A cluster analysis method for dam surface deformation based on InSAR monitoring data

Abstract

The present invention discloses a cluster analysis method for dam surface deformation situation based on InSAR monitoring data, including S1 to build a monitoring system: the monitoring system includes a data acquisition module, a database module, a data partition module and a diagnostic analysis module; S2 the data partition module analyzes the change characteristics of the monitoring data of each measuring point according to the received monitoring data, and clusters the deformation situation of the dam surface according to the change characteristics, and divides it into regions with different deformation situations; S3 the diagnostic analysis module calculates the cloud parameter values ​​of the measuring points in each region, determines whether there is an obvious difference in the distribution density of the cloud parameters in the same region, and marks the outliers as abnormal measuring points. Aiming at the characteristics of the massive monitoring data of time-series InSAR, the present invention takes the InSAR monitoring data of dam surface deformation as an example to propose a targeted and efficient diagnostic analysis method, which can comprehensively and quickly grasp the deformation situation of the dam, and has adaptability and practical value for similar multi-point monitoring data analysis.

Description

Technical Field

[0001] The invention belongs to the technical field of water conservancy project safety monitoring, and in particular relates to a dam surface deformation situation clustering analysis method based on InSAR monitoring data. Background Art

[0002] Surface deformation is the most intuitive reflection of the structural state changes of the dam, and contains important evolution information of the dam during its service. Scientific diagnosis and analysis of dam surface deformation monitoring data plays an important supporting role in understanding the dam's operating state, feeding back engineering design parameters, and guiding the safe operation of the project.

[0003] Due to limitations in observation conditions, construction and maintenance costs, the number of traditional ground measurement instruments is limited, and they cannot fully reflect the overall deformation of the dam surface. Satellite remote sensing InSAR monitoring technology, as an advanced earth observation method, can obtain a large number of permanent scatterers or dispersed scattering targets, providing convenient conditions for full coverage deformation monitoring of the dam surface, and can better reflect the actual deformation situation. It has been increasingly used in dam deformation monitoring.

[0004] Conventional diagnostic analysis of monitoring data is a single-point analysis, which does not reflect the spatial correlation of the measured values ​​of the measuring points. Some overall analysis methods are effective for spatial relationship analysis and abnormal diagnosis of a small number of measuring points on the ground. However, on the one hand, there are differences in material properties, loads and boundary conditions in different parts of the dam body, and the regional characteristics of surface deformation caused by this have not been taken into consideration. On the other hand, traditional ground measurements have fewer measuring points, and the adaptability of the corresponding methods to the analysis of massive measuring points is difficult to guarantee.

[0005] InSAR technology has a wide monitoring coverage, and the number of scatterer targets extracted is often dozens or even hundreds of times that of ground measurement points, which is beneficial to fully reflect the deformation situation of the dam. However, such a large amount of monitoring data also brings new problems of difficulty in analysis and diagnosis. Traditional methods have problems such as low computational efficiency and large noise interference in long time series data, which cannot meet the needs of InSAR massive data analysis. Therefore, it is necessary to use a method that takes into account both efficiency and accuracy to cluster and partition the massive InSAR monitoring data, and then further identify abnormal deformation in each partition, so as to fully grasp the deformation state of the dam and provide more reliable health diagnosis results. Summary of the invention

[0006] The present invention aims to provide a dam surface deformation situation clustering analysis method based on InSAR monitoring data to solve the technical problems of low accuracy and poor reliability in the existing dam health status diagnosis methods.

[0007] In order to solve the above technical problems, the present invention adopts the following solutions:

[0008] A cluster analysis method for dam surface deformation situation based on InSAR monitoring data includes the following steps:

[0009] Step S1: Building a monitoring system: The monitoring system includes a data acquisition module, a database module, a data partition module and a diagnosis and analysis module; the data acquisition module downloads SAR satellite images, uses InSAR technology to extract deformation data information of each measuring point on the dam surface, and transmits the collected monitoring data to the database module, the data partition module and the diagnosis and analysis module; the database module is used to store the monitoring data of the data acquisition module;

[0010] Step S2: The data partitioning module analyzes the change characteristics of the monitoring data of each measuring point according to the received monitoring data, and clusters and partitions the deformation of the dam surface according to the change characteristics, divides the measuring points with similar change patterns into the same area, and divides the dam surface into multiple areas with different deformation characteristics;

[0011] Step S3: The diagnostic analysis module calculates the cloud parameter values ​​of the measuring points in each region, and then determines whether there is a significant difference in the distribution density in the cloud parameter space of the same region, and marks the outliers as abnormal measuring points.

