A highway slope safety early risk coverage monitoring method and system

Abstract

This invention discloses a comprehensive monitoring method and system for early-stage risks of highway slope safety, relating to the field of slope safety monitoring technology. The method includes collecting basic slope data and constructing a three-dimensional model; based on the features of the three-dimensional model, using an improved CNN model to intelligently select monitoring equipment locations and build an adaptive networking system; collecting and fusing multi-source data, inputting it into an improved LSTM-Transformer fusion prediction model, and outputting indicators and risk prediction results; calculating the difference rate and rate of change between real-time monitored values ​​and predicted values, and combining stability coefficients to classify early warning levels and trigger response measures. This invention achieves intelligent monitoring deployment, accurate risk prediction, and differentiated early warning and response, enabling early prediction of slope risks, significantly improving the efficiency of highway slope safety management and control, and reducing the accident rate.

Description

Technical Field

[0001] This invention relates to the field of slope safety monitoring technology, and in particular to a method and system for early risk coverage monitoring of highway slope safety. Background Technology

[0002] As a crucial component of transportation infrastructure, highway slopes are susceptible to landslides and collapses due to various factors such as geological conditions, climate, and operational loads. These geological hazards severely threaten highway safety, causing significant economic losses and casualties. Currently, highway slope monitoring primarily employs the traditional "risk level-based deployment" model, which suffers from the following technical shortcomings: Low accuracy of monitoring deployment: Traditional monitoring relies heavily on manual site selection, which does not fully take into account core features such as slope topography and soil distribution, making it easy to have monitoring blind spots. In addition, unreasonable equipment deployment locations lead to high equipment wear and tear rates and unstable data transmission, making it impossible to achieve full-area coverage monitoring. Insufficient accuracy in risk prediction: Existing prediction models mostly use single time series data for fitting, ignoring the influence of three-dimensional geological features of slopes. They have weak ability to predict slope risks under the coupling of multiple factors, large prediction deviations, and cannot identify early risks in advance. The early warning and response measures are not targeted enough: the early warning triggers rely on the absolute value of a single monitoring indicator without taking into account the difference between the predicted value and the real-time value and the rate of change of the indicator. This makes it difficult to distinguish the risk level, resulting in overly crude response measures that cannot achieve accurate early warning and hierarchical control. Poor coordination throughout the entire process: The monitoring deployment, risk prediction, and early warning response are all independent of each other, lacking effective data flow and collaborative optimization mechanisms, making it impossible to form a closed-loop management and control system, which affects monitoring efficiency and risk prevention and control effectiveness.

[0003] To address these issues, there is an urgent need for a comprehensive early-stage risk monitoring method and system for highway slope safety. Summary of the Invention

[0004] To address the aforementioned issues, this application proposes a comprehensive monitoring method and system for early-stage risks to highway slope safety. This system aims to resolve the technical problems of existing slope monitoring methods, such as inaccurate deployment, low prediction accuracy, and poor early warning targeting, which prevent effective early-stage risk control.

[0005] On the one hand, this application proposes a method for early risk coverage monitoring of highway slope safety, including the following steps: S1. Collect and preprocess 3D slope data, laser scanning data, soil and rock parameters and environmental data, and construct a 3D slope model using UAV oblique photography and ground laser scanning fusion modeling technology. S2. Extract slope features based on the 3D slope model, input the slope features and monitoring requirements into the improved CNN site selection model, output the target deployment location and quantity of equipment, build an adaptive networking system, and complete the equipment installation and verification. S3. Collect real-time monitoring data from monitoring equipment, 3D model feature data, and environmental induced data; construct a multi-dimensional fusion feature vector; input the fusion feature vector into the improved LSTM-Transformer fusion prediction model; and output the predicted values ​​of monitoring indicators and slope risk level. S4. Calculate the relative difference rate between real-time monitoring values ​​and predicted values, as well as the rate of change of indicators. Combine this with the slope stability coefficient to classify the early warning level and trigger corresponding graded response measures.

