[go: up one dir, main page]

CN113688903A - Power transmission line micro-terrain classification method easy to cover ice - Google Patents

Power transmission line micro-terrain classification method easy to cover ice Download PDF

Info

Publication number
CN113688903A
CN113688903A CN202110975199.XA CN202110975199A CN113688903A CN 113688903 A CN113688903 A CN 113688903A CN 202110975199 A CN202110975199 A CN 202110975199A CN 113688903 A CN113688903 A CN 113688903A
Authority
CN
China
Prior art keywords
terrain
slope
ice
area
micro
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110975199.XA
Other languages
Chinese (zh)
Other versions
CN113688903B (en
Inventor
文屹
吴建蓉
曾华荣
何锦强
彭赤
张迅
朱登杰
黄军凯
杜昊
张伟
肖书舟
龚博
黄增浩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China South Power Grid International Co ltd
Guizhou Power Grid Co Ltd
Original Assignee
China South Power Grid International Co ltd
Guizhou Power Grid Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China South Power Grid International Co ltd, Guizhou Power Grid Co Ltd filed Critical China South Power Grid International Co ltd
Priority to CN202110975199.XA priority Critical patent/CN113688903B/en
Publication of CN113688903A publication Critical patent/CN113688903A/en
Application granted granted Critical
Publication of CN113688903B publication Critical patent/CN113688903B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Economics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a classification method for microtopography of a transmission line easy to be iced, which comprises the following steps: collecting regional DEM, power grid ice region distribution maps and regional meteorological station wind direction observation data; carrying out coarse extraction on the ridge, the valley and the puerto terrain; dividing a threshold value according to the TPI index, determining the slope type of the area, and accurately extracting the terrains of the valleys and the ridges; superposing the extracted valley, ridge, bealock and upslope landform of the windward slope with a distribution diagram of an ice area of the power grid, and extracting a landform area within a heavy ice area range of 20mm or more as a micro landform easy to cover ice; extracting terrain factors of each type of terrain areas easy to cover ice; outputting the type of the micro terrain easy to cover ice as a model, taking a terrain factor corresponding to the type of the micro terrain as an input layer, performing model training by adopting an XGboost algorithm, and establishing a classification model general of the micro terrain easy to cover ice; the power transmission line ice coating inspection and ice-preventing and ice-melting facility deployment in winter is effectively guided, and the risk of the power grid ice coating disaster is reduced.

