A method for determining the early warning criterion of landslide evolution stages by integrating multi-source heterogeneous remote sensing data

Abstract

The present invention provides a method for determining the early warning criterion for the landslide evolution stage by fusing multi-source heterogeneous remote sensing data, which relates to the technical fields of landslide disaster monitoring and early warning and landslide identification. The method comprises the following steps: acquiring and preprocessing SAR images to obtain the preprocessed SAR images; processing the preprocessed SAR images by Stacking-InSAR technology and SBAS-InSAR technology to obtain surface deformation data; extracting the surface rupture characteristics of the landslide to obtain surface rupture information; integrating multi-source data based on the surface deformation data and the surface rupture information to establish a time series sample library; based on the time series sample library, using the Apriori algorithm to mine the association relationship between deformation and rupture, and combining the hash coding strategy to refine features, so as to determine the identification factors in the landslide evolution process. The present invention can effectively improve the early recognition efficiency of landslides and provide strong support for the early warning and prevention of landslide disasters.

Description

Technical Field

[0001] The present invention provides a method for determining a landslide evolution stage warning criterion by integrating multi-source heterogeneous remote sensing data, which relates to the technical fields of landslide disaster monitoring and warning and landslide identification. Background Technique

[0002] Currently, the main methods for landslide identification include total station surveying method, GPS surveying, unmanned aerial vehicle (UAV) remote sensing monitoring, and synthetic aperture radar differential interferometry (InSAR) surveying.

[0003] Among them, total station surveying can simultaneously perform angle (horizontal angle and vertical angle) measurement and distance measurement. In landslide deformation monitoring, by setting up a station at a control point with known coordinates and observing the monitoring points on the landslide body, information such as the horizontal angle, vertical angle, and inclined distance of the monitoring points can be obtained. Then, according to the trigonometric function relationship, the three-dimensional coordinates of the monitoring points are calculated. Through multiple measurements and comparing the coordinate changes of the monitoring points at different times, the displacement conditions of the monitoring points in the horizontal and vertical directions can be obtained. This method can measure both horizontal and vertical displacements and comprehensively understand the deformation state of the landslide body. However, this method requires line-of-sight between the monitoring points and the control points, which may be restricted in areas with complex terrain, dense vegetation, or building obstructions.

[0004] GPS surveying is to use a GPS satellite signal receiving device to measure the three-dimensional coordinates of the monitoring points and obtain the deformation amount of the slope surface by comparing the elevation data at different times. It has the characteristics of high precision and high efficiency, can realize large-area and long-distance real-time dynamic monitoring, is not restricted by line-of-sight conditions, and can obtain horizontal and vertical displacement information at the same time. However, GPS signals may be affected by terrain, buildings, etc. and the accuracy will decrease in some complex environments, and the equipment cost is relatively high.

[0005] UAV remote sensing for landslide monitoring is based on aerial photogrammetry. By carrying high-resolution cameras, LiDAR (Light Detection and Ranging) or other sensors, it can obtain fine surface image data to achieve remote non-contact detection. Utilizing the low-altitude flight ability of UAVs, they can quickly reach the landslide area and take pictures and collect data of the landslide body from different angles and heights. The collected image data and other sensor data, after being processed and analyzed, can extract the characteristic information of the landslide body, such as topography, surface deformation, crack distribution, etc., so as to monitor the occurrence and development process of the landslide. When in use, UAVs can be quickly deployed according to monitoring requirements, fly at low altitude in complex terrains and inaccessible areas, obtain high-resolution images and data, and are not restricted by ground traffic and terrain conditions. Compared with traditional aerial remote sensing and satellite remote sensing, the equipment purchase, operation and maintenance costs of UAV remote sensing are relatively low, and there is no need to rent large aircraft or satellite resources, reducing the monitoring cost. However, the flight of UAVs is restricted by weather conditions (such as strong winds, rainfall, fog, etc.) and flight airspace. At the same time, its endurance is limited and the battery needs to be frequently replaced or refueled, which affects the continuity of monitoring. Although UAVs can conduct monitoring within a certain range, compared with satellite remote sensing, its monitoring range is relatively small. For large-scale landslide areas, multiple flights may be required to complete the monitoring task.

[0006] InSAR utilizes two or more SAR images acquired over the same area at different times by a synthetic aperture radar (SAR), and obtains minute surface deformation information through interferometric processing techniques. Its basic principle is based on the interference phenomenon of electromagnetic waves. When the radar wave irradiates a ground target and then reflects back to the radar antenna, due to the different distances between the targets at different positions and the radar, the echo signals will generate a phase difference. By processing SAR images of the same area at different times, the phase difference information is extracted and then converted into displacement change information of the surface. Stacking-InSAR, namely stacked synthetic aperture radar interferometry, is based on multiple SAR image data. By stacking multiple interferometric pairs in the same area, the signal is enhanced and the noise is suppressed, thereby improving the monitoring accuracy and reliability. This method mainly uses long-term SAR data, fully excavates the surface deformation information contained therein, and through comprehensive analysis and processing of the phase information of multiple interferometric pairs, separates the phase components related to landslide deformation, and then obtains the displacement change of the landslide body. It can obtain surface deformation information over a large area in a short time, but in areas with complex environmental geological conditions, other means need to be combined for more in-depth research. The small baseline subset interferometry (SBAS-InSAR) technology reduces the impact of spatio-temporal decorrelation by selecting a set of interferometric pairs with small baselines, thereby realizing the monitoring of surface deformation. The SBAS-InSAR technology increases the time sampling rate and spatial density in the observed data, and is not restricted by the linear deformation model, and can better restore the non-linear deformation characteristics of the surface and obtain the regional time series ground deformation. However, the processing flow of SBAS-InSAR is relatively complex, with a large amount of calculation, and has high requirements for the selection of baselines. If the baseline is selected improperly, errors may be introduced, affecting the accuracy of deformation monitoring. Aiming at the complex ground object scattering characteristics (uneven vegetation coverage) and deformation rate difference characteristics (measurement accuracy of different deformation rates in the same measurement area) of the landslide body.