[0012] Further optimization, in step S1, each measuring point corresponds to a unique number, and the monitoring data collected by the data acquisition module includes the deformation of each measuring point on the dam surface and the position coordinate information of each measuring point.

[0013] Further optimization, the step S2 specifically includes the following steps:

[0014] Step S21: Set the satellite remote sensing to extract the N The displacement of each measuring point, the length of the time series is T , the displacement refers to the deformation of each measuring point relative to the initial monitoring moment; then the deformation sequence of each measuring point is expressed as X i ={ x i (1), x i (2),… x i ( T )}, i =1,2,3,…, N .

[0015] Step S22: Construction N Class G i , each class contains a deformation sequence of a measurement point; calculate N The Euclidean distance between each two in the deformation sequence of the measuring points is { d ij}, forming a distance matrix D ={ d ij} N×N ; The specific calculation is shown in formula (1):

[0016] ;

[0017] in, i,j =1,2,3,…, N ,and i ≠ j .

[0018] Step S23: Merge the two classes with the smallest distance G e , G f , forming a new class G e∪f , e,f =1,2,3,…, N ,and e ≠ f .

[0019] Step S24: traverse all current classes, use formula (2) to calculate the inter-class distance between the new class and any other class, and update the distance matrix;

[0020] ;

[0021] in, G e∪f This is the new class merged in the previous step. |G e∪f | The number of deformation sequences included for the new class; G e , G f For the two classes used for merging, |G e | , |G f | They are G e , G f The number of deformation sequences included; G h Divide the current distance matrix by G e , G f Any other class except |G h | for Gh The number of deformation sequences included; d 2 ( G e , G h )for G e , G h The square of the distance between classes, d 2 ( G f , G h )for G f , G h The square of the distance between classes, d 2 ( G e , G f )for G e , G f The square of the distance between classes.

[0022] Step S25: Repeat steps S23 and S24 to merge the two classes with the smallest distance in the current distance matrix to form a new class, and calculate the inter-class distance between the new class and any other class until all classes are merged to form a cluster analysis dendrogram.

[0023] In the first round of merging, the distance between two classes is calculated according to formula (1): d ( G e , G h ), the first round of merging refers to merging the classes corresponding to the two sequences in step S22. The distance between the new class generated after merging and other classes d ( G e , G h ), calculated according to formula (2).

[0024] Step S26: Cut the cluster analysis dendrogram according to the specific conditions of the project, and project the coverage of each category onto the dam surface to obtain the cluster partitioning results of the measuring points. The following principles should be followed when cutting: ① The deformation sequences of the measuring points in the same partition are similar; ② The deformation sequences of different partitions are different; ③ The dam surface corresponding to each partition should be connected as a whole, without isolated areas, and present a roughly symmetrical distribution feature, which conforms to the deformation law of the dam surface.

[0025] Further optimization, the step S3 specifically includes the following steps:

[0026] Step S31: For any area in all the measurement point clustering partitions in step S26, assume that there is M Any measuring point a The deformation sequence can be expressed as X a ={ x a (1), x a (2),… ,x a ( T )}, a =1,2,3,…, M ; Use the cloud model to calculate the cloud parameters of all measuring points in the area. The specific calculation is shown in equations (3) to (6):

[0027] ;

[0028] in, E a For measuring point a The sample expectation of S a 2 For measuring point a The sample variance of E na For measuring point a The sample entropy of H ea For measuring point a The sample super entropy of .

