Transmission channel tree obstacle prediction and early warning method and system based on satellite remote sensing
Through the tree barrier prediction and early warning method based on satellite remote sensing, multi-source data and random forest models are used to identify and predict tree species and grow speed, and the three-dimensional model is dynamically updated, which solves the problems of high cost and low frequency of tree barrier hidden danger management in the existing technology. Efficient control and early warning of tree obstacles has been achieved.
Patent Information
- Application Number
- CN202410900668.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-05
- Publication Date
- 2024-11-12
AI Technical Summary
The existing technology cannot meet the management needs of tree barrier hidden dangers for fast-growing tree species, and the cost of multiple monitoring is high, and the manpower and material resources are invested, making it difficult to effectively control large-scale tree barrier hidden dangers.
The tree barrier prediction and early warning method of transmission channel based on satellite remote sensing is adopted, and a three-dimensional initial model is generated by collecting multi-source data, and a random forest model is used to identify tree species and predict tree growth velocity, and the three-dimensional model is dynamically updated to achieve tree barrier height prediction and early warning.
Effective control and timely handling of hidden dangers for tree barriers in transmission channels, reduce monitoring costs, and improve the timeliness and coverage of hidden dangers in tree barriers.
Smart Images

Figure CN118941976A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power system early warning, and in particular to a method and system for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing. Background Art
[0002] Tree barriers: As trees grow, the distance between them and the conductors decreases, causing the electric field on the conductor surface to distort. The electric field strength between the conductors and the trees increases, eventually causing the line to flash over and cause a power outage. Monitoring and early warning are extremely important.
[0003] The patent of the prior art publication number CN116592851B, entitled "Three-dimensional satellite monitoring method and system for tree barriers in power transmission channels integrated with line body", discloses that firstly, the power transmission tower is refined in three-dimensional modeling based on the UAV laser point cloud; then, the three-dimensional model of the power transmission line body, channel tree barriers and topography is obtained based on the satellite remote sensing stereo image; the satellite three-dimensional model is corrected based on the UAV modeling results and the power grid ledger data; finally, the power transmission channel tree barriers are monitored based on the conductor height, tree height and the preset safety distance threshold. The application of this invention takes into account the constant characteristics of the three-dimensional coordinates, call height, tower height and other attribute information of the power grid tower, and uses the existing tower information as the ground control point to correct the tree barrier monitoring model, which can avoid the interference of human errors and further improve the accuracy of tree barrier monitoring. This method uses satellite remote sensing to monitor the tree barriers in the power transmission channel on a large scale. However, affected by the satellite revisit cycle, this method can only meet the large-scale monitoring of tree barriers twice a year at most. For bamboo, fast-growing eucalyptus and other tree species that grow too fast in a large area, the time frequency of this method cannot meet the business needs.
[0004] In order to deal with tree barrier hazards, tree barrier monitoring is currently the main method used to manage tree barrier hazards. The existing tree barrier monitoring based on drone laser point cloud has high accuracy and good effect, but it is affected by factors such as high equipment cost, large restrictions on environmental conditions (meteorology, topography, traffic conditions, communication signals), and limited monitoring range. In complex environments such as remote mountainous areas and uninhabited areas, the frequency of tree barrier monitoring of power transmission lines is low, the safety risks are large, and the investment cost is high. Tree barrier monitoring based on satellite remote sensing can be free from flight approval and terrain conditions, and can achieve large-scale monitoring of tree barrier hazards. However, the frequency of satellite remote sensing tree barrier monitoring is low, and only 1-2 large-scale monitoring can be achieved each year. For fast-growing tree species such as bamboo and fast-growing eucalyptus, it cannot meet the needs of tree barrier hazard management.
[0005] The existing tree barrier hazards in power transmission channels are mainly monitored by monitoring methods, including manual inspections, drone laser point cloud scanning, and stereo satellite measurement. Manual inspections are inefficient, have a high risk factor for personal safety, and have a small monitoring range; drones are restricted by flight approval, terrain, weather, and other conditions, and are expensive, making them unsuitable for large-scale tree barrier monitoring; although stereo satellite measurement can overcome the shortcomings of the above methods, it is limited by satellite transit and revisit cycles, and large-scale tree barrier monitoring can only be carried out twice a year.
[0006] Existing technology directly predicts tree heights and then determines tree obstacle hazards. However, the power sector has to cut down trees every year, and this action cannot be predicted, that is, the predicted tree height is not consistent with the actual tree height. Therefore, directly predicting tree height is not scientific, and the resulting tree obstacle hazards are meaningless.