Description

Power transmission line micro-terrain classification method easy to cover ice
Technical Field
The invention belongs to the technical field of electric power engineering hydrographic meteorology, and particularly relates to a method for classifying microtopography easily covered with ice on a transmission line.
Background
In recent years, due to the fact that the global warming trend is increasingly obvious, extreme climatic events occur frequently, the icing probability of a part of regional power transmission lines is increased gradually, and the high-voltage power transmission lines are seriously affected in safe and stable operation. Meanwhile, the influence of the micro terrains easy to be iced on the icing of the power transmission line is very obvious, and various icing disasters such as disconnection, collapse, tower inclination, galloping, insulator flashover and the like of the ground lead are caused, so that the loss of different degrees is caused to the regional social economy.
The microtopography easy to cover ice generally refers to a terrain with comprehensive mutation of an ice covering magnitude under a local terrain, and mainly comprises relatively small-range terrains such as a windward slope, a wind gap, bealock, a canyon, a watershed, a ridge, a river and a lake in an area easy to cover ice. With the deep promotion of the national business-to-east transmission strategy, more and more power transmission lines shuttle and route in various micro-terrain areas easy to be iced, and the areas often have factors inducing severe icing, so that potential safety hazards of a power grid are caused. Therefore, the particularity of the micro-terrain needs to be considered in the processes of planning, designing, constructing, operating and maintaining the power transmission line and routing inspection, and the occurrence of major ice disasters of the power grid under severe meteorological conditions is reduced to the greatest extent. Therefore, the micro-terrain classification algorithm for the transmission line easy to cover ice tends to be constructed.
At present, the method for identifying and dividing the micro-terrain of the power transmission line mainly depends on a contour topographic map to carry out manual discrimination and vectorization and carry out on-site verification, has high identification precision but high manual cost, and is not suitable for large-scale area micro-terrain division, and along with the popularization of a geographic information technology in the application of hydrometeorology, a micro-terrain classification algorithm based on digital terrain analysis appears, such as an image processing-based method, a terrain geometric analysis-based method, a surface flow physical simulation-based method and the like. The method can identify the land feature areas such as ridges and tops of the areas to a certain degree, but has the defects of difficult threshold division, incapability of identifying ice-prone areas and the like.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the method for classifying the microtopography of the transmission line in the large-scale area, which is easy to ice, is provided to accurately identify the terrain areas such as valleys, ridges and beaks in the area easy to ice, effectively guide the arrangement of ice-coating routing inspection and anti-icing and de-icing facilities in winter of the transmission line and reduce the risk of power grid ice-coating disasters.
The technical scheme of the invention is as follows:
a classification method for power transmission line microtopography prone to icing comprises the following steps:
step 1, collecting a regional digital elevation model DEM, a power grid ice region distribution diagram and regional meteorological station wind direction observation data;
2, performing coarse extraction on the ridge, the valley and the bealock landform based on an improved surface flow physical simulation algorithm according to DEM data;
step 3, calculating a regional terrain position index TPI based on the DEM, dividing a threshold value according to the TPI index, determining the slope type of the region, and further accurately extracting the terrains of the valleys and the ridges in the step 2 by using the divided slope type;
step 4, further dividing the ascending region into a windward slope ascending terrain and a leeward slope ascending terrain;
step 5, superposing the extracted valley, ridge, bealock and upslope landform of the windward slope and the distribution map of the ice area of the power grid, and extracting a micro-landform area within the range of a heavy ice area of 20mm or more as an ice-covering-prone micro-landform;
step 6, carrying out digital terrain analysis on the DEM to extract elevation, gradient, slope direction, terrain relief degree and terrain surface roughness terrain factors of each type of easily-iced terrain areas in the step 5;
and 7, outputting the result of the micro-terrain classification easy to cover ice in the step 5 as a model, taking the terrain factor corresponding to the micro-terrain type in the step 6 as a model input layer, performing model training by adopting an XGboost algorithm, and establishing a micro-terrain classification model easy to cover ice.