[0029] Step S32: Perform Z-Score normalization on the sample expectation, sample entropy and sample super entropy of all measurement points in the area to obtain the normalized sample expectation , sample entropy and sample super entropy , the specific calculation is shown in formula (7)~(9):

[0030] ;

[0031] in, ;

[0032] ;

[0033] Step S33: Construct a three-dimensional cloud vector for each measuring point in the area , calculate the area M The Euclidean distance between two vectors of the point cloud dist ( Ya , Y q ), as shown in formula (10):

[0034] ;

[0035] in, a,q =1,2,3,…, M ,and a ≠ q .

[0036] Step S34: Any point cloud vector Y a The Euclidean distances to all other point cloud vectors in the area are arranged from small to large, where Y a No. k The Euclidean distance is k_dist ( Y a ); take all Y a The Euclidean distance does not exceed k_ dist ( Y a ) constitutes a set of corresponding point cloud vectors, as Y a of k Distance Neighborhood N k ( Y a ), as shown in formula (11):

[0037] .

[0038] Step S35: Calculation Y a The local outlier factor LOF ( Y a ), as shown in formula (12):

[0039] ;

[0040] in, Y a arrive Y q No. k Reachable distance ; Y a The local reachable density , Y q yes Y aof k Distance Neighborhood N k ( Y a ) in any point cloud vector; Y q The local reachable density , similarly Y p yes Y q of k Distance Neighborhood N k ( Y p ) in any point cloud vector; | N k ( Y a )| N k ( Y a )The number of cloud vectors in | N k ( Y q )| N k ( Y p ) is the number of cloud vectors in .

[0041] Step S36: Repeat the above steps to calculate the local outlier factor of all the point cloud vectors in the area, whose value represents the ratio of the mean of the local reachable density of the neighboring points to the point; when the local outlier factor value is greater than 1, and the larger the value, the lower the density at the point is than the density of the neighboring points; the points with larger local outlier factor values ​​in the area are the more suspected abnormal points in the area and should be further identified.

[0042] Chinese patent, title: Multi-measurement point joint diagnosis system and method for sluice deformation, application number: 2024101723800, is the applicant's prior application. This application is further developed on the basis of the scheme of the prior application, and there are significant differences between the two.

[0043] As an artificial structure under strict control during construction and operation, the deformation of a dam is completely different from that of a natural soil-rock body such as a landslide. The deformation is destined to have overall synergy, but due to the huge differences in boundary conditions and load conditions, the deformation of different parts is also different. For example, the dam crest and dam toe, the dam crest has weak constraints, simple boundary conditions, large deformation fluctuations at the measuring points, and the deformation fluctuations between measuring points are relatively synchronized; while the dam toe has strong constraints, complex boundary conditions, and suppressed deformation fluctuations at the measuring points, but there are varying degrees of delay in the response to load changes, so the deformation process line is relatively complex. Therefore, the surface deformation of the dam has both overall synergy and regional characteristics.

[0044] Based on this understanding, this application first uses a clustering algorithm to divide the dam surface into several areas. The deformation of the measuring points in each area is similar, and the deformation of the measuring points between areas is different. This is to use the regional characteristics of deformation. Within each area, there is an overall coordination of deformation between the measuring points. The local anomaly factor algorithm is introduced to calculate the degree to which each measuring point is different from the deformation law of other measuring points, identify the measuring point with the most different deformation law from most measuring points, and consider that this measuring point deviates the farthest from the overall law and has the highest probability of being an abnormal measuring point.

[0045] The starting point of the prior application is the same as that of the present application, but the final result is different for two reasons. First, the prior application did not quantify the degree to which the measuring points deviate from the overall law, but set a fixed threshold. Those exceeding the threshold are judged as abnormal. If most of the area is abnormal, the area is judged to be abnormal, otherwise it is judged to be normal. Therefore, the prior application can only be applied to the situation where abnormal deformation occurs in one area, while the technical solution of the present application is more flexible and adaptable. As long as the measuring point with the highest probability of abnormality is identified, whether it is abnormal or not is left to professionals to judge. Second, although the prior application divides several areas with similar deformation laws through clustering, the regional characteristics of deformation are not adequately applied. As mentioned above, the deformation fluctuations of various areas on the surface of the dam are different. Therefore, in the solution described in the present application, whether the measuring point is normal or not is determined by comparison with other measuring points, while in the prior application, it is determined whether the fluctuation size of the measuring point exceeds the threshold. Therefore, whether the threshold value is appropriate and accurate will affect the accuracy of the final result.