[0007] Fusing InSAR and UAV for landslide monitoring can give full play to the advantages of both, improve the accuracy and efficiency of landslide monitoring. Stacking-InSAR and SBAS-InSAR technologies can obtain surface deformation information over a large spatial range. They can monitor surface deformations at the millimeter level or even smaller, providing key data for the early identification and dynamic monitoring of landslides. UAVs can acquire high-resolution optical images, which can clearly show the topographic details of the landslide area. By processing UAV images, the landslide boundary and surface feature information can be accurately extracted. Moreover, InSAR technology is not restricted by weather and lighting conditions and can monitor all day and all weather. It can obtain continuous deformation data over a large area and can effectively monitor the slow deformation of landslides. UAVs can take pictures of the landslide area at close range and can accurately determine the boundary range of the landslide. Combined with InSAR technology, it can more accurately locate and analyze the deformed area of the landslide. InSAR technology and UAV monitoring technology can verify the accuracy of the monitoring results with each other. By comparing the data obtained by the two technologies, potential errors and uncertainties can be found, improving the reliability of the monitoring results. Therefore, fusing InSAR and UAV for landslide deformation monitoring has very important engineering significance.

[0008] In addition, the fused data is very extensive. How to find the criteria that can truly be used for landslide evolution stage early warning is also a significant challenge. Summary of the Invention

[0009] The present invention provides a method for determining the early warning criteria for landslide evolution stages by fusing multi-source heterogeneous remote sensing data, effectively obtaining the evolution law of landslide change identification indicators, accurately and efficiently finding the criteria for landslide evolution stage early warning, and improving the efficiency of early landslide identification.

[0010] To solve the above problems, the present invention provides the following technical solutions:

[0011] In a first aspect, the present invention provides a method for determining the early warning criteria for landslide evolution stages by fusing multi-source heterogeneous remote sensing data, which includes the following steps:

[0012] Obtain SAR images and preprocess them to obtain preprocessed SAR images;

[0013] After processing the preprocessed SAR images by Stacking-InSAR technology and SBAS-InSAR technology, obtain surface deformation data;

[0014] Use UAV photogrammetry and LiDAR data to extract the surface rupture characteristics of the landslide, construct a three-dimensional virtual reality model, and obtain surface rupture information;

[0015] Based on surface deformation data and surface rupture information, integrate multi-source data to establish a time-series sample library;

[0016] Based on the time-series sample library, use the Apriori algorithm to mine the correlation between deformation and rupture, combine the hash coding strategy to refine features, and determine the identification factors in the landslide evolution process.

[0017] Furthermore, the preprocessing method of SAR images includes:

[0018] For image registration, use the main image as the registration reference image, register the remaining images to the imaging space of the selected main image, so that the azimuth and range offsets of the imaging pixels of the same ground object are less than 0.25 pixels.

[0019] Furthermore, after preprocessing the SAR images, use the Stacking-InSAR technology and SBAS-InSAR technology to obtain surface deformation data, including:

[0020] Use the interferometric stacking method to process the SAR image set data to obtain the average surface deformation rate of the monitoring area;

[0021] Use the persistent scatterer synthetic aperture radar interferometry method to process the SAR image set data to obtain the surface deformation rate over time;

[0022] Use the weighted average method to calculate the average surface deformation rate of the monitoring area and the surface deformation rate over time obtained by Stacking-InSAR and SBAS-InSAR, and obtain the fused deformation rate, and organize to obtain the surface deformation data.

[0023] Furthermore, during the processing of the interferometric stacking method (Stacking-InSAR), the spatial baseline threshold is set to 100 meters, and the temporal baseline threshold is set to 50 days.

[0024] Furthermore, using the weighted average method to calculate the average surface deformation rate of the monitoring area and the surface deformation rate over time obtained by Stacking-InSAR and SBAS-InSAR includes:

[0025] Calculate the variance of the average deformation rate of the average surface deformation rate of the monitoring area ;

[0026] Calculate the variance of the average deformation rate of the surface deformation rate over time ;

[0027] Calculate through the weighted average calculation formula:

[0028]

[0029] Among them and are calculated respectively from the variances of the average deformation rates obtained by Stacking-InSAR and SBAS-InSAR. The calculation formula is as follows:

[0030]

[0031]

[0032] Furthermore, use UAV photogrammetry and LiDAR data to extract the surface rupture characteristics of the landslide, construct a three-dimensional virtual reality model, and obtain the surface rupture information, including:

[0033] Adopt the digital orthophoto map generated by UAV photogrammetry and the digital elevation model (DEM) and digital surface model (DSM) obtained by airborne LiDAR to extract the detailed characteristics of the surface rupture of the landslide. Borrow data recognition technology to extract the distribution law of landslide cracks, boundary range, surface coordinates and three-dimensional space elements, and construct a three-dimensional virtual reality model of the landslide to obtain surface rupture information including terrain morphology, structural characteristics, geological structure, crack type, and expansion situation;

[0034] The detailed characteristics of the surface rupture of the landslide include ground fissures, drumlins, sagging troughs, stepped benches, and landslide scarps.

[0035] Furthermore, based on the surface deformation data and surface rupture information, integrate multi-source data to establish a time-series sample library, including:

[0036] Sort out the surface deformation data, and extract the surface average deformation rate, surface deformation rate in the time series, and surface deformation information in the geographic coordinate system obtained by processing with Stacking-InSAR and SBAS-InSAR technologies;

[0037] Sort out the surface rupture information, and extract the detailed characteristics of the surface rupture of the landslide, the distribution law of cracks, boundary range, surface coordinates and three-dimensional space element content;

[0038] Match the two types of extracted data in chronological order, and integrate the surface deformation data and surface rupture information at the same time point or in a similar time period to obtain a time-series sample library covering all surface deformation data and surface rupture information.

