Power line detection data processing method and system, storage medium and inspection terminal

Abstract

The application relates to a power line detection data processing method and system, a storage medium and a patrol terminal, relates to the technical field of power line detection, and comprises the following steps: identifying power transmission line images patrolled by a unmanned aerial vehicle, and acquiring a predicted tree-line distance of a plurality of identified trees; according to the predicted meteorological information and historical patrol data of a preset time zone, the predicted distance is subjected to fluctuation compensation, and a compensated tree-line distance is output; based on the historical patrol data, the risk coefficients of the identified trees are evaluated according to preset evaluation indexes; the data importance is determined according to the compensated distance and the risk coefficient, and the line patrol data of the coverage area of the identified trees is subjected to differential processing according to the importance. The application solves the problems that, in traditional power line detection, the data processing for the influence of trees on line safety lacks pertinence, there is much redundant information, invalid data interference is great, and the processing precision and efficiency are low, and the precision and efficiency of power line detection data processing are improved.

Description

Technical Field

[0001] The present application relates to the field of power line detection, and in particular to a power line detection data processing method, system, storage medium and inspection terminal. Background Art

[0002] With the development of intelligent power system inspections, the impact of trees on transmission line safety has become increasingly prominent. Accurately processing relevant inspection data has become a key technical challenge in ensuring line safety. Currently, traditional power line inspection data processing methods only uniformly process various data types, making it difficult to adapt to dynamic factors such as tree growth and meteorological changes, and easily leading to the accumulation of redundant information and invalid data interference.

[0003] The existing processing method does not combine tree characteristics and environmental parameters, resulting in insufficient targeted data processing. This not only reduces the utilization value of power line inspection data, but also increases the complexity and cost of subsequent safety analysis. Summary of the Invention

[0004] In order to solve the above technical problems, the present application provides a power line detection data processing method, system, storage medium and inspection terminal, which improves the current shortcomings of traditional processing due to lack of specificity, excessive redundant information caused by dynamic factors, large interference from invalid data, low processing accuracy and efficiency.

[0005] The embodiments of this application disclose the following technical solutions:

[0006] In a first aspect, an embodiment of the present application provides a method for processing power line detection data, the method comprising:

[0007] Identify transmission line images inspected by drones and obtain several predicted tree-line distances for several identified trees;

[0008] Compensating for fluctuations in the predicted tree-line distances based on forecasted weather information within a preset time zone and historical inspection data, and outputting a plurality of compensated tree-line distances;

[0009] Based on the historical inspection data, obtaining a plurality of risk factors of the plurality of identified trees according to preset evaluation indicators;

[0010] A number of data importances are determined based on the several compensation tree-line distances and several risk coefficient assessments, and line inspection data of the coverage areas of the several identification trees are differentially processed based on the several data importances.

[0011] In a second aspect, an embodiment of the present application provides a power line detection data processing system, the system comprising:

[0012] The tree distance prediction module is used to identify the transmission line images inspected by drones and obtain several predicted tree-line distances for several identified trees;

[0013] a distance fluctuation compensation module, configured to compensate for fluctuations in the predicted tree-line distances based on the predicted meteorological information in a preset time zone and historical inspection data, and output a plurality of compensated tree-line distances;

[0014] A risk factor evaluation module is used to evaluate and obtain a plurality of risk factors of the plurality of identified trees based on the historical inspection data and according to preset evaluation indicators;

[0015] The data importance processing module is used to determine a number of data importances based on the several compensation tree-line distances and a number of risk coefficient assessments, and to perform differentiated processing on the line inspection data of the coverage areas of the several identification trees based on the several data importances.

[0016] In a third aspect, an embodiment of the present application provides a non-transitory computer-readable storage medium, wherein a first computer program is stored in the storage medium. When the first computer program is executed by a processor, the power line detection data processing method of the first aspect is implemented.

[0017] In a fourth aspect, an embodiment of the present application provides an inspection terminal, the inspection terminal comprising:

[0018] a memory for storing a second computer program;

[0019] The processor is configured to read and execute the second computer program, thereby implementing the power line detection data processing method of the first aspect.