[0046] Therefore, the solution of the prior application is only applicable to projects with strong regionality and a small number of measuring points, such as sluices, while the solution of the present application is applicable to projects with complex deformation and a large number of measuring points, such as dams.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. The monitoring system described in the present invention has a high degree of intelligence, and takes into account the spatial correlation and regional characteristics of the dam body deformation. The abnormal measurement point identification effect is accurate, which solves the problem of low efficiency of traditional methods and unstable performance in different parts of the dam body, and can more comprehensively reflect the deformation state of the dam.

[0049] 2. The method of the present invention can divide the dam into regions with different characteristics according to the similarity of deformation of each measuring point for research, and fully considers the regional characteristics of dam deformation. In addition, the abnormal deformation diagnosis method based on cloud model and local abnormal factor of the present invention has high calculation efficiency and is not easily affected by noise, which meets the needs of InSAR data analysis and improves the accuracy and efficiency of abnormal deformation identification.

[0050] 3. The present invention proposes a targeted and efficient diagnostic analysis method for the massive InSAR monitoring data of dam surface deformation, taking into account both efficiency and accuracy, identifying abnormal deformation in each partition, and comprehensively grasping the deformation state of the dam. It is also adaptable and practical for the analysis of similar multi-point monitoring data. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the dam surface deformation monitoring system based on InSAR of the present invention;

[0052] Figure 2 It is the distribution diagram of the dam measuring points in Example 1;

[0053] Figure 3 This is the flow chart of cluster analysis of measurement points;

[0054] Figure 4 This is the clustering and partitioning dendrogram of all the measuring points of the dam;

[0055] Figure 5 This is the clustering zoning map of all the measuring points of the dam;

[0056] Figure 6 This is a comparison diagram of the deformation process lines of some measuring points in area IV and area V in the 5-zone area;

[0057] Figure 7 This is a comparison diagram of the deformation process lines of some measuring points in area I and area II in the 6-zone area;

[0058] Figure 8 The dendrogram cropping result is that all the measurement points are clustered into 5 areas;

[0059] Fig. 9 This is the abnormal measurement point diagnosis flow chart;

[0060] Fig.10 It is the histogram of LOF values ​​of the measuring points in area III;

[0061] Fig.11This is a schematic diagram of the location of abnormal measurement points in area III;

[0062] Fig.12 This is the deformation process line of some measuring points in area III. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] A cluster analysis method for dam surface deformation situation based on InSAR monitoring data includes the following steps:

[0065] Step S1: Building a monitoring system: Figure 1 As shown, the monitoring system includes a data acquisition module, a database module, a data partition module and a diagnostic analysis module. The data acquisition module is used to collect SAR images of the dam location and calculate the deformation data of each measuring point on the dam surface. Each measuring point corresponds to a unique number. The monitoring data collected by the data acquisition module includes the deformation of each measuring point on the dam surface and the position coordinate information of each measuring point. The database module is used to store the monitoring data of the data acquisition module. The data partition module is used to divide the measuring points with similar deformation laws into the same set, and divide the dam surface into multiple areas with different deformation characteristics. The diagnostic analysis module is used to calculate the cloud parameter values ​​of the measuring points in each area, and then determine whether there is a significant difference in the distribution density in the cloud parameter space of the same area, and mark the outliers as abnormal measuring points.

[0066] In this embodiment, a large reservoir is taken as an example. The dam is a concrete face rockfill dam with a maximum dam height of 122.5m. The data acquisition module collects displacement data of 160 measuring points on the dam top and downstream side. The positions of the measuring points are distributed as follows: Figure 2 The monitoring period is from October 9, 2017 to December 30, 2022, with a total of 151 sets of data. Each set of monitoring data includes deformation data of 160 measuring points. The coordinates of the measuring points are stored in the database module after the deformation data are collected.