And 8, classifying the micro-terrain susceptible to icing through the established micro-terrain susceptible to icing classification model.
The method for roughly extracting the ridge, the valley and the bealock landform comprises the following steps: hydrologic analysis is carried out according to the DEM, the region confluence accumulation amount is calculated, the region with the confluence accumulation amount of 0 is divided into a watershed area, namely a ridge area, the DEM is turned over, the region with the confluence accumulation amount of 0 is extracted by using the inverse terrain DEM and is used as a catchment area, a valley terrain area is obtained, finally, space superposition is carried out according to the identified ridge and valley, and the region where the ridge and the valley intersect is defined as bealock terrain.
The method for calculating the regional terrain position index TPI based on the DEM, dividing the threshold value according to the TPI index and determining the slope type of the region comprises the following steps:
step 3.1, obtaining a topographic position index TPI by calculating a difference value of the target point and an elevation mean value in a neighborhood; the slope type is set with a threshold value according to the relation between the TPI index and the slope, and the elevation of the neighborhood pixels has fluctuation, so the threshold value is set according to the standard deviation of the TPI from the elevation of the neighborhood pixels, and the division standard of the threshold value is shown in table 1:
TABLE 1 grade Classification Table
Figure BDA0003227400800000031
Figure BDA0003227400800000041
Note: SD represents the standard deviation, namely the standard deviation value of TPI and the elevation of the adjacent pixel.
Step 3, the method for accurately extracting the terrains of the valleys and the ridges comprises the following steps:
and 3.2, overlapping the ridge slope position and the valley slope position in the step 3.1 by using the coarsely extracted valley and ridge topographic areas as backgrounds, and removing pseudo-topography except the ridge and valley slope positions, namely realizing the accurate extraction of the valleys and ridges.
The method for dividing the upslope terrain of the windward slope and the upslope terrain of the leeward slope comprises the following steps: and carrying out DEM gradient analysis on the ascending region to obtain a gradient value of an ascending pixel, obtaining a weather station adjacent to the ascending pixel, calculating the average winter wind direction observed by the weather station for many years, and determining that the ascending slope of the windward slope is formed when the included angle between the wind direction and the gradient of the ascending pixel is less than 90 degrees, or else, the ascending slope of the leeward slope is formed.
Step 6, the method for extracting the topographic relief degree comprises the following steps: the topographic relief degree refers to the difference between the maximum value and the minimum value of the elevation in a certain neighborhood range, and the topographic relief degree of the central pixel is calculated by adopting a 3 multiplied by 3 neighborhood sliding window to calculate the difference between the maximum value and the minimum value of the elevation in each window.
The method for extracting the surface roughness comprises the following steps: the surface roughness value refers to the ratio of the spherical surface area to the projected area in the region, the slope value of each grid unit is firstly obtained and converted into radian, and the surface roughness of the unit is obtained by dividing the area of the grid unit by the product of the area and the radian of the slope.
The method for establishing the micro-terrain classification model easy to cover ice according to the XGboost algorithm comprises the following steps:
step 7.1, initialize a classification model containing K trees, expressed as:
Figure BDA0003227400800000042
in the formula: a isiAnd biRespectively inputting and outputting the ith sample of the classification model, and f is a classification model tree; f is a function space formed by corresponding classification trees;
step 7.2, since each iteration generates a new tree for fitting the residual of the previous tree, the predicted value at the t-th iteration is assumed to be
Figure BDA0003227400800000051
Then:
Figure BDA0003227400800000052
step 7.3, defining an optimized objective function Q(t)An objective function Q(t)By a loss function
Figure BDA0003227400800000053
Constant term c and regular term D (f)t) Three parts are formed, loss function is formed
Figure BDA0003227400800000054
Expanded by a second order Taylor function and the constant term is removed, the objective function Q(t)Becomes the following formula (3) in which the regularization term D (f)t) The calculation formula is shown in formula (4):
Figure BDA0003227400800000055
in equation (3):
Figure BDA0003227400800000056
Figure BDA0003227400800000057
in equation (4): t is the number of leaf nodes; w is atWeight of the t leaf node; gamma and lambda are regularization coefficients;
step 7.4, define
Figure BDA0003227400800000058