[0020] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0021] This application proposes a method, system, storage medium and inspection terminal for processing power line inspection data, which realizes accurate and efficient processing of power line inspection data by identifying drone inspection images to obtain tree-line distance, combining meteorological and historical data for volatility compensation, evaluating risk factors, determining data importance and differential processing. First, identify the transmission line image inspected by drones, train a convolutional neural network based on historical inspection logs to obtain a distance recognition plug-in, and obtain the predicted distance of the identified trees; then, based on the predicted meteorological information and historical data of the preset time zone, perform volatility compensation on the predicted distance through tree growth prediction and sway simulation to obtain the compensated distance; then, identify the overflow height difference based on historical data, and evaluate the risk factor by combining the flammability score and wind-induced sway intensity of the tree feature database; finally, combine the compensation distance and risk factor, optimize the weight according to the probability of lodging tendency to determine the data importance, and perform differential processing on the inspection data accordingly.

[0022] The technical solution of this application solves the problems of low processing accuracy and poor efficiency caused by data redundancy and a lot of invalid information in traditional power line detection by integrating multiple steps such as distance acquisition through drone image recognition, distance compensation through fusion of meteorological and historical data, multi-dimensional risk factor assessment, and differentiated processing based on importance. It realizes targeted processing of detection data and provides technical support for ensuring the safety of transmission lines. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 A flowchart of a method for processing power line detection data provided in an embodiment of the present application;

[0025] Figure 2 A schematic diagram of the structure of a power line detection data processing system provided in an embodiment of the present application;

[0026] Figure 3 A schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of the present application;

[0027] Figure 4 A schematic diagram of the structure of the inspection terminal provided in an embodiment of the present application.

[0028] In the accompanying drawings, the components represented by the reference numerals are described as follows:

[0029] Tree distance prediction module 01, distance fluctuation compensation module 02, risk factor assessment module 03, data importance processing module 04, computer-readable storage medium 300, first computer program 311, inspection terminal 400, memory 410, processor 420, second computer program 411. DETAILED DESCRIPTION

[0030] The present application provides a power line detection data processing method, system, storage medium and inspection terminal, which are used to solve the technical problems existing in the prior art of power line detection data processing, such as lack of specificity, being affected by dynamic factors such as tree growth and meteorological changes, having a lot of redundant information, large interference from invalid data, and low processing accuracy and efficiency.

[0031] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0032] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the described features. In the description of this application, "plurality" means two or more, unless otherwise specifically specified.

[0033] In the description of this application, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art will recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0034] Example 1, as shown in the attached Figure 1 As shown, the present application provides a method for processing power line detection data, the method comprising the following steps:

[0035] S110: Identify the transmission line image inspected by the drone and obtain a plurality of predicted tree-line distances for a plurality of identified trees;

[0036] In the embodiment of the present application, in the power line detection data processing scenario, in order to accurately obtain the distance information between trees and lines, it is necessary to build a distance recognition plug-in to achieve efficient recognition of the distance between trees and lines.

[0037] Specifically, first, based on historical transmission line inspection logs, a comprehensive set of sample line images was collected. These images cover the scenes where trees and transmission lines coexist in different environments and seasons.

[0038] At the same time, the shortest relative distance between trees and transmission lines in each sample line image is accurately identified, thereby constructing a complete sample relative distance set.

[0039] Furthermore, a convolutional neural network was trained using a set of sample road image data as input and a set of sample relative distances as supervisory signals. By continuously adjusting the network parameters until the model converged, the resulting distance recognition plug-in was able to accurately identify the distance between trees and roads.

[0040] Finally, the line inspection image of the area covered by the marker trees is input into the distance recognition plug-in. The plug-in can quickly and accurately identify the distance between each marker tree and the transmission line, and output several predicted tree-line distances.

[0041] This step provides basic data support for subsequent distance compensation and data importance assessment through the technical solution of model training, ensuring the accuracy of data processing.

[0042] Step S110 of the method provided in the embodiment of the present application includes:

[0043] Based on historical transmission line inspection logs, a sample line image set is collected, and the shortest relative distances between trees and transmission lines in different sample line images are identified to obtain a sample relative distance set;

[0044] Using the sample line image set as input and the sample relative distance set as supervision, training the convolutional neural network until convergence, and obtaining a distance recognition plug-in;

[0045] The distance recognition plug-in is used to respectively recognize the line inspection images of the coverage areas of the plurality of identified trees, and output a plurality of predicted tree-line distances.

[0046] In the embodiment of the present application, in order to achieve accurate identification of the distance between trees and transmission lines, it is necessary to train a convolutional neural network based on historical inspection data, build a distance recognition plug-in, and learn the spatial correlation characteristics of trees and lines through the model to adapt to the image recognition requirements in different scenarios.

