Plateau mountainous area road slope geological disaster automatic identification method fusing multiple data

Abstract

A plateau mountain area road slope geological disaster automatic identification method fusing multiple data comprises the following steps: S1, acquiring multi-source heterogeneous data of a target slope area, including a time sequence satellite optical image, synthetic aperture radar interferometry (InSAR) deformation data, laser radar (LiDAR) point cloud data, meteorological and hydrological data and basic geological data, and generating fusion data; s2, constructing a space-time registration module, unifying the multi-source data obtained in the S1, establishing a unified space-time coordinate system for the multi-source data, and generating fusion data; and S3, processing synthetic aperture radar interferometry (InSAR) deformation data through a freeze-thaw cycle noise compensation model, and eliminating deformation noise caused by seasonal frost heaving and thaw collapse. According to the plateau slope disaster identification method, the plateau freeze-thaw compensation model is combined with a multi-modal dynamic decision-making mechanism, and physical mechanism complementation of InSAR and LiDAR is adopted, so that'earth surface deformation-rock mass structure 'two-dimensional verification is realized, and multi-data fusion is realized, and therefore, the automatic identification precision of plateau slope disasters is improved.

Description

Technical Field

[0001] This invention relates to the field of geological disaster identification technology, specifically to an automatic identification method for geological disasters on highway slopes in plateau and mountainous areas that integrates multiple data sources. Background Technology

[0002] Geological disasters on highway slopes in plateau and mountainous areas are characterized by their suddenness and destructiveness, seriously threatening highway traffic safety and regional economic development. Traditional methods for monitoring geological disasters on slopes mainly rely on single data sources such as satellite imagery or ground sensors. In addition, existing identification methods mostly use manual analysis or simple statistical models. With the development of remote sensing technology, the Internet of Things, and artificial intelligence, multimodal data fusion and intelligent analysis have become important directions for geological disaster monitoring.

[0003] However, satellite remote sensing identification has limitations. It is difficult to fully reflect complex geological environment changes. Information from different data sources is isolated and lacks an effective fusion mechanism, resulting in one-sided and incomplete monitoring results. Manual identification lacks the ability to efficiently fuse and automate multi-source data, is inefficient, and is easily affected by subjective factors. Therefore, it does not meet the existing needs. In response, we propose an automatic identification method for geological hazards on highway slopes in plateau and mountainous areas that integrates multiple data. Summary of the Invention

[0004] The purpose of this invention is to provide an automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an automatic identification method for geological hazards on highway slopes in plateau and mountainous areas that integrates multiple data sources, comprising the following steps:

[0006] S1: Acquire multi-source heterogeneous data of the target slope area, including: time-series satellite optical imagery, synthetic aperture radar interferometry (InSAR) deformation data, lidar (LiDAR) point cloud data, meteorological and hydrological data, and basic geological data;

[0007] S2: Construct a spatiotemporal registration module to unify the multi-source data obtained in S1, establish a unified spatiotemporal coordinate system for the multi-source data, and generate fused data;

[0008] S3: The deformation data of synthetic aperture radar interferometry (InSAR) is processed by the freeze-thaw cycle noise compensation model to eliminate deformation noise caused by seasonal freeze-thaw cycles;

[0009] S4: Use a 3D point cloud segmentation algorithm to extract rock mass structural parameters from LiDAR point cloud data and calculate slope stability factors;

[0010] S5: Employs a multimodal feature dynamic weighting module to fuse optical image features, denoised deformation features, and rock mass structure features to output a disaster probability score;

[0011] S6: Generate a three-dimensional digital risk map with stability levels based on disaster probability scores.

[0012] Preferably, the satellite optical imagery in S1 originates from the GF-1 satellite, with a spatial resolution ≥10m and a temporal resolution ≤7 days;

[0013] The deformation data from the Synthetic Aperture Radar Interferometry (InSAR) satellite were obtained, with a revisit period of ≤4 days and a deformation measurement accuracy of ≤3mm.

[0014] The LiDAR point cloud data was collected by a drone equipped with a LiDAR, with a point cloud density of ≥50 points / square meter.

[0015] Preferably, the freeze-thaw cycle noise compensation model in S3 satisfies:

[0016] δcorrected=δobs−[k1⋅exp(−k2T−Tf)+k3⋅θ].

[0017] Preferably, the 3D point cloud segmentation algorithm in S4 is an improved PointNet++ network.