[0067] Step S2: The data partitioning module analyzes the change characteristics of the monitoring data of each measuring point based on the received monitoring data, and clusters and partitions the dam surface deformation according to the change characteristics, classifies the measuring points with similar change patterns into the same area, and divides the dam surface into multiple areas with different deformation characteristics.

[0068] like Figure 3As shown, the specific steps include:

[0069] Step S21: Satellite remote sensing extracts the displacement of 160 measuring points on the dam surface. The monitoring period is from October 9, 2017 to December 30, 2022. The deformation sequence of each measuring point includes 151 data, that is, the sequence length is 151. The displacement refers to the deformation of each measuring point relative to the initial moment of monitoring; the deformation sequence of each measuring point is expressed as X i ={ x i (1), x i (2),… x i (151)}, i =1,2,3,…,160.

[0070] Step S22: Construct 160 classes G i , each class contains a deformation sequence of a measurement point; calculate the Euclidean distance between each of the 160 measurement point deformation sequences { d ij}, forming a distance matrix D ={ d ij} N×N ; The specific calculation is shown in formula (1). Due to space limitations, some data of the distance matrix of the dam measurement point deformation sequence are shown in Table 1.

[0071] ;

[0072] in, i,j =1,2,3,…, N ,and i ≠ j .

[0073] Table 1 Partial data of the distance matrix of the deformation sequence of the dam measuring points

[0074] ;

[0075] Step S23: Merge the two classes with the smallest distance into a new class. At this time, the Euclidean distance of the deformation sequence of the measuring points numbered 56 and 71 in the matrix is ​​the smallest, which is 4.09, so the measuring points 56 and 71 are merged into a new class.

[0076] Step S24: traverse all current classes, use formula (2) to calculate the inter-class distance between the new class and any other class, and update the distance matrix;

[0077]

[0078] in, G e∪f This is the new class merged in the previous step. |G e∪f | The number of deformation sequences included for the new class; G e , G f For the two classes used for merging, |G e | , |G f | They are G e , G f The number of deformation sequences included; G h Divide the current distance matrix by G e , G f Any other class except |G h | for G h The number of deformation sequences included; d 2 ( G e , G h )for G e , G h The square of the distance between classes, d 2 ( G f , G h )for G f , G h The square of the distance between classes, d 2 ( G e , G f )for G e , G f The square of the distance between classes.

[0079] Step S25: Repeat steps S23 and S24 to merge the two classes with the smallest distance in the current distance matrix to form a new class, and calculate the inter-class distance between the new class and any other class until all classes are merged to form a cluster analysis tree diagram, such as Figure 4 shown.

[0080] In the first round of merging, the distance between two classes is calculated according to formula (1): d ( G e , G h ), the first round of merging refers to merging the classes corresponding to the two sequences in step S22. The distance between the new class generated after merging and other classes d ( G e , G h ), calculated according to formula (2).

[0081] Due to space limitations, only part of the inter-class distance matrix data after the first update is given, as shown in Table 2, where the class numbered 161 corresponds to the new class merged in the previous step.

[0082] Table 2 Partial data of the first update of the distance matrix between classes of the dam measurement point deformation sequence

[0083] .

[0084] Step S26: Reasonably cut the cluster analysis tree diagram according to the specific conditions of the project, and project the coverage of each category onto the dam surface to obtain the cluster partition results of the measuring points, such as Figure 5 The following principles should be followed when cutting: ① The deformation sequences of the measuring points in the same partition are similar; ② The deformation sequences of different partitions are different; ③ The dam surface corresponding to each partition should be connected as a whole, without isolated areas, and present a roughly symmetrical distribution feature, which conforms to the deformation law of the dam surface.

[0085] Figure 5 Different colors represent different partitions, where (a) all measuring points on the dam surface are divided into 4 areas, (b) all measuring points on the dam surface are divided into 5 areas, and (c) all measuring points on the dam surface are divided into 6 areas. Finally, the corresponding measuring point clustering partition results are obtained.