[0007] In order to dynamically grasp the growth of trees in power transmission corridors, simply relying on regular monitoring methods can no longer meet the management requirements of fast-growing tree species. In addition, multiple monitoring is costly and requires a lot of manpower and material resources, which is not suitable for large-scale tree barrier hidden danger management in power transmission corridors. Therefore, it is urgent to explore a tree barrier hidden danger prediction and early warning method to fundamentally improve the real-time control capabilities of tree barrier hidden dangers and further significantly reduce the cost of tree barrier hidden danger investigation. Summary of the invention
[0008] In order to solve the problem that the existing technology can no longer meet the management requirements of fast-growing tree species simply relying on regular monitoring means, and that multiple monitoring is costly, requires large amounts of manpower and material resources, and is not suitable for large-scale tree barrier hidden danger management in power transmission channels, the present invention proposes a power transmission channel tree barrier prediction and early warning method based on satellite remote sensing, comprising:
[0009] Collect multi-source data along the transmission channel, and generate a three-dimensional initial model including the transmission line body and channel tree barriers based on the multi-source data;
[0010] Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers;
[0011] Based on a tree growth rate prediction model corresponding to the tree species, predict the growth rate of the tree species within a prediction time period, and determine the tree height within the prediction time period based on the growth rate;
[0012] The three-dimensional initial model is updated based on the tree heights within the prediction time period to obtain a dynamic three-dimensional model, and an early warning is issued based on the dynamic three-dimensional model.
[0013] Optionally, the multi-source data include: multispectral satellite remote sensing data, global ecosystem dynamics survey data and ground field survey data.
[0014] Optionally, generating a three-dimensional initial model including a transmission line body and channel tree barriers based on the multi-source data includes:
[0015] Input stereo satellite remote sensing images into 3D modeling software and output digital surface models within the image coverage area;
[0016] Based on the power grid ledger data, using stereo satellite remote sensing images and 3DsMax software, we built three-dimensional models of trees, transmission towers and power lines.
[0017] The digital surface model and the monomeric three-dimensional model are unified into the same coordinate system, and the two are merged in space to generate a three-dimensional initial model of the transmission line body and the channel tree barrier.
[0018] Optionally, the method of identifying tree species of tree barriers based on multi-source data using a pre-trained random forest model includes:
[0019] Extract spectral features and texture features using multispectral satellite remote sensing data;
[0020] Extracting three-dimensional structural features using the Global Ecosystem Dynamics Survey;
[0021] Inputting the spectral feature, the texture feature and the three-dimensional structural feature into a pre-trained random forest model for identification to obtain a tree species category;
[0022] Among them, the random forest model is obtained by training the machine learning algorithm based on the remote sensing image features of tree species and the features related to tree species extracted from ground field survey data.
[0023] Optionally, the training of the random forest model includes:
[0024] Extract tree species-related features from ground sample survey data;
[0025] Constructing a sample set based on remote sensing image features of the tree species and features related to actual ground surveys of the tree species;
[0026] Taking the features related to tree species in the sample set as input and the tree species category as output, training is performed using a machine learning algorithm to obtain a random forest model;
[0027] The tree species-related features include: spectral features, texture features, time-varying features, and observed height, tree age, and canopy structure information.
[0028] Optionally, the construction of the tree growth rate prediction model includes:
[0029] Build a sample library of tree ages and growth rates of the same tree species;
[0030] Dividing the sample library into modeling data and test data according to a set ratio;
[0031] Taking tree age as an independent variable and growth rate as a dependent variable, a linear regression model is used to fit the modeling data to obtain an initial tree growth rate prediction model;
[0032] A tree growth rate prediction model is determined based on the test data and the initial tree growth rate prediction model using a root mean square error method.
[0033] Optionally, the determining the tree growth rate prediction model based on the test data and the initial tree growth rate prediction model using a root mean square error method includes:
[0034] Substituting the tree age in the test data into the initial tree growth rate prediction model to obtain a tree growth rate prediction value;
[0035] The root mean square error method is used to calculate the root mean square error between the tree growth rate prediction value and the tree growth rate in the test data, and an initial tree growth rate prediction model whose root mean square error is less than a set threshold is selected as the tree growth rate prediction model.
[0036] Optionally, the providing of early warning based on the dynamic three-dimensional model includes:
[0037] The dynamic 3D model outputs the 3D clearance distance between the wire and the trees;
[0038] Whether to issue an early warning message is determined based on the three-dimensional clearance distance between the wire and the tree and the overhead power transmission line operation regulations.
[0039] Optionally, the determining whether to issue warning information based on the three-dimensional clearance distance between the wire and the tree and the overhead power line operation regulations includes:
[0040] Determining the minimum vertical distance between the conductor and the tree from the overhead power line operation regulations based on the voltage level of the power transmission channel;
[0041] Determining whether the three-dimensional clearance distance between the wire and the tree is less than the minimum vertical distance between the wire and the tree;
[0042] If it is less than, a warning message is issued, otherwise no warning message is issued.
[0043] Optionally, also include:
[0044] Regularly collect multi-source data along the transmission corridor;
[0045] Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers;
[0046] Predicting the tree height of the tree species based on a tree growth rate prediction model corresponding to the tree species;
[0047] The dynamic three-dimensional model is updated based on the tree heights.