IjA set of samples at the jth leaf node; then the objective function Q(t)Can be further reduced to formula (5), an objective function Q(t)The smaller the tree, the optimal structure of the whole tree is shown:
Figure BDA0003227400800000059
step 7.5, the XGBoost model parameters are optimized by using a grid search method, and the Log loss f of the predicted value and the actual value are respectively comparedlossAnd a classification error rate e; the smaller the loss, the better the performance of the classification model. Log loss flossAnd the calculation formula of the classification error rate e are respectively shown in formula (6) and formula (7):
Figure BDA00032274008000000510
Figure BDA00032274008000000511
in formula (6) and formula (7), M is the total number of samples of the model input, biIs the true category of the model; p is a radical ofiClassifying the probability that the sample belongs to class 1 for the model;
Figure BDA0003227400800000061
is the classification result of the model.
The invention has the beneficial effects that:
according to the method, micro-terrain samples are extracted based on a DEM and a surface flow physical simulation algorithm, the algorithm is improved so as to accurately identify ice-prone areas such as valleys, ridges and slopes ascending on a windward slope, the ice-prone micro-terrain samples in heavy ice areas are established by combining a power grid ice area diagram, a machine learning model XGboost (extreme Gradient boosting) is adopted to train the micro-terrain samples, and an algorithm of an ice-prone micro-terrain classification model is established. The method can identify and divide the micro-terrain easy to cover ice of the power transmission line in a large-scale area, and the result can effectively guide the ice-covering inspection of the power transmission line in winter, the anti-icing and ice-melting facilities of the power grid, and has important guiding significance for scientifically evaluating the risk of the ice-covering disaster of the power transmission line.
The TPI index is introduced into the extraction of the microtopography sample, so that the extraction method of the valleys and the ridges is improved, and meanwhile, the upgoing terrain area of the windward slope is extracted by combining wind direction data. A machine learning algorithm XGboost model is adopted in the establishment of the micro-terrain classification model, a network search method is utilized to train micro-terrain samples and obtain the optimal parameters of the model, and finally the classification model of the micro-terrain which is easy to cover ice is determined; the classification model divides the micro-terrain types of the large-scale ice-prone areas through various terrain factors, and the classification efficiency and precision can effectively identify the micro-terrain types of the heavy ice areas of the large-scale power transmission lines due to the similar machine learning models.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 shows the distribution of STRMs in the experimental region at 90 meters DEM in the embodiment;
fig. 3 is a detailed embodiment of the microtopography classification results in the experimental area based on the present invention.
Detailed Description
The invention provides a XGboost model-based classification method for transmission line microtopography prone to icing. The invention adopts the following technical scheme:
step 1, collecting a regional Digital Elevation Model (DEM), a power grid ice region distribution map and regional meteorological station wind direction observation data;
step 2, according to the DEM data in the step 1, carrying out coarse extraction on the ridge, the valley and the bealock landform based on an improved surface water physical simulation algorithm;
step 3, calculating a regional Terrain Position Index (TPI) based on the DEM, dividing a threshold value according to the TPI, determining the slope type of the region, and further accurately extracting the valley and ridge terrains in the step 2 by using the divided slope type;
step 4, further dividing the upgoing area of the hillside extracted in the step 3 into a windward slope and a leeward slope, performing DEM (dynamic effect model) slope analysis on the upgoing area to obtain a slope value of an upgoing pixel, obtaining wind direction data of an upgoing pixel adjacent to a meteorological station, calculating the average winter wind direction of the meteorological station for many years, and determining that the upgoing slope of the windward slope is determined when an included angle between the wind direction and the slope of the upgoing pixel is smaller than 90 degrees, or else, determining that the upgoing slope of the leeward slope is determined;
step 5, superposing the terrains of valleys, ridges, beaks and upslopes on the windward slopes extracted in the step 2 to the step 4 with the distribution map of the ice area of the power grid, extracting the terrain area within the range of the heavy ice area of 20mm or more as the micro terrain easy to cover ice,
and 6, carrying out digital terrain analysis on the terrain DEM, and extracting the elevation, the gradient, the slope direction, the terrain relief degree and the surface roughness factor of each type of easily-iced terrain area in the step 5.