[0047] Specifically, we first collected a set of sample line images based on historical transmission line inspection logs, covering scenes of transmission lines and surrounding trees in different seasons, different terrains, and different lighting conditions. At the same time, we accurately identified the shortest relative distance between trees and transmission lines in each sample image to form a sample relative distance set, providing basic data for model training.

[0048] For example, in a certain sample image, it can be identified that the shortest straight-line distance between the trunk of a tree and the road is 5.2 meters, and in another backlit scene sample, the shortest distance between the side branches of a tree and the road is 3.8 meters, and other specific data.

[0049] Furthermore, with the sample line image set as input and the sample relative distance set as supervision, a convolutional neural network is trained as the basic framework to construct a distance recognition plug-in to realize the automatic recognition of tree-line distance.

[0050] Specifically, we first build a convolutional neural network framework consisting of an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The convolutional layer uses a 3×3 convolution kernel to extract local features such as tree outlines and road directions. The pooling layer compresses feature dimensions through maximum pooling, and the fully connected layer maps high-dimensional features into distance prediction values.

[0051] Furthermore, the sample line image set and the sample relative distance set are divided in a ratio of 7:2:1. For example, 700 groups of samples are selected from 1,000 groups as training sets, 200 groups as validation sets, and 100 groups as test sets, covering a variety of scenarios to ensure the comprehensiveness of data distribution.

[0052] Furthermore, during the model training phase, sample route images from the training set were fed into the network. The network parameters were optimized using a backpropagation algorithm, gradually reducing the mean squared error between the predicted distance and the sample distance. The initial learning rate was set at 0.001, and after every 50 iterations, the learning rate decayed to 1 / 10 of the previous value.

[0053] At the same time, the validation set is used to monitor the model performance in real time. If the mean square error of the validation set decreases by less than 0.001 meters for 10 consecutive rounds, the model is judged to have converged.

[0054] Furthermore, during the testing phase, the test set is input into the model. If the average deviation between the predicted distance and the actual identification distance is less than 0.2 meters, the distance recognition plug-in is judged to be qualified and can be put into practical use.

[0055] Finally, the trained distance recognition plug-in can output an accurate prediction of the tree-line distance based on the input line inspection image.

[0056] For example, when a line inspection image of an area covered by a landmark tree is fed into the plug-in, the model, through multi-layer feature extraction and calculation, outputs a predicted distance of 4.5 meters between the tree and the line. After manual verification, the actual minimum distance is 4.6 meters, with a deviation of only 0.1 meter, meeting the accuracy requirement.

[0057] Furthermore, after the distance recognition plug-in is harvested, it is applied to the recognition of actual line inspection images to obtain the predicted distance of the marked trees.

[0058] Specifically, the line inspection images of the area covered by the marking trees collected by the drone inspection are preprocessed according to a preset format (such as normalizing the size to 224×224 pixels) and input into the distance recognition plug-in. The plug-in uses the built-in convolutional neural network to quickly calculate and output the predicted tree-line distance corresponding to each marking tree.

[0059] At the same time, during the recognition process, the image acquisition time, location and other data are recorded in real time and stored in association with the predicted distance, providing complete basic data for subsequent volatility compensation.

[0060] Finally, the predicted tree-line distances output by the distance identification plug-in provide a quantitative basis for subsequent analysis of the safety impact of trees on transmission lines.

[0061] S120: Compensating for fluctuations in the plurality of predicted tree-line distances based on predicted weather information in a preset time zone and historical inspection data, and outputting a plurality of compensated tree-line distances;

[0062] In the embodiment of the present application, in the power line detection data processing scenario, in order to eliminate the influence of dynamic factors on distance data, it is necessary to achieve accurate compensation of the predicted distance by combining meteorological information and historical data.

[0063] Specifically, the predicted meteorological information within the preset time zone is first obtained, including the predicted wind direction, predicted wind speed, predicted temperature mean, predicted temperature range and predicted rainfall.

[0064] Among them, the preset time zone is the time interval between the current time and the next adjacent inspection time point, the predicted wind speed defaults to the maximum predicted wind speed in the preset time zone, and the predicted wind direction is the wind direction of the predicted wind speed.

[0065] At the same time, the first identified tree is randomly selected, and its first predicted tree-line distance and first historical inspection data are extracted.