[0048] On the other hand, the present invention also provides a tree obstacle prediction and early warning system for power transmission channels based on satellite remote sensing, comprising:
[0049] An initial model building module, used to collect multi-source data along the transmission channel, and generate a three-dimensional initial model including the transmission line body and channel tree barriers based on the multi-source data;
[0050] A tree species identification module, used to identify tree species of tree barriers using a pre-trained random forest model based on the multi-source data;
[0051] A tree height prediction module, used to predict the growth rate of the tree species within a prediction period based on a tree growth rate prediction model corresponding to the tree species, and determine the tree height within the prediction period based on the growth rate;
[0052] The early warning module is used to update the three-dimensional initial model based on the tree heights within the predicted time period to obtain a dynamic three-dimensional model, and issue an early warning based on the dynamic three-dimensional model.
[0053] Optionally, the multi-source data include: multispectral satellite remote sensing data, global ecosystem dynamics survey data and ground field survey data.
[0054] Optional, initial model building module, specifically for:
[0055] Input stereo satellite remote sensing images into 3D modeling software and output digital surface models within the image coverage area;
[0056] Based on the power grid ledger data, using stereo satellite remote sensing images and 3DsMax software, we built three-dimensional models of trees, transmission towers and power lines.
[0057] The digital surface model and the monomeric three-dimensional model are unified into the same coordinate system, and the two are merged in space to generate a three-dimensional initial model of the transmission line body and the channel tree barrier.
[0058] Optionally, the tree species identification module is specifically used to:
[0059] Extract spectral features and texture features using multispectral satellite remote sensing data;
[0060] Extracting three-dimensional structural features using the Global Ecosystem Dynamics Survey;
[0061] Inputting the spectral feature, the texture feature and the three-dimensional structural feature into a pre-trained random forest model for identification to obtain a tree species category;
[0062] Among them, the random forest model is obtained by training the machine learning algorithm based on the remote sensing image features of tree species and the features related to tree species extracted from ground field survey data.
[0063] Optionally, a training module is also included to:
[0064] Extract tree species-related features from ground sample survey data;
[0065] Constructing a sample set based on tree species and the features related to the tree species;
[0066] Taking the features related to tree species in the sample set as input and the tree species category as output, training is performed using a machine learning algorithm to obtain a random forest model;
[0067] The tree species-related features include: spectral features, texture features, time-varying features, and observed height, tree age, and canopy structure information.
[0068] Optionally, a growth model building module is also included to:
[0069] Build a sample library of tree ages and growth rates of the same tree species;
[0070] Dividing the sample library into modeling data and test data according to a set ratio;
[0071] Taking tree age as an independent variable and growth rate as a dependent variable, a linear regression model is used to fit the modeling data to obtain an initial tree growth rate prediction model;
[0072] A tree growth rate prediction model is determined based on the test data and the initial tree growth rate prediction model using a root mean square error method.
[0073] Optionally, the specific implementation steps of determining the tree growth rate prediction model by using the root mean square error method based on the test data and the initial tree growth rate prediction model in the growth model construction module include:
[0074] Substituting the tree age in the test data into the initial tree growth rate prediction model to obtain a tree growth rate prediction value;
[0075] The root mean square error method is used to calculate the root mean square error between the tree growth rate prediction value and the tree growth rate in the test data, and an initial tree growth rate prediction model whose root mean square error is less than a set threshold is selected as the tree growth rate prediction model.
[0076] Optional, early warning module, specifically used for:
[0077] The dynamic 3D model outputs the 3D clearance distance between the wire and the trees;
[0078] Whether to issue an early warning message is determined based on the three-dimensional clearance distance between the wire and the tree and the overhead power transmission line operation regulations.
[0079] Optionally, the steps of determining whether to issue a warning message based on the three-dimensional clearance distance between the wire and the tree and the operation procedures of the overhead transmission line in the warning module include:
[0080] Determining the minimum vertical distance between the conductor and the tree from the overhead power line operation regulations based on the voltage level of the power transmission channel;
[0081] Determining whether the three-dimensional clearance distance between the wire and the tree is less than the minimum vertical distance between the wire and the tree;
[0082] If it is less than, a warning message is issued, otherwise no warning message is issued.
[0083] Optionally, also include update modules for:
[0084] Regularly collect multi-source data along the transmission corridor;
[0085] Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers;
[0086] Predicting the tree height of the tree species based on a tree growth rate prediction model corresponding to the tree species;
[0087] The dynamic three-dimensional model is updated based on the tree heights.
[0088] In another aspect, the present application further provides a computing device, comprising: at least one processor and a memory;
[0089] The memory is used to store one or more programs;
[0090] When the one or more programs are executed by the at least one processor, the above-mentioned method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing is implemented.
[0091] On the other hand, the present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed, it implements the above-mentioned method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing.