[0039] Furthermore, based on the time-series sample library, use the Apriori algorithm to mine the correlation relationship between deformation and rupture, and combine the hash coding strategy to refine the features to determine the identification factors in the landslide evolution process, including:

[0040] Based on the time-series sample library, the Apriori algorithm for mining spatial association rules is used to process and analyze the monitored deformation data, and to mine the correlation between the spatio-temporal evolution mechanism of landslides and influencing factors;

[0041] The influencing factors include terrain morphology, structural characteristics, geological structure, fracture type, and expansion situation;

[0042] Each landslide monitoring point or area is regarded as a transaction. Each transaction contains the deformation rate data of the monitoring point or area and the data of the corresponding influencing factors of terrain morphology, structural characteristics, geological structure, fracture type, and expansion situation;

[0043] Through the preliminary statistical analysis of the data, understand the frequency of each item set in the data set, select a value to start trying to set the minimum support threshold. In all data samples, an item set will be considered a meaningful item set only if its occurrence frequency reaches at least the minimum support threshold.

[0044] According to the subsequent calculation results and analysis requirements, gradually adjust the support threshold. According to the set minimum confidence threshold, filter out the association rules with a confidence greater than or equal to the minimum confidence.

[0045] For the ground deformation data obtained by fusing InSAR and the surface rupture information constructed based on UAV and LiDAR, feature extraction methods are used to extract features. The features extracted from the ground deformation data include the change trend of deformation rate and the spatial distribution characteristics of the deformation area; the features extracted from the surface rupture information include the geometric features of surface cracks, terrain slope, and slope aspect;

[0046] Hash encoding is performed on the extracted features to map the high-dimensional feature vectors to a low-dimensional hash space, and the hash values obtained after hash encoding of different types of data are combined to construct a combined hash encoding.

[0047] Based on the combined hash encoding, co-occurrence analysis is carried out to mine the internal connections and synergistic relationships between different data. By calculating the similarity between the combined hash encoding vectors, a similarity matrix is constructed. Based on the similarity matrix, a clustering algorithm is used to cluster the data, and the samples with similar features are grouped into one category to obtain the results of co-occurrence analysis;

[0048] Combined with the results of co-occurrence analysis, collaborative representation learning is carried out to obtain a low-dimensional representation;

[0049] From the low-dimensional representation obtained by collaborative representation learning, the features that have an important impact on landslide evolution are screened out as the identification factors in the landslide evolution process.

[0050] Furthermore, the features that have an important impact on landslide evolution are screened out, including:

[0051] Use a feature selection algorithm to screen and sort features, and select features with importance scores exceeding a preset threshold as identification factors.

[0052] Furthermore, based on the time-series sample library, use the Apriori algorithm to mine the association relationship between deformation and rupture, and combine the hash coding strategy to refine features to determine the identification factors in the landslide evolution process, which also includes:

[0053] Verify the selected identification factors, and use the cross-validation method to evaluate the identification factors.

[0054] Furthermore, the method also includes:

[0055] Use a feature clustering algorithm and a correlation analysis method to quantify the weight allocation of identification factors.

[0056] Furthermore, using a feature clustering algorithm and a correlation analysis method to quantify the weight allocation of identification factors includes:

[0057] Perform standardization processing on the data of historical landslide cases and corresponding identification factors in the study area to make the data of different indicators comparable;

[0058] Use the K-Means clustering algorithm to cluster the features of identification factors, and determine the appropriate number of clusters K according to experience or the elbow method. Then, randomly initialize K cluster centers, calculate the distance from each data point to these cluster centers, and assign the data points to the category of the cluster center with the closest distance; continuously iterate and update the cluster centers until the cluster centers no longer change significantly or reach the preset number of iterations. Cluster similar indicators into one category;

[0059] Calculate the correlation coefficients between each identification factor. For each pair of indicators, calculate their correlation coefficients to obtain the correlation results, identify the group of identification factors with correlation coefficients higher than the preset value, and perform comprehensive processing on each group of identification factors during weight allocation to avoid double calculation of weights.

[0060] Use the analytic hierarchy process for weight allocation, construct a judgment matrix, determine the relative importance of different identification factors in landslide catastrophes through data analysis to form a judgment matrix. Then, calculate the maximum eigenvalue and the corresponding eigenvector of the judgment matrix, and perform normalization processing on the eigenvector to obtain the weights of each identification factor.

[0061] During the process of determining weights, perform a consistency test to ensure the rationality of the judgment matrix.

[0062] In a second aspect, the present invention provides a landslide evolution stage early warning criterion determination system that fuses multi-source heterogeneous remote sensing data, including a processor and a storage medium;

[0063] The storage medium is used to store instructions;

[0064] The processor is used to operate according to the instructions to execute the steps of the method according to the first aspect.

[0065] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0066] (1) The present invention integrates two technologies, InSAR and UAV, gives full play to their respective advantages, can comprehensively and accurately monitor landslide deformation, and improves the accuracy and reliability of monitoring.

[0067] (2) Through the spatial association rule algorithm and data mining algorithm, the internal relationship between landslide deformation and evolution law and evolution characteristics is mined, providing a scientific basis for the early warning and prevention of landslide disasters.