[0066] Furthermore, based on the first historical inspection data, the sample temperature mean set, sample temperature interval set, sample rainfall set and historical tree growth distance after the corresponding historical time zone of the tree are collected to form a sample tree growth distance set.

[0067] Furthermore, the machine learning model is trained until convergence to obtain the first tree growth prediction plug-in. Combined with the temperature and rainfall information in the predicted meteorological information, the first predicted tree growth distance in the preset time zone is analyzed and obtained. The first predicted distance is compensated inward to obtain the first initial compensated tree-line distance.

[0068] Furthermore, a simulation constructs a first simulated sway space for the first identified tree and line, and parameters are identified based on the first initial compensation distance. Specifically, multiple sway simulations are performed within the space based on the predicted wind direction and speed. The average of the shortest distances from these simulations is taken as the first compensation tree-line distance and added to the compensation distance set.

[0069] This step accurately corrects the distance deviation caused by meteorological factors through a technical solution that combines growth compensation with swing compensation, providing dynamically adapted distance data for subsequent data importance assessment.

[0070] Step S120 in the method provided in the embodiment of the present application includes:

[0071] Obtaining forecasted weather information within a preset time zone, wherein the preset time zone is the time interval between the current time and the next adjacent inspection time point, and the forecasted weather information includes forecasted wind direction, forecasted wind speed, forecasted mean temperature, forecasted temperature range, and forecasted rainfall. The forecasted wind speed defaults to the maximum forecasted wind speed within the preset time zone, and the forecasted wind direction is the wind direction of the forecasted wind speed;

[0072] Randomly selecting a first identified tree, and obtaining a first predicted tree-line distance and first historical inspection data of the first identified tree;

[0073] Based on the first historical inspection data, as well as the predicted temperature mean, the predicted temperature range, and the predicted rainfall, analyzing to obtain a first predicted tree growth distance for the preset time zone, performing inward fluctuation compensation on the first predicted tree-line distance to obtain a first initial compensated tree-line distance;

[0074] The first initial compensation tree-line distance is simulated for fluctuation compensation according to the predicted wind direction and predicted wind speed, and a first compensation tree-line distance is output, which is added to the plurality of compensation tree-line distances.

[0075] In the embodiment of the present application, in order to achieve dynamic correction of the predicted tree-line distance, it is necessary to combine the meteorological information of the preset time zone and the historical inspection data, and perform tree-line distance compensation that fits the actual environmental changes through growth compensation and swing compensation to eliminate the interference of dynamic factors on the distance data.

[0076] Specifically, the system first obtains forecasted weather information for a preset time zone. The preset time zone is the time interval between the current time and the next adjacent inspection time point. The forecasted weather information includes forecasted wind direction, forecasted wind speed (the maximum forecasted wind speed within the preset time zone is used by default), forecasted mean temperature, forecasted temperature range, and forecasted rainfall. The forecasted wind direction corresponds to the maximum forecasted wind speed, providing comprehensive meteorological parameter support for subsequent compensation calculations.

[0077] Furthermore, a first identified tree is randomly selected, and its corresponding first predicted tree-line distance and first historical inspection data are obtained based on the constructed distance identifier plug-in as basic data for the tree distance compensation.

[0078] Furthermore, a growth distance analysis is performed based on the first historical inspection data.

[0079] In the method provided in the embodiment of the present application, the step of “analyzing and obtaining the first predicted tree growth distance in the preset time zone based on the first historical inspection data, the predicted temperature mean, the predicted temperature range, and the predicted rainfall” includes:

[0080] Based on the first historical inspection data, a sample temperature mean set, a sample temperature interval set, and a sample rainfall set of the first identified tree are collected, and historical tree growth distances of different sample temperature means, sample temperature intervals, and sample rainfall in historical time zones are obtained to obtain a sample tree growth distance set;

[0081] Using the sample temperature mean set, the sample temperature interval set, the sample rainfall set, and the sample tree growth distance set, a machine learning model is trained until convergence to obtain a first tree growth prediction plug-in;

[0082] The first tree growth prediction plug-in is used to obtain the first predicted tree growth distance in the preset time zone based on the predicted temperature mean, predicted temperature range and predicted rainfall analysis.

[0083] In an embodiment of the present application, in order to accurately predict the growth distance of trees in a preset time zone, it is necessary to construct a first tree growth prediction plug-in based on historical environmental data and growth patterns, and learn the correlation characteristics between meteorological factors and tree growth through a machine learning algorithm to achieve quantitative analysis of the growth distance.