[0018] Preferably, the optimization of the PointNet++ network includes:

[0019] Add a height difference weighted sampling layer (weight=1 / ∣zi−zˉ∣weight=1 / ∣zi−zˉ∣) to adapt to steep terrain;

[0020] Output the rock mass structural plane orientation (dip / dip angle), fracture density (fractures / 10m²), and rock mass integrity coefficient Kv.

[0021] Preferably, the generation of the three-dimensional risk digital map in S6 includes: a base layer, a disaster layer, and a risk layer.

[0022] Preferably, the base layer integrates the DEM with geological structural lines, with a resolution ≤1m;

[0023] Hazard layer: Mark the landslide boundary (accuracy error ≤ 2m) and collapse volume (error ≤ 10%).

[0024] Risk level: Based on stability, it is divided into three levels: red (P≥0.7), yellow (0.4≤P<0.7), and green (P<0.4).

[0025] Preferably, it also includes step (g) field verification mechanism:

[0026] When the system identifies a disaster probability P≥0.7, it automatically dispatches a drone to perform oblique photography verification.

[0027] By comparing historical point clouds with a change detection algorithm, the disaster body with a displacement greater than 10cm was verified.

[0028] Compared with the prior art, the beneficial effects of the present invention are:

[0029] This invention combines a plateau freeze-thaw compensation model with a multimodal dynamic decision-making mechanism and complements the physical mechanisms of InSAR and LiDAR to achieve dual-dimensional verification of "surface deformation-rock structure" and realize the fusion of multiple data, thereby improving the accuracy and effectiveness of automated identification of geological hazards on highway slopes in plateau and mountainous areas. Attached Figure Description

[0030] Figure 1 This is a diagram illustrating the geological hazard identification method of the present invention. Detailed Implementation

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0032] like Figure 1 As shown, the automatic identification method for geological hazards on highway slopes in plateau and mountainous areas, which integrates multiple data sources, includes the following steps:

[0033] S1: Acquire multi-source heterogeneous data of the target slope area, including: time-series satellite optical imagery, synthetic aperture radar interferometry (InSAR) deformation data, lidar (LiDAR) point cloud data, meteorological and hydrological data, and basic geological data;

[0034] S2: Construct a spatiotemporal registration module to unify the multi-source data obtained in S1, establish a unified spatiotemporal coordinate system for the multi-source data, and generate fused data;

[0035] S3: The deformation data of synthetic aperture radar interferometry (InSAR) is processed by the freeze-thaw cycle noise compensation model to eliminate deformation noise caused by seasonal freeze-thaw cycles;

[0036] S4: Use a 3D point cloud segmentation algorithm to extract rock mass structural parameters from LiDAR point cloud data and calculate slope stability factors;

[0037] S5: Employs a multimodal feature dynamic weighting module to fuse optical image features, denoised deformation features, and rock mass structure features to output a disaster probability score;

[0038] S6: Generate a three-dimensional digital risk map with stability levels based on disaster probability scores.

[0039] Among them, the satellite optical imagery in S1 comes from the GF-1 satellite, with a spatial resolution of ≥10m and a temporal resolution of ≤7 days;

[0040] The deformation data from the Synthetic Aperture Radar Interferometry (InSAR) satellite were obtained, with a revisit period of ≤4 days and a deformation measurement accuracy of ≤3mm.

[0041] The LiDAR point cloud data was collected by a drone equipped with a LiDAR, with a point cloud density of ≥50 points / square meter.

[0042] further,.

[0043] Among them, the freeze-thaw cycle noise compensation model in S3 satisfies:

[0044] δcorrected=δobs−[k1⋅exp(−k2T−Tf)+k3⋅θ];

[0045] Where δobs is the observed deformation value, T is the surface temperature, Tf is the freezing point, θ is the soil moisture content, and k1, k2, k3 are calibration parameters for the plateau region.

[0046] Among them, the 3D point cloud segmentation algorithm in S4 is an improved PointNet++ network.

[0047] The optimizations of the PointNet++ network include:

[0048] Add a height difference weighted sampling layer (weight=1 / ∣zi−zˉ∣weight=1 / ∣zi−zˉ∣) to adapt to steep terrain;

[0049] Output the rock mass structural plane orientation (dip / dip angle), fracture density (fractures / 10m²), and rock mass integrity coefficient Kv.

[0050] The weight allocation rules for the dynamic weighting module in S5 are as follows:

[0051] Plateau climate model Optical weight Wo Deformation weight Wd Structural weights WsWs Rainy season (June-September) 0.1-0.3 0.5-0.7 0.2-0.4 Freeze-thaw period (October to May) 0.3-0.5 0.2-0.4 0.4-0.6

[0052] The final disaster probability score is calculated as: P = Wo⋅Fo + Wd⋅Fd + Ws⋅Fs, where Fo, Fd, and Fs are normalized eigenvalues.