And 7, outputting the micro-terrain types which are easy to cover ice and divided in the step 5 as models, processing terrain factors under the corresponding micro-terrain types in the step 6 as input layers, performing model training by adopting an XGboost algorithm, and establishing a micro-terrain classification model easy to cover ice.
The specific steps for dividing the rough extraction microtopography type in the step 2 comprise:
and 2.1, hydrologic analysis is carried out according to the DEM, the region convergence accumulation amount is calculated, the region with the convergence accumulation amount of 0 is divided into watersheds, namely ridge regions, the DEM is turned over in the same way, the region with the convergence accumulation amount of 0 is extracted by using an inverse terrain DEM and is used as a catchment region, valley terrain regions are obtained, finally, space superposition is carried out according to the identified ridge and valley, and the region where the ridge and the valley intersect is defined as beaut terrain.
In the step 3, the concrete steps include:
step 3.1, the terrain position index TPI can calculate the difference value of the target point and the average value of the elevations in the neighborhood, the slope type can set a threshold value according to the relation between the TPI index and the slope, the elevation of the neighborhood pixels has fluctuation, the threshold value is set according to the standard deviation of the TPI from the elevation of the neighborhood pixels, and the threshold division standard is shown in table 1.
TABLE 1 grade Classification Table
Numbering Type of slope Threshold (SD)
1 Mountain ridge TPI＞1
2 Ascending slope 0.5＜TPI＜1
3 Middle slope -0.5＜TPI＜0.5
4 Downhill slope -1＜TPI≤-0.5
5 Mountain valley TPI＜-1
Note: SD represents standard deviation, i.e. standard deviation value of TPI and neighbor pixel elevation
And 3.2, with the valleys and ridge terrain areas extracted in the step 2 as backgrounds, superposing the ridge slope positions and the valley slope positions in the step 3.1, removing pseudo-terrain ridge and valley grids outside the ridge and valley slope positions, and accurately extracting the valleys and the ridges.
In step 6, the specific method comprises the following steps:
and 6.1, obtaining the surface roughness of the unit by firstly calculating the gradient value of each grid unit, converting the gradient value into radian and dividing the area of each grid unit by the product of the area and the gradient radian according to the ratio of the spherical surface area in the region indicated by the surface roughness value to the projection area of the spherical surface area.
And 6.2, calculating the terrain relief degree by adopting a 3 x 3 neighborhood sliding window to only take the difference between the maximum elevation value and the minimum elevation value in each window as a central pixel.
The step 7 of establishing the micro-terrain classification model easy to cover ice specifically comprises the following steps:
step 7.1, a classification model containing K trees is initialized, expressed as:
Figure BDA0003227400800000091
in the formula: a isiAnd biRespectively inputting and outputting the ith sample of the model, and f is a classification model tree; f is the function space composed by the corresponding classification tree.
Step 7.2, since each iteration generates a new tree for fitting the residual of the previous tree, the predicted value at the t-th iteration is assumed to be
Figure BDA0003227400800000092
Then
Figure BDA0003227400800000093
Step 7.3, define the optimized objective function Q(t)An objective function Q(t)By a loss function
Figure BDA0003227400800000094
Constant term c and regular term D (f)t) Three parts are formed, loss function is formed
Figure BDA0003227400800000095
Expanded by a second order Taylor function and the constant term is removed, the objective function Q(t)Becomes the following formula (3), the regularization term D (f)t) See formula (4):
Figure BDA0003227400800000096
in equation (3):
Figure BDA0003227400800000097
Figure BDA0003227400800000101
in equation (4): t is the number of leaf nodes; w is atWeight of the t leaf node; gamma and lambda are regularization coefficients;
step 7.4, define
Figure BDA0003227400800000102
IjSample set on jth leaf node. Then the objective function Q(t)Can be further reduced to formula (5), an objective function Q(t)The smaller the tree, the optimal structure of the whole tree is shown:
Figure BDA0003227400800000103
and 7.5, optimizing the XGBoost model parameters by utilizing a grid search method, and respectively comparing the Log loss f of the predicted value and the measured valuelossAnd a classification error rate e; the smaller the loss, the better the performance of the classification model. Log loss flossAnd the calculation formula of the classification error rate e are respectively shown in formula (6) and formula (7):
Figure BDA0003227400800000104
Figure BDA0003227400800000105
in formula (6) and formula (7), M is the total number of samples of the model input, biIs the true category of the model; p is a radical ofiClassifying the probability that the sample belongs to class 1 for the model;
Figure BDA0003227400800000106
is the classification result of the model.