[0086] contrast Figure 5 In (a) and (b), the 4-zone and 5-zone are basically the same, the difference is that the areas IV and V in the 5-zone are merged in the 4-zone. However, it can be seen from the zoning map that the area IV is located at the top of the dam, and the area V is located at the junction of the top of the dam and the left bank. The two are not strictly symmetrical.

[0087] like Figure 6As shown in the figure, it is a comparison diagram of the deformation process lines of some measuring points in areas IV and V in the 5-zone area. Figure 6 (a) is the deformation process line of some measuring points in area IV; (b) is the deformation process line of some measuring points in area V. It can be seen that the deformation trends of areas IV and V are basically the same, and they show a certain periodicity, but the measured values ​​of the measuring points in area IV are closer, the synchronization of fluctuations is higher, but the fluctuation degree of the measured values ​​is much higher than that of area V. Combined with the partition map, it can be judged that the reason for this phenomenon is that areas IV and V are located in different positions, and the boundary conditions and loads are also different. Area IV is not connected to the right bank, so the boundary conditions are simpler, and the deformation synchronization degree of each measuring point in the area is higher; area V is adjacent to the left bank, so it is more constrained and the deformation fluctuation is smaller. In summary, the deformation trends of areas IV and V are different, and the fluctuation degree of area IV tends to intensify, so they should not be merged into one partition for unified diagnosis.

[0088] contrast Figure 5 In (b) and (c), the 5-partition and 6-partition diagrams are basically the same. The difference is that the regions I and II in the 6-partition diagram are merged into region I in the 5-partition diagram. From the 6-partition diagram, we can see that region I is located in the inner circle and region II is located in the outer circle.

[0089] like Figure 7 As shown in the figure, it is the deformation process line of the measuring points in areas I and II in the 6-zone area. Figure 7 (a) is the deformation process line of the measuring point in area I; (b) is the deformation process line of the measuring point in area II. It can be seen that the deformation trends of areas I and II are almost the same. The deformation amount increases steadily over time as a whole, and also shows a certain periodicity. The only difference is that the deformation fluctuation degree of area II is slightly higher, but there is no aggravation trend. In addition, from Figure 5 As can be seen from (c) in the figure, region II is divided into two disconnected regions. This partitioning result does not have good interpretability, so region I and region II are suitable to be merged into one partition.

[0090] In summary, the dam measuring points should be divided into five areas, and the tree diagram cutting method is as follows: Figure 8 The dam measuring point zoning diagram is shown in Figure 5 As shown in (b), regions I~V have 43, 35, 32, 22 and 28 measuring points respectively.

[0091] Step S3: The diagnostic analysis module calculates the cloud parameter values ​​of the measuring points in each region, and then determines whether there is a significant difference in the distribution density in the cloud parameter space of the same region, and marks the outliers as abnormal measuring points.

[0092] In this embodiment, taking 32 measuring points in area III as an example, Fig. 9 As shown, the specific diagnostic analysis method includes the following steps:

[0093] Step S31: For the 32 measuring points in region III, the deformation sequence of any measuring point can be expressed as X a ={ x a (1), x a (2),… ,x a ( T )}, a =1,2,3,…,32; the cloud model is used to calculate the cloud parameters of all measuring points in the area. The specific calculation is shown in equations (3) to (6):

[0094] ;

[0095] in, E a For measuring point a The sample expectation of x a ( b )express b Time measurement point a Deformation value ，S a 2 For measuring point a The sample variance of E na For measuring point a The sample entropy of H ea For measuring point a The calculation results of cloud parameters for all measuring points in region III are shown in Table 3.

[0096] Table 3 Calculation results of cloud parameters for all measuring points in region III

[0097] .

[0098] Step S32: Perform Z-Score normalization on the sample expectation, sample entropy and sample super entropy of all measurement points in the area to obtain the normalized sample expectation , sample entropy and sample super entropy , the specific calculation is shown in formula (7)~(9):

[0099] ;

[0100] in, ;

[0101] ;

[0102] Step S33: Construct a three-dimensional cloud vector for each measuring point in the area , calculate the area 32 The Euclidean distance between two vectors of the point cloud dist ( Y a , Y q ), as shown in formula (10):

[0103] ;

[0104] in, a,q =1,2,3,…,32, and a ≠ q .

[0105] The calculation results of cloud vector distances of some measuring points in area III are shown in Table 4.

[0106] Table 4. Distance matrix data of the measured point cloud vectors in area III

[0107]

[0108] Step S34: Any point cloud vector Y a The Euclidean distances to all other point cloud vectors in the area are arranged from small to large, where Y a No. k The Euclidean distance is k_dist ( Y a ); take all Y a The Euclidean distance does not exceed k_ dist ( Y a ) constitutes a set of corresponding point cloud vectors, as Y a of k Distance Neighborhood N k ( Y a ), set here k The value is 4, as shown in formula (11):

[0109] .

[0110] Step S35: Calculation Y a The local outlier factor LOF ( Y a ), as shown in formula (12):

[0111] ;

[0112] in, Y a arrive Y q No. k Reachable distance ; Y a The local reachable density , Y q yes Y a of k Distance Neighborhood N k ( Y a ) in any point cloud vector; Y q The local reachable density , similarly Y p yes Y q of k Distance Neighborhood N k ( Y p ) in any point cloud vector; | N k ( Y a )| N k ( Y a )The number of cloud vectors in | N k ( Y q )| N k ( Y p ) is the number of cloud vectors in .

[0113] The calculation results of the LOF values ​​of all the point cloud vectors in area III are shown in Table 5.

[0114] Table 5 LOF values ​​of the cloud vectors of the dam points in region III

[0115]

[0116] Step S36: Calculate the local outlier factor of all point cloud vectors in the area LOF ( Y a), measuring point factor LOF Value histogram see Fig.10 , LOF The 42# measuring point with the largest value is determined to be a suspected abnormal measuring point. Its location is as follows Fig.11 The measured value process lines of some measuring points in area III are as follows: Fig.12 As shown, the deformation of measuring point 42# has been significantly higher than that of other measuring points since September 2019, and shows a trend of rapid increase, which is significantly different from other measuring points, indicating that the judgment is accurate and can be further analyzed in combination with the project operation status.

[0117] Similarly, follow the above steps to diagnose whether there are frequently measured points in other areas, and no further details will be given.

[0118] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of this invention patent through the above description. The technical scope of this invention patent is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A cluster analysis method for dam surface deformation situation based on InSAR monitoring data, characterized in that: The steps include: Step S1: Building a monitoring system: The monitoring system includes a data acquisition module, a database module, a data partition module and a diagnosis and analysis module; the data acquisition module downloads SAR satellite images, uses InSAR technology to extract deformation data information of each measuring point on the dam surface, and transmits the collected monitoring data to the database module, the data partition module and the diagnosis and analysis module; the database module is used to store the monitoring data of the data acquisition module; Step S2: The data partitioning module analyzes the change characteristics of the monitoring data of each measuring point according to the received monitoring data, and clusters and partitions the deformation of the dam surface according to the change characteristics, divides the measuring points with similar change patterns into the same area, and divides the dam surface into multiple areas with different deformation characteristics; Step S3: The diagnostic analysis module calculates the cloud parameter values ​​of the measuring points in each region, and then determines whether there is a significant difference in the distribution density in the cloud parameter space of the same region, and marks the outliers as abnormal measuring points; specifically, the following steps are included: Step S31: For any area in all the measurement point cluster partitions, assume that there is M Any measuring point a The deformation sequence can be expressed as X a ={ x a (1), x a (2),… ,x a ( T )}, a =1,2,3,…, M ; Use the cloud model to calculate the cloud parameters of all measuring points in the area. The specific calculation is shown in equations (3) to (6): ； in, E a For measuring point a The sample expectation of S a 2 For measuring point a The sample variance of E na For measuring point a The sample entropy of H ea For measuring point a The sample super entropy of Step S32: Perform Z-Score normalization on the sample expectation, sample entropy and sample super entropy of all measurement points in the area to obtain the normalized sample expectation , sample entropy and sample super entropy , the specific calculation is shown in formula (7)~(9): ； in, ; ； Step S33: Construct a three-dimensional cloud vector for each measuring point in the area , calculate the area M The Euclidean distance between two vectors of the point cloud dist ( Y a , Y q ), as shown in formula (10): ； in, a,q =1,2,3,…, M ,and a ≠ q ; Step S34: Any point cloud vector Y a The Euclidean distances to all other point cloud vectors in the area are arranged from small to large, where Y a No. k The Euclidean distance is k_dist ( Y a ); take all Y a The Euclidean distance does not exceed k_dist ( Y a ) constitutes a set of corresponding point cloud vectors, as Y a of k Distance Neighborhood N k ( Y a ), as shown in formula (11): ； Step S35: Calculation Y a The local outlier factor LOF ( Y a ), as shown in formula (12): ； in, Y a arrive Y q No. k Reachable distance ; Y a The local reachable density , Y q yes Y a of k Distance Neighborhood N k ( Y a ) in any point cloud vector; Y q The local reachable density , similarly Y p yes Y q of k Distance Neighborhood N k ( Y p ) in any point cloud vector; | N k ( Y a )| N k ( Y a )The number of cloud vectors in | N k ( Y q )| N k ( Y p )The number of cloud vectors; Step S36: Repeat the above steps to calculate the local outlier factor of all the point cloud vectors in the area, whose value represents the ratio of the mean of the local reachable density of the neighboring points to the point; when the local outlier factor value is greater than 1, and the larger the value, the lower the density at the point is than the density of the neighboring points; the point with the largest local outlier factor value in the area is the point that is more suspected to be abnormal in the area.

2. The dam surface deformation situation clustering analysis method based on InSAR monitoring data according to claim 1 is characterized in that: In the step S1, each measuring point corresponds to a unique number, and the monitoring data collected by the data acquisition module includes the deformation of each measuring point on the dam surface and the position coordinate information of each measuring point.

3. The dam surface deformation situation clustering analysis method based on InSAR monitoring data according to claim 2 is characterized in that: The step S2 specifically includes the following steps: Step S21: Set the satellite remote sensing to extract the N The displacement data of the measuring points, the time series length is T , the displacement refers to the deformation of each measuring point relative to the initial monitoring moment; then the deformation sequence of each measuring point is expressed as X i ={ x i (1), x i (2),… x i ( T )}, i =1,2,3,…, N ; Step S22: Construction N Class G i , each class contains a deformation sequence of a measurement point; calculate N The Euclidean distance between each two in the deformation sequence of the measuring points is { d ij }, forming a distance matrix D ={ d ij } N×N ; The specific calculation is shown in formula (1): ； in, i,j =1,2,3,…, N ,and i ≠ j ; Step S23: Merge the two classes with the smallest distance G e , G f , forming a new class G e∪f , e,f =1,2,3,…, N ,and e ≠ f ; Step S24: traverse all current classes, use formula (2) to calculate the inter-class distance between the new class and any other class, and update the distance matrix; ； in, G e∪f This is the new class merged in the previous step. |G e∪f | The number of deformation sequences included for the new class; G e , G f For the two classes used for merging, |G e |、 |G f | They are G e , G f The number of deformation sequences included; G h Divide the current distance matrix by G e , G f Any other class except |G h | for G h The number of deformation sequences included; d 2 ( G e , G h )for G e , G h The square of the distance between classes, d 2 ( G f , G h )for G f , G h The square of the distance between classes, d 2 ( G e , G f )for G e , G f The square of the distance between classes; Step S25: repeat steps S23 and S24, merge the two classes with the smallest distance in the current distance matrix to form a new class, and calculate the inter-class distance between the new class and any other class, until all classes are merged to form a cluster analysis dendrogram; Step S26: cutting the cluster analysis dendrogram according to the specific conditions of the project, and projecting the coverage of each category onto the dam surface to obtain the clustering and partitioning results of the measuring points.

Images