[0006] Preferably, the expression for the point cloud density of the laser scanning data in S1 is: ; in, Here, N is the point cloud density, N is the total number of scanned points, and S is the area of ​​the scanned region, representing the slope surface. 50 Hidden areas 80 ; In the construction of the 3D model of the slope in S1, CloudCompare software was used to fuse imagery and laser data to generate a 3D model that includes the distribution of soil and rock mass, slope, and crack locations. Weak areas were automatically marked, and the slope gradient was calculated. The expression is: ; in, The vertical elevation difference of the slope. This is the horizontal length of the slope; when At that time, it was marked as a high, steep, and weak area.

[0007] Preferably, the specific content of S2 includes: The slope features include Topographic slope (used to identify steep and vulnerable areas); Rock and soil stability (Related soil and rock parameters and slope safety factor); Concealment (This corresponds to the monitoring needs of concealed areas such as the foot of a slope). Transportation accessibility (Ensuring ease of equipment installation and maintenance); Signal transmission conditions (Adapts to adaptive networking to ensure stable data transmission); Slope integrity (Based on 3D modeling and laser scanning technology, it reflects potential hazards such as slope cracks and heaves, and refers to conventional characteristics of slope monitoring.) Rock and soil types (Parameters such as soil cohesion and internal friction angle affect slope stability and are key monitoring points); Impact on the surrounding environment (This relates to factors such as highway operating load and extreme weather, reflecting the impact of external factors on monitoring deployment.) Constructing a feature dataset based on slope characteristics Among them, terrain slope, soil and rock stability, concealment, accessibility, and signal transmission conditions are the core weighted features, and the weight allocation expression is as follows: ; in, The weights of the i-th type of features are... Let i be the influence index of the i-th type of feature on site selection, satisfying Core feature weights ; The objective function of the improved CNN location selection model is: ; Where L is the equipment loss rate (%), and D is the monitoring blind zone area (m²). 2 ), where T is the data transmission delay (s); , , The weighting coefficients for equipment loss rate, monitoring blind zone area, and data transmission delay are respectively. =0.3, =0.4, =0.3); The improved CNN location selection model introduces an attention mechanism, outputting the target deployment location coordinates (x, y, z) and deployment quantity N for each device, while also labeling the device protection level and installation angle.

[0008] Preferably, the improved LSTM-Transformer fusion prediction model introduces a Transformer self-attention mechanism in the LSTM hidden layer, and the expression for the attention weights is: ; Where Q is the query matrix, K is the key matrix, and V is the value matrix, all derived from the fusion vector of monitoring data and three-dimensional features. The dimension is K; The improved LSTM-Transformer fusion prediction model adds a three-dimensional feature fusion layer, which transforms the three-dimensional model features into feature vectors. , and monitoring data vector Environmental data vector The fusion expression is: ; in, , , These are the fusion weights for the feature vector, monitoring data vector, and environmental data vector, respectively. ; The improved LSTM-Transformer fusion prediction model introduces an adaptive error correction module to correct prediction bias in real time. The correction expression is as follows: ; in, This is the corrected predicted value. These are the initial predicted values. This is the historical true value. The correction factor is 0.05 ≤ ≤0.15).

[0009] Preferably, the three-dimensional model feature data consists of terrain slope and soil stability parameters extracted from the three-dimensional model. The expression for evaluating the stability of soil and rock masses is as follows: ; Where A is the area of ​​the sliding surface (m²) 2 W is the weight of the slope soil and rock mass, and c is the cohesion. It is the internal friction angle; Environmentally induced data includes historical extreme weather data and highway operating load data.

