a single probability value with a sigmoid activation function. The output is the short-term overall instability probability. The calculation formula is: ; in, This is the network weight matrix. For bias vectors, These are the feature outputs of two LSTM layers, respectively. The sigmoid function is used to map the hidden layer outputs to the [0,1] interval.

[0013] Preferably, the slope height layered GIS map displays the regional instability index of each slope height segment unit by layering and overlaying slope height segments. At the same time, the slope height segment unit area contained in the collaborative influence unit is covered with a semi-transparent color block, and it supports clicking to view the slope height segment distribution and its corresponding short-term overall instability probability within the collaborative influence unit. The slope height segment refers to the slope height segment unit divided according to the vertical height gradient from the top of the slope to the bottom of the slope, and each slope height segment corresponds to a fixed vertical height range.

[0014] Preferably, the rule for landslide risk assessment is as follows: When the regional instability index of a single slope segment unit is ≥4 or the short-term overall instability probability of a single synergistic influence unit is ≥0.3, the landslide risk is judged to be low risk. When the regional instability index of a single slope segment unit is ≥6 or the short-term overall instability probability of a single synergistic influence unit is ≥0.5, the landslide risk is judged to be medium risk. When the regional instability index of a single slope segment unit is ≥8 or the short-term overall instability probability of a single synergistic influence unit is ≥0.7, the landslide risk is judged to be high risk. When the short-term overall instability probability of a single synergistic influence unit is ≥0.9, the landslide risk is judged to be extremely high.

[0015] Due to the adoption of the above technical solution, the technical progress achieved by this invention compared to the prior art is as follows: 1. This invention solves the problem of unbalanced monitoring resource allocation caused by the lack of slope height-specific zoning in existing systems by using an innovative slope height segmentation method. Existing systems are mostly based on macro-geological or engineering functional zoning, without considering the stability differences of different slope height segments; while this invention divides the basic monitoring units according to the vertical height gradient from the slope top to the slope toe, further densifies the segmentation of high-risk slopes with frequent historical landslides, and deploys multi-parameter sensor groups in each unit to accurately collect data, so that monitoring resources are focused on the risk characteristics of different slope height segments, achieving refined monitoring and rational resource allocation.

[0016] 2. This invention improves the efficiency of multi-source data utilization and the accuracy of regional instability indicators through a multi-stage collaborative data fusion and indicator calculation system, solving the problems of crude data fusion and distorted instability indicators in existing systems. This invention first ensures data integrity and validity through edge-end preprocessing, then selectively removes abnormal data, fills in missing values, and extracts key features. Finally, it combines stability coefficients calculated by the physical model, correction factors that reflect the influence of slope height, and data-driven correction values ​​to generate instability indicators. This not only reflects the physical stability of the slope but also corrects for abnormal disturbances, significantly reducing false alarms and missed alarms.

[0017] 3. This invention overcomes the bottlenecks of existing systems, which lack short-term local prediction capabilities and fail to consider risk transmission, by combining a prediction scheme that integrates collaborative influence unit segmentation with long-term and short-term neural networks. This invention first constructs spatial adjacency relationships based on geographic coordinates, then combines correlation to screen and segment collaborative influence units. Subsequently, it inputs the overall unit indicators and environmental parameters into a pre-trained model, outputting the short-term instability probability, accurately capturing short-term local instability trends and risk correlations, and solving the problem of delayed early warning.

[0018] 4. This invention enhances the intuitiveness of risk display and the scientific rigor of risk assessment by employing slope-height layered GIS visualization and dual-dimensional, multi-level risk judgment rules, thus addressing the problems of unintuitive visualization and weak decision support in existing systems. This invention displays data using a slope-height layered map, supporting the viewing of risk details within a unit. Simultaneously, it classifies multiple risk levels based on both instability indicators and instability probabilities, helping managers quickly locate high-risk areas and enhancing decision support capabilities. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0020] Figure 1 This is a schematic diagram of the system functional modules of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Examples, such as Figure 1 The aforementioned slope multi-parameter fusion monitoring and data analysis system includes a zone sensing module, a data transmission module, a fusion calculation module, a prediction module, and a visualization application module, which work together to predict and determine slope landslide risks.

[0023] The partitioned sensing module divides the slope into n slope height segment units and collects multi-source monitoring data of each slope height segment unit through a multi-parameter sensor group deployed within the slope height segment unit. The data transmission module transmits the multi-source monitoring data of each slope height segment unit to the cloud in real time after preprocessing it at the edge terminal; The fusion computing module performs anomaly removal and feature extraction on the multi-source monitoring data received from the cloud and preprocessed at the edge. Then, it uses the limit balance method to calculate the stability coefficient of each slope height segment unit. Combined with the slope height correction factor and data-driven anomaly correction, it generates the regional instability index of each slope height segment unit. The prediction module is based on the regional instability index of each high-segment unit of the slope. It divides the slope into m co-influence units through the Moran index and combines the overall regional instability index and environmental parameters of each co-influence unit to output the short-term overall instability probability of each co-influence unit through a long short-term neural network. The visualization application module displays the short-term overall instability probability of each collaboratively affected unit and the regional instability index of each slope segment unit using a slope height-layered GIS map, and makes a landslide risk assessment based on the short-term overall instability probability and regional instability index.

[0024] Furthermore, the working principle of the present invention will be illustrated below through embodiments: This embodiment takes the rock and soil mixed slope of a mountain expressway from K12+300 to K12+500 as the application object. The vertical height from the top to the bottom of the slope is about 45m, and there have been two small landslides caused by rainfall in the past 3 years. According to the slope height segmentation unit division rules, the slope height is divided into 9 segments based on the vertical height gradient from the top to the bottom of the slope, with 5m as the base. The fixed vertical height range and specific geological characteristics of each slope height segmentation unit are shown in Table 1 below. Table 1: Slope Height Segmentation Unit Division Table , Meanwhile, a multi-parameter sensor group is deployed in each slope height segment unit: for deformation monitoring, each slope height segment unit is equipped with 1 GNSS displacement meter, 1 tilt sensor and 1 crack gauge; for stress monitoring, each slope height segment unit is equipped with 1 earth pressure cell and 1 anchor cable force gauge; for environmental monitoring, each slope height segment unit is equipped with 1 rain gauge, 1 groundwater level gauge and 1 soil moisture sensor, and all sensors are connected to the corresponding unit's local data acquisition terminal to realize real-time acquisition and temporary storage of multi-source monitoring data.

[0025] Based on the data transmission module, the local data acquisition terminal of each unit first performs edge preprocessing on the multi-source monitoring data. The multi-source monitoring data is locally cached for 72 hours using a 32GB SD card in the format of "unit number-sensor type-time" to prevent data loss. A Kalman filter algorithm with a filtering gain of 0.6 is used to remove high-frequency noise with a frequency >0.1Hz, such as GNSS displacement and tilt data. A unified timestamp with an error ≤10ms is added to the multi-source monitoring data with the help of the built-in Beidou positioning module. Then, the LZ4 algorithm is used to compress the multi-source monitoring data with the unified timestamp to reduce the transmission bandwidth usage, with a compression ratio of approximately 1:4. Since the slope area is covered by a 5G network, the multi-source monitoring data after edge preprocessing is transmitted to the highway management center cloud, which includes a fusion computing module, a prediction module, and a visualization application module, through a 5G module that supports SA standalone networking. Before transmission, a slope height segment unit identifier and a Beidou timestamp are added to each edge preprocessed multi-source monitoring data. At the same time, the breakpoint resume function is enabled. When the network is interrupted, the data is temporarily stored. After the network is restored, the data during the interruption is transmitted first. The transmission frequency is consistent with the sensor acquisition frequency.

[0026] The highway management center's cloud platform processes the pre-processed multi-source monitoring data from the edge end according to the rules of the fusion computing module, and calculates the quartiles Q1, Q3, and interquartile range of the pre-processed deformation and stress detection data from the edge end. Remove those smaller than or greater than The abnormal mutation points are used to calculate the mean of the environmental data after preprocessing at the edge. with standard deviation Remove those smaller than or greater than For abnormal fluctuation points, missing values ​​were calculated and filled using linear interpolation, with the two valid data points before and after the missing point as the baseline. Based on this, deformation characteristics, stress characteristics, and environmental characteristics were extracted from the multi-source monitoring data after anomaly removal. Deformation characteristics include GNSS displacement rate (e.g., in Unit 3, the GNSS displacement increased from 2.1 mm to 2.7 mm in a certain period, with a time interval of 1 hour and a rate of 0.6 mm / h), dip angle change rate (e.g., in Unit 5, the dip angle increased from 0.2° to 0.3° in a time interval of 2 hours and a change rate of 0.05° / h), and crack width propagation rate (e.g., in Unit 7, the crack width increased from 1.2 mm to 1.5 mm in a time interval of 3 hours and a propagation rate of 0.1 mm / h). Stress characteristics include peak earth pressure (e.g., the maximum earth pressure data for the day in Unit 7 is 28 kPa) and anchor cable load. The variance of the load (e.g., the mean of the anchor cable load data for Unit 3 on that day is 150kN, and the variance is 25kN²); environmental characteristics include daily rainfall intensity (e.g., the total rainfall for Unit 1 on that day is 24mm, and the daily rainfall intensity is 1mm / h), the daily rise slope of the groundwater level (e.g., the groundwater level for Unit 7 rises from 2.1m to 2.7m on that day, over a 24-hour time interval, with a rise slope of 0.025m / h), and the daily average increase in soil moisture content (e.g., the average soil moisture content for Unit 5 on that day is 22%, the average for the previous day is 20%, and the daily average increase is 2%). Based on this, the calculation of regional instability indicators is initiated. First, the geometric characteristics and hydrological conditions of each unit are obtained through geological surveys, and mechanical properties are obtained through indoor geotechnical tests. The stability coefficient is calculated using the Bishop method, combined with the average slope height of each unit and the overall average slope height of 657.5m, according to the formula... Calculate the slope height correction factor; then input the real-time features of each unit into a pre-trained random forest regression model (trained with historical landslide event samples of 1200 groups of similar slopes, R²=0.88), outputting data-driven anomaly correction values; finally, follow the formula... Calculated regional instability indicators (e.g., Unit 5) Unit 7 ).

[0027] According to the requirements of the prediction module, the highway management center first obtained the geographic coordinates of the center points of 9 slope height segment units through total station surveying. Then, it constructed a spatial adjacency diagram using the Delaunay triangulation algorithm. Based on this, it calculated the correlation between adjacent units using the Moran's index formula, substituting the regional instability index of each unit (e.g., Moran's index for units 2 and 3 is 0.78, and for units 7 and 8 it is 0.82). Adjacent units with a Moran's index > 0.7 were merged, ultimately dividing the region into 6 co-influence units (e.g., co-influence unit B consists of units 2, 3, and 4, and co-influence unit E consists of units 7 and 8). Subsequently, it calculated the overall regional instability index for each co-influence unit. Taking co-influence unit E as an example, unit 7 has an area of ​​80 m² and a weight of 0.44, and unit 8 has an area of ​​100 m² and a weight of 0.56. The overall regional instability index is... Simultaneously, environmental parameters including the highest groundwater level, average rainfall intensity, and average soil moisture content within the co-influence unit are obtained. The overall regional instability index and environmental parameters are input into a pre-trained LSTM model (500 sets of samples for training, 6-dimensional input layer, 2-dimensional hidden layer, and 1-dimensional output layer) to output the short-term overall instability probability in the next 0-24 hours (e.g., the short-term overall instability probability of co-influence unit E = 0.25, and the short-term overall instability probability of co-influence unit B = 0.18).

[0028] The visualization application module uses ArcGIS Pro to create a layered GIS map of slope height, divided into three layers according to vertical height: upper (670m-680m), middle (655m-670m), and lower (635m-655m). The transparency of each layer is 60%, 50%, and 40% respectively. Each slope height segment unit is color-coded according to the regional instability index. blue, (e.g., green), the co-influence units are covered with semi-transparent blocks and labeled with their numbers and short-term instability probabilities. Mouse clicks are supported to view details of the co-influence units and to trace back the short-term instability probability over the past 72 hours. Risk assessment is based on a dual dimension of regional instability indicators and short-term instability probabilities, classifying risks into four levels: low, medium, high, and extremely high. In this embodiment, all slope height segmented units are initially... Furthermore, the short-term instability probability of all collaboratively affected units is less than 0.3, indicating a low risk, and routine data tracking is initiated. If the short-term instability probability of collaboratively affected unit E rises to 0.52 after rainfall, and the slope height segmentation unit... If the value is 6.1, it is considered a medium risk and a manual inspection will be triggered within 24 hours.

[0029] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A slope multi-parameter fusion monitoring and data analysis system, characterized in that, The application relates to a slope stability prediction method and system. The application comprises: a partition perception module, which is used for dividing a slope into n slope height segmentation units and collecting multi-source monitoring data of each slope height segmentation unit through a multi-parameter sensor group arranged in the slope height segmentation unit; a data transmission module, which is used for transmitting the multi-source monitoring data of each slope height segmentation unit to a cloud end in real time after the multi-source monitoring data is preprocessed by an edge end; a fusion calculation module, which is used for calculating the stability coefficient of each slope height segmentation unit by using an extreme limit balance method after the multi-source monitoring data received by the cloud end is preprocessed by the edge end, and generating a regional instability index of each slope height segmentation unit by combining the stability coefficient and data-driven abnormal correction with a slope height correction factor; a prediction module, which is used for dividing the slope into m cooperative influence units by using a Moran index based on the regional instability index of each slope height segmentation unit, and outputting a short-term overall instability probability of each cooperative influence unit by using a long short-term neural network in combination with the overall regional instability index and environmental parameters of each cooperative influence unit; 2.The slope multi-parameter fusion monitoring and data analysis system according to claim 1, characterized in that, a visual application module, which is used for displaying the short-term overall instability probability of each cooperative influence unit and the regional instability index of each slope height segmentation unit on a slope height layered GIS map, and judging the landslide risk based on the short-term overall instability probability and the regional instability index.

3. The slope multi-parameter fusion monitoring and data analysis system according to claim 2, characterized in that, The slope height segmentation unit is divided according to the vertical height gradient of the slope from the slope top to the slope foot, and the slope is divided into n slope height segmentation units at a segmentation interval of 5-10 m; if the number of landslides occurring in the slope in three years is greater than 2, the segmentation interval is shortened to 3-5 m; the multi-source monitoring data comprises deformation monitoring data, stress monitoring data and environmental monitoring data, wherein the deformation monitoring data comprises GNSS displacement, inclination data and crack width, the stress monitoring data comprises soil pressure data and anchor cable dynamometer load, and the environmental monitoring data comprises rainfall, underground water level and soil moisture content.

4. The slope multi-parameter fusion monitoring and data analysis system according to claim 3, characterized in that, The abnormality rejection is: adopting IQR quantile method to reject abnormal mutation points for the edge-end preprocessed deformation detection data and force detection data, adopting criteria to reject abnormal fluctuation points, and filling in missing values of the rejected abnormal mutation points and abnormal fluctuation points through linear interpolation method. The edge end preprocessing comprises local caching, high-frequency noise elimination based on Kalman filtering, Beidou timing synchronization and LZ4 algorithm compression of the multi-source monitoring data; the real-time transmission adopts one of 5G network, 4G network and LoRa wireless communication to transmit the multi-source monitoring data preprocessed by the edge end, and adds corresponding slope height segmentation unit identification and time stamp to the multi-source monitoring data preprocessed by the edge end.

5. The slope multi-parameter fusion monitoring and data analysis system according to claim 4, characterized in that, The feature extraction comprises deformation features including GNSS displacement rate, inclination change rate and crack width expansion rate, stress features including soil pressure peak value and anchor cable dynamometer load variance, and environmental features including daily rainfall intensity, underground water level daily rising slope and soil moisture content daily average increment. The method for obtaining the regional instability index comprises the following steps: S31. Based on the geometric characteristics, hydrological state and mechanical properties of each slope height segmentation unit obtained through geological exploration and indoor test, the stability coefficient of each slope height segmentation unit is calculated by using the limit equilibrium method, and the average slope height of each slope height segmentation unit and the overall average slope height of the slope are brought into a slope height correction factor formula to calculate the slope height correction factor of each slope height segmentation unit. S32. The deformation characteristics, stress characteristics and environmental characteristics of each slope height segmentation unit are input into a pre-trained random forest regression model for processing, and the data-driven abnormal correction value of each slope height segmentation unit is output; S33. The stability coefficient, slope height correction factor and data-driven abnormal correction value of each slope height segmentation unit are brought into the regional instability index formula to obtain the regional instability index of each slope height segmentation unit.

6. The slope multi-parameter fusion monitoring and data analysis system according to claim 5, characterized in that, The division method of the synergistic influence unit is: A spatial adjacency graph is constructed based on the geographical coordinates of the slope height segmentation unit using the Delaunay triangulation algorithm. Based on the spatial adjacency graph, all slope height segmentation units adjacent to the position of each slope height segmentation unit are determined. Based on the regional instability index of the slope height segmentation unit, the Moran index of each slope height segmentation unit and its adjacent slope height segmentation unit is calculated. The slope height segmentation unit with a Moran index > 0.7 is screened and merged into a synergistic influence unit. Accordingly, the slope is re-divided into m synergistic influence units, where m≤n.

7. The slope multi-parameter fusion monitoring and data analysis system according to claim 6, characterized in that, The overall regional instability index is the weighted average of the regional instability indexes of the slope height segmentation units in the synergistic influence unit, and the weight of each slope height segmentation unit is the proportion of the unit area of each slope height segmentation unit to the unit area of the synergistic influence unit. The environmental parameters include the highest groundwater level in the unit, the average rainfall intensity in the unit, and the average soil moisture content in the unit.

8. The slope multi-parameter fusion monitoring and data analysis system according to claim 7, characterized in that, The slope height layer GIS map displays the regional instability indexes of each slope height segmentation unit by layering and superimposing the slope height segmentation units according to the slope height layer. The synergistic influence unit covers the slope height segmentation unit area it contains with a semi-transparent block, and supports clicking to view the slope height segment distribution in the synergistic influence unit and its corresponding short-term overall instability probability.

9. The slope multi-parameter fusion monitoring and data analysis system of claim 8, wherein, The rules for landslide risk judgment are: When the regional instability index of a single slope height segmentation unit is ≥4 or the short-term overall instability probability of a single synergistic influence unit is ≥0.3, the landslide risk is judged to be low risk; When the regional instability index of a single slope height segmentation unit is ≥6 or the short-term overall instability probability of a single synergistic influence unit is ≥0.5, the landslide risk is judged to be medium risk; When the regional instability index of a single slope height segmentation unit is ≥8 or the short-term overall instability probability of a single synergistic influence unit is ≥0.7, the landslide risk is judged to be high risk; When the short-term overall instability probability of a single synergistic influence unit is ≥0.9, the landslide risk is judged to be extremely high risk.

Images