Multi-source data fusion slope monitoring and early warning method, system, equipment and medium

Through multi-source data fusion technology, a three-dimensional model of slopes and surface deformation map are constructed, combined with radar data to analyze potential sliding surfaces and stress strains, the problem of poor reliability of slope monitoring and early warning data in the existing technology is solved, and accurate and timely Safety warning.

CN118758225BActive Publication Date: 2024-11-12SHANDONG GUANGAN INTELLIGENT TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411237185.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-05
Publication Date
2024-11-12
Estimated Expiration
2044-09-05

AI Technical Summary

Technical Problem

The existing slope monitoring methods are single and the data sources are few, resulting in poor reliability of early warning data and ineffectively reducing safety losses caused by slope instability.

Method used

采用多源数据融合的方法,通过获取边坡区域的坐标数据和高程数据进行拼接处理,构建三维模型,并结合雷达数据分析地表形变,进行目标检测和数据融合,确定潜在滑动面和应力应变, Conduct accurate safety warnings.

Benefits of technology

It provides accurate, timely and reliable slope warning information, reduces safety losses caused by slope instability, and improves monitoring accuracy and reliability through multi-source data fusion.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present application relates to the technical field of slope monitoring and early warning, specifically to a slope monitoring and early warning method, system, equipment and medium with multi-source data fusion. In order to solve the problem of poor reliability of slope early warning in the prior art, the present application first splices the inclined images of the slope area to be inspected, and converts them into point clouds based on coordinate data and elevation data to obtain a three-dimensional model of the slope; then, based on the radar data of the slope area to be inspected at two moments (current and historical), the surface deformation of the current slope area to be inspected is analyzed, and the current slope safety and stability is analyzed in combination with the ground object information on the slope; finally, the slope deformation map and the slope three-dimensional model are data-fused, and the potential sliding surface of the current medium and low risk areas of the slope is determined based on the geological information of the slope, and the slope stress and strain analysis of the potential sliding surface is performed in combination with future meteorological information, the slope safety and stability is predicted, and accurate, timely and reliable safety warning information is released.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of slope monitoring and early warning, and in particular to a slope monitoring and early warning method, system, equipment and medium for multi-source data fusion. Background Art

[0002] The hazards caused by slope instability not only affect the safety of life and property, but also involve the ecological environment, social stability and economic development. The slope safety risk early warning system plays a vital role in preventing geological disasters such as landslides. It can not only reduce the direct hazards caused by geological disasters such as landslides, but also provide strong support for disaster management, resource allocation and social emergency preparedness. Through scientific slope early warning and management, the risk of landslide disasters can be effectively reduced and the sustainable development of society can be promoted.

[0003] At present, the displacement and velocity of slopes are often monitored by radar and other equipment, and some pre-processing measures are taken in combination with the monitoring data to reduce or avoid slope safety accidents. Due to the single slope monitoring method, few data sources, and poor reliability of early warning data, it is easy to cause the unreliable monitoring results and fail to reduce the adverse consequences of slope instability losses. Summary of the invention

[0004] The purpose of the present invention is to provide an accurate, timely and reliable multi-source data fusion slope monitoring and early warning method, system, equipment and medium.

[0005] The technical solution of the present invention is as follows:

[0006] A slope monitoring and early warning method based on multi-source data fusion includes the following operations:

[0007] S1. Obtain the coordinate data and elevation data of the current slope area to be inspected; based on the coordinate data, stitch several inclined images of the slope area to be inspected to obtain a slope area map; based on the elevation data, convert the pixels in the slope area map into a point cloud to obtain a slope point cloud model; map the color information in the slope area map to the slope point cloud model to obtain a three-dimensional slope model;

[0008] S2, respectively obtain the radar data of the slope area to be inspected at the current and historical neighborhood moments, obtain the phase difference of the corresponding position points after image registration processing, and obtain the slope phase difference map; the slope phase difference map is converted into a surface deformation map after least squares processing to obtain a slope deformation map; the slope deformation map is processed by target detection to obtain the slope building distribution information and slope vegetation distribution information; based on the slope building distribution information, slope vegetation distribution information and surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas; slope safety warning information is issued to the high-risk area; and the operation in S3 is performed on the low-risk area;

[0009] S3. After registering and density matching the slope deformation map and the three-dimensional slope model, data fusion processing is performed based on the corresponding weights to obtain the three-dimensional deformation model of the slope; the potential sliding surface of the three-dimensional slope deformation model is determined based on the stratigraphic structure information and lithology information, geological structure information, slope direction, slope gradient and slope angle of the slope area to be inspected; based on the rainfall duration and rainfall amount in the future neighborhood time, and the corresponding rock and soil information of the potential sliding surface, the slope stress and strain analysis is performed to obtain the slope safety factor; slope safety warning information is issued for areas where the slope safety factor is less than the safety factor threshold.