[0010] Preferably, the improved LSTM-Transformer fusion prediction model uses a two-dimensional prediction output to provide accurate basis for differential early warning, including indicator prediction output and output indicators and optimization methods as follows: The indicator forecast output is used to output the future predicted values ​​of each monitoring indicator. Mark the prediction deviation The expression for the rate of change is: ; in, For the prediction time interval; The indicator prediction output combines indicator prediction with three-dimensional features to output a stability level (stable, potential risk, relatively high risk, high risk), indicate the credibility of the prediction, and predict the development trend of the risk.

[0011] The preferred expression for the relative difference rate is: ; in, For real-time monitoring values, This is a forecast value for the same period; The rate of change of the indicator serves as an auxiliary early warning basis, and the expression for the rate of change of the indicator is: ; The principle for determining differences is as follows: like and If the slope is stable, then it is determined to be indifferent. For normal rates of change, such as displacement ; like and If so, it is determined to be a slight difference, indicating a slight slope anomaly; like and If the difference is moderate, it indicates a potential risk to the slope. like and If the slope is significantly different, it is considered a severe difference, increasing the risk of slope displacement. .

[0012] On the other hand, this application proposes an early risk coverage monitoring system for highway slope safety, comprising: Data acquisition and processing unit: Collects and preprocesses 3D slope data, laser scan data, soil and rock parameters and environmental data, and constructs a 3D slope model using UAV oblique photography and ground laser scanning fusion modeling technology; Equipment deployment unit: Based on the 3D model of the slope, the slope features are extracted, the slope features and monitoring requirements are input into the improved CNN site selection model, the target deployment location and quantity of equipment are output, an adaptive networking system is built, and the equipment installation and verification are completed; Indicator prediction unit: Collects real-time monitoring data from monitoring equipment, 3D model feature data, and environmental induced data, constructs a multi-dimensional fusion feature vector, inputs the fusion feature vector into the improved LSTM-Transformer fusion prediction model, and outputs the predicted values ​​of monitoring indicators and slope risk levels; Monitoring and Judgment Unit: Calculates the relative difference rate between real-time monitoring values ​​and predicted values, as well as the rate of change of indicators, and classifies early warning levels based on slope stability coefficients, triggering corresponding graded response measures.

[0013] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor invokes the computer program in the memory to implement the content of the method for early risk coverage monitoring of highway slope safety.

[0014] A storage medium storing computer-executable instructions, which, when loaded and executed by a processor, implement the content of an early risk coverage monitoring method for highway slope safety.

[0015] In summary, the early-stage risk monitoring method and system for highway slope safety of the present invention has the following advantages compared with traditional technologies: 1. By adopting 3D modeling, improved CNN intelligent site selection, and adaptive networking technology, combined with the 3D features of the slope and multi-factor optimization of equipment deployment, we can ensure that there are no blind spots in the entire area, reduce equipment wear rate, reduce data transmission latency, and significantly improve deployment accuracy and reliability. 2. An improved LSTM-Transformer fusion prediction model was constructed, incorporating three-dimensional geological features and multi-source data. An attention mechanism and an adaptive error correction module were introduced to improve prediction accuracy and predict early slope risks in advance, thus solving the shortcomings of traditional models that ignore geological features and have large prediction biases. 3. Using the difference between real-time monitored values ​​and predicted values ​​as the core triggering condition, and combining the rate of change and stability coefficient, four levels of early warning are divided, and graded handling is implemented to avoid crude handling measures and improve the pertinence and efficiency of risk prevention and control.

[0016] The technical method of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of a method for early risk monitoring of highway slope safety using a comprehensive monitoring approach. Figure 2 This is a unit diagram of an early risk coverage monitoring system for highway slope safety according to the present invention. Detailed Implementation

[0018] The technical method of the present invention will be further described below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0019] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0020] Techniques, systems, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the instruction manual.