[0092] Compared with the prior art, the present invention has the following beneficial effects:
[0093] The present invention provides a method for predicting and warning tree barriers in power transmission channels based on satellite remote sensing, including: collecting multi-source data along the power transmission channel, and generating a three-dimensional initial model including the power transmission line body and the channel tree barriers based on the multi-source data; using a pre-trained random forest model to identify the tree species of the tree barriers based on the multi-source data; predicting the growth rate of the tree species within a prediction time period based on a tree growth rate prediction model corresponding to the tree species, and determining the tree height within the prediction time period based on the growth rate; updating the three-dimensional initial model based on the tree height within the prediction time period to obtain a dynamic three-dimensional model, and issuing an early warning based on the dynamic three-dimensional model. The present invention can realize the effective control and timely disposal of tree barrier hazards in power transmission channels, and solves the problem that relying solely on regular monitoring methods can no longer meet the management of fast-growing tree species, and that multiple monitoring is costly and requires large human and material resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0094] Figure 1 This is a flow chart of a method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing of the present invention;
[0095] Figure 2 This is an overall flow chart of a method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing according to an embodiment of the present invention;
[0096] Figure 3 The figure is a schematic diagram of a computer structure of the present invention. DETAILED DESCRIPTION
[0097] The present invention proposes a method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing. By using satellite remote sensing image data, through key steps such as tree species identification, three-dimensional modeling, and tree growth height prediction, the clearance distance between power transmission lines and trees can be predicted, ultimately achieving large-scale and high-frequency prediction and warning of tree obstacle hazards.
[0098] Embodiment 1:
[0099] A method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing, such as Figure 1 As shown, including:
[0100] Step 1: Collect multi-source data along the transmission channel, and generate a three-dimensional initial model including the transmission line body and channel tree barriers based on the multi-source data;
[0101] Step 2: Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers;
[0102] Step 3: Based on the tree growth rate prediction model corresponding to the tree species, predict the growth rate of the tree species within the prediction time period, and determine the tree height within the prediction time period based on the growth rate;
[0103] Step 4: Update the three-dimensional initial model based on the tree heights within the prediction time period to obtain a dynamic three-dimensional model, and issue an early warning based on the dynamic three-dimensional model.
[0104] The present invention proposes a method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing. Satellite remote sensing images are used to perform initial monitoring of tree obstacles in power transmission channels and identify tree species. A tree species growth height prediction model is proposed to continuously predict the growth height of trees in power transmission channels. In combination with the "Overhead Power Transmission Line Operation Regulations", early warnings and disposal are carried out for trees whose growth height exceeds the warning line. The present invention can achieve effective control and timely disposal of hidden dangers of tree obstacles in power transmission channels. Figure 2 The present invention is described in detail:
[0105] Step 1: Collect multi-source data along the transmission channel, and generate a three-dimensional initial model including the transmission line body and channel tree barriers based on the multi-source data, specifically including:
[0106] 1. Data Collection
[0107] First, the data required for this method are collected, including power grid data, stereo satellite remote sensing data, multi-spectral satellite remote sensing data, global ecosystem dynamics survey data, and ground field survey data. The data are described as follows:
[0108] (1) The power grid ledger data includes attribute field information such as tower number, longitude, latitude, line name, and call height;
[0109] (2) Stereo satellite remote sensing images include forward and backward images, which are acquired every six months and used to construct the initial three-dimensional model of the transmission line body and the tree barriers in the channel;
[0110] (3) Multispectral satellite remote sensing data: used to extract spectral features and texture features, which are subsequently used for tree species identification. The multispectral satellite remote sensing data used in the present invention has a wide coverage area, which is convenient for detecting the growth of trees along the power transmission lines over a large area. The multispectral satellite remote sensing data has a large coverage area at one time, which reduces the number of calculations and improves calculation efficiency.
[0111] (4) Global ecosystem dynamics survey data: used to extract ground elevation, canopy top height and relative height indicators, and canopy structure-related indicators, which are subsequently used for tree species identification.
[0112] (5) Ground survey data, including information on tree species, tree height, tree coordinates, tree age, crown width, and breast diameter, are used to construct training samples.
[0113] (2) 3D model construction
[0114] Firstly, based on the principle of oblique photogrammetry, the stereo satellite remote sensing images are input into the existing 3D modeling software (Mirauge3D) to output the digital surface model (DSM) of the image coverage area, which can reflect the 3D terrain information of the area.
[0115] Then, using stereo satellite remote sensing images and 3DsMax software, individual three-dimensional models of trees, transmission towers and power lines were constructed.
[0116] The DSM and the monomer model are unified into the same coordinate system (CGCS2000), and the two are merged in space to generate a three-dimensional initial model of the transmission line body and the channel tree barrier.
[0117] Step 2: Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers, including:
[0118] Extract spectral features and texture features using multispectral satellite remote sensing data;
[0119] Extracting three-dimensional structural features using the Global Ecosystem Dynamics Survey;
[0120] Inputting the spectral features, the texture features and the three-dimensional structural features into a pre-trained random forest model for identification to obtain tree species;
[0121] Among them, the random forest model is obtained by training the machine learning algorithm based on tree species and extracting tree species-related features from ground field survey data.
[0122] Before introducing step 2, let’s first introduce the training of the random forest model for tree species identification.