[0010] The stitching operation in S1 is as follows: in the current oblique image, the pixel values ​​of which are extreme values ​​after Gaussian blur processing at different scales and whose neighborhood contrast is less than the contrast threshold are taken as feature points; the pixel points at the corresponding positions of the feature points in the standard image are taken as standard feature points; based on the pixel values ​​of the feature points and the pixel values ​​of the standard feature points, a pixel loss function is constructed, and a pixel loss matrix is ​​obtained by least squares processing; based on the pixel loss matrix, the current oblique image is pixel adjusted to obtain the current adjusted image; each adjusted image is sequentially connected with the adjusted images corresponding to the adjacent acquisition times by pixel mixing in the same area to obtain a slope area map.

[0011] After the image registration processing operation in S2, interpolation processing is also included to obtain the current slope radar interpolation map and the historical slope radar interpolation map, which are used to perform the operation of obtaining the phase difference; the interpolation processing operation is specifically: based on each target pixel point in the current slope radar registration map or the historical slope radar registration map, the neighborhood pixel value, neighborhood pixel position and target pixel position at the corresponding position in the current slope radar registration map or the historical slope radar registration map, linear interpolation processing is performed on the corresponding target pixel points in the current slope radar registration map or the historical slope radar registration map in the horizontal direction and the vertical direction in turn to obtain the current slope radar interpolation map or the historical slope radar interpolation map.

[0012] The specific operations of data fusion processing in S3 are as follows: based on the coordinate data and elevation data of the slope area to be inspected, the slope deformation matching map is converted into point cloud data to obtain an initial slope deformation model; the initial slope deformation model and the slope three-dimensional matching model are converted into Bayesian functions respectively to obtain an initial slope deformation Bayesian model and a slope three-dimensional Bayesian model; based on the corresponding weights and Gaussian distribution function, the likelihood functions of the initial slope deformation Bayesian model and the slope three-dimensional Bayesian model are obtained to obtain an initial slope deformation likelihood model and a slope three-dimensional likelihood model; the initial slope deformation likelihood model and the slope three-dimensional likelihood model are processed by maximum a posteriori estimation to obtain a slope three-dimensional deformation model.

[0013] The S2 medium and high risk areas are: areas in the slope deformation map where the average surface deformation value is greater than the first surface deformation threshold and the surface deformation variance is less than the variance threshold; or areas in the slope deformation map where the average surface deformation value is not greater than the first surface deformation threshold but greater than the second surface deformation threshold and the building distribution area is greater than the building distribution area threshold.

[0014] Before the stitching operation in S1, it also includes image enhancement processing on several oblique images to obtain several oblique enhanced images for performing the stitching operation; the image enhancement operation is specifically as follows: the current oblique image is smoothed at different scales to obtain several smoothed images; the logarithmic difference of the pixel values ​​between each smoothed image and the current oblique image is obtained to obtain several frequency feature maps; the several frequency feature maps are weighted fused to obtain the current initial enhanced image; the current initial enhanced image is color restored to obtain the current oblique enhanced image.

[0015] The color restoration processing operation is specifically as follows: based on the pixel average value of the current initial enhanced image and the color channel pixel value of the current oblique image, a color restoration coefficient is obtained; based on the color restoration coefficient, pixel points in the current initial enhanced image are adjusted to obtain the current oblique enhanced image.

[0016] A multi-source data fusion slope monitoring and early warning system, used to implement the multi-source data fusion slope monitoring and early warning method, comprising:

[0017] The slope 3D model generation module is used to obtain the coordinate data and elevation data of the current slope area to be inspected; based on the coordinate data, several inclined images of the slope area to be inspected are spliced ​​to obtain a slope area map; based on the elevation data, the pixels in the slope area map are converted into a point cloud to obtain a slope point cloud model; the color information in the slope area map is mapped to the slope point cloud model to obtain a slope 3D model;

[0018] The slope deformation map generation module is used to obtain the radar data of the slope area to be inspected at the current and historical neighborhood moments, and obtain the phase difference of the corresponding position points after image registration processing to obtain the slope phase difference map; the slope phase difference map is converted into a surface deformation map after least squares processing to obtain the slope deformation map;

[0019] The current slope monitoring result generation module is used to process the slope deformation map through target detection to obtain the slope building distribution information and slope vegetation distribution information; based on the slope building distribution information, slope vegetation distribution information and surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas; the operations of the early warning module are executed for the high-risk areas, and the operations in the future slope monitoring result generation module are executed for the low-risk areas;

[0020] The future slope monitoring result generation module aligns and density-matches the slope deformation map and the three-dimensional slope model, and then performs data fusion processing based on the corresponding weights to obtain the three-dimensional deformation model of the slope; based on the stratigraphic structure information and lithology information, geological structure information, slope direction, slope gradient and slope angle of the slope area to be inspected, the potential sliding surface of the three-dimensional deformation model of the slope is determined; based on the rainfall duration and rainfall amount in the future neighborhood time, and the corresponding geotechnical information of the potential sliding surface, the slope stress and strain analysis is performed to obtain the slope safety factor; the operations in the early warning module are executed for areas where the slope safety factor is less than the safety factor threshold;

[0021] Early warning module, used for issuing slope safety early warning information.