Claims (8)

1. A classification method for power transmission line microtopography prone to icing comprises the following steps:
step 1, collecting a regional digital elevation model DEM, a power grid ice region distribution diagram and regional meteorological station wind direction observation data;
2, performing coarse extraction on the ridge, the valley and the bealock landform based on an improved surface flow physical simulation algorithm according to DEM data;
step 3, calculating a regional terrain position index TPI based on the DEM, dividing a threshold value according to the TPI index, determining the slope type of the region, and further accurately extracting the terrains of the valleys and the ridges in the step 2 by using the divided slope type;
step 4, further dividing the ascending region into a windward slope ascending terrain and a leeward slope ascending terrain;
step 5, superposing the extracted valley, ridge, bealock and upslope landform of the windward slope and the distribution map of the ice area of the power grid, and extracting a landform area within the range of a heavy ice area of 20mm or more as a micro landform easy to cover ice;
step 6, carrying out digital terrain analysis on the DEM to extract elevation, gradient, slope direction, terrain relief degree and terrain surface roughness terrain factors of each type of easily-iced terrain areas in the step 5;
and 7, outputting the type of the micro terrain easy to cover ice in the step 5 as a model, taking the terrain factor corresponding to the type of the micro terrain in the step 6 as an input layer, performing model training by adopting an XGboost algorithm, and establishing a classification model of the micro terrain easy to cover ice.
And 8, classifying the micro-terrain susceptible to icing through the established micro-terrain susceptible to icing classification model.
2. The method for classifying transmission line microtopography prone to icing according to claim 1, wherein the method comprises the following steps: the method for roughly extracting the ridge, the valley and the bealock landform comprises the following steps: hydrologic analysis is carried out according to the DEM, the region confluence accumulation amount is calculated, the region with the confluence accumulation amount of 0 is divided into a watershed area, namely a ridge area, the DEM is turned over, the region with the confluence accumulation amount of 0 is extracted by using the inverse terrain DEM and is used as a catchment area, a valley terrain area is obtained, finally, space superposition is carried out according to the identified ridge and valley, and the region where the ridge and the valley intersect is defined as bealock terrain.
3. The method for classifying transmission line microtopography prone to icing according to claim 1, wherein the method comprises the following steps: calculating a regional terrain position index TPI based on DEM, dividing a threshold value according to the TPI index, and determining the slope type of the region by the following steps:
step 3.1, obtaining a topographic position index TPI by calculating a difference value of the target point and an elevation mean value in a neighborhood; the slope type is set with a threshold value according to the relation between the TPI index and the slope, and the threshold value is set according to the standard deviation of the TPI from the elevations of the adjacent pixels because the elevations of the adjacent pixels have fluctuation, and the division standard of the threshold value is shown in table 1:
TABLE 1 grade Classification Table
Figure FDA0003227400790000021
Note: SD represents the standard deviation, namely the standard deviation value of TPI and the elevation of the adjacent pixel.
4. The method for classifying transmission line microtopography prone to icing according to claim 3, wherein the method comprises the following steps: step 3, the method for accurately extracting the terrains of the valleys and the ridges comprises the following steps:
and 3.2, overlapping the ridge slope position and the valley slope position in the step 3.1 by taking the extracted valley and ridge topographic areas as backgrounds, and removing pseudo-topography except the ridge and valley slope positions, namely realizing accurate extraction of the valley and the ridge.
5. The method for classifying transmission line microtopography prone to icing according to claim 1, wherein the method comprises the following steps: the method for dividing the upslope terrain of the windward slope and the upslope terrain of the leeward slope comprises the following steps: and carrying out DEM gradient analysis on the ascending region to obtain the gradient value of the ascending pixel, obtaining wind direction data of the ascending pixel adjacent to the meteorological station, calculating the average winter wind direction observed by the meteorological station for many years, and determining that the upwind slope is the windward slope if the included angle between the wind direction and the gradient of the ascending pixel is less than 90 degrees, or else, the upwind slope is the leeward slope.
6. The method for classifying transmission line microtopography prone to icing according to claim 1, wherein the method comprises the following steps: step 6, the method for extracting the topographic relief degree comprises the following steps: the topographic relief degree refers to the difference between the maximum value and the minimum value of the elevation in a certain neighborhood range, and the topographic relief degree of the central pixel is calculated by adopting a 3 multiplied by 3 neighborhood sliding window to calculate the difference between the maximum value and the minimum value of the elevation in each window.
7. The method for classifying transmission line microtopography prone to icing according to claim 5, wherein the method comprises the following steps: the method for extracting the surface roughness comprises the following steps: the surface roughness value refers to the ratio of the spherical surface area to the projected area in the region, the slope value of each grid unit is firstly obtained and converted into radian, and the surface roughness of the unit is obtained by dividing the area of the grid unit by the product of the area and the radian of the slope.
8. The method for classifying transmission line microtopography prone to icing according to claim 1, wherein the method comprises the following steps: the method for establishing the micro-terrain classification model easy to cover ice comprises the following steps:
step 7.1, initialize a classification model containing K trees, expressed as:
Figure FDA0003227400790000031
in the formula: a isiAnd biRespectively inputting and outputting the ith sample of the classification model, and f is a classification model tree; f is a function space formed by corresponding classification trees;
step 7.2, since each iteration generates a new tree for fitting the residual of the previous tree, the predicted value at the t-th iteration is assumed to be
Figure FDA0003227400790000041
Then
Figure FDA0003227400790000042
Figure FDA0003227400790000043
Step 7.3, defining an optimized objective function Q(t)An objective function Q(t)By a loss function
Figure FDA0003227400790000044
Constant term c and regular term D (f)t) Three parts are formed, loss function is formed
Figure FDA0003227400790000045
Expanded by a second order Taylor function and the constant term is removed, the objective function Q(t)Become as followsEquation (3) where the regularization term D (f)t) The calculation formula is shown in formula (4):
Figure FDA0003227400790000046
in equation (3):
Figure FDA0003227400790000047
Figure FDA0003227400790000048
in equation (4): t is the number of leaf nodes; w is atWeight of the t leaf node; gamma and lambda are regularization coefficients;
step 7.4, define
Figure FDA0003227400790000049
IjA set of samples at the jth leaf node; then the objective function Q(t)Can be further reduced to formula (5), an objective function Q(t)The smaller the tree, the optimal structure of the whole tree is shown:
Figure FDA00032274007900000410
step 7.5, the XGBoost model parameters are optimized by using a grid search method, and the Log loss f of the predicted value and the actual value are respectively comparedlossAnd a classification error rate e; the smaller the loss, the better the performance of the classification model; log loss flossAnd the calculation formula of the classification error rate e are respectively shown in formula (6) and formula (7):
Figure FDA0003227400790000051
Figure FDA0003227400790000052
in formula (6) and formula (7), M is the total number of samples of the model input, biIs the true category of the model; p is a radical ofiClassifying the probability that the sample belongs to class 1 for the model;
Figure FDA0003227400790000053
is the classification result of the model.
CN202110975199.XA 2021-08-24 2021-08-24 Method for classifying ice-covered micro-topography of power transmission line Louis Active CN113688903B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110975199.XA CN113688903B (en) 2021-08-24 2021-08-24 Method for classifying ice-covered micro-topography of power transmission line Louis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110975199.XA CN113688903B (en) 2021-08-24 2021-08-24 Method for classifying ice-covered micro-topography of power transmission line Louis