[0123] (III) Tree species identification
[0124] Use multispectral satellite remote sensing data to extract its spectral characteristics and texture characteristics; use the global ecosystem dynamics survey to extract three-dimensional structural characteristics; collect ground survey data to make samples; and use machine learning algorithms to identify tree species. The specific steps are as follows:
[0125] (1) Spectral feature extraction: Spectral feature extraction is achieved by using multispectral image band combination. The extracted spectral features include normalized difference vegetation index (NDVI), difference vegetation index (DVI), enhanced vegetation index (EVI), ratio vegetation index (RVI), green ratio vegetation index (GRVI), greenness index (GI), adjusted soil adjusted vegetation index (MSAVI), red-shortwave infrared band difference (RedSWIR1), etc.
[0126] (2) Texture feature extraction: Different tree species show great differences in texture. For example: the texture density of fast-growing eucalyptus is usually lower, while the texture density of bamboo is usually higher; the texture direction of fast-growing eucalyptus is diverse, while the texture direction of bamboo is generally vertical; the texture color of fast-growing eucalyptus is usually gray and brown, while the texture color of bamboo tends to be yellow and green. The present invention intends to use the gray-level co-occurrence matrix to calculate the relationship between pixel values, including angular second moment, contrast, correlation, heterogeneity, homogeneity, mean, standard deviation (StdDev) and entropy.
[0127] (3) Extraction of three-dimensional structural information: Using the Global Ecosystem Dynamics Investigation (GEDI) LiDAR Level 2A and Level 2B datasets, we extracted ground elevation, canopy top height and relative height indicators, and canopy structure-related indicators, and removed low-quality data using quality marks, information degradation marks, and terrain quality marks. At the same time, considering the rapid growth of fast-growing eucalyptus, that is, the growth conditions of fast-growing eucalyptus at different ages show obvious differences, we can combine the sample site survey data in the continuous national forest resources inventory to obtain the tree age information corresponding to the tree height extracted based on the image data of different phases.
[0128] (4) Obtaining ground survey data: Obtain survey plot data near the transmission line, which mainly includes fast-growing eucalyptus, bamboo, etc. Sample plots are arranged in each plot according to accessibility and representativeness, and the corner points of the sample plots are calculated using GPS. Single trees with a breast diameter greater than or equal to 5 cm in the sample plot are measured, and information such as tree species, tree height, breast diameter, and crown width are recorded. If there is no ground survey data, the second-class forest resource survey data can be used, combined with the multi-phase remote sensing images provided by Google Earth, and sample plots of tree species such as fast-growing eucalyptus and bamboo can be selected through visual interpretation, and the interpretation generates sample data.
[0129] (5) Construction of training samples and model training: Extract spectral features, texture features, time-varying features, and observed height, tree age, and canopy structure information corresponding to the ground sample survey data to construct training samples. In order to reduce feature redundancy, it is necessary to select several features with relatively large contributions from the above features to participate in the actual classification work. The random forest algorithm is used. While performing classification, the algorithm can also calculate the relative importance of each feature, that is, the contribution of different feature variables to the final classification result. Set up a combination of different feature variables to participate in the training of random forest models for forest land extraction and tree species classification, respectively, and use the feature combination with the highest classification accuracy to participate in the actual forest land extraction and tree species classification work.
[0130] The specific implementation process of step 2 is as follows:
[0131] (6) Tree species identification and classification: The optimal feature variable combination selected in the above feature optimization work is used to extract the forest land using the random forest algorithm. Similarly, a random forest tree species identification model is constructed based on the optimal variable combination of tree species selected in step (5) to achieve the final classification of tree species. Finally, the user accuracy, producer accuracy, overall classification accuracy and Kappa coefficient provided by the confusion matrix are used to evaluate the classification results of the image.
[0132] Before introducing step 3, the tree growth rate prediction model involved in this step is first introduced.
[0133] The construction of the tree growth rate prediction model includes:
[0134] Build a sample library of tree ages and growth rates of the same tree species;
[0135] Dividing the sample library into modeling data and test data according to a set ratio;
[0136] Taking tree age as an independent variable and growth rate as a dependent variable, a linear regression model is used to fit the modeling data to obtain an initial tree growth rate prediction model;
[0137] A tree growth rate prediction model is determined based on the test data and the initial tree growth rate prediction model using a root mean square error method.
[0138] The specific construction process of the tree growth rate prediction model is as follows:
[0139] (1) Sample construction: Based on the sample plot survey data from the national forest resources continuous inventory, several sample plots in the study area were selected, the number of trees of each species in the sample plots was recorded, and the growth rate and age of the trees were counted. After establishing the corresponding sample library of tree age-growth rate for different tree species, the samples were divided into modeling data and test data in a ratio of 4:1, which were used for the establishment of the tree growth rate prediction model and the independent test of the model, respectively.
[0140] (2) Construction of tree growth rate prediction model: Using the Logistic model, the independent variables of the model are tree age, and the dependent variables are tree growth rate. The modeling data are used to fit these data, and the adjusted coefficient of determination and root mean square error are used to compare the fitting results of the model and select the optimal model suitable for different tree species to obtain the growth rate values of different tree species at different ages.