[0022] A multi-source data fusion slope monitoring and early warning device comprises a processor and a memory, wherein the processor implements the multi-source data fusion slope monitoring and early warning method when executing a computer program stored in the memory.

[0023] A computer-readable storage medium is used to store a computer program, wherein the computer program, when executed by a processor, implements the above-mentioned slope monitoring and early warning method of multi-source data fusion.

[0024] The beneficial effects of the present invention are:

[0025] The slope monitoring and early warning method of multi-source data fusion provided by the present invention first splices the inclined images of the slope area to be inspected, converts them into point clouds based on coordinate data and elevation data, and maps color information to the point cloud model to obtain a high-precision three-dimensional slope model that can clearly and accurately reflect the actual terrain characteristics of the slope, which is convenient for subsequent identification of potential sliding surfaces of the slope and slope stability analysis; then, based on the radar data of the slope area to be inspected at two moments (current and historical), the surface deformation of the slope area to be inspected is analyzed to obtain a slope deformation map, and target detection processing is performed based on the slope deformation map to obtain the ground object information on the surface of the slope area to be inspected, and the safety of the current slope area to be inspected is preliminarily analyzed by combining the ground object information with the slope deformation map. Analysis is performed to divide high-risk areas and low-risk areas. Slope safety warning information is immediately issued for high-risk areas, while further future slope safety prediction analysis is performed for low-risk areas. Finally, the slope deformation map and the slope three-dimensional model are fused to obtain a three-dimensional slope deformation model with high spatial resolution that can quickly and accurately reflect surface deformation. Based on the slope geological information, the potential sliding surface is determined, and the slope stress and strain analysis of the potential sliding surface is performed in combination with future meteorological information to predict the safety and stability of the slope, and to issue accurate and timely safety warning information. The slope monitoring and early warning method has rich data sources and accurate information analysis, and can provide accurate, timely and reliable slope warning information to reduce safety losses caused by slope instability. DETAILED DESCRIPTION

[0026] This embodiment provides a slope monitoring and early warning method based on multi-source data fusion, including the following operations:

[0027] S1. Obtain the coordinate data and elevation data of the current slope area to be inspected; based on the coordinate data, stitch several inclined images of the slope area to be inspected to obtain a slope area map; based on the elevation data, convert the pixels in the slope area map into a point cloud to obtain a slope point cloud model; map the color information in the slope area map to the slope point cloud model to obtain a three-dimensional slope model;

[0028] S2, respectively obtain the radar data of the slope area to be inspected at the current and historical neighborhood moments, obtain the phase difference of the corresponding position points after image registration processing, and obtain the slope phase difference map; the slope phase difference map is converted into a surface deformation map after least squares processing to obtain a slope deformation map; the slope deformation map is processed by target detection to obtain the slope building distribution information and slope vegetation distribution information; based on the slope building distribution information, slope vegetation distribution information and surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas; slope safety warning information is issued to the high-risk area; and the operation in S3 is performed on the low-risk area;

[0029] S3. After registering and density matching the slope deformation map and the three-dimensional slope model, data fusion processing is performed based on the corresponding weights to obtain the three-dimensional deformation model of the slope; the potential sliding surface of the three-dimensional slope deformation model is determined based on the stratigraphic structure information and lithology information, geological structure information, slope direction, slope gradient and slope angle of the slope area to be inspected; based on the rainfall duration and rainfall amount in the future neighborhood time, and the corresponding rock and soil information of the potential sliding surface, the slope stress and strain analysis is performed to obtain the slope safety factor; slope safety warning information is issued for areas where the slope safety factor is less than the safety factor threshold.

[0030] S1. Obtain the coordinate data and elevation data of the current slope area to be inspected; based on the coordinate data, stitch several inclined images of the slope area to be inspected to obtain a slope area map; based on the elevation data, convert the pixels in the slope area map into a point cloud to obtain a slope point cloud model; map the color information in the slope area map to the slope point cloud model to obtain a three-dimensional slope model.

[0031] The inclined images of the slope area to be inspected are stitched, the coordinate data and elevation data are converted into point clouds, and the color information is mapped to the point cloud model to obtain a high-precision three-dimensional slope model that can clearly and accurately reflect the actual terrain characteristics of the slope, facilitating the subsequent identification of the potential sliding surface of the slope and the slope stability analysis.

[0032] First, the coordinate data and elevation data of the current slope area to be inspected are obtained, which are used as the X-axis, Y-axis and Z-axis data for the subsequent three-dimensional model construction. The edge clearly reflects the position information of the slope.

[0033] Then, based on the coordinate data, several oblique images of the slope area to be inspected are spliced ​​to obtain a complete slope area map. Several oblique images can be achieved through drone oblique photography.

[0034] Among them, the stitching processing operation is specifically as follows: in the current oblique image, the pixel values ​​are all extreme values ​​after Gaussian blur processing at different scales, and the pixel points whose neighborhood contrast is less than the contrast threshold are taken as feature points; the pixel points at the corresponding positions of the feature points in the standard image are taken as standard feature points; based on the pixel values ​​of the feature points and the pixel values ​​of the standard feature points, a pixel loss function is constructed, and a pixel loss matrix is ​​obtained by least squares processing; based on the pixel loss matrix, the current oblique image is pixel adjusted to obtain the current adjusted image; each adjusted image is connected with the adjacent coordinate adjusted image in turn, and pixel mixing in the same area is performed (which can be achieved through pixel gradient mixing) to obtain a slope area map.

[0035] In order to further improve the quality of the spliced ​​slope area map, after obtaining the slope area map, the splicing area optimization processing of the slope area map is also included to make the image more continuous at the splicing point, and the optimized slope area map is obtained for performing the operation of converting into a point cloud.

[0036] The specific operation of stitching area optimization processing is as follows: when the sliding window traverses the stitching area in the slope area map, the pixel mean square error of each sliding area is obtained to determine whether the pixel distribution is uniform; the sliding area corresponding to the minimum pixel mean square error is used as the standard stitching area; the sliding area in the slope area map where the pixel mean square error is greater than the error threshold is replaced with the standard stitching area to obtain the optimized slope area map.

[0037] In order to prevent the oblique images from having poor image quality due to weather reasons (thick fog and low visibility), to improve the quality of the subsequently obtained slope area map and to enhance the detail expression capability of the slope area map, before the above-mentioned stitching operation, it also includes performing image enhancement processing on several oblique images respectively to obtain several oblique enhanced images for performing the stitching operation.

[0038] The specific operation of image enhancement processing is as follows: the current tilted image is smoothed at different scales (preferably by a Gaussian filter) to obtain several smoothed images; the logarithmic difference of the pixel values ​​between each smoothed image and the current tilted image is obtained to obtain several frequency feature maps; the several frequency feature maps are weighted fused to obtain the current initial enhanced image; the current initial enhanced image is color restored to prevent the color distortion of the current initial enhanced image, to obtain the current tilted enhanced image.

[0039] The color restoration processing operation is specifically as follows: based on the pixel average value of the current initial enhanced image and the color channel pixel value of the current oblique image, a color restoration coefficient is obtained; based on the color restoration coefficient, pixel points in the current initial enhanced image are adjusted to obtain the current oblique enhanced image.

[0040] Next, based on the elevation data, the pixels in the slope area map are converted into point clouds. Each point cloud includes both position data (coordinates and elevation information) and pixel information, and a slope point cloud model is obtained.

[0041] Finally, in order to further improve the realism of the slope point cloud model and facilitate the subsequent division of risk areas, the color information in the slope area map is mapped to the slope point cloud model to obtain the three-dimensional slope model.

[0042] S2. Obtain radar data of the slope area to be inspected at the current and historical neighborhood moments respectively, obtain the phase difference of the corresponding position points after image registration processing, and obtain the slope phase difference map; the slope phase difference map is converted into a surface deformation map after least squares processing to obtain the slope deformation map; the slope deformation map is processed by target detection to obtain the slope building distribution information and the slope vegetation distribution information; based on the slope building distribution information, the slope vegetation distribution information and the surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas; slope safety warning information is issued to the high-risk area; and the operation in S3 is performed on the low-risk area.

[0043] Based on the radar data of the slope area to be inspected at two moments (current and historical), the surface deformation of the slope area to be inspected is analyzed to obtain the slope deformation map, and target detection processing is performed based on the slope deformation map to obtain the ground object information on the surface of the slope area to be inspected. Combining the ground object information with the slope deformation map, a preliminary analysis of the safety of the slope area to be inspected is performed, and high-risk areas and low-risk areas are divided. Slope safety warning information is immediately issued for high-risk areas, while further slope safety future prediction analysis in S3 is performed for low-risk areas.

[0044] First, in order to obtain the current deformation distribution of the slope, radar data (preferably InSAR data) of the slope area to be inspected at the current and historical neighborhood moments covering the same area are obtained respectively to obtain the current slope radar map and the historical slope radar map; the current slope radar map and the historical slope radar map are processed by image registration to ensure that each position point represents the same ground point in the two images, and the current slope radar registration map and the historical slope radar registration map are obtained; the phase difference (pixel difference) of the corresponding position points in the current slope radar registration map and the historical slope radar registration map is obtained to obtain a slope phase difference map that can reflect comprehensive information such as surface deformation and terrain height difference between the current slope radar map and the historical slope radar map.