[0068] (3) The present invention can predict the dynamic change law of the discrimination index during the landslide evolution process, helps to timely discover the potential risks of landslide disasters, and provides a time guarantee for taking effective prevention and control measures. Description of the Drawings

[0069] Figure 1 is the flowchart of the method of the present invention;

[0070] Figure 2 is the three-stage evolution curve of landslide deformation;

[0071] Figure 3 is the schematic diagram of the evolution stage division during the landslide disaster process. Detailed Embodiments

[0072] Embodiment 1:

[0073] This embodiment provides a method for determining the early warning criterion for the landslide evolution stage by integrating multi-source heterogeneous remote sensing data, as Figure 1 shown, which includes the following steps:

[0074] (1) Obtain multiple SAR image data of Sentinel-1A satellite within the monitoring area. When registering the images, select one image as the master image, and use the image registration algorithm based on feature matching to register the remaining images to the imaging space of the master image, ensuring that the azimuth and range offsets of the imaging pixels of the same ground object are less than 0.25 pixels. After registration, according to the boundary vector file of the monitoring area, use the image cropping tool to crop the images. Subsequently, perform orbit correction on the images through the orbit parameter file, perform radiometric correction using the radiometric calibration coefficient, and use Goldstein filtering to remove the noise in the images;

[0075] (2)Set the spatial baseline threshold to 100 m and the temporal baseline threshold to 50 days. Use InSAR data processing software to stack and interfere with SAR images that meet the baseline threshold conditions to generate a landslide differential interference phase map;

[0076] (3)Perform weighted averaging on the differential interference phase map to obtain the average surface deformation rate of the monitoring area and calculate the variance of the average deformation rate . Based on the orbital information of satellite imaging and the geometric information of the reference ellipsoid, generate a lookup table in the data processing software, and calculate and convert the coordinates of each pixel through the lookup table to obtain the surface deformation information in the geographic coordinate system;

[0077] and can be calculated by the following formula

[0078]

[0079] where is the average deformation rate; n is the number of interferograms participating in the weighted average calculation; and are the temporal baseline of the i-th interferogram and its corresponding phase value, respectively;

[0080]

[0081] where is the variance of the average deformation rate, n is the number of interferograms participating in the weighted average calculation, is the rate estimate value corresponding to each interferogram;

[0082] (4)Adopt the SBAS-InSAR technology, use the data processing software to generate a connection map, and determine the image pairs participating in the interference. Perform interference processing on the selected image pairs to generate interferograms. Perform orbit refinement and re-flattening operations to remove interference factors such as orbit errors and topographic phases. Through the inversion algorithm, obtain the surface deformation rate in the time series, draw a time series curve, and calculate the variance of the average deformation rate . Based on the method in step (3), convert it into the surface deformation information in the geographic coordinate system;

[0083] The calculation formula is:

[0084]

[0085] where is the deformation rate, is the total deformation amount, is the total time interval;

[0086] Variance of average deformation rate The calculation method refers to the calculation formula given in step (3);

[0087] (5) The deformation rates obtained by fusing Stacking-InSAR and SBAS-InSAR are averaged using a weighted method. According to the variance of the average deformation rate obtained from both, weighted calculation is performed according to the formula to obtain the fused deformation rate ;

[0088] The weighted average calculation formula is:

[0089]

[0090] where and are calculated from the variance of the deformation rates obtained by Stacking-InSAR and SBAS-InSAR respectively, and their calculation formula is:

[0091]

[0092]

[0093] (6) Use an unmanned aerial vehicle (UAV) to fly and take pictures in the landslide area. During the flight, ensure that the entire landslide area is covered, and at the same time ensure sufficient forward and lateral overlap rates. By setting an appropriate flight altitude, high-resolution image data is obtained (the specific flight altitude should be dynamically adjusted according to the size and terrain complexity of the landslide area). Use the LiDAR device carried on the aircraft to scan the landslide area to obtain high-precision three-dimensional point cloud data (during the LiDAR scanning process, parameters such as flight altitude, scanning angle, and point cloud density need to be considered to ensure that the obtained data can accurately reflect the terrain characteristics of the landslide area). Then, geometric correction is performed on the collected UAV images to eliminate image deformation caused by factors such as camera lens distortion and unstable flight attitude. Through the image matching algorithm, adjacent images are stitched together to generate a complete landslide area image. Use digital photogrammetry software to process the images to generate a digital orthophoto map (DOM). After that, the LiDAR point cloud data is filtered to remove noise points and non-ground points to obtain pure ground point cloud data. Use the filtered ground point cloud data to generate a digital elevation model (DEM), and through interpolation processing of the LiDAR point cloud data, a digital surface model (DSM) is generated;

[0094] According to the detailed characteristics of the surface rupture of the landslide, such as ground fissures, drumlins, depression troughs, stepped benches, landslide scarps, etc., manual recognition is carried out on the digital orthophoto map and digital surface model, or automatic recognition is carried out using image recognition algorithms. For ground fissures, they can be recognized by analyzing the texture features and color changes of the image. For drumlins and stepped benches, they can be recognized by analyzing the elevation changes of the terrain. For landslide scarps, they can be recognized by analyzing the edge features of the image. Once the detailed characteristics of the surface rupture of the landslide are recognized, relevant information such as location, shape, size, elevation, etc. of these features is extracted using data extraction techniques. For ground fissures, information such as the length, width, and depth of the fissures can be extracted; for drumlins, information such as the height and diameter of the drumlins can be extracted; for stepped benches, information such as the elevation difference of the stepped benches can be extracted; for landslide scarps, information such as the height and slope of the scarps can be extracted.

[0095] Analyze the extracted landslide fissure data and count parameters such as the number, length, and width of the fissures. Using the digital orthophoto map and digital surface model, combined with the detailed characteristics of the surface rupture of the landslide, determine the boundary range of the landslide (the boundary of the landslide can be determined by analyzing factors such as terrain changes and vegetation distribution). Using the digital orthophoto map and digital surface model, extract the coordinate information of the landslide surface, including longitude, latitude, elevation, etc. Integrate the data such as the detailed characteristics of the surface rupture of the landslide, the distribution law of fissures, the boundary range, the surface coordinates, and three-dimensional spatial elements. Use 3D modeling software to construct a 3D virtual reality model of the landslide based on the integrated data.

[0096] (7) Integrate the surface deformation data obtained by fusing InSAR and the surface rupture information extracted by drones, establish a time-series sample library of landslide evolution characteristics, and classify and label the data in the sample library.