Publications (2)

Publication Number Publication Date
CN113688903A true CN113688903A (en) 2021-11-23
CN113688903B CN113688903B (en) 2024-03-22

Family

ID=78581898

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110975199.XA Active CN113688903B (en) 2021-08-24 2021-08-24 Method for classifying ice-covered micro-topography of power transmission line Louis

Country Status (1)

Country Link
CN (1) CN113688903B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114118291A (en) * 2021-12-04 2022-03-01 国网湖南省电力有限公司 Method and system for power grid micro-topography recognition based on spatial feature clustering analysis
CN119622420A (en) * 2024-12-05 2025-03-14 重庆大学 DEM window optimization selection method for power grid micro-topography identification
CN120745331A (en) * 2025-08-25 2025-10-03 四川电力设计咨询有限责任公司 Method for predicting icing thickness of power transmission line

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180058932A1 (en) * 2016-08-12 2018-03-01 China Institute Of Water Resources And Hydropower Research Method for analyzing the types of water sources based on natural geographical features
CN109800905A (en) * 2018-12-19 2019-05-24 国网重庆市电力公司检修分公司 The powerline ice-covering analysis method that mountain environment mima type microrelief microclimate influences
US20200249674A1 (en) * 2019-02-05 2020-08-06 Nvidia Corporation Combined prediction and path planning for autonomous objects using neural networks
CN111738104A (en) * 2020-06-04 2020-10-02 中国电力工程顾问集团西南电力设计院有限公司 Method for extracting terrain category based on geographic information system
CN113240172A (en) * 2021-05-11 2021-08-10 国网湖南省电力有限公司 Micro-terrain icing numerical prediction method and system
CN113269254A (en) * 2021-05-26 2021-08-17 安徽理工大学 Coal and gangue identification method for particle swarm optimization XGboost algorithm
CN113284136A (en) * 2021-06-22 2021-08-20 南京信息工程大学 Medical image classification method of residual error network and XGboost of double-loss function training

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180058932A1 (en) * 2016-08-12 2018-03-01 China Institute Of Water Resources And Hydropower Research Method for analyzing the types of water sources based on natural geographical features
CN109800905A (en) * 2018-12-19 2019-05-24 国网重庆市电力公司检修分公司 The powerline ice-covering analysis method that mountain environment mima type microrelief microclimate influences
US20200249674A1 (en) * 2019-02-05 2020-08-06 Nvidia Corporation Combined prediction and path planning for autonomous objects using neural networks
CN111738104A (en) * 2020-06-04 2020-10-02 中国电力工程顾问集团西南电力设计院有限公司 Method for extracting terrain category based on geographic information system
CN113240172A (en) * 2021-05-11 2021-08-10 国网湖南省电力有限公司 Micro-terrain icing numerical prediction method and system
CN113269254A (en) * 2021-05-26 2021-08-17 安徽理工大学 Coal and gangue identification method for particle swarm optimization XGboost algorithm
CN113284136A (en) * 2021-06-22 2021-08-20 南京信息工程大学 Medical image classification method of residual error network and XGboost of double-loss function training