[0141] Step 3: Based on the tree growth rate prediction model corresponding to the tree species, predict the growth rate of the tree species within the prediction time period, and determine the tree height within the prediction time period based on the growth rate, including:
[0142] For the identified tree species, the corresponding optimal tree growth rate prediction models are fitted respectively to obtain the growth rates of different tree species.
[0143] Step 4: updating the three-dimensional initial model based on the tree heights within the prediction time period to obtain a dynamic three-dimensional model, and issuing an early warning based on the dynamic three-dimensional model, including:
[0144] By combining the constructed three-dimensional initial model of the transmission line body and the channel tree barrier, the tree height is dynamically updated to obtain the three-dimensional dynamic model of the transmission line body and the channel tree barrier.
[0145] (IV) Dynamic update of tree barrier 3D model
[0146] The specific steps are as follows:
[0147] Dynamic update of tree height: The initial height of trees is obtained in the constructed 3D initial model of the transmission line and channel tree barriers. The tree species and age are obtained by combining the tree species identification results and ground survey data. The growth rate of different tree species is then obtained through the tree growth rate prediction model, so that the tree height is dynamically updated. Finally, a 3D dynamic model of the transmission line and channel tree barriers is obtained.
[0148] The present application provides a method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing, which requires regular collection of multi-source data along the power transmission channel;
[0149] Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers;
[0150] Predicting the tree height of the tree species based on a tree growth rate prediction model corresponding to the tree species;
[0151] The dynamic three-dimensional model is updated based on the tree heights.
[0152] (V) Tree obstacle warning
[0153] In combination with the "Overhead Transmission Line Operation Regulations DL / T 741-2019", the classification standards for satellite remote sensing potential tree barrier risks at different voltage levels are established, as shown in Table 1. When the difference between the line height and the tree height is less than or equal to the vertical distance in the table, a tree barrier warning is issued.
[0154] Table 1 Minimum vertical distance between wires and trees based on satellite remote sensing
[0155]
[0156] (1) The present invention breaks through the technical bottleneck of insufficient timeliness brought by traditional tree barrier monitoring and early warning, and innovatively proposes a tree barrier prediction and early warning technology based on satellite remote sensing. On the basis of three-dimensional monitoring of tree barriers, the present invention realizes the dynamic update of tree barrier height by proposing tree species identification and tree growth rate prediction methods.
[0157] (2) Through the present invention, only one large-scale tree barrier monitoring needs to be completed each year to conduct real-time control and prevention of tree height, which has a good effect on the management of hidden dangers of fast-growing tree barriers such as fast-growing eucalyptus and bamboo.
[0158] Compared with traditional tree barrier monitoring methods, the present invention has lower investment costs, better tree barrier hidden danger control effects, and higher feasibility.
[0159] Embodiment 2:
[0160] The present invention based on the same inventive concept also provides a tree obstacle prediction and early warning system for power transmission channels based on satellite remote sensing, comprising:
[0161] An initial model building module, used to collect multi-source data along the transmission channel, and generate a three-dimensional initial model including the transmission line body and channel tree barriers based on the multi-source data;
[0162] A tree species identification module, used to identify tree species of tree barriers using a pre-trained random forest model based on the multi-source data;
[0163] A tree height prediction module, used to predict the growth rate of the tree species within a prediction period based on a tree growth rate prediction model corresponding to the tree species, and determine the tree height within the prediction period based on the growth rate;
[0164] The early warning module is used to update the three-dimensional initial model based on the tree heights within the predicted time period to obtain a dynamic three-dimensional model, and issue an early warning based on the dynamic three-dimensional model.
[0165] Optionally, the multi-source data include: multispectral satellite remote sensing data, global ecosystem dynamics survey data and ground field survey data.
[0166] Optional, initial model building module, specifically for:
[0167] Input stereo satellite remote sensing images into 3D modeling software and output digital surface models within the image coverage area;
[0168] Based on the power grid ledger data, using stereo satellite remote sensing images and 3DsMax software, we built three-dimensional models of trees, transmission towers and power lines.
[0169] The digital surface model and the monomeric three-dimensional model are unified into the same coordinate system, and the two are merged in space to generate a three-dimensional initial model of the transmission line body and the channel tree barrier.
[0170] Optionally, the tree species identification module is specifically used to:
[0171] Extract spectral features and texture features using multispectral satellite remote sensing data;
[0172] Extracting three-dimensional structural features using the Global Ecosystem Dynamics Survey;
[0173] Inputting the spectral feature, the texture feature and the three-dimensional structural feature into a pre-trained random forest model for identification to obtain a tree species category;
[0174] Among them, the random forest model is obtained by training the machine learning algorithm based on the remote sensing image features of tree species and the features related to tree species extracted from ground field survey data.
[0175] Optionally, a training module is also included to:
[0176] Extract tree species-related features from ground sample survey data;
[0177] Constructing a sample set based on remote sensing image features of tree species and features related to the tree species;
[0178] Taking the features related to tree species in the sample set as input and the tree species category as output, training is performed using a machine learning algorithm to obtain a random forest model;
[0179] The tree species-related features include: spectral features, texture features, time-varying features, and observed height, tree age, and canopy structure information.
[0180] Optionally, a growth model building module is also included to:
[0181] Build a sample library of tree ages and growth rates of the same tree species;
[0182] Dividing the sample library into modeling data and test data according to a set ratio;
[0183] Taking tree age as an independent variable and growth rate as a dependent variable, a linear regression model is used to fit the modeling data to obtain an initial tree growth rate prediction model;
[0184] A tree growth rate prediction model is determined based on the test data and the initial tree growth rate prediction model using a root mean square error method.
[0185] Optionally, the specific implementation steps of determining the tree growth rate prediction model by using the root mean square error method based on the test data and the initial tree growth rate prediction model in the growth model construction module include:
[0186] Substituting the tree age in the test data into the initial tree growth rate prediction model to obtain a tree growth rate prediction value;
[0187] The root mean square error method is used to calculate the root mean square error between the tree growth rate prediction value and the tree growth rate in the test data, and an initial tree growth rate prediction model whose root mean square error is less than a set threshold is selected as the tree growth rate prediction model.
[0188] Optional, early warning module, specifically used for:
[0189] The dynamic 3D model outputs the 3D clearance distance between the wire and the trees;
[0190] Whether to issue an early warning message is determined based on the three-dimensional clearance distance between the wire and the tree and the overhead power transmission line operation regulations.
[0191] Optionally, the steps of determining whether to issue a warning message based on the three-dimensional clearance distance between the wire and the tree and the operation procedures of the overhead transmission line in the warning module include:
[0192] Determining the minimum vertical distance between the conductor and the tree from the overhead power line operation regulations based on the voltage level of the power transmission channel;
[0193] Determining whether the three-dimensional clearance distance between the wire and the tree is less than the minimum vertical distance between the wire and the tree;
[0194] If it is less than, a warning message is issued, otherwise no warning message is issued.
[0195] Optionally, also include update modules for:
[0196] Regularly collect multi-source data along the transmission corridor;
[0197] Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers;
[0198] Predicting the tree height of the tree species based on a tree growth rate prediction model corresponding to the tree species;
[0199] The dynamic three-dimensional model is updated based on the tree heights.
[0200] Example 3
[0201] like Figure 3As shown, the present invention also provides an electronic device, which may be a computer device, a single-chip device, an intelligent mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, the processor, and the transceiver component are connected via a bus; the memory may be used to store an execution program, and an exemplary execution program may include instructions; the processor is used to execute the instructions stored in the memory. The memory may also be used to store data, which may be called and / or modified when the instructions are executed.
[0202] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in a storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of a tree obstacle prediction and early warning method for a power transmission channel based on satellite remote sensing in the above-mentioned embodiment.
[0203] Example 4
[0204] Based on the same inventive concept, the present invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory), which is a memory device in the electronic device for storing programs and data. It can be understood that the storage medium here can include both the built-in storage medium in the electronic device and the extended storage medium supported by the electronic device. The storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more execution programs (including program codes). It should be noted that the storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor loads and executes one or more instructions stored in the storage medium, which can implement the steps of a tree obstacle prediction and early warning method for a power transmission channel based on satellite remote sensing in the above embodiment.
[0205] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0206] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0207] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0208] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0209] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.
Claims
1. A method for predicting and warning tree obstacles in power transmission channels based on satellite remote sensing. It is characterized in that include: Collect multi-source data along the transmission channel, and generate a three-dimensional initial model including the transmission line body and channel tree barriers based on the multi-source data; Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers; Based on a tree growth rate prediction model corresponding to the tree species, predict the growth rate of the tree species within a prediction time period, and determine the tree height within the prediction time period based on the growth rate; The three-dimensional initial model is updated based on the tree heights within the prediction time period to obtain a dynamic three-dimensional model, and an early warning is issued based on the dynamic three-dimensional model.
2. The method according to claim 1, It is characterized in that The multi-source data include: multi-spectral satellite remote sensing data, global ecosystem dynamics survey data and ground field survey data.
3. The method according to claim 2, It is characterized in that The generating of a three-dimensional initial model including a transmission line body and channel tree barriers based on the multi-source data comprises: Input stereo satellite remote sensing images into 3D modeling software and output digital surface models within the image coverage area; Based on the power grid ledger data, using stereo satellite remote sensing images and 3DsMax software, we built three-dimensional models of trees, transmission towers and power lines. The digital surface model and the monomeric three-dimensional model are unified into the same coordinate system, and the two are merged in space to generate a three-dimensional initial model of the transmission line body and the channel tree barrier.
4. The method according to claim 2, It is characterized in that Based on the multi-source data, using a pre-trained random forest model to identify tree species of tree barriers includes: Extract spectral features and texture features using multispectral satellite remote sensing data; Extracting three-dimensional structural features using the Global Ecosystem Dynamics Survey; Inputting the spectral feature, the texture feature and the three-dimensional structural feature into a pre-trained random forest model for identification to obtain a tree species category; Among them, the random forest model is obtained by training the machine learning algorithm based on the remote sensing image features of tree species and the features related to tree species extracted from ground field survey data.