[0097] The integration method specifically includes:

[0098] First, organize the surface deformation data obtained by InSAR. These data include the average surface deformation rate, the surface deformation rate in the time series, and the surface deformation information in the geographic coordinate system, etc., which are processed by Stacking-InSAR and SBAS-InSAR techniques. At the same time, sort out the surface rupture information extracted by the UAV, covering the detailed features of the landslide surface rupture (such as ground fissures, drumlins, etc.), the distribution law of cracks, the boundary range, the surface coordinates, and three-dimensional space elements, etc. Then, match the above two types of data in chronological order, and integrate the surface deformation data and the surface rupture information at the same time point or in a similar time period. For example, associate the surface deformation rate data at a certain moment with the data such as the length and width of the landslide cracks corresponding to that moment to form a record containing various characteristic information. Repeat the above process, integrate the relevant data at different times in turn, and construct a data set that can reflect the evolutionary characteristics of the landslide over time.

[0099] The time-series sample library, from the perspective of data structure, is similar to a two-dimensional table, where each row represents a record of the landslide state at a time point. Each row contains multiple columns, which store different types of characteristic data respectively, including various surface deformation data obtained by InSAR (such as numerical values of average deformation rate, deformation rate variance, etc.), and surface rupture information extracted by the UAV (such as the number of cracks, length, boundary range coordinates, etc.). From the perspective of data type, there are both numerical data (such as numerical values of various rates, lengths, heights) and text or categorical data (such as crack types, etc.).

[0100] (8) Apply the integrated InSAR technology to obtain the ground deformation information of the landslide, calculate the deformation-time monitoring data of the study area, and judge the evolutionary stage of the landslide ground deformation from the time perspective;

[0101] Combine the optical image and the three-dimensional virtual reality model in step 6 to interpret the spatial development and distribution of the landslide deformation, and identify the evolutionary characteristics of the ground deformation from the space perspective:

[0102] Through the analysis of the deformation-time monitoring data, the evolutionary stage of the landslide ground deformation can be judged. Generally speaking, the evolutionary process of the landslide can be divided into the initial deformation stage, the constant velocity deformation stage, and the accelerated deformation stage. In the initial deformation stage, the deformation of the landslide body is small and the deformation rate is relatively low. As time goes by, it enters the constant velocity deformation stage, and the deformation rate remains relatively stable. When the landslide body enters the accelerated deformation stage, the deformation rate increases rapidly, indicating that the landslide may be about to undergo instability failure, as Figure 3 shown, the three-stage evolutionary curve of the landslide. By combining the optical image and the three-dimensional virtual reality model, the spatial development and distribution of the landslide deformation can be analyzed, and the characteristics such as the sliding direction of the landslide, the position of the sliding surface, and the distribution of the accumulation can be identified.

[0103] From Figure 2 It can be seen that there are usually no absolutely fixed thresholds for the several stages of landslide evolution, but it can be judged from the displacement rate-time curve. Initial deformation stage: The displacement rate basically remains stable and at a low level, showing a relatively gentle straight line on the displacement rate-time curve, with a slope almost zero or very small; Constant velocity deformation stage: Appears as a straight line with a fixed slope on the displacement rate-time curve; Accelerated deformation stage: The displacement rate shows a rapid upward trend, and on the displacement rate-time curve, it is manifested as a rapid increase in the curve slope and an obvious upward bend of the curve.

[0104] Such as Figure 2 , the deformation values corresponding to different landslides vary greatly, there is no unified threshold, and the judgment criterion is the "cumulative displacement-time curve".

[0105] (9) Use spatial association rule mining - Apriori algorithm to process and analyze the monitored deformation data, and mine the correlation relationships between the spatio-temporal evolution process of the landslide and influencing factors such as terrain morphology, structural characteristics, geological structure, crack type, and expansion situation;

[0106] Take each landslide monitoring point or area as a transaction. Each transaction contains the deformation rate data of the monitoring point or area and the data of influencing factors such as the corresponding terrain morphology, structural characteristics, geological structure, crack type, and expansion situation;

[0107] The Apriori algorithm usually processes discrete data, while some features in ground deformation and surface rupture data may be continuous, such as deformation rate, crack width, etc.

[0108] According to the deformation rate obtained by fusing InSAR, it is divided into four deformation evolution stages according to the deformation rate. These four deformation stages are respectively marked as "initial deformation stage", "constant velocity deformation stage", "accelerated deformation stage", "sliding initiation deformation stage", such as Figure 3 shown. For continuous surface rupture data such as crack width obtained from the three-dimensional virtual reality model, an equal-frequency division method is used. First, count the frequency distribution of the crack width data, and divide the crack width into three intervals: narrow, medium, and wide, so that the number of cracks in each interval is approximately the same. For categorical variables, such as crack type, one-hot encoding is used. If there are three types of cracks: tensile crack, shear crack, and mixed crack, then the tensile crack can be encoded as [1, 0, 0], the shear crack is encoded as [0, 1, 0], and the mixed crack is encoded as [0, 0, 1]. For the discretized variables, encoding is performed. The four intervals of the above deformation rate can be respectively encoded as 1, 2, 3, 4 for convenient subsequent algorithm calculation.

[0109] First, through the preliminary statistical analysis of the data, understand the approximate frequencies of various item sets in the dataset. Then, start trying to set the minimum support threshold from a relatively moderate value, such as 0.1. This means that in all data samples, an item set (such as "the deformation rate is medium-speed positive deformation and the crack width is wide") will be considered a meaningful item set only if its frequency of occurrence reaches at least 10%.

[0110] According to the subsequent calculation results and analysis requirements, gradually adjust the support threshold. If the number of generated frequent item sets is too small, it may be that the support threshold is set too high and needs to be appropriately reduced; if the number of frequent item sets is too large and contains a lot of irrelevant information, then the support threshold may need to be increased. According to the set minimum confidence threshold, filter out the association rules whose confidence is greater than or equal to the minimum confidence. These strong association rules are the association relationships between the landslide evolution laws and evolution characteristics and ground deformation and surface rupture that we hope to find. The confidence threshold will change with the data, and specific analysis should be carried out during specific projects.