[0021] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0022] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0023] Example 1 This application proposes a comprehensive monitoring method for early-stage safety risks of highway slopes, such as... Figure 1 As shown, it includes the following steps: S1. Collect 3D slope data, laser scanning data, soil and rock parameters (cohesion, internal friction angle, etc.) and environmental data. After noise reduction and anomaly removal of the collected data, use UAV oblique photography and ground laser scanning fusion modeling technology to construct a 1:1 3D refined model of the slope. Preferably, the expression for the point cloud density of the laser scanning data in S1 is: ; in, Point cloud density (points / m) 2 N is the total number of scan points, and S is the area of ​​the scanned region (m²). 2 ), slope surface 50 Hidden areas 80 To ensure that no terrain information is missed; In the construction of the 3D model of the slope in S1, CloudCompare software was used to fuse imagery and laser data to generate a 3D model that includes the distribution of soil and rock mass, slope, and crack locations. Weak areas were automatically marked, and the slope gradient was calculated. The expression is: ; in, The vertical elevation difference of the slope (m) is the vertical difference between the slope and the elevation of the slope. The horizontal length of the slope is (m). when At that time, areas marked as high, steep, and vulnerable should be prioritized for the deployment of monitoring equipment.

[0024] S2. Intelligent deployment and networking of monitoring equipment: Based on the three-dimensional model of the slope, the slope features are extracted, the slope features and monitoring requirements are input into the improved CNN site selection model, the target deployment location and quantity of equipment are output, an adaptive networking system is built, and the equipment installation and verification are completed. Preferably, the specific content of S2 includes: The slope features include Topographic slope (used to identify steep and vulnerable areas); Rock and soil stability (Related soil and rock parameters and slope safety factor); Concealment (This corresponds to the monitoring needs of concealed areas such as the foot of a slope). Transportation accessibility (Ensuring ease of equipment installation and maintenance); Signal transmission conditions (Adapts to adaptive networking to ensure stable data transmission); Slope integrity (Based on 3D modeling and laser scanning technology, it reflects potential hazards such as slope cracks and heaves, and refers to conventional characteristics of slope monitoring.) Rock and soil types (Parameters such as soil cohesion and internal friction angle affect slope stability and are key monitoring points); Impact on the surrounding environment (This relates to factors such as highway operating load and extreme weather, reflecting the impact of external factors on monitoring deployment.) Constructing a feature dataset based on slope characteristics Among them, terrain slope, soil and rock stability, concealment, accessibility, and signal transmission conditions are the core weighted features, and the weight allocation expression is as follows: ; in, The weights of the i-th type of features are... Let i be the influence index of the i-th type of feature on location selection (with a value of 0-1), satisfying Core feature weights ; The improved CNN location selection model aims to achieve the highest monitoring coverage efficiency, lowest equipment loss, and most stable data transmission. The objective function is: ; Where L is the equipment loss rate (%), and D is the monitoring blind zone area (m²). 2 ), where T is the data transmission delay (s); , , The weighting coefficients for equipment loss rate, monitoring blind zone area, and data transmission delay are respectively. =0.3, =0.4, =0.3); An improved CNN site selection model introduces an attention mechanism to output the target deployment location coordinates (x, y, z) and deployment quantity N of each device. It also labels the protection level and installation angle of the devices, installs monitoring equipment according to the site selection scheme, and builds an adaptive networking system of "5G combined with Beidou and Mesh".

[0025] The system debugs equipment operation and data transmission, verifies monitoring coverage based on 3D model, deploys supplementary points for blind spots, calibrates equipment accuracy, starts system trial operation, monitors equipment operation status during trial operation, and optimizes deployment location and networking method through model iteration.