[0084] Specifically, firstly, based on the first historical inspection data, a sample temperature mean set, a sample temperature range set, and a sample rainfall set of the first identified tree at different growth stages are collected.

[0085] At the same time, the historical tree growth distances corresponding to these sample conditions after passing through historical time zones are recorded and integrated to form a sample tree growth distance set, providing comprehensive input features and supervision data for model training.

[0086] For example, if the first identified tree is a willow tree in a certain area, its sample temperature mean set may include the average temperature in different seasons, such as 18℃ in spring, 26℃ in summer, and 15℃ in autumn; the sample temperature interval set covers the daytime temperature range of each season, such as 10-25℃ in spring, 20-32℃ in summer, and 8-22℃ in autumn; the sample rainfall set includes the cumulative rainfall in each season, such as 150mm in spring, 280mm in summer, and 100mm in autumn, and the corresponding historical tree growth distances are 0.5m, 1.1m, and 0.3m, respectively. These data together constitute the sample tree growth distance set.

[0087] Furthermore, the first tree growth prediction plug-in was constructed with the sample temperature mean set, sample temperature interval set, and sample rainfall set as input features, the sample tree growth distance set as the supervision signal, and the gradient boosting regression tree as the basic framework.

[0088] Specifically, the sample temperature mean set, sample temperature interval set, sample rainfall set and the corresponding sample tree growth distance set are first divided into a training set and a validation set in a ratio of 8:2 to ensure that the distribution of the two sets of data under different meteorological conditions and growth stages is consistent, so as to ensure the generalization ability of model training.

[0089] During the training process, using a gradient boosting regression tree as a framework, the prediction results are gradually optimized by iteratively generating multiple regression trees. The first tree constructs an initial prediction model based on the training set data and calculates the residual between its predicted value and the actual historical tree growth distance. The second tree is trained using this residual as a target, fitting the residual pattern to correct the previous prediction deviation. This process is repeated for each subsequent tree, continuously reducing the overall prediction error.

[0090] For example, the actual tree growth distance of a certain group of samples is 1.2m, the initial tree prediction value is 0.9m, and the residual is 0.3m; after the second tree is trained based on this residual, the prediction value can be corrected to 1.1m, and the residual is reduced to 0.1m. After multiple rounds of iterations, the prediction deviation gradually converges.

[0091] At the same time, a learning rate of 0.1 is introduced to control the contribution weight of each tree, that is, the prediction results of each tree only participate in the final prediction at a ratio of 10%, avoiding a single tree from having an excessive impact on the overall model and effectively preventing overfitting.

[0092] During training, the validation set is used to evaluate the model performance in real time, and the mean square error between the predicted growth distance and the actual value is calculated. When the error decreases by less than 0.005m for five consecutive iterations, the model is considered to have converged and training is stopped.

[0093] Finally, the first tree growth prediction plug-in trained through the above steps can accurately capture the intrinsic relationship between meteorological characteristics and tree growth distance, providing reliable model support for growth distance prediction in the preset time zone.

[0094] Furthermore, after obtaining the first predicted tree growth distance, the first predicted tree-line distance is compensated for inward fluctuations based on the growth distance, thereby obtaining a first initial compensated tree-line distance.

[0095] Specifically, the inward fluctuation compensation refers to subtracting the first predicted tree growth distance from the first predicted tree-line distance to correct the distance shortening that may be caused by tree growth.

[0096] For example, if the first predicted tree-line distance of the first identified tree is 5.0m, and the first predicted tree growth distance in the preset time zone obtained by the first tree growth prediction plug-in is 0.6m, then after inward volatility compensation, the first initial compensated tree-line distance is 5.0-0.6=4.4m.

[0097] This compensation process fully considers the growth trend of trees within the preset time zone, so that the initial compensation distance is more in line with the actual distance between trees and transmission lines at the end of the preset time zone, ensuring that the distance data can more accurately reflect potential safety risks.

[0098] In the method provided in the embodiment of the present application, the step of “performing simulated fluctuation compensation for the first initial compensation tree-line distance according to the predicted wind direction and predicted wind speed, and outputting the first compensation tree-line distance” includes:

[0099] Simulating and constructing a first simulated swaying space between the first identified tree and the transmission line, and performing parameter identification on the first simulated swaying space according to the first initial compensation tree-line distance;

[0100] In the first simulated swaying space, multiple swaying simulations are performed based on the predicted wind direction and predicted wind speed, and multiple simulated shortest distances between the first identification tree and the transmission line are output, and the average of the multiple simulated shortest distances is set as the first compensation tree-line distance.