[0045] In order to prevent the pixel position from shifting during the above-mentioned image registration process and affecting the accuracy of the slope deformation analysis, the image registration process also includes interpolating the current slope radar registration map and the historical slope radar registration map to obtain the current slope radar interpolation map and the historical slope radar interpolation map, which are used to perform the operation of obtaining the phase difference.

[0046] The interpolation operation is specifically as follows: based on each target pixel point in the current slope radar registration map or the historical slope radar registration map, the neighborhood pixel value, neighborhood pixel position and target pixel position at the corresponding position in the current slope radar registration map or the historical slope radar registration map are linearly interpolated in the horizontal and vertical directions respectively to obtain the current slope radar interpolation map or the historical slope radar interpolation map, so that the image maintains continuity and smoothness, making the subsequently obtained slope phase difference map more accurate.

[0047] In order to further improve the accuracy of the slope phase difference map, the interpolation process also includes a sharpening process to obtain the current slope radar sharpening map and the historical slope radar sharpening map for performing the phase difference acquisition process.

[0048] Taking the current slope radar interpolation map as an example, the sharpening operation is as follows: grayscale processing and convolution processing are performed on the current slope radar interpolation map to obtain the current slope radar grayscale map; the current slope radar grayscale map and the current slope radar interpolation map are superimposed after weighted processing to obtain the current slope radar feature map; the contrast in the current slope radar data feature map is adjusted to the preset contrast and then denoising is performed to obtain the current slope radar sharpened map.

[0049] The sharpening process of the historical slope radar interpolation image is the same as above, so it will not be described here in detail to save space.

[0050] Then, in order to facilitate the intuitive analysis of the slope surface movement, the slope phase difference map is converted into a surface deformation map after being processed by the least squares method. Color or vector field is used to represent different deformation quantities and directions, and a slope deformation map is obtained that can analyze the temporal and spatial distribution of surface changes and facilitate the identification of potential geological disasters.

[0051] Next, target detection processing is performed on the slope deformation map to obtain the slope building distribution information and slope vegetation distribution information.

[0052] Among them, the specific operations of target detection processing are: performing bounding box regression and classification processing on the slope deformation map to obtain an initial target detection map containing multiple detection frames in the same area, each detection frame containing multiple classification types and corresponding classification probabilities; in the initial detection map, the classification type corresponding to the maximum classification probability is used as the detection result of the corresponding detection frame, and the detection results of all detection frames form the slope target detection results, which include slope building distribution information and slope vegetation distribution information.

[0053] Subsequently, based on the slope building distribution information, slope vegetation distribution information and surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas. Among them, the high-risk area is the area in the slope deformation map where the average surface deformation is greater than the first surface deformation threshold and the surface deformation variance is less than the variance threshold. Such areas have large surface deformation and are prone to geological disasters; or the area in the slope deformation map where the average surface deformation is not greater than the first surface deformation threshold but greater than the second surface deformation threshold, and the building distribution area is greater than the building distribution area threshold. Such areas have moderate surface deformation, but are also prone to geological disasters due to the increase in gravity caused by large-scale buildings above.

[0054] The first ground surface deformation threshold is greater than the second ground surface deformation threshold.

[0055] Finally, slope safety warning information is released for high-risk areas, and the operations in S3 are performed for low-risk areas to further analyze the future safety of the slope.

[0056] S3. After registering and density matching the slope deformation map and the three-dimensional slope model, data fusion processing is performed based on the corresponding weights to obtain the three-dimensional deformation model of the slope; the potential sliding surface of the three-dimensional slope deformation model is determined based on the stratigraphic structure information, geological structure information, slope direction, slope gradient and slope angle of the slope area to be inspected; based on the rainfall duration and rainfall amount in the future neighborhood time, as well as the corresponding geotechnical information of the potential sliding surface, the slope stress-strain analysis is performed to obtain the slope safety factor; slope safety warning information is issued for areas where the slope safety factor is less than the safety factor threshold.

[0057] The slope deformation map and the slope three-dimensional model are fused to obtain a slope three-dimensional deformation model with high spatial resolution that can quickly and accurately reflect surface deformation; the potential sliding surface is determined based on the slope geological information; and the slope stress and strain analysis of the potential sliding surface is carried out in combination with future meteorological information to predict the safety and stability of the slope and issue accurate and timely safety warning information.