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
吴琼;余文铖;洪海生;喻蕾;段炼;尚明远;刘哲;: "基于XGBoost算法的配网台区低压跳闸概率预测", 中国电力, no. 04, pages 105 - 113 *
夏智宏;周月华;刘敏;刘来林;任永建;: "湖北省电线积冰微地形因子影响识别研究", 气象, pages 103 - 108 *
李浩;朱焱;: "基于梯度分布调节策略的Xgboost算法优化", 计算机应用, no. 06, pages 1633 - 1637 *
杨华;: "输电线路冰区勘测方法", 电力建设, no. 02 *
韦金丽;王国波;凌子燕;: "基于高分辨率DEM的地形特征提取与分析", 测绘与空间地理信息, pages 33 - 36 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114118291A (en) * 2021-12-04 2022-03-01 国网湖南省电力有限公司 Method and system for power grid micro-topography recognition based on spatial feature clustering analysis
CN114118291B (en) * 2021-12-04 2025-07-15 国网湖南省电力有限公司 Power grid micro-terrain identification method and system based on spatial feature clustering analysis
CN119622420A (en) * 2024-12-05 2025-03-14 重庆大学 DEM window optimization selection method for power grid micro-topography identification
CN120745331A (en) * 2025-08-25 2025-10-03 四川电力设计咨询有限责任公司 Method for predicting icing thickness of power transmission line

Also Published As

Publication number Publication date
CN113688903B (en) 2024-03-22

Similar Documents

Publication Publication Date Title
CN111666918B (en) A Multi-factor Based Coastline Change Identification Method
CN113591572A (en) Water and soil loss quantitative monitoring method based on multi-source data and multi-temporal data
CN113688903A (en) Power transmission line micro-terrain classification method easy to cover ice
CN108257142A (en) Ramp unit extracting method in DEM
CN119130243B (en) A Method for Identifying Key Influencing Factors of Ecological Benefits of Urban Blue-Green Spaces
CN111538798B (en) A refined extraction method for urban watersheds taking into account DSM and DLG
CN118114800A (en) Transmission line icing monitoring terminal optimization layout method and system
Peng et al. Study on the contributions of 2D and 3D urban morphologies to the thermal environment under local climate zones
CN117973247A (en) Urban street tree disaster-bearing capacity simulation method based on digital twinning
CN111738104A (en) Method for extracting terrain category based on geographic information system
CN112070056A (en) Sensitive land use identification method based on object-oriented and deep learning
CN106991247A (en) The method for drafting and system of a kind of power network windburn distribution map
Jaroenchai et al. Transfer learning with convolutional neural networks for hydrological streamline delineation
CN117710141A (en) GIS-based power grid icing monitoring site optimization site selection method
CN120974910A (en) A dynamic DEM spatial interpolation method
CN112699599A (en) Flood disaster early warning method based on BP-GEO
CN119649222A (en) River section division method, device, electronic device and storage medium based on runoff flux
Liu et al. Road density analysis based on skeleton partitioning for road generalization
CN118864991A (en) Transmission line tree obstacle prediction method, device, storage medium and computer equipment
CN117274660A (en) Method and system for establishing micro-topography micro-meteorological database for preventing ice and reducing disaster of power grid
CN113963219A (en) A method for automatic recognition of micro-terrain scenes
CN116563610A (en) Plateau mountain area ice-covering micro-topography identification method based on surface running water physical simulation
Qi et al. Environmental impact assessment method of highway reconstruction project based on the integration of remote sensing image and GIS
Nouri-Sangarab et al. Investigation of subsidence potential of the Ajabshir plain using artificial intelligence models and radar interferometric technique
CN120745331A (en) Method for predicting icing thickness of power transmission line

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant