A method and system for analyzing geological hazards of power transmission lines based on deformation detection
Through differential interference technology and artificial neural network model, the geological damage degree of the pole base was identified, and the problem of low data acquisition efficiency after earthquake was solved, and a method of rapid assessment of the degree of earthquake damage was realized, which improved the accuracy and efficiency of post-disaster rescue.
Patent Information
- Application Number
- CN202111340434.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-12
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2041-11-12
AI Technical Summary
The prior art is difficult to quickly and accurately obtain real data on the collapse of the tower after an earthquake, especially when the distribution of transmission lines in mountainous areas is scattered, diverse in types and different degrees of damage, resulting in hindering rescue and reconstruction in the disaster area.
Using a method based on deformation detection, the deformation field diagram is established through differential interference technology, combined with an artificial neural network model, the image is segmented according to the offset value and offset direction, the geological damage degree of the tower foundation is identified, and the damage situation distribution is formed, and SAR and high The change detection results of spectroscopic remote sensing images were evaluated.
It has achieved rapid identification of the disaster range as soon as possible after the earthquake, accurately assessed the degree of earthquake damage, and improved the efficiency and accuracy of post-disaster rescue and reconstruction.
Smart Images

Figure CN114239379B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of remote sensing application, and in particular relates to a method and system for analyzing geological hazards of power transmission lines based on deformation detection. Background Art
[0002] The earthquake caused a large number of towers to collapse, resulting in huge losses of life and property. The number and extent of tower base damage after the earthquake is necessary information before the start of rescue and reconstruction in the disaster area.
[0003] At present, the number of collapsed towers is usually determined by manually reported data and field survey data. However, this method that relies solely on manual surveys is inefficient and has the problem of not being able to obtain real data in the first place.
[0004] High spatial resolution remote sensing technology is real, objective, and less affected by ground communication conditions. It can obtain a large amount of effective data in the first place, providing effective support for timely determination of the number of tower collapses. It is difficult to determine the collapse of towers based on image recognition, especially in mountainous areas, where transmission line towers are scattered, of various types, and with varying degrees of damage. It is impossible to accurately identify the basic geology of the towers and to accurately assess the degree of damage. Therefore, how to overcome the shortcomings of existing technologies is an urgent problem to be solved in the field of remote sensing application technology. Summary of the invention
[0005] The purpose of the present invention is to solve the deficiencies of the prior art and to provide a disaster analysis method and system based on deformation detection, to establish the relationship between the deformation field and the degree of geological damage to the tower foundation, to obtain a deformation field map using differential interference, and then to obtain the degree of geological damage to the tower foundation, so as to achieve rapid delineation of the disaster-stricken area or accurate assessment of the degree of earthquake damage.
[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:
[0007] A method for analyzing geological hazards of power transmission lines based on deformation detection, comprising:
[0008] The distribution of geological damage degree of tower foundation is obtained by comparing several groups of remote sensing images before and after the earthquake. The deformation field map is established by differential interferometry and the distribution of geological damage degree of tower foundation is marked. The sample library is formed by segmentation according to the offset value range and offset direction.
[0009] Several artificial neural network models are constructed, and samples with different offset value ranges and offset directions are used for training and packaging respectively;
[0010] Remote sensing images before and after the earthquake to be analyzed are obtained, and the deformation field is established by differential interference. Segmentation is performed based on the offset value and offset direction. The artificial neural network model corresponding to the offset value range and offset direction is selected to identify the degree of geological damage to the tower foundation, and form a distribution of geological damage to the tower foundation.
[0011] Further, preferably, the method further includes evaluating the earthquake intensity based on the obtained damage situation and outputting the earthquake intensity distribution.
[0012] Further, preferably, the distribution of the degree of geological damage to the tower foundation includes the proportion of each type of damage, which is determined by the change detection results of the SAR remote sensing images and the corresponding high-resolution optical remote sensing images before and after the earthquake, and is qualitatively marked and confirmed through on-site visual investigation;
[0013] The damage types include destroyed and collapsed, destroyed but not collapsed, severely damaged, slightly damaged, and basically intact, corresponding to the regional change rate of the before and after images ≥80%, 60%≤regional change rate<80%, 40%≤regional change rate<60%, 20%≤regional change rate<40%, and regional change rate<20% respectively; the regional change rate is the root mean square error of the tilt displacement of the serial images of two phases in the same area.
[0014] Further, preferably, establishing the deformation field by differential interference includes: obtaining a deformation point map covering the earthquake-affected area for quantitative annotation, and calculating the tilt displacement information of the tower area, thereby quantitatively reflecting the impact of the geological disaster.
[0015] Furthermore, preferably, the artificial neural network model includes a convolution layer, a pooling layer, an activation layer and a fully connected layer; the convolution layer performs feature extraction including extracting image features based on the interference pattern, extracting deformation value distribution based on deformation contour lines, and extracting whether it is located in the fault zone based on the cross-fault profile line; the pooling layer performs dimensionality reduction on the extracted features; the activation layer adopts the Relu function; and the fully connected layer outputs the distribution results of geological damage to the tower foundation.
[0016] Further, preferably, segmenting based on the offset value and the offset direction includes: segmenting the region into an offset value and an offset direction having a certain range.
[0017] The present invention also provides a transmission line geological disaster analysis system based on deformation detection, including an input module, a segmentation module, a selection module, several damage degree identification modules and an output module;
[0018] The input module is used to obtain the deformation field;
[0019] The segmentation module is used to segment the deformation field based on the offset value and the offset direction;
[0020] The selection module is used to select a corresponding damage degree identification module based on the offset value and the offset direction to identify the geological damage degree of the tower foundation;
[0021] The output module is used to quantitatively output the degree of geological damage to the tower foundation, and to form a distribution result of geological damage to the tower foundation in combination with corresponding geographic information.
[0022] Furthermore, preferably, the damage degree identification module is constructed and trained by an artificial neural network model; based on the comparison of several groups of remote sensing images before and after the earthquake, the distribution of geological damage degree of the tower foundation is obtained, and differential interference is used to establish a deformation field map and mark the distribution of geological damage degree of the tower foundation, and a sample library is formed by segmentation according to the offset value range and offset direction. The sample library selects training samples to train the artificial neural network model to meet the error level and encapsulate it.
[0023] Further, preferably, the artificial neural network model includes a convolution layer, a pooling layer, an activation layer and a fully connected layer; the convolution layer performs feature extraction including extracting image features based on the interference pattern, extracting deformation value distribution based on deformation contour lines, and extracting whether it is located in the fault zone based on the cross-fault profile line; the pooling layer performs dimensionality reduction on the extracted features; the activation layer uses the Relu function for nonlinear calculation; the fully connected layer outputs the distribution results of geological damage to the tower foundation;
[0024] The distribution of geological damage degree of tower foundation includes the proportion of each damage type, including destroyed and collapsed, destroyed but not collapsed, severely damaged, slightly damaged and basically intact;
[0025] The deformation field includes: a deformation point map covering the earthquake-affected area, with quantitative annotations based on their offsets.
[0026] Preferably, the activation layer of the present invention adopts a Relu function, and utilizes the Relu function to perform nonlinear calculations on the characteristic vectors of geometry and texture provided by the high-resolution optical remote sensing image, and polarization and tilt displacement provided by the SAR remote sensing image.
[0027] The present invention does not impose any restrictions on the specific division of the offset value and the offset direction, and can be divided accordingly according to the actual amount of computation that can be run. For example, if the amount of computation that can be carried is large, a finer division can be performed. For another example, the offset direction is divided into 8 ranges according to 360°, representing 8 directions; the offset is divided into 8 ranges according to no offset to the maximum offset.
[0028] Compared with the prior art, the present invention has the following beneficial effects:
[0029] (1) The present invention utilizes the characteristics of remote sensing images, such as wide coverage, no influence from ground conditions, and the ability to be obtained immediately, to achieve the goal of defining the affected area and assessing the extent of earthquake damage immediately after an earthquake occurs.
[0030] (2) As a new type of geodetic technology, InSAR technology has the advantage of covering a large area and can image the disaster area and obtain its deformation information. The present invention establishes the relationship between the deformation field and the degree of geological damage to the tower foundation. It uses the deformation field, based on the different fault strikes, deformation directions on both sides of the fault, and deformation values, combined with the corresponding high-resolution optical remote sensing images, to comprehensively form a standard unified regional change rate, which is used to evaluate the corresponding degree of geological damage to the tower foundation, making the evaluation more accurate.
[0031] (3) The present invention establishes multiple groups of artificial neural network models for different deformation directions and deformation values, and selects the corresponding artificial neural network model according to the deformation direction and deformation value, thereby ensuring the accuracy of recognition. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0033] Figure 1 It is a flow chart of the geological disaster analysis method of transmission lines based on deformation detection;
[0034] Figure 2 It is a schematic diagram of the artificial neural network model;
[0035] Figure 3 It is a schematic diagram of the geological disaster analysis system for power transmission lines based on deformation detection;
[0036] Figure 4 Surface deformation map of the study area
[0037] Figure 5 This is a schematic diagram of image segmentation of the surface deformation field map according to different offset values; different thicknesses of external boundaries correspond to different offset values;
[0038] Figure 6 Schematic diagram of inputting SAR remote sensing images and high-resolution optical remote sensing images of the tower area for training in order to label samples based on on-site verification;
[0039] Figure 7 In order to carry out tower change monitoring and identification, a result map of the geological damage degree of the tower foundation was obtained. DETAILED DESCRIPTION
[0040] The present invention is further described in detail below in conjunction with embodiments.
[0041] Those skilled in the art will appreciate that the following examples are only used to illustrate the present invention and should not be considered as limiting the scope of the present invention. In the examples, if no specific techniques or conditions are indicated, the techniques or conditions described in the literature in this area or the product specifications are carried out. If the materials or equipment used do not indicate the manufacturer, they are all conventional products that can be purchased. In addition, in the following description, the description of known structures and techniques is omitted to avoid unnecessary confusion of the concept of the present invention.
[0042] like Figure 1 As shown, a method for analyzing geological hazards of power transmission lines based on deformation detection includes:
[0043] The distribution of geological damage degree of tower foundation is obtained by comparing several groups of remote sensing images before and after the earthquake. The deformation field map is established by differential interferometry and the distribution of geological damage degree of tower foundation is marked. The sample library is formed by segmentation according to the offset value range and offset direction.
[0044] Several artificial neural network models are constructed, and samples with different offset value ranges and offset directions are used for training and packaging respectively;
[0045] Remote sensing images before and after the earthquake to be analyzed are obtained, and the deformation field is established by differential interference. Segmentation is performed based on the offset value and offset direction. The artificial neural network model corresponding to the offset value range and offset direction is selected to identify the degree of geological damage to the tower foundation, and form a distribution of geological damage to the tower foundation.
[0046] Preferably, the method further includes evaluating the earthquake intensity based on the obtained damage conditions and outputting the earthquake intensity distribution.
[0047] Preferably, the distribution of the degree of geological damage to the tower foundation includes the proportion of each type of damage, which is determined by the change detection results of the SAR remote sensing images and the corresponding high-resolution optical remote sensing images before and after the earthquake, and is qualitatively marked and confirmed through on-site visual investigation;
[0048] The damage types include destroyed and collapsed, destroyed but not collapsed, severely damaged, slightly damaged, and basically intact, corresponding to the regional change rate of the before and after images ≥80%, 60%≤regional change rate<80%, 40%≤regional change rate<60%, 20%≤regional change rate<40%, and regional change rate<20% respectively; the regional change rate is the root mean square error of the tilt displacement of the serial images of two phases in the same area.
[0049] Preferably, establishing the deformation field by using differential interference includes: obtaining a deformation point map covering the earthquake-affected area for quantitative annotation, and calculating the tilt displacement information of the tower area.
[0050] Preferably, the artificial neural network model includes a convolution layer, a pooling layer, an activation layer and a fully connected layer; the convolution layer performs feature extraction including extracting image features based on the interference pattern, extracting deformation value distribution based on deformation contour lines, and extracting whether it is located in the fault zone based on the cross-fault profile line; the pooling layer performs dimensionality reduction on the extracted features; the activation layer uses the Relu function; and the fully connected layer outputs the distribution results of geological damage to the tower foundation.
[0051] Preferably, segmenting based on the offset value and the offset direction includes: segmenting the region into an offset value and an offset direction having a certain range.
[0052] like Figure 3 As shown, a transmission line geological disaster analysis system based on deformation detection includes an input module, a segmentation module, a selection module, several damage degree identification modules and an output module;
[0053] The input module is used to obtain the deformation field;
[0054] The segmentation module is used to segment the deformation field based on the offset value and the offset direction;
[0055] The selection module is used to select a corresponding damage degree identification module based on the offset value and the offset direction to identify the geological damage degree of the tower foundation;
[0056] The output module is used to quantitatively output the degree of geological damage to the tower foundation, and to form a distribution result of geological damage to the tower foundation in combination with corresponding geographic information.
[0057] Preferably, the damage degree identification module is constructed and trained by an artificial neural network model; based on the comparison of several groups of remote sensing images before and after the earthquake, the distribution of geological damage degree of the tower foundation is obtained, and differential interference is used to establish a deformation field map and mark the distribution of geological damage degree of the tower foundation. A sample library is formed by segmentation according to the offset value range and offset direction, and training samples are selected from the sample library to train the artificial neural network model to meet the error level and encapsulate it.
[0058] Preferably, the artificial neural network model includes a convolution layer, a pooling layer, an activation layer and a fully connected layer; the convolution layer performs feature extraction including extracting image features based on interference fringe patterns, extracting deformation value distribution based on deformation contour lines, and extracting whether it is located in the fault zone based on cross-fault profile lines; the pooling layer performs dimensionality reduction on the extracted features; the activation layer uses Relu function for nonlinear calculation; the fully connected layer outputs the distribution results of geological damage to the tower foundation;
[0059] The distribution of geological damage degree of tower foundation includes the proportion of each damage type, including destroyed and collapsed, destroyed but not collapsed, severely damaged, slightly damaged and basically intact;
[0060] The deformation field includes: a deformation point map covering the earthquake-affected area, with quantitative annotations based on their offsets.
[0061] Specifically, the present invention provides a method for analyzing geological hazards of power transmission lines based on deformation detection, Figure 1 , including the following steps:
[0062] (1) The distribution of geological damage degree of tower foundation is obtained by comparing several groups of remote sensing images before and after the earthquake. The terrain phase difference of the area is obtained by differential interferometry, and a deformation field map is established. The map is segmented according to the offset value range and offset direction. The corresponding high-resolution optical remote sensing images are combined to form a sample database, and the degree of geological damage of the tower foundation is marked.
[0063] SAR interferometry technology is used to monitor surface deformation and evaluate earthquake damage index. The deformation field is obtained through image registration, noise filtering and phase unwrapping in interference processing. The spatial dynamic change characteristics of the deformation field are extracted through quantitative analysis of the offset information of each deformation point.
[0064] Differential interferometry uses two SAR complex images observed in the same area for interference processing, and obtains surface elevation information and deformation information through phase information. According to the imaging time, InSAR can be divided into two modes: single-pass and repeat-pass. Single-pass interference refers to the installation of two antennas on the same airborne or satellite platform, one of which transmits a signal, and both antennas receive ground echo signals, and the obtained data is used for interference processing. Repeated-pass interference refers to the same sensor or similar sensor imaging the ground twice according to parallel orbits, and using the obtained data for interference processing. The spatial distance between the SAR systems during the two imaging is called the spatial baseline distance, and the time interval is called the time baseline. Ignoring noise, assuming that the atmospheric conditions are basically the same during the two imagings, the deformation information of the ground target point can be obtained by removing the flat ground phase and the terrain phase. The present invention uses the time-series InSAR images before and after the earthquake, and the deformation monitoring points obtained by processing have three-dimensional position coordinates, deformation direction and rate information. Deformation refers to the deformation in the direction of the radar line of sight (LOS). Positive values represent movement toward the radar, and negative values represent movement away from the radar. For ease of expression, they are expressed as uplift and downlift. Green represents stability, the left side of the scale represents downlift, and the right side represents uplift. The darker the color, the greater the deformation. Figure 4 shown.
[0065] The quantitative evaluation of the distribution of geological damage to the tower foundation includes the proportion of each type of damage. The change detection results of the remote sensing images before and after the geology are used as the indicator, that is, the root mean square error of the tilt displacement of the two images in the same area. The tilt direction is the direction away from the satellite radar or the direction close to the satellite radar. The root mean square error of the tilt displacement of each tower point in the area in two directions (unit: cm, ), divided by the threshold value of the area (such as 10cm), to obtain the ratio of the regional change rate. (X, Y) is the coordinate after displacement, (X', Y') is the coordinate before displacement, and the damage types include destroyed and collapsed, destroyed but not collapsed, severely damaged, slightly damaged, and basically intact, corresponding to the regional change rate of the before and after images ≥ 80%, 60% ≤ regional change rate < 80%, 40% ≤ regional change rate < 60%, 20% ≤ regional change rate < 40%, and regional change rate < 20%.
[0066] Artificial neural networks require a large number of training samples. The present invention divides the geological deformation field map according to the offset value range and offset direction (the offset direction is away from the satellite radar direction or close to the satellite radar direction, and the offset value is usually 1mm level), and combines the corresponding high-resolution optical remote sensing image to form a sample library. The image corresponding to the geological deformation field map is manually visually interpreted and combined with the published information to determine the degree of damage to the tower unit, and is corrected in combination with the published information, such as Figure 5 and Figure 6 .
[0067] Furthermore, in order to ensure the richness of the samples and the comprehensiveness of the coverage, deformation field maps of multiple geological regions can be established and annotated to form a sample library.
[0068] (2) Construct several artificial neural network models, and train and encapsulate them using samples with different offset value ranges and offset directions.
[0069] In order to obtain more accurate evaluation results, the offset value is divided into multiple ranges, and different sample training is used for different offset value ranges and offset directions, so as to identify different offset value ranges and offset directions during use.
[0070] Artificial neural network model, combined with Figure 2 , including convolutional layer, pooling layer, activation layer and fully connected layer; the input of the convolutional layer is the geological deformation field map segmented into a single offset and direction. Feature extraction includes the geometry and texture of high-resolution optical remote sensing images, as well as the polarization and tilt displacement features of SAR remote sensing images, the deformation value distribution is extracted based on the deformation contour line, and whether it is located in the fault zone is extracted based on the cross-fault profile line. After extraction, features 1 to M are obtained; the pooling layer reduces the dimension of the extracted features, such as Figure 2As shown, feature 1 to feature M are reduced to nodes a21 to a2n by weight W1n, nodes a21 to a2n are reduced to nodes a31 to a3n by weight W2n, and reduced to output nodes by weight W3n; the activation layer uses the Relu function to perform nonlinear calculations on the required feature vectors; the fully connected layer outputs the distribution results of geological damage to the tower foundation. During the training process, W1n, W2n and W3n are continuously updated to complete the classification.
[0071] Each artificial neural network model is obtained through training, which includes:
[0072] Select training samples and validation samples from the corresponding sub-sample library; use the training samples to perform a set number of training rounds; use the validation samples for validation, and determine whether the error level is acceptable. If acceptable, complete the training; otherwise, perform another set number of training rounds. The set number of rounds is, for example, 50.
[0073] The trained artificial neural network is encapsulated for use in field recognition.
[0074] (3) Remote sensing images before and after the earthquake are obtained, and the deformation field is established by differential interferometry. Segmentation is performed based on the offset value and offset direction. Combined with high-resolution optical remote sensing images, the corresponding artificial neural network model is selected to obtain the degree of geological damage to the broken tower foundation and form a distribution of geological damage to the tower foundation.
[0075] Segmentation based on offset value and offset direction includes segmenting the area into a single offset value and offset direction. If the size exceeds the size recognized by the artificial neural network model, it is further segmented into a recognizable size. If the image is too small, it is enlarged to meet the recognition size. The corresponding artificial neural network model is input with a single offset value and offset direction for recognition, and the degree of damage is marked to ensure the accuracy of the marking. The marking of each part forms the distribution of geological damage to the tower foundation.
[0076] Using multiple artificial neural networks for recognition can improve the accuracy on the one hand, and can also achieve higher efficiency through parallel recognition on the other hand.
[0077] Another aspect of the present invention provides a transmission line geological disaster analysis system based on deformation detection, combined with Figure 3 , including an input module, a segmentation module, a selection module, several damage degree identification modules and an output module;
[0078] The input module acquires the geological deformation field;
[0079] The segmentation module segments the geological deformation field based on the offset value and the offset direction;
[0080] The selection module selects a corresponding damage degree identification module based on the offset value and the offset direction to identify the geological damage degree of the tower foundation;
[0081] The damage degree identification module is constructed and trained by an artificial neural network model; based on the comparison of several groups of remote sensing images before and after the earthquake, the distribution of the geological damage degree of the tower foundation is obtained, and the deformation field map is established by differential interference and the distribution of the geological damage degree of the tower foundation is marked. The sample library is segmented according to the offset value range and the offset direction. The training samples are selected from the sample library to train the artificial neural network model to meet the error level and encapsulate it. The present invention segments the image based on the two indicators of offset value and offset direction, and inputs the SAR remote sensing images and high-resolution optical remote sensing images with different damage degree annotations into the neural network as training samples. The network outputs the distribution map of the geological damage of the tower foundation, such as Figure 7 .
[0082] The output module matches the degree of geological damage to the tower foundation with the geographical location to form a distribution of geological damage to the tower foundation.
[0083] 1) The geological deformation field map is the distribution map of the surface deformation field obtained by D-InSAR technology. It is obtained by pre-processing the two complex SAR images before and after the geology through orbital parameter correction and pre-filtering, and then through image precision registration, interference map and coherence map generation, post-filtering, phase unwrapping and other generation processes. Finally, the surface deformation information can be obtained through the calculation of interference phase to geographic information, mainly including phase to elevation conversion, geographic coding, and DEM construction.
[0084] 2) The distribution map of geological damage degree of tower foundation is obtained by analyzing the relevant geological data in the geological area collected in the early stage, or it can be obtained by drawing the surface deformation map obtained manually in the early stage according to the geographical coordinates. The geological deformation field map and the distribution map of geological damage degree of tower foundation are input into the network according to the geographical location relationship for feature extraction.
[0085] 3) The surface position and strike change data of the fault can be obtained from the local geological department through relevant vector or raster data, or it can be vectorized from the local geological atlas according to actual needs. The geographic coordinate system used when vectorizing the layer should be the same as the previous geological deformation field distribution map to ensure that overlay analysis can be performed.
[0086] 4) The offset value needs to be calculated based on the surface deformation map obtained in the early stage. The surface deformation value directly obtained by D-InSAR is the deformation observed in the distance direction. According to the relationship with the satellite incident angle, the vertical deformation and the deformation in a certain direction of the plane, namely the offset value, can be solved. At this time, the offset direction obtained is perpendicular to the satellite flight direction; different orbit data can also be used to perform three-dimensional deformation calculation using two observation angles to obtain the offset in the north-south and east-west directions. Since the monitoring accuracy can reach the millimeter level, the offset value range can be divided according to the offset caused by the magnitude, and the offset value range and magnitude are associated to quantitatively evaluate the degree of tower damage.
[0087] 5) The model structure includes convolutional layer, pooling layer, activation layer, and fully connected layer, which are described in detail in the manual. Different initial weight values can be assigned to the offset, offset direction, degree of geological damage to the tower foundation, and fault direction at the beginning of training, which can be set to 0.4, 0.1, 0.4, and 0.1. The specific value setting depends on the geological conditions of the earthquake range.
[0088] In summary, the present invention provides a method and system for analyzing geological disasters of power transmission lines based on deformation detection, which obtains the distribution of geological damage degree of tower foundation based on the comparison of several groups of remote sensing images before and after earthquakes, adopts differential interference to obtain and establish deformation field map and mark the distribution of geological damage degree of tower foundation, and forms a sample library by segmentation according to the offset value range and offset direction, and trains several artificial neural network models; remote sensing images before and after earthquakes are obtained, and differential interference is used to obtain and establish deformation field, which is segmented based on offset value and offset direction, and the corresponding artificial neural network model is selected to obtain the geological damage degree of broken tower foundation, so as to form the distribution of geological damage of tower foundation. The present invention adopts differential interference to obtain and establish deformation field map to obtain the geological damage degree of tower foundation, so as to realize the rapid delineation of disaster-affected area or accurate assessment of earthquake damage degree.
[0089] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.
Claims
1. A method for analyzing geological hazards of power transmission lines based on deformation detection, characterized in that: include: The distribution of geological damage degree of tower foundation is obtained by comparing several groups of remote sensing images before and after the earthquake. The deformation field map is established by differential interferometry and the distribution of geological damage degree of tower foundation is marked. The sample library is formed by segmentation according to the offset value range and offset direction. Several artificial neural network models are constructed, and samples with different offset value ranges and offset directions are used for training and packaging respectively; Acquire remote sensing images before and after the earthquake to be analyzed, use differential interferometry to establish the deformation field, segment based on the offset value and offset direction, select the artificial neural network model corresponding to the offset value range and offset direction to identify the degree of geological damage to the tower foundation, and form the distribution of geological damage to the tower foundation; The differential interferometry deformation field establishment includes: obtaining the deformation point map covering the earthquake-affected area for quantitative annotation, and calculating the tilt displacement information of the tower area; The artificial neural network model includes a convolution layer, a pooling layer, an activation layer and a fully connected layer; the convolution layer performs feature extraction, including extracting image features based on interference fringe patterns, extracting deformation value distribution based on deformation contour lines, and extracting whether it is located in the fault zone based on cross-fault profile lines; the pooling layer reduces the dimension of the extracted features; the activation layer uses the Relu function; and the fully connected layer outputs the distribution results of geological damage to the tower foundation.
2. The method for analyzing geological hazards of power transmission lines based on deformation detection according to claim 1, characterized in that: It also includes evaluating the earthquake intensity based on the obtained damage situation and outputting the earthquake intensity distribution.
3. The method for analyzing geological hazards of power transmission lines based on deformation detection according to claim 1 or 2, characterized in that: The distribution of the degree of geological damage to the tower foundation includes the proportion of each type of damage, which is determined by the change detection results of the SAR remote sensing images and the corresponding high-resolution optical remote sensing images before and after the earthquake, and is qualitatively marked and confirmed through on-site visual investigation; The damage types include destroyed and collapsed, destroyed but not collapsed, severely damaged, slightly damaged, and basically intact, corresponding to the regional change rate of the before and after images ≥80%, 60%≤regional change rate<80%, 40%≤regional change rate<60%, 20%≤regional change rate<40%, and regional change rate<20% respectively; the regional change rate is the root mean square error of the tilt displacement of the serial images of two phases in the same area.
4. The method for analyzing geological hazards of power transmission lines based on deformation detection according to claim 1, characterized in that: The segmentation based on the offset value and the offset direction includes: segmenting the region into a region having a single offset value and an offset direction.
5. A transmission line geological disaster analysis system based on deformation detection, using the transmission line geological disaster analysis method based on deformation detection according to any one of claims 1 to 4, characterized in that: It includes an input module, a segmentation module, a selection module, several damage degree identification modules and an output module; The input module is used to obtain the deformation field; The segmentation module is used to segment the deformation field based on the offset value and the offset direction; The selection module is used to select a corresponding damage degree identification module based on the offset value and the offset direction to identify the geological damage degree of the tower foundation; The output module is used to quantitatively output the degree of geological damage to the tower foundation, and to form a distribution result of geological damage to the tower foundation in combination with corresponding geographic information.
6. The transmission line geological disaster analysis system based on deformation detection according to claim 5 is characterized in that: The damage degree identification module is constructed and trained by an artificial neural network model; based on the comparison of several groups of remote sensing images before and after the earthquake, the distribution of geological damage degree of the tower foundation is obtained, and the deformation field map is established and the distribution of geological damage degree of the tower foundation is marked by differential interference. According to the offset value range and offset direction, the sample library is segmented to form a sample library, and the training samples are selected from the sample library to train the artificial neural network model to meet the error level and encapsulate it.
7. The transmission line geological disaster analysis system based on deformation detection according to claim 6 is characterized in that: The artificial neural network model includes a convolution layer, a pooling layer, an activation layer and a fully connected layer; the convolution layer performs feature extraction, including extracting image features based on interference fringe patterns, extracting deformation value distribution based on deformation contour lines, and extracting whether it is located in the fault zone based on cross-fault profile lines; the pooling layer performs dimensionality reduction on the extracted features; The activation layer uses the Relu function to perform nonlinear calculations; the fully connected layer outputs the distribution results of geological damage to the tower foundation; The distribution of geological damage degree of tower foundation includes the proportion of each damage type, including destroyed and collapsed, destroyed but not collapsed, severely damaged, slightly damaged and basically intact; The deformation field includes: a deformation point map covering the earthquake-affected area, with quantitative annotations based on their offsets.
Citation Information
Patent Citations
Method for monitoring tilt of transmission tower based on InSAR technology
CN110456346A
Geological disaster hidden danger judging, analyzing, evaluating and governing method
CN111090954A
Ground disaster monitoring processing method and device of power transmission line, early warning system and equipment
CN111880210A
Landslide deformation accumulation area prediction model generation method and landslide deformation accumulation area prediction method
CN111929683A