[0058] First, in order to more accurately describe the three-dimensional morphological changes of the slope surface and facilitate subsequent analysis to obtain accurate slope safety factors, the slope deformation map that can accurately reflect the surface deformation information and the three-dimensional slope model with high spatial resolution and accuracy are successively registered (which can be achieved through control points or automated methods) and density matched. Then, data fusion processing is performed based on the corresponding weights to obtain a three-dimensional slope deformation model with high spatial resolution and accurate surface deformation, which is beneficial to enhance the monitoring capability of slope surface deformation.

[0059] The above registration can be achieved by geometrically correcting the three-dimensional slope model and the slope deformation map so that the two data are aligned in space; the above registration can also be achieved by spatially registering the three-dimensional slope model and the slope deformation map in three-dimensional space through control points or automated methods.

[0060] The density matching can be achieved by performing point cloud thinning processing on the registered three-dimensional slope model and / or performing interpolation processing on the registered slope deformation map to achieve data density matching.

[0061] The operations of the above data fusion processing are specifically as follows: based on the coordinate data and elevation data of the slope area to be inspected, the slope deformation matching map is converted into point cloud data to obtain an initial slope deformation model; the initial slope deformation model and the slope three-dimensional matching model are converted into Bayesian functions respectively to obtain an initial slope deformation Bayesian model and a slope three-dimensional Bayesian model; based on the corresponding weights and Gaussian distribution function, the likelihood functions of the initial slope deformation Bayesian model and the slope three-dimensional Bayesian model are obtained to obtain an initial slope deformation likelihood model and a slope three-dimensional likelihood model; the initial slope deformation likelihood model and the slope three-dimensional likelihood model are processed by maximum a posteriori estimation to obtain a slope three-dimensional deformation model in which each point integrates point cloud and radar information, thereby providing more accurate surface deformation analysis results.

[0062] Then, based on the stratigraphic structure information of the slope area to be inspected (including the lithology, bedding, joints and other characteristics of different strata, which can be obtained through drilling sampling), geological structure information (faults, folds, cracks and rock layer inclination, etc.), slope direction, slope gradient and slope angle, the weak structural surfaces where sliding may occur are preliminarily identified, and a limit equilibrium analysis is performed to determine the potential sliding surface of the three-dimensional deformation model of the slope.

[0063] Next, based on the rainfall duration and amount in the future neighborhood, as well as the corresponding geotechnical information of the potential sliding surface, slope stress-strain analysis is performed (which can be achieved through the slope finite element analysis method) to obtain the slope safety factor for a period of time in the future and conduct a future assessment of the slope safety and stability.

[0064] Finally, slope safety warning information is issued for areas where the slope safety factor is less than the safety factor threshold.

[0065] This embodiment also provides a multi-source data fusion slope monitoring and early warning system, which is used to implement the multi-source data fusion slope monitoring and early warning method, including:

[0066] The slope 3D model generation module is used to obtain the coordinate data and elevation data of the current slope area to be inspected; based on the coordinate data, several inclined images of the slope area to be inspected are spliced ​​to obtain a slope area map; based on the elevation data, the pixels in the slope area map are converted into a point cloud to obtain a slope point cloud model; the color information in the slope area map is mapped to the slope point cloud model to obtain a slope 3D model;

[0067] The slope deformation map generation module is used to obtain the radar data of the slope area to be inspected at the current and historical neighborhood moments, and obtain the phase difference of the corresponding position points after image registration processing to obtain the slope phase difference map; the slope phase difference map is converted into a surface deformation map after least squares processing to obtain the slope deformation map;

[0068] The current slope monitoring result generation module is used to process the slope deformation map through target detection to obtain the slope building distribution information and slope vegetation distribution information; based on the slope building distribution information, slope vegetation distribution information and surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas; the operations of the early warning module are executed for the high-risk areas, and the operations in the future slope monitoring result generation module are executed for the low-risk areas;

[0069] The future slope monitoring result generation module aligns and density-matches the slope deformation map and the three-dimensional slope model, and then performs data fusion processing based on the corresponding weights to obtain the three-dimensional deformation model of the slope; based on the stratigraphic structure information and lithology information, geological structure information, slope direction, slope gradient and slope angle of the slope area to be inspected, the potential sliding surface of the three-dimensional deformation model of the slope is determined; based on the rainfall duration and rainfall amount in the future neighborhood time, and the corresponding geotechnical information of the potential sliding surface, the slope stress and strain analysis is performed to obtain the slope safety factor; the operations in the early warning module are executed for areas where the slope safety factor is less than the safety factor threshold;

[0070] Early warning module, used for issuing slope safety early warning information.

[0071] This embodiment also provides a multi-source data fusion slope monitoring and early warning device, including a processor and a memory, wherein the processor implements the above-mentioned multi-source data fusion slope monitoring and early warning method when executing the computer program stored in the memory.

[0072] This embodiment further provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the above-mentioned slope monitoring and early warning method of multi-source data fusion.

[0073] The multi-source data fusion slope monitoring and early warning method provided in this embodiment first splices the inclined images of the slope area to be inspected, converts them into point clouds based on coordinate data and elevation data, and maps color information to the point cloud model to obtain a high-precision three-dimensional slope model that can clearly and accurately reflect the actual terrain characteristics of the slope, which is convenient for subsequent identification of potential sliding surfaces of the slope and slope stability analysis; then, based on the radar data of the slope area to be inspected at two moments (current and historical), the surface deformation of the slope area to be inspected is analyzed to obtain a slope deformation map, and target detection processing is performed based on the slope deformation map to obtain the ground object information on the surface of the slope area to be inspected, and the safety of the current slope area to be inspected is preliminarily evaluated by combining the ground object information with the slope deformation map. Analysis is performed to divide high-risk areas and low-risk areas. Slope safety warning information is immediately issued for high-risk areas, while further future slope safety prediction analysis is performed for low-risk areas. Finally, the slope deformation map and the slope three-dimensional model are fused to obtain a three-dimensional slope deformation model with high spatial resolution that can quickly and accurately reflect surface deformation. Based on the slope geological information, the potential sliding surface is determined, and the slope stress and strain analysis of the potential sliding surface is performed in combination with future meteorological information to predict the safety and stability of the slope, and to issue accurate and timely safety warning information. The slope monitoring and early warning method has rich data sources and accurate information analysis, and can provide accurate, timely and reliable slope warning information to reduce safety losses caused by slope instability.

Claims

1. A slope monitoring and early warning method based on multi-source data fusion, characterized in that: The following operations are included: S1. Obtain the coordinate data and elevation data of the current slope area to be inspected; based on the coordinate data, stitch several oblique images of the slope area to be inspected to obtain a slope area map; the stitching operation is specifically as follows: in the current oblique image, the pixel values ​​are all extreme values ​​after Gaussian blur processing at different scales, and the pixel points whose neighborhood contrast is less than the contrast threshold are taken as feature points; the pixel points at the corresponding positions of the feature points in the standard image are taken as standard feature points; based on the pixel values ​​of the feature points and the pixel values ​​of the standard feature points, a pixel loss function is constructed, and a pixel loss matrix is ​​obtained by least squares processing; Based on the pixel loss matrix, the current tilted image is pixel-adjusted to obtain the current adjusted image; each adjusted image is sequentially connected with the adjusted images corresponding to the adjacent acquisition time, and the pixel mixing in the same area is processed to obtain the slope area map; based on the elevation data, the pixel points in the slope area map are converted into a point cloud to obtain a slope point cloud model; the color information in the slope area map is mapped to the slope point cloud model to obtain a three-dimensional slope model; S2, respectively obtaining radar data of the slope area to be inspected at the current and historical neighborhood moments, obtaining the phase difference of the corresponding position points after image registration processing, and obtaining a slope phase difference map; the slope phase difference map is converted into a surface deformation map after least squares processing to obtain a slope deformation map; The slope deformation map is processed by target detection to obtain slope building distribution information and slope vegetation distribution information; Based on the slope building distribution information, slope vegetation distribution information and surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas; The high-risk area is: in the slope deformation map, the area where the average surface deformation is greater than the first surface deformation threshold and the surface deformation variance is less than the variance threshold; or in the slope deformation map, the area where the average surface deformation is not greater than the first surface deformation threshold but greater than the second surface deformation threshold, and the building distribution area is greater than the building distribution area threshold; The first ground deformation threshold is greater than the second ground deformation threshold; Issue slope safety warning information for the high-risk areas; Perform the operation in S3 on the low-risk area; S3, after registering and density matching the slope deformation map and the three-dimensional slope model, data fusion processing is performed based on corresponding weights to obtain a three-dimensional slope deformation model; the specific operation of the data fusion processing is as follows: based on the coordinate data and elevation data of the slope area to be inspected, the slope deformation matching map is converted into point cloud data to obtain an initial slope deformation model; the initial slope deformation model and the three-dimensional slope matching model are converted into Bayesian functions respectively to obtain an initial slope deformation Bayesian model and a three-dimensional slope Bayesian model; based on the corresponding weights and Gaussian distribution function, the likelihood function of the initial slope deformation Bayesian model and the three-dimensional slope Bayesian model is obtained to obtain an initial slope deformation likelihood model and a three-dimensional slope likelihood model; the initial slope deformation likelihood model and the three-dimensional slope likelihood model are subjected to maximum a posteriori estimation processing to obtain the three-dimensional slope deformation model; Based on the stratigraphic structure information, geological structure information, slope direction, slope gradient and slope angle of the slope area to be inspected, the potential sliding surface of the slope three-dimensional deformation model is determined; based on the rainfall duration and rainfall amount in the future neighborhood time, as well as the geotechnical information corresponding to the potential sliding surface, the slope stress and strain analysis is performed to obtain the slope safety factor; Slope safety warning information will be issued for areas where the slope safety factor is less than the safety factor threshold.

2. The slope monitoring and early warning method based on multi-source data fusion according to claim 1 is characterized in that: After the image registration process in S2, interpolation processing is also included to obtain the current slope radar interpolation map and the historical slope radar interpolation map for performing the operation of obtaining the phase difference; The interpolation operation is specifically as follows: based on each target pixel point in the current slope radar registration map or the historical slope radar registration map, the neighborhood pixel value, the neighborhood pixel position and the target pixel position at the corresponding position in the current slope radar registration map or the historical slope radar registration map, linear interpolation processing is performed on the corresponding target pixel points in the current slope radar registration map or the historical slope radar registration map in the horizontal direction and the vertical direction in turn to obtain the current slope radar interpolation map or the historical slope radar interpolation map.

3. The slope monitoring and early warning method based on multi-source data fusion according to claim 1 is characterized in that: Before the stitching operation in S1, the process further includes performing image enhancement processing on the plurality of oblique images to obtain a plurality of oblique enhanced images for performing the stitching operation; The operations of the image enhancement processing are specifically as follows: the current tilted image is smoothed at different scales to obtain a number of smoothed images; the logarithmic difference of the pixel values ​​between each smoothed image and the current tilted image is obtained to obtain a number of frequency feature maps; the number of frequency feature maps are weighted fused to obtain the current initial enhanced image; the current initial enhanced image is color restored to obtain the current tilted enhanced image.

4. The slope monitoring and early warning method based on multi-source data fusion according to claim 3 is characterized in that: The color restoration process is specifically performed as follows: A color restoration coefficient is obtained based on the pixel average of the current initial enhanced image and the color channel pixel value of the current oblique image; based on the color restoration coefficient, pixel points in the current initial enhanced image are adjusted to obtain the current oblique enhanced image.

5. A multi-source data fusion slope monitoring and early warning system, used to implement the multi-source data fusion slope monitoring and early warning method according to claim 1, characterized in that: include: The slope three-dimensional model generation module is used to obtain the coordinate data and elevation data of the current slope area to be inspected; Based on the coordinate data, several inclined images of the slope area to be inspected are spliced ​​to obtain a slope area map; based on the elevation data, the pixels in the slope area map are converted into a point cloud to obtain a slope point cloud model; the color information in the slope area map is mapped to the slope point cloud model to obtain a three-dimensional slope model; The slope deformation map generation module is used to obtain the radar data of the slope area to be inspected at the current and historical neighborhood moments respectively, obtain the phase difference of the corresponding position points after image registration processing, and obtain the slope phase difference map; the slope phase difference map is converted into a surface deformation map after least squares processing to obtain the slope deformation map; The current slope monitoring result generation module is used to process the slope deformation map through target detection to obtain slope building distribution information and slope vegetation distribution information; Based on the slope building distribution information, slope vegetation distribution information and surface deformation threshold, the slope deformation map is divided into high-risk areas and low-risk areas; Execute the operations of the early warning module for the high-risk area, and execute the operations of the future slope monitoring result generation module for the low-risk area; The future slope monitoring result generation module performs registration and density matching on the slope deformation map and the three-dimensional slope model, and then performs data fusion processing based on the corresponding weights to obtain the three-dimensional slope deformation model; determines the potential sliding surface of the three-dimensional slope deformation model based on the stratigraphic structure information and lithology information, geological structure information, slope direction, slope gradient and slope angle of the slope area to be inspected; performs slope stress and strain analysis based on the rainfall duration and rainfall amount in the future neighborhood time, and the corresponding geotechnical information of the potential sliding surface to obtain the slope safety factor; Execute the operation in the early warning module for the area where the slope safety factor is less than the safety factor threshold; Early warning module, used for issuing slope safety early warning information.

6. A slope monitoring and early warning device with multi-source data fusion, characterized in that: The method comprises a processor and a memory, wherein the processor implements the slope monitoring and early warning method of multi-source data fusion as described in any one of claims 1 to 4 when executing the computer program stored in the memory.

7. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the slope monitoring and early warning method based on multi-source data fusion as described in any one of claims 1 to 4 is implemented.

Citation Information

Patent Citations

  • Multi-dimensional information labeling method and device, electronic equipment and storage medium

    CN116958858A

  • Landslide trailing edge crack identification method based on multi-source data

    CN117348022A

  • Slope monitoring and early warning method, computer readable storage medium and computer equipment

    CN117739879A

  • Method and system for realizing disaster early warning by means of deformation identification of river channel landslide

    WO2023284344A1

  • Measurement method and system based on interferometric synthetic aperture radar

    WO2024175129A1