[0101] In the embodiment of the present application, in order to fully consider the impact of the swing of trees under the action of wind on their distance from the transmission line, it is necessary to perform dynamic compensation by constructing a simulated swing space to correct the initial compensation distance and obtain a compensation distance that is more in line with the actual wind conditions.

[0102] Specifically, the first simulated swing space of the first identified tree and the transmission line is simulated and constructed. The space needs to accurately restore the morphological characteristics of the tree and the spatial position of the transmission line, and perform parameter identification of the relative position relationship in the space based on the first initial compensation tree-line distance to ensure the consistency between the simulated scene and the actual environment.

[0103] For example, if the first identified tree is a poplar tree with a height of 12m and a crown width of 5m, the average height of the transmission line is 15m, and the first initial compensation tree-line distance is 4.4m, then in the simulated swing space, the initial relative distance between the tree trunk and the line needs to be set to 4.4m, and the spatial distribution range of the tree crown must be restored proportionally.

[0104] Furthermore, in the first simulated swaying space, multiple swaying simulations are performed based on the predicted wind direction and predicted wind speed in the preset time zone.

[0105] During the simulation process, it is necessary to calculate the swing amplitude and frequency of branches in different parts based on the material properties of the trees (such as wood hardness and branch flexibility) and wind speed levels, and focus on tracking the changes in the shortest distance between the trees and the lines.

[0106] Specifically, the number of simulations can be set to 50, with each simulation based on a random wind load perturbation (fluctuating within ±10% of the predicted wind speed) and outputting the shortest distance value from each simulation. For example, under a southeast wind of 10 m / s, the shortest distances obtained from 50 simulations fluctuated between 3.8 m and 4.3 m, with multiple simulations converging around 4.0 m.

[0107] Finally, the average of these 50 simulated shortest distances is obtained, and the result is the first compensation tree-line distance.

[0108] For example, if the average of the above 50 simulations is 4.05m, then this average is used as the first compensation tree-line distance, which further corrects the distance reduction risk caused by wind compared to the first initial compensation distance, and provides a more accurate dynamic distance basis for subsequent data importance evaluation.

[0109] Furthermore, after completing the simulated fluctuation compensation, the obtained first compensated tree-line distance is added to a plurality of compensated tree-line distance sets to achieve systematic compensation for the distances corresponding to all identified trees.

[0110] Specifically, for each marked tree, the above-mentioned simulated sway compensation process based on the predicted wind direction and predicted wind speed is repeated, and after the corresponding compensation distance is calculated, it is included in the compensation distance set.

[0111] For example, if there are 100 identified trees, after compensation one by one, the compensated tree-line distance set will contain 100 corresponding values, including short-distance compensation results such as 4.05m, as well as different values ​​such as 5.2m and 3.7m due to differences in tree position, shape and meteorological conditions.

[0112] Through the processing method of the above steps, the distance data between all marked trees and transmission lines are corrected for wind effects, ensuring that the compensated distance set can fully reflect the dynamic position relationship between various types of trees and lines at the end of the preset time zone, laying a reliable foundation for the subsequent risk factor assessment and data importance analysis based on these distance data.

[0113] S130: Based on the historical inspection data, obtain a plurality of risk factors of the plurality of identified trees according to preset evaluation indicators;

[0114] In the embodiment of the present application, in the power line detection data processing scenario, in order to quantitatively evaluate the potential threat of trees to transmission lines, it is necessary to combine historical inspection data with the inherent characteristics of trees to conduct a multi-dimensional risk factor assessment.

[0115] Specifically, first, based on historical inspection data, several overflow height differences of several marked trees are identified, that is, the vertical distance by which the trees exceeded the height of the transmission line in the most recent inspection. This indicator directly reflects the safety redundancy between the trees and the line in the vertical direction.

[0116] Furthermore, a tree feature database is used to obtain a number of flammability scores and a number of wind-induced sway intensities of a number of identified trees.

[0117] Among them, the flammability score is determined based on factors such as the tree species characteristics, the ignition point of leaves and branches, and the wind-induced sway intensity refers to data such as the swaying amplitude and frequency of the tree branches under specific wind force.

[0118] Furthermore, the overflow height difference, flammability score and wind-induced sway intensity were dimensionlessly processed to eliminate the dimensional differences between different indicators.