[0026] S3. Collect real-time monitoring data (deformation, seepage, stress, etc.) from the monitoring equipment, feature data of the three-dimensional model in S1, environmental induced data (extreme weather, operational load), and support structure data. After preprocessing (normalization and denoising), construct a multi-dimensional fusion feature vector. Input the fusion feature vector into the improved LSTM-Transformer fusion prediction model to output the predicted values ​​of monitoring indicators and slope risk level. Preferably, the improved LSTM-Transformer fusion prediction model introduces a Transformer self-attention mechanism in the LSTM hidden layer, and the expression for the attention weights is: ; Where Q is the query matrix, K is the key matrix, and V is the value matrix, all derived from the fusion vector of monitoring data and three-dimensional features. The dimension is K; The improved LSTM-Transformer fusion prediction model adds a three-dimensional feature fusion layer, which transforms the three-dimensional model features into feature vectors. , and monitoring data vector Environmental data vector The fusion expression is: ; in, , , These are the fusion weights for the feature vector, monitoring data vector, and environmental data vector, respectively. ; The improved LSTM-Transformer fusion prediction model introduces an adaptive error correction module to correct prediction bias in real time. The correction expression is as follows: ; in, This is the corrected predicted value. These are the initial predicted values. This is the historical true value. The correction factor is 0.05 ≤ ≤0.15).

[0027] The model outputs the future predicted values, prediction bias, rate of change, and slope risk level (stable, potential risk, relatively high risk, high risk) of each monitoring indicator, as well as the prediction credibility. The latest data is incorporated every quarter to complete the model iteration, ensuring the prediction accuracy. The trained model is then embedded into the monitoring platform to achieve fully automated prediction.

[0028] Preferably, the three-dimensional model feature data consists of terrain slope and soil stability parameters extracted from the three-dimensional model. The expression for evaluating the stability of soil and rock masses is as follows: ; Where A is the area of ​​the sliding surface (m²) 2 W is the weight of the slope soil and rock mass, and c is the cohesion. It is the internal friction angle; Environmentally induced data includes historical extreme weather data and highway operating load data.

[0029] Preferably, the improved LSTM-Transformer fusion prediction model uses a two-dimensional prediction output to provide accurate basis for differential early warning, including indicator prediction output and output indicators and optimization methods as follows: The indicator forecast output is used to output the future predicted values ​​of each monitoring indicator. Mark the prediction deviation The expression for the rate of change is: ; in, For the prediction time interval; The indicator prediction output combines indicator prediction with three-dimensional features to output a stability level (stable, potential risk, relatively high risk, high risk), indicate the credibility of the prediction, and predict the development trend of the risk.

[0030] S4. Calculate the relative difference rate between real-time monitoring values ​​and predicted values, as well as the rate of change of indicators. Combine this with the slope stability coefficient to classify the early warning level and trigger corresponding graded response measures.

[0031] The preferred expression for the relative difference rate is: ; in, For real-time monitoring values, This is a forecast value for the same period; The rate of change of the indicator serves as an auxiliary early warning basis, and the expression for the rate of change of the indicator is: ; The principle for determining differences is as follows: like and If the slope is stable, then it is determined to be indifferent. For normal rates of change, such as displacement ; like and If so, it is determined to be a slight difference, indicating a slight slope anomaly; like and If the difference is moderate, it indicates a potential risk to the slope. like and If the slope is significantly different, it is considered a severe difference, increasing the risk of slope displacement. .

[0032] The monitoring system is ensured to operate stably through a three-pronged approach: technical, managerial, and emergency response. Technical support includes establishing a professional technical team responsible for equipment deployment, calibration, maintenance, model training, and iteration; collaborating with universities and research institutions to optimize core technologies; and building a technology reserve. Managerial support includes establishing and improving monitoring management systems, clarifying job responsibilities, conducting regular personnel training, and creating full-process archives to ensure data traceability and process control. Emergency response support includes developing comprehensive emergency response plans, equipping sufficient emergency equipment and supplies, conducting regular emergency drills, and establishing a collaborative mechanism involving transportation, emergency response, and geological departments to form a unified response force. The 3D model is updated every six months, and based on monitoring data feedback, the deployment plan is iteratively optimized using an improved CNN model to adapt to slope deformation and environmental changes, achieving closed-loop management throughout the entire process.

[0033] Example 2 This application proposes a comprehensive early-stage risk monitoring system for highway slope safety, such as... Figure 2 As shown, it includes: Data acquisition and processing unit: Collects and preprocesses 3D slope data, laser scan data, soil and rock parameters and environmental data, and constructs a 3D slope model using UAV oblique photography and ground laser scanning fusion modeling technology; Equipment deployment unit: Based on the 3D model of the slope, the slope features are extracted, the slope features and monitoring requirements are input into the improved CNN site selection model, the target deployment location and quantity of equipment are output, an adaptive networking system is built, and the equipment installation and verification are completed; Indicator prediction unit: Collects real-time monitoring data from monitoring equipment, 3D model feature data, and environmental induced data, constructs a multi-dimensional fusion feature vector, inputs the fusion feature vector into the improved LSTM-Transformer fusion prediction model, and outputs the predicted values ​​of monitoring indicators and slope risk levels; Monitoring and Judgment Unit: Calculates the relative difference rate between real-time monitoring values ​​and predicted values, as well as the rate of change of indicators, and classifies early warning levels based on slope stability coefficients, triggering corresponding graded response measures.

[0034] Finally, it should be noted that the above embodiments are only used to illustrate the technical methods of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical methods of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical methods to deviate from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for early-stage risk monitoring of highway slope safety, characterized in that, Includes the following steps: S1. Collect and preprocess 3D slope data, laser scanning data, soil and rock parameters and environmental data, and construct a 3D slope model using UAV oblique photography and ground laser scanning fusion modeling technology. S2. Extract slope features based on the 3D slope model, input the slope features and monitoring requirements into the improved CNN site selection model, output the target deployment location and quantity of equipment, build an adaptive networking system, and complete the equipment installation and verification. S3. Collect real-time monitoring data from monitoring equipment, 3D model feature data, and environmental induced data; construct a multi-dimensional fusion feature vector; input the fusion feature vector into the improved LSTM-Transformer fusion prediction model; and output the predicted values ​​of monitoring indicators and slope risk level. S4. Calculate the relative difference rate between real-time monitoring values ​​and predicted values, as well as the rate of change of indicators. Combine this with the slope stability coefficient to classify the early warning level and trigger corresponding graded response measures.

2. The method for early risk coverage monitoring of highway slope safety according to claim 1, characterized in that, The expression for the point cloud density of the laser scanning data mentioned in S1 is: ； in, Here, N is the point cloud density, N is the total number of scanned points, and S is the area of ​​the scanned region, representing the slope surface. 50 Hidden areas 80 ; In the construction of the 3D model of the slope in S1, CloudCompare software was used to fuse imagery and laser data to generate a 3D model that includes the distribution of soil and rock mass, slope, and crack locations. Weak areas were automatically marked, and the slope gradient was calculated. The expression is: ； in, The vertical elevation difference of the slope. This is the horizontal length of the slope; when At that time, it was marked as a high, steep, and weak area.

3. The method for early risk coverage monitoring of highway slope safety according to claim 1, characterized in that, The specific content of S2 includes: The slope characteristics include topographic slope. Rock and soil stability Concealment Accessibility Signal transmission conditions Slope integrity Types of rock and soil Impact on the surrounding environment ; Constructing a feature dataset based on slope characteristics Among them, terrain slope, soil and rock stability, concealment, accessibility, and signal transmission conditions are the core weighted features, and the weight allocation expression is as follows: ； in, The weights of the i-th type of features are... Let i be the influence index of the i-th type of feature on site selection, satisfying Core feature weights ; The objective function of the improved CNN location selection model is: ； Where L is the equipment loss rate, D is the monitoring blind zone area, and T is the data transmission delay; , , These are the weighting coefficients for equipment loss rate, monitoring blind zone area, and data transmission delay, respectively. The improved CNN location selection model introduces an attention mechanism, outputting the target deployment location coordinates (x, y, z) and deployment quantity N for each device, while also labeling the device protection level and installation angle.

4. The method for early risk coverage monitoring of highway slope safety according to claim 1, characterized in that, The improved LSTM-Transformer fusion prediction model introduces a Transformer self-attention mechanism into the LSTM hidden layer. The expression for the attention weights is as follows: ； Where Q is the query matrix, K is the key matrix, and V is the value matrix, all derived from the fusion vector of monitoring data and three-dimensional features. The dimension is K; The improved LSTM-Transformer fusion prediction model adds a three-dimensional feature fusion layer, which transforms the three-dimensional model features into feature vectors. , and monitoring data vector Environmental data vector The fusion expression is: ； in, , , These are the fusion weights for the feature vector, monitoring data vector, and environmental data vector, respectively. ; The improved LSTM-Transformer fusion prediction model introduces an adaptive error correction module to correct prediction bias in real time. The correction expression is as follows: ； in, This is the corrected predicted value. These are the initial predicted values. This is the historical true value. This is a correction factor.

5. The method for early risk coverage monitoring of highway slope safety according to claim 1, characterized in that, The 3D model feature data includes terrain slope and soil stability parameters extracted from the 3D model. The expression for evaluating the stability of soil and rock masses is as follows: ； Where A is the sliding surface area, W is the weight of the slope soil and rock mass, and c is the cohesion. It is the internal friction angle; Environmentally induced data includes historical extreme weather data and highway operating load data.

6. The method for early risk coverage monitoring of highway slope safety according to claim 1, characterized in that, The improved LSTM-Transformer fusion prediction model adopts a two-dimensional prediction output, including indicator prediction output and risk prediction output; The indicator forecast output is used to output the future predicted values ​​of each monitoring indicator. Mark the prediction deviation The expression for the rate of change is: ； in, For the prediction time interval; The indicator prediction output is used to combine indicator prediction with three-dimensional features to output the stability level, label the prediction credibility, and predict the risk development trend.

7. The method for early risk coverage monitoring of highway slope safety according to claim 1, characterized in that, The expression for the relative difference rate is: ； in, For real-time monitoring values, This is a forecast value for the same period; The rate of change of the indicator serves as an auxiliary early warning basis, and the expression for the rate of change of the indicator is: ； The principle for determining differences is as follows: like and If the slope is stable, then it is determined to be indifferent. For normal rates of change, such as displacement ; like and If so, it is determined to be a slight difference, indicating a slight slope anomaly; like and If the difference is moderate, it indicates a potential risk to the slope. like and If the slope is significantly different, it is considered a severe difference, increasing the risk of slope displacement. .

8. A comprehensive early-stage risk monitoring system for highway slope safety, characterized in that, include: Data acquisition and processing unit: Collects and preprocesses 3D slope data, laser scan data, soil and rock parameters and environmental data, and constructs a 3D slope model using UAV oblique photography and ground laser scanning fusion modeling technology; Equipment deployment unit: Based on the 3D model of the slope, the slope features are extracted, the slope features and monitoring requirements are input into the improved CNN site selection model, the target deployment location and quantity of equipment are output, an adaptive networking system is built, and the equipment installation and verification are completed; Indicator prediction unit: Collects real-time monitoring data from monitoring equipment, 3D model feature data, and environmental induced data, constructs a multi-dimensional fusion feature vector, inputs the fusion feature vector into the improved LSTM-Transformer fusion prediction model, and outputs the predicted values ​​of monitoring indicators and slope risk levels; Monitoring and Judgment Unit: Calculates the relative difference rate between real-time monitoring values ​​and predicted values, as well as the rate of change of indicators, and classifies early warning levels based on slope stability coefficients, triggering corresponding graded response measures.

9. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor, when calling the computer program in the memory, implements the content of the early risk coverage monitoring method for highway slope safety as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores computer-executable instructions, which, when loaded and executed by a processor, implement the content of the early risk coverage monitoring method for highway slope safety as described in any one of claims 1 to 7.

Images