[0111] (10) Use the joint hashing coding and co-occurrence-privacy-driven strategy to refine the common characteristics and collaborative representation among data, and obtain the identification factors in the landslide evolution process. The specific process is as follows:

[0112] For the ground deformation data obtained by fusing InSAR and the 3D virtual reality model data constructed based on UAV and LiDAR, adopt corresponding feature extraction methods. For the ground deformation data, the change trend of the deformation rate, the spatial distribution characteristics of the deformation area, etc. can be extracted; for the 3D virtual reality model data, extract the geometric characteristics of the surface cracks (such as length, width, curvature, etc.), terrain slope and aspect, etc.;

[0113] Perform hashing coding (locality-sensitive hashing) on the extracted features to map the high-dimensional feature vectors to a low-dimensional hashing space. By constructing a hashing function, it is ensured that feature vectors that are close in the original high-dimensional space also have a high probability of being mapped to the same or similar hash values in the hashing space. Combine the hash values obtained after hashing coding of different types of data to construct a joint hashing coding. Using a simple concatenation method, concatenate the hash values of different data in order into a new hash vector. The joint hashing coding can integrate the feature information of multi-source data and lay a foundation for subsequent refinement of common characteristics and collaborative representation.

[0114] Based on joint hash coding, co - occurrence analysis is carried out. Co - occurrence analysis aims to explore the internal connections and collaborative relationships between different data. By calculating the similarity (such as cosine similarity, Euclidean distance, etc.) between joint hash coding vectors, a similarity matrix is constructed. Based on the similarity matrix, clustering algorithms are used to cluster the data, and samples with similar characteristics are grouped into one class. During the clustering process, the data characteristics of different classes are analyzed to discover the common characteristics and collaborative patterns among the data.

[0115] During the data processing and analysis process, the issue of data privacy protection is considered. The differential privacy - driven strategy is adopted, and data is perturbed by adding a certain amount of noise to the data, so that attackers cannot accurately infer the original sensitive information from the processed data. When applying differential privacy, it is necessary to reasonably set the privacy budget parameter according to the sensitivity of the data and privacy requirements to balance the relationship between privacy protection and data availability.

[0116] Combining the results of co - occurrence analysis and privacy protection, collaborative representation learning is carried out. By constructing a collaborative representation model, the collaborative representation method between different data is learned, enabling the model to better capture the common characteristics and collaborative relationships among the data. The collaborative representation model is constructed based on the deep learning framework, adopting model structures such as autoencoders and generative adversarial networks, and learning the low - dimensional representation of the data by training the model. In this low - dimensional representation, the collaborative relationships between different data are effectively reflected.

[0117] From the low - dimensional representation obtained by collaborative representation learning, the features that have an important impact on landslide evolution are screened out, and these features are the identification factors in the landslide evolution process. Feature selection algorithms can be used to screen and rank the features, and select the features with higher importance scores as the identification factors.

[0118] The features screened out from the low - dimensional representation obtained by collaborative representation learning that have an important impact on landslide evolution generally cover aspects such as deformation rate, cumulative displacement, acceleration, and feature identification indicators. These features play a key role in the landslide evolution process and can provide important basis for landslide monitoring, early warning, and prevention.

[0119] Through analysis, it is found that features such as the change rate of deformation rate, density and length of surface cracks have relatively high importance in the landslide evolution process and can be used as identification factors.

[0120] The selected identification factors are verified to ensure their actual correlation with the landslide evolution process. The cross - validation method is used to evaluate the identification factors. By dividing the data into a training set and a test set, the model is trained on the training set and the performance of the model is verified on the test set, and the influence of the identification factors on the accuracy and reliability of the model predicting the landslide evolution state is observed.

[0121] (11) Use the feature clustering algorithm and correlation analysis method to quantify the weight assignment and sensitivity of the catastrophe identification indicators:

[0122] Standardize the detailed data such as historical landslide cases and corresponding geological environment indicators, topographic and geomorphic indicators, lithology indicators, rock mass development indicators, rock mass weathering degree indicators, structural plane characteristics, type indicators, slope form indicators, etc. within the study area to make the data of different indicators comparable.

[0123] Use the K-Means clustering algorithm to cluster the features. First, determine the appropriate number of clusters K according to experience or the elbow method. Then, randomly initialize K cluster centers, calculate the distance from each data point to these cluster centers (such as Euclidean distance), and assign the data point to the category of the nearest cluster center. Continuously iterate and update the cluster centers until the cluster centers no longer change significantly or reach the preset number of iterations. Through cluster analysis, similar indicators are grouped into one category for subsequent analysis of the impact of different categories of indicators on catastrophe identification. Calculate the correlation coefficient (Pearson correlation coefficient) between each pair of indicators. For each pair of indicators, calculate its correlation coefficient through the formula. The closer the absolute value of this coefficient is to 1, the stronger the correlation between the two indicators. Analyze the correlation results, identify highly correlated indicator groups, and consider comprehensively processing these correlated indicators during weight assignment to avoid double counting of weights.

[0124] Use the Analytic Hierarchy Process (AHP) for weight assignment. First, construct a judgment matrix to determine the relative importance of different indicators in landslide catastrophes through data analysis, forming a judgment matrix. Then, calculate the maximum eigenvalue and the corresponding eigenvector of the judgment matrix, and normalize the eigenvector to obtain the weights of each indicator. During the process of determining the weights, a consistency test is required to ensure the rationality of the judgment matrix. For the assessment of sensitivity, it can be determined by analyzing the magnitude of the indicator weights and the degree of influence of indicator changes on the catastrophe identification results in different landslide cases. Indicators with larger weights and significant influence on the results due to changes can be regarded as indicators with higher sensitivity.

[0125] Adopt a multi-index and multi-weight multi-layer nested ISODATA clustering method to reveal the variation law of discriminant indexes in the landslide disaster process: By determining the key parameters of the ISODATA clustering algorithm (the number of initial clustering centers, the expected range of the number of clusters, the thresholds for splitting and merging). The setting of these parameters needs to be adjusted in combination with the characteristics of the research area and the data distribution. In areas with complex terrain and diverse landslide types, a larger number of initial clustering centers and a wider expected range of the number of clusters may need to be set to ensure that the variation laws of discriminant indexes of different types of landslides can be captured. Then, according to the nature and hierarchical relationship of the indexes, the data is stratified. First, the geological environment indexes are taken as one layer, the topographic and geomorphic indexes are taken as another layer, etc. Apply the ISODATA clustering algorithm to each layer for preliminary clustering to obtain the clustering results of each layer. Then, combine and nest the clustering results of different layers to form a multi-layer nested clustering structure. During the nesting process, the association and interaction between the clustering results of different layers need to be considered. By defining appropriate connection rules and fusion methods, the information of each layer is integrated to more comprehensively reveal the variation law of discriminant indexes in the landslide disaster process.

[0126] (12) Utilize the inherent relevance of multi-source data, taking the surface deformation rate, crack distribution characteristics, terrain slope change, etc. as key dimensions. By establishing a mathematical model or mapping function, map the multi-source data into a unified multi-dimensional space to form a multi-dimensional common coupling space shared at the bottom layer. During the construction process, principal component analysis (PCA) can be used for dimensionality reduction and feature extraction to reduce data redundancy and complexity while retaining the information that is of great significance for landslide disaster identification.

[0127] The core goal of this method is not simply to obtain the features related to landslide disaster identification, but to achieve the prediction of the evolution law of landslide disaster identification indexes, and then serve for early landslide identification and disaster warning and prevention. However, in the process of achieving this goal, obtaining relevant features is an important intermediate link.

[0128] Design a multi-task learning model architecture, define multiple tasks related to landslide disaster identification, such as: prediction of the possibility of landslide occurrence, prediction of the disaster time, prediction of landslide scale, etc. During the model training process, utilize the correlation and shared information between different tasks, simultaneously learn the model parameters of multiple tasks, and realize the transfer and sharing of data between different tasks and domains. Adopt a structure with a shared bottom neural network layer, set specific output layers and loss functions on the branches of different tasks, and simultaneously optimize the loss functions of multiple tasks through the backpropagation algorithm, so that the model can learn the commonalities and differences of multi-source data in different landslide disaster identification tasks, and improve the generalization ability and prediction accuracy of the model.

[0129] Design a multi-task learning model architecture and define multiple tasks related to landslide disaster identification. Ultimately, instead of obtaining multiple models, a unified model is constructed. In this model, different tasks share part of the network structure, especially the underlying neural network layers.

[0130] The purpose is to simultaneously optimize the loss functions of multiple tasks through the backpropagation algorithm during model training. This joint training method enables the model to promote each other when learning different tasks, utilize the correlation and shared information between tasks, and improve the generalization ability and prediction accuracy of the model.

[0131] For example, the information about terrain and deformation features learned when predicting the likelihood of landslide occurrence can also help the model better predict the disaster time and landslide scale.

[0132] Based on multi-task learning, a contrastive learning mechanism is introduced. By constructing positive and negative sample pairs, the model can learn the differences between different landslide states and data features. Use data of the same landslide area at different times but with similar disaster trends as positive sample pairs, and use data of different landslide areas or with significantly different disaster characteristics as negative sample pairs. During the model training process, through the optimization of the contrastive learning loss function, the model can better distinguish different landslide situations, further optimize the model parameters, and improve the data mining ability of the model for landslide disaster identification indicators and the accuracy of predicting the dynamic change law of the identification indicators during the landslide evolution process. During the model training process, hyperparameters such as the learning rate, batch size, and number of training epochs need to be reasonably set, and a validation set is used to monitor and adjust the model performance to prevent overfitting. Through continuous experiments and optimizations, separate the characteristic identification indicators that affect landslide instability and failure in the disaster identification indicators, divide the deformation evolution stages of the landslide, extract the thresholds of each evolution stage, and use them as early warning criteria for dividing the evolution stages in the landslide disaster process.

[0133] Embodiment 2:

[0134] This embodiment provides a landslide evolution stage early warning criterion determination system that integrates multi-source heterogeneous remote sensing data, including a processor and a storage medium;

[0135] The storage medium is used to store instructions;

[0136] The processor is used to operate according to the instructions to execute the steps of the method according to Embodiment 1.