5. The method according to claim 2, It is characterized in that The training of the random forest model includes: Extract tree species-related features from ground sample survey data; Constructing a sample set based on remote sensing image features of tree species and features related to the tree species; Taking the features related to tree species in the sample set as input and the tree species category as output, training is performed using a machine learning algorithm to obtain a random forest model; The tree species-related features include: spectral features, texture features, time-varying features, and observed height, tree age, and canopy structure information.
6. The method according to claim 1, It is characterized in that The construction of the tree growth rate prediction model includes: Build a sample library of tree ages and growth rates of the same tree species; Dividing the sample library into modeling data and test data according to a set ratio; Taking tree age as an independent variable and growth rate as a dependent variable, a linear regression model is used to fit the modeling data to obtain an initial tree growth rate prediction model; A tree growth rate prediction model is determined based on the test data and the initial tree growth rate prediction model using a root mean square error method.
7. The method according to claim 6, It is characterized in that The method of determining the tree growth rate prediction model based on the test data and the initial tree growth rate prediction model by using a root mean square error method comprises: Substituting the tree age in the test data into the initial tree growth rate prediction model to obtain a tree growth rate prediction value; The root mean square error method is used to calculate the root mean square error between the tree growth rate prediction value and the tree growth rate in the test data, and an initial tree growth rate prediction model whose root mean square error is less than a set threshold is selected as the tree growth rate prediction model.
8. The method according to claim 1, It is characterized in that The early warning based on the dynamic three-dimensional model includes: The dynamic 3D model outputs the 3D clearance distance between the wire and the trees; Whether to issue an early warning message is determined based on the three-dimensional clearance distance between the wire and the tree and the overhead power transmission line operation regulations.
9. The method according to claim 8, It is characterized in that The determining whether to issue warning information based on the three-dimensional clearance distance between the wire and the tree and the overhead transmission line operation regulations includes: Determining the minimum vertical distance between the conductor and the tree from the overhead power line operation regulations based on the voltage level of the power transmission channel; Determining whether the three-dimensional clearance distance between the wire and the tree is less than the minimum vertical distance between the wire and the tree; If it is less than, a warning message is issued, otherwise no warning message is issued.
10. The method according to claim 1, It is characterized in that Also includes: Regularly collect multi-source data along the transmission corridor; Based on the multi-source data, a pre-trained random forest model is used to identify tree species of tree barriers; Predicting the tree height of the tree species based on a tree growth rate prediction model corresponding to the tree species; The dynamic three-dimensional model is updated based on the tree heights.
11. A tree obstacle prediction and early warning system for power transmission channels based on satellite remote sensing. It is characterized in that include: An initial model building module, used to collect multi-source data along the transmission channel, and generate a three-dimensional initial model including the transmission line body and channel tree barriers based on the multi-source data; A tree species identification module, used to identify tree species of tree barriers using a pre-trained random forest model based on the multi-source data; A tree height prediction module, used to predict the growth rate of the tree species within a prediction period based on a tree growth rate prediction model corresponding to the tree species, and determine the tree height within the prediction period based on the growth rate; The early warning module is used to update the three-dimensional initial model based on the tree heights within the predicted time period to obtain a dynamic three-dimensional model, and issue an early warning based on the dynamic three-dimensional model.
12. The system according to claim 11, It is characterized in that The multi-source data include: multi-spectral satellite remote sensing data, global ecosystem dynamics survey data and ground field survey data.
13. The system of claim 12, It is characterized in that Initial model building module, specifically for: Input stereo satellite remote sensing images into 3D modeling software and output digital surface models within the image coverage area; Based on the power grid ledger data, using stereo satellite remote sensing images and 3DsMax software, we built three-dimensional models of trees, transmission towers and power lines. The digital surface model and the monomeric three-dimensional model are unified into the same coordinate system, and the two are merged in space to generate a three-dimensional initial model of the transmission line body and the channel tree barrier.
14. The system of claim 12, It is characterized in that The tree species identification module is specifically used for: Extract spectral features and texture features using multispectral satellite remote sensing data; Extracting three-dimensional structural features using the Global Ecosystem Dynamics Survey; Inputting the spectral feature, the texture feature and the three-dimensional structural feature into a pre-trained random forest model for identification to obtain a tree species category; Among them, the random forest model is obtained by training the machine learning algorithm based on the remote sensing image features of tree species and the features related to tree species extracted from ground field survey data.
15. An electronic device, It is characterized in that include: at least one processor and memory; The memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, a method for predicting and warning tree obstacles in a power transmission channel based on satellite remote sensing as described in any one of claims 1 to 10 is implemented.
16. A readable storage medium, It is characterized in that An execution program is stored thereon, and when the execution program is executed, a method for predicting and warning tree obstacles in a power transmission channel based on satellite remote sensing as described in any one of claims 1 to 10 is implemented.
Citation Information
Cited By
Tree obstacle assessment method and system
CN119228225A
