Highway slope deformation risk assessment system based on multi-modal data

Abstract

The present application relates to the technical field of highway slope monitoring, and discloses a highway slope deformation risk assessment system based on multi-modal data, comprising a multi-dimensional acquisition module and an intelligent assessment module. The system obtains remote sensing monitoring data of the highway network, three-dimensional surveying and mapping data of the highway slope and management data of the protection project through the multi-dimensional acquisition module, classifies and constructs a data set, the intelligent assessment module assesses the credibility of the remote sensing delineation range of each region, generates a delineation index, accurately quantifies the quality of the remote sensing delineation result, re-evaluates the accuracy of the deformation points on each highway slope, generates a calibration index, the multi-dimensional assessment has high accuracy, the intelligent assessment module analyzes the risk degree of landslide disaster caused by construction disturbance, generates a disaster score, and then combines the double threshold interval to judge the credibility level, hidden danger level and disaster risk, and trigger targeted response measures, realize the closed-loop management from data quality verification, risk level determination and emergency response, and the intelligent management has strong adaptability.

Description

Technical Field

[0001] This invention relates to the field of highway slope monitoring technology, specifically a highway slope deformation risk assessment system based on multimodal data. Background Technology

[0002] Highway slopes are artificial rock and soil structures formed by excavating mountains or filling roadbeds during road construction to ensure smooth road surfaces and meet driving safety requirements. As a crucial component of highway subgrades, they are widely distributed in mountainous and hilly terrains with complex topography. These slopes directly bear the combined effects of their own weight, traffic loads, rainwater erosion, earthquakes, and other natural and man-made factors, making them critical components of highway engineering with relatively poor stability and prone to disasters. Deformation risk assessment is of irreplaceable importance for the safe operation and sustainable development of highways. From a safety perspective, if slope deformation is not controlled in a timely manner, it may trigger landslides, collapses, and other disasters, directly threatening the lives and property of passing vehicles and pedestrians, and even causing traffic disruptions and major safety accidents. From an economic perspective, slope disasters not only lead to road damage and facility destruction, resulting in high repair costs, but also cause significant indirect economic losses due to traffic disruptions affecting regional logistics and public travel. From an operational perspective, scientific deformation risk assessment can identify potential hazards in advance, providing accurate data for slope protection and reinforcement, and routine maintenance, thereby ensuring the smooth and long-term operation of the highway network. Especially in the construction and operation of mountain expressways, deformation risk assessment is a key link in improving the overall safety level of highway projects and has important practical significance.

[0003] Currently, traditional highway slope deformation risk assessment systems based on multimodal data have significant limitations in adaptability to complex application environments, making it difficult to achieve large-scale and accurate identification of potential slope hazards along highway networks. In addition, the accuracy of 3D real-scene models acquired by drones equipped with high-definition cameras has large errors, failing to meet the actual needs of precise slope positioning and dynamic monitoring. Furthermore, the systems lack adaptability to different types of slopes, such as soil, rock, and mixed soil-rock, resulting in a high misjudgment rate in risk assessments. Moreover, the corresponding response measures lack specificity and are difficult to effectively address the differentiated risks of various slope types. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a highway slope deformation risk assessment system based on multimodal data. This system has advantages such as high accuracy in multidimensional assessment and strong adaptability in intelligent management. It solves the problems of traditional highway slope deformation risk assessment systems based on multimodal data being unable to achieve accurate identification over a wide range and having insufficient adaptability.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a highway slope deformation risk assessment system based on multimodal data, comprising a multidimensional acquisition module and an intelligent assessment module; The multidimensional acquisition module connects remote sensing satellites, drones, and soil and rock monitoring equipment to acquire remote sensing monitoring data of the highway network, three-dimensional mapping data of highway slopes, and management data of protection projects, and classifies them into remote sensing datasets, slope datasets, and protection datasets. The intelligent assessment module includes a surface identification unit, an inspection and assessment unit, a mechanical analysis unit, and a risk management unit. The surface identification unit assesses the reliability of the remote sensing delineation range for each area based on the remote sensing dataset and generates a corresponding delineation index. The inspection and evaluation unit assesses the accuracy of deformation points on each highway slope based on the slope dataset and generates corresponding calibration indices. The mechanical analysis unit analyzes the risk level of landslide disaster caused by construction disturbance based on the protection dataset and generates a corresponding disaster score. The risk management unit is configured with a fixed range of reliable thresholds. and the threshold range of hidden dangers Combined with the defined index , calibration index and disaster rating It determines the credibility level, hazard level, and landslide disaster risk, outputs the assessment results, and triggers corresponding response measures.

[0006] Preferably, the remote sensing dataset includes the remote sensing delineated area, measured area, coordinates of the geometric center point within the delineated area, coordinates of the measured geometric center point, length of the delineated boundary line, length of the measured boundary line, false delineated area, measured deformation-sensitive area, number of independent patches, delineation error area, delineation time delay, and radar backscattering coefficient for each region of the highway network.

[0007] Preferably, the slope dataset includes the type of each highway slope, the total number of point clouds, the unit normal vector, the cumulative displacement of the parallel slope, the cumulative displacement of the vertical slope, the cumulative displacement of the vertical ground plane, the number of cracks, the surveyed area, the number of gullies, the green area, the rock surface area, the specular emissivity, the specular reflection area, the diffuse reflectivity, the diffuse reflection area, and the laser echo intensity. The slope types include soil, rock, and soil-rock mixture.

[0008] Preferably, the protection dataset includes the average daily vertical settlement rate, the average daily shear deformation rate along the fracture surface, the soil-rock cohesion, and the average daily heave at the toe of the slope for each slope protection project.

[0009] Preferably, the delineation index The calculation process is as follows: S11. Based on the remote sensing dataset, extract the first [item] of the highway network. The remote sensing monitoring data of each region at the current timestamp, based on the current timestamp, will be used to... The remote sensing delineated area of ​​each region is denoted as . , will the The measured area of ​​each region is denoted as . Then calculate the first section of the highway network. The overlap rate of the delineated boundaries of each region ; S12. Based on the current timestamp, the first... The coordinates of the geometric center point within the demarcated area are: , will the The coordinates of the measured geometric center points of each region are as follows: Then calculate the first section of the highway network. Geometric center point offset error of each region ; S13. Based on the current timestamp, in the... On the demarcated and measured boundaries of each region, several inflection points are selected respectively. After connecting all the inflection points to close the line, the first... The length of the boundary line delineated for each region is denoted as . , will the The measured length of the boundary line of each region is denoted as . Then calculate the first section of the highway network. Boundary line length error of each region ; S14. Based on S11-S13, calculate the first section of the highway network. Spatial error index of each region ; S15. Based on the current timestamp, the first... The falsely delineated area resulting from misjudgment or omission in each region is denoted as . , will the The measured deformation-sensitive area of ​​each region is denoted as . Then calculate the first section of the highway network. Percentage of effective sensitive areas in each region ; S16. Based on the current timestamp, the first... The number of independent patches within the delineated area is denoted as . Then calculate the first section of the highway network. Patch density in each region ; S17. Based on the current timestamp, the first... The area of ​​delineation error caused by cloud shadows in each region is denoted as . Then calculate the first section of the highway network. The percentage of delineated error area in each region ; S18. Based on S15-S17, calculate the first section of the highway network. Effective identification index of each region ; S19. Based on the current timestamp, the first... The demarcation time for each region is recorded as follows: , will the The radar backscattering coefficient of each region is denoted as . Then, based on S11-S18, calculate the first section of the highway network. The delineation index of each region .

[0010] Preferably, the calibration index The calculation process is as follows: S21. Based on the slope dataset, extract the first... The first expressway The 3D mapping data of the slope at the current timestamp is known. This expressway belongs to the highway network. In each region, the UAV LiDAR point cloud and the satellite remote sensing terrain point cloud are spatially registered, and the two sets of point clouds are in the same coordinate system. S22. Based on the current timestamp, the first... The unit normal vector of the LiDAR point cloud of a slope from a UAV is denoted as... , This represents the total number of point clouds, and the number of points is... The unit normal vector of a slope satellite remote sensing topographic point cloud is denoted as... Then calculate the first The first expressway Mean angle residual between the slope UAV LiDAR point cloud and the satellite remote sensing terrain point cloud ; S23. Based on the current timestamp, the first... The cumulative displacement of the parallel slope surface from the LiDAR point cloud of a slope drone is denoted as: , will the The cumulative vertical slope displacement of the LiDAR point cloud from a slope drone is denoted as: , will the The cumulative vertical ground plane displacement of the LiDAR point cloud of a slope drone is denoted as: Then calculate the first The first expressway Cumulative deformation in LiDAR point cloud of a slope drone ; S24. Based on the current timestamp, the first... The number of cracks in the LiDAR point cloud of a slope drone is denoted as . , will the The area mapped from the LiDAR point cloud of a slope by a drone is denoted as Then calculate the first The first expressway Crack density of each slope ; S25. Based on the current timestamp, the first... The number of gullies in the LiDAR point cloud of a slope drone is denoted as Then calculate the first The first expressway Gullage density of individual slopes ; S26. Based on the current timestamp, the first... The green area in the LiDAR point cloud of a slope drone is denoted as... Then calculate the first The first expressway Green coverage rate of each slope ; S27. Based on the current timestamp, the first... The rock surface area in the LiDAR point cloud of a slope drone is denoted as , will the The specular reflectance of the LiDAR point cloud of a slope drone is denoted as . , will the The area of ​​the specular reflection region in the LiDAR point cloud of a slope drone is denoted as . , will the The diffuse reflectance of the LiDAR point cloud of a slope drone is denoted as . , will the The area of ​​the diffuse reflectance region in the LiDAR point cloud of a slope drone is denoted as . Then calculate the first The first expressway Average reflectivity of the rock surface on each slope ; S28. Based on the current timestamp, the first... The laser echo intensity in the LiDAR point cloud of a slope drone is denoted as Then calculate the first The first expressway Point cloud intensity dispersion of each slope ; S29. Based on S21-S28, calculate the first slope type using a weighted method. The first expressway The calibration index of each slope .

[0011] Preferably, the disaster rating The calculation process is as follows: S31. Assess the disaster risk of each slope protection project. The initial value is set to 0 points, and then the first value is extracted based on the project dataset. The first expressway Management data for individual slope protection projects; S32. Trigger the corresponding single scoring condition according to the slope type; Condition 1: No. If, during the construction of the protective engineering project, the average daily vertical settlement rate of a slope is not continuously reduced to below 0.1 mm / d over time and remains stable for 7 consecutive days, it indicates that the slope deformation has not converged, and the risk of landslide disaster is increasing. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 2: No. The slope type is rock. During the construction of the protective project, if the average daily shear deformation rate along the fracture surface does not decrease to below 0.1 mm / d over time and remains stable for 7 days, it indicates that the slope deformation has not converged, and the risk of landslide disaster is increasing. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 3: If The slope type is mixed soil and rock. During the construction of the protective engineering, if the average daily vertical settlement rate of the upper soil layer does not decrease to below 0.1 mm / d and remain stable for 7 consecutive days in a positive sequence over time, or if the average daily shear deformation rate of the lower rock layer along the fracture surface does not decrease to below 0.1 mm / d and remain stable for 7 consecutive days in a positive sequence over time, it indicates that the slope deformation has not converged, and the risk of landslide disaster has increased. The first expressway Disaster rating of individual slope protection projects Add 1 point; S33. Trigger the corresponding cumulative scoring for each slope; Condition 4: If The slope type is mixed soil and rock. During the construction of the protective engineering, the vertical settlement rate of the upper soil layer and the shear deformation rate of the lower rock layer along the fracture surface show inconsistent trends over time. This indicates that the slope deformation has not converged, and there is a disjointed effect of the protective engineering on the upper and lower parts of the slope, increasing the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 5: During the construction of the protective project, if the first The first expressway If the cohesion of the soil and rock mass on a slope is less than 30% of the design value for anti-slide, it indicates that construction disturbance has led to the failure of interparticle cohesion in the soil and rock mass, increasing the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition Six: During the construction of the protective engineering project, if the first The first expressway If the average daily uplift at the toe of a slope exceeds 2 mm / d for three consecutive days, it indicates an increase in slope sliding force, causing uplift at the toe and raising the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point.

[0012] Preferably, the trust level determination process is as follows: Let the upper limit of the confidence threshold range be denoted as The lower limit of the confidence threshold interval is denoted as ; If the highway network is number The delineation index of each region < , indicating the first The remote sensing delineation of the area has no credibility level, with a credibility rating of Level 1. Response measures include suspending the [number] [section / section / etc.]. The process for assessing slope deformation risk in a specific area involves replacing remote sensing satellites, acquiring new high spatial resolution remote sensing images, and delineating the corresponding areas. If the highway network... The delineation index of each region , indicating the first The reliability of the remote sensing delineation of the area is low, with a reliability level of 2. Response measures include full-range remote sensing image correction, filtering based on historical cloudless period images of the corresponding area, and continuing the process after correction and filtering. The process of assessing slope deformation risk in a given area, if <Highway Network No. 1> The delineation index of each region ≤ , indicating the first The reliability of the remote sensing delineation of the area is moderate, with a reliability level of 3. Response measures include local remote sensing image enhancement processing; for discrepancies between the delineated and measured areas, pixel mean interpolation is used to replace abnormal pixel values. If the highway network... The delineation index of each region > , indicating the first The remote sensing delineation of the area has a high degree of reliability, with a reliability level of 4. Response measures include storing backup data and conducting verification work to check for the presence of new cloud shadows and temporary obstructions.

[0013] Preferably, the hazard level determination process is as follows: The upper limit of the hazard threshold range is denoted as... The lower limit of the hazard threshold range is denoted as ; If the first The first expressway The calibration index of each slope < , indicating the first The first expressway The slope showed no obvious deformation risks, with a risk level of Level 1. Response measures included quarterly LiDAR drone inspections, synchronously updating point cloud data, and clearing surface soil and weeds to ensure a clear monitoring view. If the... The first expressway The calibration index of each slope , indicating the first The first expressway The slope has a minor potential deformation risk, classified as Level 2. Response measures include monthly LiDAR drone inspections, replanting soil-stabilizing vegetation, and optimizing slope drainage facilities. <No. The first expressway The calibration index of each slope ≤ , indicating the first The first expressway One slope has significant potential deformation, classified as level 3. Response measures include weekly LiDAR drone inspections, injecting sealant into cracks, adding retaining structures at the slope toe, and restricting construction activities and vehicle traffic around the slope. The first expressway The calibration index of each slope > , indicating the first The first expressway The slope has a serious potential for deformation, with a risk level of 4. The response measures include delineating a danger warning area, evacuating personnel and equipment from the danger area, activating emergency monitoring equipment to collect three-dimensional mapping data in real time, and implementing slope cutting and load reduction and anchor bolt support reinforcement projects.

[0014] Preferably, the disaster rating A score greater than 0 indicates that the risk of landslide disaster has exceeded the standard due to construction disturbance. Response measures include immediately suspending the slope protection project, conducting on-site verification, and formulating a temporary reinforcement plan.

[0015] Compared with existing technologies, this invention provides a highway slope deformation risk assessment system based on multimodal data, which has the following advantages: 1. This invention acquires remote sensing monitoring data of highway networks, 3D mapping data of highway slopes, and management data of protective engineering through a multi-dimensional acquisition module. It then classifies and constructs remote sensing datasets, slope datasets, and protective datasets, overcoming the limitations of single data sources. This provides a structured, high-quality data foundation for subsequent intelligent assessment, avoiding assessment biases caused by data clutter, and significantly improving data utilization efficiency. It lays a solid data foundation for the accuracy and comprehensiveness of slope deformation risk assessment. Based on the remote sensing dataset, the intelligent assessment module evaluates the reliability of the remote sensing delineation range for each area and generates a corresponding delineation index. Taking into account core dimensions such as spatial error and effective identification, the quality of remote sensing delineation results is accurately quantified. Then, based on the slope dataset, the accuracy of deformation points on each highway slope is evaluated, and corresponding calibration indices are generated. The multidimensional assessment has high accuracy.

[0016] 2. This invention analyzes the risk level of landslide disasters caused by construction disturbances through an intelligent assessment module, and generates a corresponding disaster score. By combining dual threshold ranges, it enables the scientific determination of credibility level, hazard level and disaster risk, and triggers targeted response measures. It realizes closed-loop management of the entire process from data quality verification to risk level determination and emergency response, which greatly improves the scientificity and timeliness of risk assessment, provides reliable decision-making basis for slope safety management, reinforcement and protection and emergency response, effectively reduces the losses caused by landslide disasters, and has strong intelligent management adaptability. Attached Figure Description

[0017] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0018] 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, and 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.

[0019] Example Please see Figure 1Table 1 shows the experimental data of the delineation index, Table 2 shows the experimental data of the calibration index, and Table 3 shows the experimental data of the disaster scoring. This invention provides a highway slope deformation risk assessment system based on multimodal data, including a multidimensional acquisition module and an intelligent assessment module. The multidimensional acquisition module connects remote sensing satellites, drones, and soil and rock monitoring equipment to acquire remote sensing monitoring data of the highway network, three-dimensional mapping data of highway slopes, and management data of protection projects, and classifies them into remote sensing datasets, slope datasets, and protection datasets. The remote sensing dataset includes the remote sensing delineated area, measured area, coordinates of the geometric center point within the delineated area, coordinates of the measured geometric center point, length of the delineated boundary line, length of the measured boundary line, false delineated area, measured deformation-sensitive area, number of independent patches, delineation error area, delineation time delay, and radar backscattering coefficient for each region of the highway network. The slope dataset includes the type of each highway slope, the total number of point clouds, the unit normal vector, the cumulative displacement of the parallel slope, the cumulative displacement of the vertical slope, the cumulative displacement of the vertical ground plane, the number of cracks, the surveyed area, the number of gullies, the green area, the rock surface area, the specular reflectivity, the area of ​​the specular reflection area, the diffuse reflectivity, the area of ​​the diffuse reflection area, and the laser echo intensity. The slope types include soil, rock, and soil-rock mixture. The protection dataset includes the average daily vertical settlement rate, the average daily shear deformation rate along the fracture surface, the soil-rock cohesion, and the average daily heave at the toe of the slope for each slope protection project. The intelligent assessment module includes a surface identification unit, an inspection and assessment unit, a mechanical analysis unit, and a risk management unit. The surface identification unit assesses the reliability of the remote sensing delineation range for each area based on the remote sensing dataset and generates a corresponding delineation index. The calculation process is as follows: S11. Based on the remote sensing dataset, extract the first [item] of the highway network. The remote sensing monitoring data of each region at the current timestamp, based on the current timestamp, will be used to... The remote sensing delineated area of ​​each region is denoted as . , will the The measured area of ​​each region is denoted as . Then calculate the first section of the highway network. The overlap rate of the delineated boundaries of each region Its expression is as follows: Specifically, the ratio of the absolute difference to the sum of the areas is a normalized value of the absolute difference, ensuring that the result is not affected by the order of the areas. This facilitates horizontal comparisons between different regions and helps determine the boundary overlap rate. The closer it is to 1, the greater the deviation between the remotely sensed area and the measured area, and the lower the degree of boundary overlap. S12. Based on the current timestamp, the first... The coordinates of the geometric center point within the demarcated area are: , will the The coordinates of the measured geometric center points of each region are as follows: Then calculate the first section of the highway network. Geometric center point offset error of each region Its expression is as follows: Specifically, by calculating the straight-line distance between the geometric center of the remotely sensed area and the measured geometric center using the Euclidean distance formula, the overall spatial offset of the remotely sensed area relative to the measured area can be accurately quantified, including the geometric center point offset error. The larger the value, the worse the spatial consistency of the delineation results; S13. Based on the current timestamp, in the... On the demarcated and measured boundaries of each region, several inflection points are selected (inflection points can be points with clear geometric characteristics, such as the shoulder or toe of a highway, or a road intersection). After connecting all the inflection points to form a closed line, the first... The length of the boundary line delineated for each region is denoted as . , will the The measured length of the boundary line of each region is denoted as . Then calculate the first section of the highway network. Boundary line length error of each region Its expression is as follows: Specifically, boundary line length error By eliminating the influence of the boundary line's own length, it can intuitively reflect local boundary offset issues, making results from different regions horizontally comparable, and reducing boundary line length errors. The closer it is to 1, the greater the deviation in length between the delineated and measured boundary lines, and the lower the accuracy of the delineation. S14. Based on S11-S13, calculate the first section of the highway network. Spatial error index of each region Its expression is as follows: In the formula, The weights representing the overlap rate of the demarcated boundaries. The weights representing the offset error of the geometric center point. The weights representing the boundary line length error are... , and All are constants, and ; Specifically, by using the three spatial error dimensions of coverage area, location, and boundary line, the weight of each indicator can be flexibly adjusted according to the focus of different business scenarios, thereby more comprehensively reflecting the spatial accuracy of the delineation results of each area of ​​the highway network. This not only avoids the one-sidedness of a single indicator, but also provides convenience for subsequent automated quality grading and screening. S15. Based on the current timestamp, the first... The falsely delineated area resulting from misjudgment or omission in each region is denoted as . , will the The measured deformation-sensitive area of ​​each region is denoted as . Then calculate the first section of the highway network. Percentage of effective sensitive areas in each region Its expression is as follows: In the formula, This represents the actual and correctly identified sensitive area, and the percentage of the effective sensitive area. The closer it is to 1, the smaller the false area caused by misjudgment or omission, the higher the accuracy of sensitive area identification, and the less invalid investigation work is required for subsequent drone precision investigation; S16. Based on the current timestamp, the first... The number of independent patches within the delineated area is denoted as . Then calculate the first section of the highway network. Patch density in each region Its expression is as follows: Specifically, large-scale remote sensing data is susceptible to issues such as cloud cover and radar shadows, leading to fragmentation of the delineated area. Independent patches refer to fragmented areas within the deformation-sensitive range delineated by optical satellites that are independently segmented. This fragmentation increases the difficulty and cost of subsequent drone-based reconnaissance flight path planning. The number of independent patches... The fewer the values, the more continuous the delineated area, and the stronger the practicality of the engineering project. S17. Based on the current timestamp, the first... The area of ​​delineation error caused by cloud shadows in each region is denoted as . Then calculate the first section of the highway network. The percentage of delineated error area in each region Its expression is as follows: Specifically, the percentage of the delineated error area. The influence of cloud shadow interference on the delineation results can be quantified. The higher the value, the greater the interference of cloud shadow on the delineation results and the lower the reliability of the delineation results. S18. Based on S15-S17, calculate the first section of the highway network. Effective identification index of each region Its expression is as follows: In the formula, The weight representing the percentage of effective sensitive area. The weight representing the inverse of patch density, The weight represents the reciprocal of the proportion of the delineated error area. , and All are constants, and ; Specifically, the three effective identification indicators of weighted fusion identification accuracy, regional continuity, and interference impact can accurately quantify the data credibility of remote sensing-delineated areas; S19. Based on the current timestamp, the first... The demarcation time for each region is recorded as follows: , will the The radar backscattering coefficient of each region is denoted as . Then, based on S11-S18, calculate the first section of the highway network. The delineation index of each region Its expression is as follows: In the formula, The weights representing the complement of the spatial error exponent. This indicates the weight of the index for effective identification. This represents a reference value indicating the timing delay of the remote sensing satellite sphere. This indicates the weight of the reference value and the timing delay ratio. The reference value representing the backscattering coefficient of remote sensing satellite radar. The weights representing the ratio of the radar backscattering coefficient to the reference value are: , , and All are constants, and ; Specifically, the reference values ​​for remote sensing satellite circumference delay and radar backscattering coefficient are determined by combining industry standard requirements with the statistical average of historical monitoring data, and the circumference index is used for delineation. By focusing on four core dimensions—spatial error, effective identification, timeliness, and radar characteristics—a comprehensive evaluation of the results of delineating highway network areas was achieved in terms of spatial accuracy, identification efficiency, response timeliness, and equipment data quality. This ensured complete and thorough coverage of dimensions while accurately reflecting the overall quality level of the delineation results. The following is the experimental data for the delineation index, as shown in Table 1: Table 1: Experimental Data for the Delineation Index In Table 1, the highway network area A was selected as the experimental target in the delineation index experimental data. For spatial error index Weight , , Spatial error index During the calculation, the normalized geometric center point offset error The value is 0.867; For effective identification index Weight , , Effective identification index During the calculation, the normalized patch density The reciprocal value is 0.4, and after normalization, the percentage of the error area is determined. The reciprocal value is 0.4; Targeting the index Weight , , , Delineate the index During the calculation, the ratio of the normalized reference value to the timing delay is 1; The risk management unit has a fixed range of confidence thresholds. This method is used to quickly determine the reliability of the remote sensing delineation range for each region, providing a reliable guarantee for the orderly advancement of highway network slope deformation risk assessment. The scientific validity of the range directly affects the effectiveness of the risk assessment results and the targeted nature of subsequent intervention measures. An excessively narrow reliability threshold range... This could lead to overly stringent credibility level criteria, misclassifying areas with moderate credibility as low or no credibility, increasing unnecessary costs for image re-acquisition and full-range correction, and delaying the overall assessment progress. Conversely, overly stringent criteria would blur the credibility level definition, making it impossible to distinguish the reliability differences between low and moderate, or high and moderate, potentially resulting in low-quality remote sensing data entering subsequent risk assessment processes, creating hidden risks of biased assessment results and inappropriate intervention measures. Therefore, the optimal range of this credibility threshold interval needs to be determined through calibration experiments of the following systems: Credibility Threshold Interval The calibration method is as follows: Samples were selected using remote sensing monitoring databases of different areas of the highway network, covering scenarios with no, low, medium, and high credibility levels within the remote sensing delineated areas. Remote sensing image parameters (such as spatial resolution, cloud cover percentage, and signal-to-noise ratio), measured range data (such as field survey boundaries and topographic measurement results), and subsequent application feedback from risk assessments (such as the consistency between assessment results and actual deformation, and the effectiveness of intervention measures) were extracted for the sample areas. Different candidate credibility threshold ranges were set. In each calibration experiment, the credibility level of the sample areas was classified based on the candidate credibility threshold range, and the matching degree between the classification results and the actual reliability verification results was recorded. Then, combined with dynamic monitoring data, the remote sensing data quality was simulated after the implementation of response measures corresponding to different credibility levels. To assess the improvement effect and impact on risk assessment efficiency, the upper and lower limits of the candidate confidence threshold intervals were adjusted. Multiple verification experiments were conducted to record the comprehensive impact of threshold interval settings on the overall assessment quality and efficiency. For each candidate confidence threshold interval, the collected sample data and dynamic simulation results were used as inputs to count the number of times low confidence levels were misjudged as high confidence levels due to improper interval range settings (counted as over-trust), the number of times high confidence levels were misjudged as low confidence levels (counted as under-trust), and the degree of fit between the confidence level classification results and the subsequent risk assessment application effects (such as data utilization rate, assessment accuracy, intervention cost control effect, etc.). Finally, the interval range that minimizes both over-trust and under-trust rates and has the highest degree of fit with the application effect was selected as the confidence threshold interval. The preferred range; Table 1 defines the confidence threshold range in the index experiment data. The preferred range is 0.7 to 0.9. It was determined that 0.7 < the delineation index of highway network region A. <0.9 indicates that the reliability of the remote sensing delineation of area A is moderate, with a reliability level of 3. Response measures include local remote sensing image enhancement processing and replacing abnormal pixel values ​​with pixel mean interpolation to address the deviation between the delineated area and the measured area. Based on the slope dataset, the inspection and evaluation unit assesses the accuracy of deformation points on each highway slope and generates corresponding calibration indices. The calculation process is as follows: S21. Based on the slope dataset, extract the first... The first expressway The 3D mapping data of the slope at the current timestamp is known. This expressway belongs to the highway network. In each region, the UAV LiDAR point cloud and the satellite remote sensing terrain point cloud are spatially registered, and the two sets of point clouds are in the same coordinate system. S22. Based on the current timestamp, the first... The unit normal vector of the LiDAR point cloud of a slope from a UAV is denoted as... , This represents the total number of point clouds, and the number of points is... The unit normal vector of a slope satellite remote sensing topographic point cloud is denoted as... Then calculate the first The first expressway Mean angle residual between the slope UAV LiDAR point cloud and the satellite remote sensing terrain point cloud Its expression is as follows: In the formula, Indicates the first point cloud of the drone's LiDAR The unit normal vector at each point , The first point in the satellite remote sensing terrain point cloud The unit normal vector at each point Indicates the first The included angle residual at each point; Specifically, the smaller the normal vector residual, the higher the consistency between the local terrain orientation of the UAV point cloud and the satellite reference terrain, and the higher the degree of restoration of the terrain tilt features after calibration. S23. Based on the current timestamp, the first... The cumulative displacement of the parallel slope surface from the LiDAR point cloud of a slope drone is denoted as: , will the The cumulative vertical slope displacement of the LiDAR point cloud from a slope drone is denoted as: , will the The cumulative vertical ground plane displacement of the LiDAR point cloud of a slope drone is denoted as: Then calculate the first The first expressway Cumulative deformation in LiDAR point cloud of a slope drone Its expression is as follows: Specifically, cumulative deformation It can comprehensively reflect the actual movement range of the slope in three-dimensional space, avoiding misjudgment of the severity of deformation that may be caused by displacement analysis in one direction; S24. Based on the current timestamp, the first... The number of cracks in the LiDAR point cloud of a slope drone is denoted as . , will the The area mapped from the LiDAR point cloud of a slope by a drone is denoted as Then calculate the first The first expressway Crack density of each slope Its expression is as follows: Specifically, crack density It is a core indicator for measuring the degree of damage to the slope under internal force. By statistically analyzing the density of tensile and shear cracks on the slope surface, it directly reflects the stress release and structural deterioration trend inside the slope. It can serve as a key basis for early warning and help managers determine whether reinforcement or early warning measures need to be taken in a timely manner. S25. Based on the current timestamp, the first... The number of gullies in the LiDAR point cloud of a slope drone is denoted as Then calculate the first The first expressway Gullage density of individual slopes Its expression is as follows: Specifically, gully density It is a key indicator reflecting the degree of damage to slopes driven by external forces. It quantifies the development of erosion grooves formed on the slope surface under the action of rainwater scouring and runoff erosion. It is a direct manifestation of the damage caused to the slope surface by external hydrodynamic action. S26. Based on the current timestamp, the first... The green area in the LiDAR point cloud of a slope drone is denoted as... Then calculate the first The first expressway Green coverage rate of each slope Its expression is as follows: Specifically, through green coverage rate The quantifiable vegetation coverage of soil slopes provides a key basis for assessing the impact of vegetation on slope stability and the degree of interference with deformation monitoring. S27. Based on the current timestamp, the first... The rock surface area in the LiDAR point cloud of a slope drone is denoted as , will the The specular reflectance of the LiDAR point cloud of a slope drone is denoted as . , will the The area of ​​the specular reflection region in the LiDAR point cloud of a slope drone is denoted as . , will the The diffuse reflectance of the LiDAR point cloud of a slope drone is denoted as . , will the The area of ​​the diffuse reflectance region in the LiDAR point cloud of a slope drone is denoted as . Then calculate the first The first expressway Average reflectivity of the rock surface on each slope Its expression is as follows: In the formula, This indicates the percentage of the area of ​​the specular reflection region. Indicates the specular reflection calibration coefficient. Indicates the area ratio of the diffuse reflection region. Indicates the diffuse reflection calibration coefficient; Specifically, rock slopes exhibit significant differences in reflectivity. Smooth or jointed surfaces are primarily specular reflective, while fractured rock surfaces are primarily diffuse reflective. For smooth granite surfaces, which are primarily specular reflective, the specular reflection calibration coefficient is set to 0.7. By setting specific calibration coefficients for different lithologies, the original reflectivity can be corrected to a reflectivity range that conforms to the inherent characteristics of the lithology, thereby providing a more accurate and reliable quantitative basis for subsequent slope stability analysis. S28. Based on the current timestamp, the first... The laser echo intensity in the LiDAR point cloud of a slope drone is denoted as Then calculate the first The first expressway Point cloud intensity dispersion of each slope Its expression is as follows: In the formula, Indicates the first point cloud of the drone's LiDAR Laser echo intensity at each point , This represents the average intensity of the laser echo; Specifically, due to the heterogeneous distribution of obstructions on soil-rock mixed slopes, the slope surface is often covered with weeds, loose soil, gravel piles, shrubs, and other materials. Among these, the laser echo intensity of the obstructions (loose soil, gravel, weeds) is relatively low and fluctuates significantly, while the echo intensity of the bedrock and undisturbed soil is relatively high and stable, exhibiting a high point cloud intensity dispersion. The magnitude of the dispersion value can directly reflect the uniformity of the medium in the region. The larger the dispersion value, the more violent the fluctuation of the echo intensity, indicating that there are a lot of obstructions in the region and the higher the degree of heterogeneity of the medium. S29. Based on S21-S28, calculate the first slope type using a weighted method. The first expressway The calibration index of each slope Its expression is as follows: If the first The slope type is soil. In the formula, The weight representing the reciprocal of the mean of the included angle residuals. The weight represents the cumulative deformation. The weight representing the crack density, The weight representing the density of gullies, The weight representing the reciprocal of the green coverage rate. , , , and All are constants, and , Indicates the first The first expressway The calibration index of a soil slope; If the first The slope type is rocky. In the formula, The weight representing the reciprocal of the mean of the included angle residuals. The weight represents the cumulative deformation. The weight representing the crack density, The weight representing the density of gullies, The weight representing the reciprocal of the average reflectivity of the rock surface. , , , and All are constants, and , Indicates the first The first expressway The calibration index of a rock slope; If the first The slope type is mixed rock and soil. In the formula, The weight representing the reciprocal of the mean of the included angle residuals. The weight represents the cumulative deformation. The weight representing the crack density, The weight representing the density of gullies, The weights represent the reciprocal of the point cloud intensity dispersion. , , , and All are constants, and , Indicates the first The first expressway The calibration index of a soil-rock mixed slope; Specifically, the UAV LiDAR point cloud is spatially registered with the satellite remote sensing terrain point cloud, and then core indicators such as the mean of the included angle residual and the cumulative deformation are extracted. Adaptation indicators such as green coverage rate and point cloud intensity dispersion are introduced in combination with slope type, and the calibration index is calculated through weighted fusion. It comprehensively covers key dimensions such as slope data reliability and three-dimensional deformation, and improves the accuracy and adaptability of the results. It can provide reliable quantitative basis for the stability analysis, early warning and reinforcement of high-speed slopes, while avoiding the limitations and risks of misjudgment of single indicators or unified logic. The following are the experimental data for the scaling index, as shown in Table 2: Table 2: Experimental Data for Scaling Index In Table 2, the soil slope of Section A of the 403 Expressway was selected as the experimental target in the calibration index experimental data. For the calibration index Weight , , , , ; The risk management unit has a fixed range of hazard thresholds. This method is used to quickly and accurately determine the location of deformation points on highway slopes, providing quantitative support for classifying slope deformation hazard levels and formulating targeted prevention and control measures. The rationality of the range directly affects the accuracy of hazard assessment and the timeliness of prevention and control intervention. An excessively narrow hazard threshold range... Overly strict hazard classification can lead to slopes with minor deformation risks being misclassified as having significant or severe deformation risks, increasing unnecessary engineering costs such as sealing and grouting, and adding retaining structures. It can also excessively restrict traffic and construction, impacting traffic efficiency. Conversely, overly lenient hazard classification can blur the lines between minor and significant deformation risks, and between significant and severe deformation risks, potentially leading to high-risk slopes not being reinforced in time, creating the potential for slope collapse, traffic disruption, and other safety hazards. Therefore, the optimal range for this hazard threshold interval needs to be determined through calibration experiments using the following system: Hazard Threshold Interval. The calibration method is as follows: Slope samples with different hazard levels were screened using a highway slope deformation monitoring database, covering various situations including no obvious deformation hazard, slight deformation hazard, obvious deformation hazard, and severe deformation hazard. Deformation monitoring data (such as UAV LiDAR point cloud deviation values, crack width changes, and slope displacement rates), field investigation records (such as slope soil and rock properties, slope drainage conditions, and vegetation cover), and feedback on prevention and control effects (such as deformation control effects after intervention, hazard escalation rate, and traffic operation support status) were extracted from the sample slopes. Different candidate hazard threshold ranges were set. In each calibration experiment, the hazard levels of the sample slopes were classified according to the candidate hazard threshold ranges. The matching degree between the classification results and the actual on-site hazard assessment standards was recorded. Combined with dynamic monitoring data, slope calibration indices were simulated under different prevention and control intervention rhythms. Adjust the candidate hazard threshold range based on the changing trend. The upper and lower limits of the threshold range were determined, and multiple verification experiments were conducted to record the comprehensive impact of the threshold range setting on the prevention and control effect and operating cost. For each candidate hazard threshold range, the collected sample data and dynamic simulation results were used as inputs to count the number of times low-level hazards were misjudged as high-level hazards due to improper range setting (counted as over-warning), the number of times high-level hazards were misjudged as low-level hazards (counted as under-warning), and the degree of fit between the hazard level classification results and the actual prevention and control effect (such as hazard management compliance rate, safety accident incidence rate, operating cost control level, etc.). Finally, the range that minimizes both the over-warning rate and the under-warning rate and has the highest degree of fit with the prevention and control effect was selected as the hazard threshold range. The preferred range; In the calibration index experimental data in Table 2, the hazard threshold range The preferred range is 18 to 25. Based on this, the calibration index for the soil slope of Section A of the 403 Expressway is determined. >25 indicates that there is a serious potential deformation hazard on the soil slope of Section A of the 403 Expressway, with a hazard level of 4. The response measures include delineating the danger warning area, evacuating personnel and equipment in the danger area, activating emergency monitoring equipment to collect three-dimensional mapping data in real time, and implementing slope cutting and load reduction and anchor bolt support reinforcement projects. The mechanical analysis unit analyzes the risk level of landslide disaster caused by construction disturbance based on the protection dataset and generates a corresponding disaster score. The calculation process is as follows: S31. Assess the disaster risk of each slope protection project. The initial value is set to 0 points, and then the first value is extracted based on the project dataset. The first expressway Management data for individual slope protection projects; S32. Trigger the corresponding single scoring condition according to the slope type; Condition 1: No. If, during the construction of the protective engineering project, the average daily vertical settlement rate of a slope is not continuously reduced to below 0.1 mm / d over time and remains stable for 7 consecutive days, it indicates that the slope deformation has not converged, and the risk of landslide disaster is increasing. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 2: No. The slope type is rock. During the construction of the protective project, if the average daily shear deformation rate along the fracture surface does not decrease to below 0.1 mm / d over time and remains stable for 7 days, it indicates that the slope deformation has not converged, and the risk of landslide disaster is increasing. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 3: If The slope type is mixed soil and rock. During the construction of the protective engineering, if the average daily vertical settlement rate of the upper soil layer does not decrease to below 0.1 mm / d and remain stable for 7 consecutive days in a positive sequence over time, or if the average daily shear deformation rate of the lower rock layer along the fracture surface does not decrease to below 0.1 mm / d and remain stable for 7 consecutive days in a positive sequence over time, it indicates that the slope deformation has not converged, and the risk of landslide disaster has increased. The first expressway Disaster rating of individual slope protection projects Add 1 point; S33. Trigger the corresponding cumulative scoring for each slope; Condition 4: If The slope type is mixed soil and rock. During the construction of the protective engineering, the vertical settlement rate of the upper soil layer and the shear deformation rate of the lower rock layer along the fracture surface show inconsistent trends over time. This indicates that the slope deformation has not converged, and there is a disjointed effect of the protective engineering on the upper and lower parts of the slope, increasing the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 5: During the construction of the protective project, if the first The first expressway If the cohesion of the soil and rock mass on a slope is less than 30% of the design value for anti-sliding, it indicates that construction disturbance has led to the failure of the bond between soil and rock particles, making the slope prone to sliding along the weak surface and increasing the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition Six: During the construction of the protective engineering project, if the first The first expressway If the average daily uplift at the toe of a slope exceeds 2 mm / d for three consecutive days, it indicates an increase in slope sliding force, causing uplift at the toe and raising the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point; Specifically, the core design of protective engineering is based on the original mechanical properties of the soil and rock. If the mechanical properties of the soil and rock drop sharply due to disturbance during construction, the anti-sliding design of the protective structure will be seriously mismatched with the actual bearing capacity of the slope. The protective engineering will naturally fail to achieve the expected results, and the slope deformation will enter an accelerated development stage, thus forming a vicious cycle of "deformation aggravation → protection failure → greater deformation", which will eventually induce landslide disasters. The following is the experimental data for disaster rating, as shown in Table 3: Table 3: Experimental Data for Disaster Rating In Table 3, the experimental data of the calibration index were selected as the experimental target for the mixed soil and rock slope of Section B of the 403 Expressway, with a sliding design value of 80 kPa. Based on the assessment, the average daily vertical settlement rate of the upper soil layer of the soil-rock mixed slope of Section B of Expressway 403 did not decrease to below 0.1 mm / d over time and remain stable for 7 days, and the average daily shear deformation rate of the lower rock layer along the fracture surface did not decrease to below 0.1 mm / d over time and remain stable for 7 days. Therefore, the catastrophic assessment of the soil-rock mixed slope of Section B of Expressway 403 is determined. Add 1 point; During the construction of the protective engineering project, the vertical settlement rate of the upper soil layer and the shear deformation rate of the lower rock layer along the fracture surface showed inconsistent trends over time. This led to the determination of the catastrophic impact of the soil-rock composite slope on Section B of the 403 Expressway. Add 1 point; During the construction of the protective engineering project, the cohesion of the soil and rock mass was below 24 kPa (30% of the anti-sliding design value of 80 kPa) on days 8-10, which affected the catastrophic assessment of the soil-rock composite slope of Section B of the 403 Expressway. Add 1 point; During the construction of the protective engineering project, the average daily uplift at the toe of the soil-rock composite slope of Section B of the 403 Expressway exceeded 2 mm / d for three consecutive days, which affected the disaster assessment of the soil-rock composite slope of Section B of the 403 Expressway. Add 1 point; Ultimately, the catastrophic assessment of the soil-rock mixed slope of Section B of the 403 Expressway was... A score of 4 indicates that the construction disturbance has caused the risk of landslide disaster to exceed the standard. The response measures include immediately suspending the slope protection project, conducting on-site verification, and formulating a temporary reinforcement plan. The risk management unit has a fixed range of confidence thresholds. and the threshold range of hidden dangers Combined with the defined index , calibration index and disaster rating It determines the credibility level, hazard level, and landslide disaster risk, outputs the assessment results, and triggers corresponding response measures. The trust level determination process is as follows: Let the upper limit of the confidence threshold range be denoted as The lower limit of the confidence threshold interval is denoted as ; If the highway network is number The delineation index of each region < , indicating the first The remote sensing delineation of the area has no credibility level, with a credibility rating of Level 1. Response measures include suspending the [number] [section / section / etc.]. The process for assessing slope deformation risk in a specific area involves replacing remote sensing satellites, acquiring new high spatial resolution remote sensing images, and delineating the corresponding areas. If the highway network... The delineation index of each region , indicating the first The reliability of the remote sensing delineation of the area is low, with a reliability level of 2. Response measures include full-range remote sensing image correction, filtering based on historical cloudless period images of the corresponding area, and continuing the process after correction and filtering. The process of assessing slope deformation risk in a given area, if <Highway Network No. 1> The delineation index of each region ≤ , indicating the first The reliability of the remote sensing delineation of the area is moderate, with a reliability level of 3. Response measures include local remote sensing image enhancement processing; for discrepancies between the delineated and measured areas, pixel mean interpolation is used to replace abnormal pixel values. If the highway network... The delineation index of each region > , indicating the first The remote sensing delineation of the area has a high degree of credibility, with a credibility level of 4. The response measures include storing backup data and conducting a review to check for any new cloud shadows or temporary obstructions. The procedure for determining the level of hazard is as follows: The upper limit of the hazard threshold range is denoted as... The lower limit of the hazard threshold range is denoted as ; If the first The first expressway The calibration index of each slope < , indicating the first The first expressway The slope showed no obvious deformation risks, with a risk level of Level 1. Response measures included quarterly LiDAR drone inspections, synchronously updating point cloud data, and clearing surface soil and weeds to ensure a clear monitoring view. If the... The first expressway The calibration index of each slope , indicating the first The first expressway The slope has a slight potential deformation risk, classified as Level 2. Response measures include monthly LiDAR drone inspections, replanting soil-stabilizing vegetation, and optimizing slope drainage facilities to prevent rainwater erosion from exacerbating the risk. <No. The first expressway The calibration index of each slope ≤ , indicating the first The first expressway One slope has significant potential deformation, classified as level 3. Response measures include weekly LiDAR drone inspections, injecting sealant into cracks, adding retaining structures at the slope toe, and restricting construction activities and vehicle traffic around the slope. The first expressway The calibration index of each slope > , indicating the first The first expressway The slope has a serious potential for deformation, with a risk level of 4. The response measures include delineating a danger warning area, evacuating personnel and equipment from the danger area, activating emergency monitoring equipment to collect three-dimensional mapping data in real time, and implementing slope cutting and load reduction and anchor bolt support reinforcement projects.

[0020] In this embodiment, the multi-dimensional acquisition module acquires remote sensing monitoring data of the highway network, 3D mapping data of highway slopes, and management data of protective engineering projects. It then classifies and constructs remote sensing datasets, slope datasets, and protective datasets, overcoming the limitations of single data sources. This provides a structured, high-quality data foundation for subsequent intelligent assessment, avoiding assessment biases caused by data clutter, and significantly improving data utilization efficiency. This lays a solid data foundation for the accuracy and comprehensiveness of slope deformation risk assessment. The intelligent assessment module, through the coordinated operation of four units—surface identification, inspection assessment, mechanical analysis, and risk management—achieves multi-dimensional, accurate assessment and efficient response to slope deformation risks. The delineation index generated by the surface identification unit... Taking into account core dimensions such as spatial error and effective identification, the quality of remote sensing delineation results is accurately quantified. The inspection and evaluation unit calculates the calibration index by weighted fusion of multiple core indicators for different slope types. To improve the adaptability and accuracy of deformation point assessment, the catastrophic scoring of the mechanical analysis unit focuses on key factors of construction disturbance, effectively identifies potential landslide disaster risks, and generates corresponding catastrophic scores. The risk management unit, combined with dual threshold ranges, enables the scientific determination of credibility level, hazard level, and disaster risk, and triggers targeted response measures. It achieves closed-loop management of the entire process from data quality verification to risk level determination and emergency response, greatly improving the scientific nature and timeliness of risk assessment. This provides a reliable basis for decision-making in slope safety management, reinforcement and protection, and emergency response, effectively reducing losses caused by landslide disasters.

[0021] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.

[0022] The above formulas are all derived from software simulation using a large amount of data and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art according to the actual situation. The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the protection scope of the present invention.

Claims

1. A highway slope deformation risk assessment system based on multimodal data, characterized in that: Includes a multi-dimensional data acquisition module and an intelligent evaluation module; The multidimensional acquisition module connects remote sensing satellites, drones, and soil and rock monitoring equipment to acquire remote sensing monitoring data of the highway network, three-dimensional mapping data of highway slopes, and management data of protection projects, and classifies them into remote sensing datasets, slope datasets, and protection datasets. The intelligent assessment module includes a surface identification unit, an inspection and assessment unit, a mechanical analysis unit, and a risk management unit. The surface identification unit assesses the reliability of the remote sensing delineation range for each area based on the remote sensing dataset and generates a corresponding delineation index. The inspection and evaluation unit assesses the accuracy of deformation points on each highway slope based on the slope dataset and generates corresponding calibration indices. The mechanical analysis unit analyzes the risk level of landslide disaster caused by construction disturbance based on the protection dataset and generates a corresponding disaster score. The risk management unit is configured with a fixed range of reliable thresholds. and the threshold range of hidden dangers Combined with the defined index , calibration index and disaster rating It determines the credibility level, hazard level, and landslide disaster risk, outputs the assessment results, and triggers corresponding response measures.

2. The highway slope deformation risk assessment system based on multimodal data according to claim 1, characterized in that: The remote sensing dataset includes the remote sensing delineated area, measured area, coordinates of the geometric center point within the delineated area, coordinates of the measured geometric center point, length of the delineated boundary line, length of the measured boundary line, false delineated area, measured deformation-sensitive area, number of independent patches, delineation error area, delineation time delay, and radar backscattering coefficient for each region of the highway network.

3. The highway slope deformation risk assessment system based on multimodal data according to claim 2, characterized in that: The slope dataset includes the type of each highway slope, the total number of point clouds, the unit normal vector, the cumulative displacement of the parallel slope, the cumulative displacement of the vertical slope, the cumulative displacement of the vertical ground plane, the number of cracks, the surveyed area, the number of gullies, the green area, the rock surface area, the specular emissivity, the area of ​​the specular reflection region, the diffuse reflectivity, the area of ​​the diffuse reflection region, and the laser echo intensity. The slope types include soil, rock, and soil-rock mixture.

4. The highway slope deformation risk assessment system based on multimodal data according to claim 3, characterized in that: The protection dataset includes the average daily vertical settlement rate, the average daily shear deformation rate along the fracture surface, the soil-rock cohesion, and the average daily heave at the toe of the slope for each slope protection project.

5. The highway slope deformation risk assessment system based on multimodal data according to claim 4, characterized in that: The delineation index The calculation process is as follows: S11. Based on the remote sensing dataset, extract the first [item] of the highway network. The remote sensing monitoring data of each region at the current timestamp, based on the current timestamp, will be used to... The remote sensing delineated area of ​​each region is denoted as . , will the The measured area of ​​each region is denoted as . Then calculate the first section of the highway network. The overlap rate of the delineated boundaries of each region ; S12. Based on the current timestamp, the first... The coordinates of the geometric center point within the demarcated area are: , will the The coordinates of the measured geometric center points of each region are as follows: Then calculate the first section of the highway network. Geometric center point offset error of each region ; S13. Based on the current timestamp, in the... On the demarcated and measured boundaries of each region, several inflection points are selected respectively. After connecting all the inflection points to close the line, the first... The length of the boundary line delineated for each region is denoted as . , will the The measured length of the boundary line of each region is denoted as . Then calculate the first section of the highway network. Boundary line length error of each region ; S14. Based on S11-S13, calculate the first section of the highway network. Spatial error index of each region ; S15. Based on the current timestamp, the first... The falsely delineated area resulting from misjudgment or omission in each region is denoted as . , will the The measured deformation-sensitive area of ​​each region is denoted as . Then calculate the first section of the highway network. Percentage of effective sensitive areas in each region ; S16. Based on the current timestamp, the first... The number of independent patches within the delineated area is denoted as . Then calculate the first section of the highway network. Patch density in each region ; S17. Based on the current timestamp, the first... The area of ​​delineation error caused by cloud shadows in each region is denoted as . Then calculate the first section of the highway network. The percentage of delineated error area in each region ; S18. Based on S15-S17, calculate the first section of the highway network. Effective identification index of each region ; S19. Based on the current timestamp, the first... The demarcation time for each region is recorded as follows: , will the The radar backscattering coefficient of each region is denoted as . Then, based on S11-S18, calculate the first section of the highway network. The delineation index of each region .

6. The highway slope deformation risk assessment system based on multimodal data according to claim 5, characterized in that: The calibration index The calculation process is as follows: S21. Based on the slope dataset, extract the first... The first expressway The 3D mapping data of the slope at the current timestamp is known. This expressway belongs to the highway network. In each region, the UAV LiDAR point cloud and the satellite remote sensing terrain point cloud are spatially registered, and the two sets of point clouds are in the same coordinate system. S22. Based on the current timestamp, the first... The unit normal vector of the LiDAR point cloud of a slope from a UAV is denoted as... , This represents the total number of point clouds, and the number of points is... The unit normal vector of a slope satellite remote sensing topographic point cloud is denoted as... Then calculate the first The first expressway Mean angle residual between the slope UAV LiDAR point cloud and the satellite remote sensing terrain point cloud ; S23. Based on the current timestamp, the first... The cumulative displacement of the parallel slope surface from the LiDAR point cloud of a slope drone is denoted as: , will the The cumulative vertical slope displacement of the LiDAR point cloud from a slope drone is denoted as: , will the The cumulative vertical ground plane displacement of the LiDAR point cloud of a slope drone is denoted as: Then calculate the first The first expressway Cumulative deformation in LiDAR point cloud of a slope drone ; S24. Based on the current timestamp, the first... The number of cracks in the LiDAR point cloud of a slope drone is denoted as . , will the The area mapped from the LiDAR point cloud of a slope by a drone is denoted as Then calculate the first The first expressway Crack density of each slope ; S25. Based on the current timestamp, the first... The number of gullies in the LiDAR point cloud of a slope drone is denoted as Then calculate the first The first expressway Gullage density of individual slopes ; S26. Based on the current timestamp, the first... The green area in the LiDAR point cloud of a slope drone is denoted as... Then calculate the first The first expressway Green coverage rate of each slope ; S27. Based on the current timestamp, the first... The rock surface area in the LiDAR point cloud of a slope drone is denoted as , will the The specular reflectance of the LiDAR point cloud of a slope drone is denoted as . , will the The area of ​​the specular reflection region in the LiDAR point cloud of a slope drone is denoted as . , will the The diffuse reflectance of the LiDAR point cloud of a slope drone is denoted as . , will the The area of ​​the diffuse reflectance region in the LiDAR point cloud of a slope drone is denoted as . Then calculate the first The first expressway Average reflectivity of the rock surface on each slope ; S28. Based on the current timestamp, the first... The laser echo intensity in the LiDAR point cloud of a slope drone is denoted as Then calculate the first The first expressway Point cloud intensity dispersion of each slope ; S29. Based on S21-S28, calculate the first slope type using a weighted method. The first expressway The calibration index of each slope .

7. The highway slope deformation risk assessment system based on multimodal data according to claim 6, characterized in that: The disaster rating The calculation process is as follows: S31. Assess the disaster risk of each slope protection project. The initial value is set to 0 points, and then the first value is extracted based on the project dataset. The first expressway Management data for individual slope protection projects; S32. Trigger the corresponding single scoring condition according to the slope type; Condition 1: No. If, during the construction of the protective engineering project, the average daily vertical settlement rate of a slope is not continuously reduced to below 0.1 mm / d over time and remains stable for 7 consecutive days, it indicates that the slope deformation has not converged, and the risk of landslide disaster is increasing. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 2: No. The slope type is rock. During the construction of the protective project, if the average daily shear deformation rate along the fracture surface does not decrease to below 0.1 mm / d over time and remains stable for 7 days, it indicates that the slope deformation has not converged, and the risk of landslide disaster is increasing. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 3: If The slope type is mixed soil and rock. During the construction of the protective engineering, if the average daily vertical settlement rate of the upper soil layer does not decrease to below 0.1 mm / d and remain stable for 7 consecutive days in a positive sequence over time, or if the average daily shear deformation rate of the lower rock layer along the fracture surface does not decrease to below 0.1 mm / d and remain stable for 7 consecutive days in a positive sequence over time, it indicates that the slope deformation has not converged, and the risk of landslide disaster has increased. The first expressway Disaster rating of individual slope protection projects Add 1 point; S33. Trigger the corresponding cumulative scoring for each slope; Condition 4: If The slope type is mixed soil and rock. During the construction of the protective engineering, the vertical settlement rate of the upper soil layer and the shear deformation rate of the lower rock layer along the fracture surface show inconsistent trends over time. This indicates that the slope deformation has not converged, and there is a disjointed effect of the protective engineering on the upper and lower parts of the slope, increasing the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition 5: During the construction of the protective project, if the first The first expressway If the cohesion of the soil and rock mass on a slope is less than 30% of the design value for anti-slide, it indicates that construction disturbance has led to the failure of interparticle cohesion in the soil and rock mass, increasing the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point; Condition Six: During the construction of the protective engineering project, if the first The first expressway If the average daily uplift at the toe of a slope exceeds 2 mm / d for three consecutive days, it indicates an increase in slope sliding force, causing uplift at the toe and raising the risk of landslide disaster. The first expressway Disaster rating of individual slope protection projects Add 1 point.

8. The highway slope deformation risk assessment system based on multimodal data according to claim 7, characterized in that: The trust level determination process is as follows: Let the upper limit of the confidence threshold range be denoted as The lower limit of the confidence threshold interval is denoted as ; If the highway network is number The delineation index of each region < , indicating the first The remote sensing delineation of the area has no credibility level, with a credibility rating of Level 1. Response measures include suspending the [number] [section / section / etc.]. The process for assessing slope deformation risk in a specific area involves replacing remote sensing satellites, acquiring new high spatial resolution remote sensing images, and delineating the corresponding areas. If the highway network... The delineation index of each region , indicating the first The reliability of the remote sensing delineation of the area is low, with a reliability level of 2. Response measures include full-range remote sensing image correction, filtering based on historical cloudless period images of the corresponding area, and continuing the process after correction and filtering. The process of assessing slope deformation risk in a given area, if <Highway Network No. 1> The delineation index of each region ≤ , indicating the first The reliability of the remote sensing delineation of the area is moderate, with a reliability level of 3. Response measures include local remote sensing image enhancement processing; for discrepancies between the delineated and measured areas, pixel mean interpolation is used to replace abnormal pixel values. If the highway network... The delineation index of each region > , indicating the first The remote sensing delineation of the area has a high degree of reliability, with a reliability level of 4. Response measures include storing backup data and conducting verification work to check for the presence of new cloud shadows and temporary obstructions.

9. The highway slope deformation risk assessment system based on multimodal data according to claim 8, characterized in that: The procedure for determining the level of hazard is as follows: The upper limit of the hazard threshold range is denoted as... The lower limit of the hazard threshold range is denoted as ; If the first The first expressway The calibration index of each slope < , indicating the first The first expressway The slope showed no obvious deformation risks, with a risk level of Level 1. Response measures included quarterly LiDAR drone inspections, synchronously updating point cloud data, and clearing surface soil and weeds to ensure a clear monitoring view. If the... The first expressway The calibration index of each slope , indicating the first The first expressway The slope has a minor potential deformation risk, classified as Level 2. Response measures include monthly LiDAR drone inspections, replanting soil-stabilizing vegetation, and optimizing slope drainage facilities. <No. The first expressway The calibration index of each slope ≤ , indicating the first The first expressway One slope has significant potential deformation, classified as level 3. Response measures include weekly LiDAR drone inspections, injecting sealant into cracks, adding retaining structures at the slope toe, and restricting construction activities and vehicle traffic around the slope. The first expressway The calibration index of each slope > , indicating the first The first expressway The slope has a serious potential for deformation, with a risk level of 4. The response measures include delineating a danger warning area, evacuating personnel and equipment from the danger area, activating emergency monitoring equipment to collect three-dimensional mapping data in real time, and implementing slope cutting and load reduction and anchor bolt support reinforcement projects.

10. The highway slope deformation risk assessment system based on multimodal data according to claim 9, characterized in that: The disaster rating A score greater than 0 indicates that the risk of landslide disaster has exceeded the standard due to construction disturbance. Response measures include immediately suspending the slope protection project, conducting on-site verification, and formulating a temporary reinforcement plan.

Images