[0053] A red alert is triggered when all of the following conditions are met:

[0054] P≥0.85

[0055] Crack density ≥ 2 cracks / 10m2

[0056] Deformation acceleration |dt2d2δ| ≥ 5 mm / month2.

[0057] The generation of the three-dimensional risk digital map in S6 includes: a basic layer, a disaster layer, and a risk layer.

[0058] Base layer: Integrating DEM and geological structural lines, with a resolution of ≤1m;

[0059] Hazard layer: Mark the landslide boundary (accuracy error ≤ 2m) and collapse volume (error ≤ 10%).

[0060] Risk level: Based on stability, it is divided into three levels: red (P≥0.7), yellow (0.4≤P<0.7), and green (P<0.4).

[0061] It also includes step (g) field verification mechanism:

[0062] When the system identifies a disaster probability P≥0.7, it automatically dispatches a drone to perform oblique photography verification.

[0063] By comparing historical point clouds with a change detection algorithm, the disaster body with a displacement greater than 10cm was verified.

[0064] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. An automatic identification method for geological hazards on highway slopes in plateau and mountainous areas, integrating multiple data sources, characterized in that... Includes the following steps: S1: Acquire multi-source heterogeneous data of the target slope area, including: time-series satellite optical imagery, synthetic aperture radar interferometry (InSAR) deformation data, lidar (LiDAR) point cloud data, meteorological and hydrological data, and basic geological data; S2: Construct a spatiotemporal registration module to unify the multi-source data obtained in S1, establish a unified spatiotemporal coordinate system for the multi-source data, and generate fused data; S3: The deformation data of synthetic aperture radar interferometry (InSAR) is processed by the freeze-thaw cycle noise compensation model to eliminate deformation noise caused by seasonal freeze-thaw cycles; S4: Use a 3D point cloud segmentation algorithm to extract rock mass structural parameters from LiDAR point cloud data and calculate slope stability factors; S5: Employs a multimodal feature dynamic weighting module to fuse optical image features, denoised deformation features, and rock mass structure features to output a disaster probability score; S6: Generate a three-dimensional digital risk map with stability levels based on disaster probability scores.

2. The automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data as described in claim 1, characterized in that: The satellite optical imagery in S1 is derived from the GF-1 satellite, with a spatial resolution ≥10m and a temporal resolution ≤7 days. The deformation data from the Synthetic Aperture Radar Interferometry (InSAR) satellite were obtained, with a revisit period of ≤4 days and a deformation measurement accuracy of ≤3mm. The LiDAR point cloud data was collected by a drone equipped with a LiDAR, with a point cloud density of ≥50 points / square meter.

3. The automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data as described in claim 1, characterized in that: The freeze-thaw cycle noise compensation model in S3 satisfies: δcorrected=δobs−[k1⋅exp(−k2T−Tf)+k3⋅θ].

4. The automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data as described in claim 1, characterized in that: The 3D point cloud segmentation algorithm in S4 is an improved PointNet++ network.

5. The automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data as described in claim 4, characterized in that: The optimizations to the PointNet++ network include: Add a height difference weighted sampling layer (weight=1 / ∣zi−zˉ∣weight=1 / ∣zi−zˉ∣) to adapt to steep terrain; Output the rock mass structural plane orientation (dip / dip angle), fracture density (fractures / 10m²), and rock mass integrity coefficient Kv.

6. The automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data as described in claim 1, characterized in that: The generation of the three-dimensional risk digital map in S6 includes: a basic layer, a disaster layer, and a risk layer.

7. The automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data as described in claim 6, characterized in that: The base layer: integrates DEM and geological structural lines, with a resolution ≤1m; Hazard layer: Mark the landslide boundary (accuracy error ≤ 2m) and collapse volume (error ≤ 10%). Risk level: Based on stability, it is divided into three levels: red (P≥0.7), yellow (0.4≤P<0.7), and green (P<0.4).

8. The automatic identification method for geological hazards on highway slopes in plateau and mountainous areas by integrating multiple data as described in claim 1, characterized in that: It also includes step (g) field verification mechanism: When the system identifies a disaster probability P≥0.7, it automatically dispatches a drone to perform oblique photography verification. By comparing historical point clouds with a change detection algorithm, the disaster body with a displacement greater than 10cm was verified.

Images