[0137] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0138] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0139] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable devices provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0141] The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A method for determining early warning criteria for landslide evolution stages by integrating multi-source heterogeneous remote sensing data, characterized in that: It includes the following steps: Acquire and preprocess the SAR image to obtain a preprocessed SAR image; The pre-processed SAR images are processed by Stacking-InSAR technology and SBAS-InSAR technology to obtain surface deformation data; Use UAV photogrammetry and LiDAR data to extract the surface rupture characteristics of the landslide, build a three-dimensional virtual reality model, and obtain surface rupture information; Based on surface deformation data and surface rupture information, multi-source data are integrated to establish a time series sample library; Based on the time series sample library, the Apriori algorithm is used to mine the correlation between deformation and rupture, and the hash coding strategy is combined to refine the characteristics and determine the identification factors in the landslide evolution process. Based on the time series sample library, the Apriori algorithm is used to mine the correlation between deformation and rupture, and the hash coding strategy is combined to extract features and determine the identification factors in the landslide evolution process, including: Based on the time series sample library, the spatial association rule mining-Apriori algorithm is used to process and analyze the monitored deformation data to explore the correlation between the spatiotemporal evolution mechanism of landslides and the influencing factors; The influencing factors include topographic morphology, structural characteristics, geological structure, fracture type, and expansion; Each landslide monitoring point or area is regarded as a transaction; each transaction contains the deformation rate data of the monitoring point or area and the corresponding terrain morphology, structural characteristics, geological structure, crack type, and data on factors affecting the expansion; Through preliminary statistical analysis of the data, we can understand the frequency of each item set in the data set, and select a value to set the minimum support threshold. In all data samples, the frequency of an item set must at least reach the minimum support threshold to be considered as a meaningful item set; Gradually adjust the support threshold; according to the set minimum confidence threshold, filter out the association rules whose confidence is greater than or equal to the minimum confidence, and obtain the processed item set; For the surface deformation data and surface rupture information in the processed item set, feature extraction methods are used to extract features. The features extracted from the ground deformation data include the changing trend of the deformation rate and the spatial distribution characteristics of the deformation area; the features extracted from the surface rupture information include the geometric characteristics of the surface cracks, the terrain slope and the slope direction; Hash code the extracted features, map the high-dimensional feature vectors to the low-dimensional hash space, combine the hash values ​​obtained after hash coding of different types of data, and construct a joint hash code; On the basis of joint hash coding, symbiosis analysis is carried out to explore the intrinsic connection and synergistic relationship between different data. By calculating the similarity between joint hash coding vectors, a similarity matrix is ​​constructed. Based on the similarity matrix, a clustering algorithm is used to cluster the data, and samples with similar characteristics are clustered into one category to obtain the results of symbiosis analysis. Combined with the results of symbiosis analysis, collaborative representation learning is performed to obtain low-dimensional representation; From the low-dimensional representation obtained by collaborative representation learning, features that have an important impact on landslide evolution are screened out as discriminative factors in the landslide evolution process.

2. The method for determining landslide evolution stage early warning criteria by integrating multi-source heterogeneous remote sensing data according to claim 1 is characterized in that: The preprocessing methods of SAR images include: Image registration uses the main image as the registration reference image, and registers the remaining images to the selected main image imaging space, so that the azimuth and distance offsets of the same object imaging pixels are less than 0.25 pixels.

3. The method for determining landslide evolution stage early warning criteria by integrating multi-source heterogeneous remote sensing data according to claim 1 is characterized in that: The pre-processed SAR images are processed by Stacking-InSAR technology and SBAS-InSAR technology to obtain surface deformation data, including: The Stacking-InSAR technology is used to process the SAR image data set to obtain the average surface deformation rate of the monitoring area; The SAR image data was processed using SBAS-InSAR technology to obtain the surface deformation rate in time series. The average surface deformation rate of the monitored area and the surface deformation rate in the time series are calculated using the weighted average method to obtain the fused deformation rate and the surface deformation data are sorted out.

4. The method for determining landslide evolution stage early warning criteria by integrating multi-source heterogeneous remote sensing data according to claim 3 is characterized in that: The average surface deformation rate of the monitoring area and the surface deformation rate in the time series are calculated using the weighted average method, including: Calculate the average deformation rate variance of the average surface deformation rate in the monitoring area Calculate the average deformation rate variance of the surface deformation rate over the time series Calculated by weighted average formula: V fusion =V stacking ×σ1+V sbas ×σ2; Among them, V fusion is the deformation rate after fusion, V stacking V is the average deformation rate of the surface in the monitoring area. sbas is the surface deformation rate in the time series, σ1 and σ2 are the average deformation rate variances obtained by Stacking-InSAR and SBAS-InSAR respectively, and the calculation formula is:

5. The method for determining landslide evolution stage early warning criteria by integrating multi-source heterogeneous remote sensing data according to claim 1 is characterized in that: UAV photogrammetry and LiDAR data are used to extract the surface rupture characteristics of the landslide, build a three-dimensional virtual reality model, and obtain surface rupture information, including: The digital orthophotos generated by UAV photogrammetry and the digital ground model and digital surface model obtained by airborne LiDAR are used to extract the detailed features of the landslide surface ruptures. The distribution pattern, boundary range, surface coordinates and three-dimensional spatial elements of the landslide cracks are extracted by using data recognition technology. A three-dimensional virtual reality model of the landslide is constructed to obtain surface rupture information including terrain morphology, structural characteristics, geological structure, crack type, and extension status. The detailed features of the landslide surface rupture include ground fissures, drumlins, pull-out troughs, dislocations, and steep landslide walls.

6. The method for determining landslide evolution stage early warning criteria by integrating multi-source heterogeneous remote sensing data according to claim 1 is characterized in that: Based on surface deformation data and surface rupture information, a time series sample library is established by integrating multi-source data, including: Arrange the surface deformation data to extract the average surface deformation rate, the surface deformation rate in time series and the surface deformation information in the geographic coordinate system; Sorting out the surface rupture information, extracting the detailed features of the landslide surface rupture, crack distribution patterns, boundary ranges, surface coordinates and three-dimensional spatial elements; The two types of extracted data are matched in chronological order, and the surface deformation data and surface rupture information at the same time point or in a similar time period are integrated together to obtain a time series sample library covering all surface deformation data and surface rupture information.

7. The method for determining landslide evolution stage early warning criteria by integrating multi-source heterogeneous remote sensing data according to claim 1 is characterized in that: Screen out the features that have important influence on landslide evolution, including: The feature selection algorithm is used to screen and sort the features, and the features with importance scores exceeding the preset threshold are selected as the discriminant factors.

8. The method for determining landslide evolution stage early warning criteria by integrating multi-source heterogeneous remote sensing data according to claim 1 is characterized in that: The method further comprises: The feature clustering algorithm and correlation analysis method are used to quantify the weight distribution of the discriminant factors.

9. A landslide evolution stage early warning criterion determination system integrating multi-source heterogeneous remote sensing data, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1-8.

Images