[0119] Furthermore, a weighted assessment method was used to calculate the hazard factor. The weight of each indicator was set according to its impact on line safety, and the hazard factor was positively correlated with the overflow height difference, flammability score, and wind-induced sway intensity.

[0120] Through this step, the multi-dimensional characteristics of trees can be converted into intuitive risk coefficients, providing a quantitative risk basis for the subsequent assessment of data importance in combination with compensation distance.

[0121] Step S130 in the method provided in the embodiment of the present application includes:

[0122] Based on the historical inspection data, identifying a plurality of overflow height differences of the plurality of identified trees;

[0123] Using a tree feature database, obtaining a plurality of flammability scores and a plurality of wind-induced sway intensities of the plurality of identified trees;

[0124] After dimensionless processing of the several overflow height differences, the several flammability scores and the several wind-induced swaying intensities, weighted evaluation is performed to obtain the several hazard coefficients of the several identified trees, wherein the hazard coefficient is positively correlated with the overflow height difference, the flammability score and the wind-induced swaying intensity.

[0125] In the embodiment of the present application, in order to scientifically quantify the potential safety threat of trees to transmission lines, it is necessary to conduct a risk factor assessment from multiple dimensions such as the spatial position relationship between the trees and the lines, their own characteristics and environmental response, so as to measure the degree of risk.

[0126] Specifically, based on historical inspection data such as historical transmission line inspection logs, we first identify and extract several overflow height differences of several identified trees, that is, the vertical distance that the trees exceed the height of the transmission line. The overflow height difference directly reflects the basic potential of the trees to pose a threat to the line in the vertical direction.

[0127] Furthermore, with the help of the tree feature database, the flammability score and wind-induced sway intensity corresponding to each identified tree are obtained.

[0128] Among them, the flammability score is determined based on the tree species, combustion characteristics of leaves and branches, and is used to assess the risk of it causing line fires in specific environments; the wind-induced swing intensity refers to the swing amplitude, frequency and other data of the tree branches under the action of wind, reflecting the possibility of it coming into contact with the line due to the action of wind.

[0129] For example, historical inspection data for a certain signposted tree showed a 2.3-meter overflow height difference, indicating that some of its branches were significantly higher than the transmission line. A database query of tree characteristics revealed that the tree is a flammable species with a flammability score of 82 out of 100. Furthermore, its branches are highly flexible, with a maximum swing amplitude of 1.5 meters in a force 8 wind, giving it a high wind-induced swing intensity rating.

[0130] Furthermore, the overflow height difference, flammability score and wind-induced sway intensity were dimensionlessly processed to eliminate the dimensional differences between different indicators and uniformly map them to the numerical range of 0-1.

[0131] For example, if the maximum value of the overflow height difference is set to 5 meters, then the dimensionless value corresponding to 2.3 meters is 0.46 (2.3 / 5=0.46); the flammability score of 82 points corresponds to 0.82 (80 / 100=0.82); and the high wind-induced sway intensity corresponds to 0.9.

[0132] Subsequently, a comprehensive assessment was conducted using a weighted summation method to obtain the hazard factor. The weight of each indicator was set based on its impact on line safety, and the hazard factor was positively correlated with the overflow height difference, flammability score, and wind-induced sway intensity.

[0133] For example, in a scenario where there are high-risk trees next to a power transmission line in a mountainous area, if the weight of the overflow height difference is 0.3, the weight of the flammability score is 0.3, and the weight of the wind-induced sway intensity is 0.4, then the risk factor of the tree in the above example is 0.46×0.3+0.82×0.3+0.9×0.4=0.744.

[0134] Through the above steps, the multi-dimensional characteristics of trees are converted into intuitive risk coefficients, providing a clear risk reference for the subsequent assessment of data importance in combination with compensation distance. This helps to more accurately identify trees that pose a high threat to transmission line safety, thereby taking targeted treatment measures.

[0135] S140: Determine a plurality of data importances according to the plurality of compensation tree-line distances and a plurality of risk coefficient assessments, and perform differential processing on the line inspection data of the coverage areas of the plurality of identification trees according to the plurality of data importances.

[0136] In the embodiment of the present application, in the power line detection data processing scenario, in order to achieve accurate and efficient processing of line inspection data, it is necessary to combine the compensation distance and the risk factor to assess the importance of the data, and then carry out differentiated processing to improve the pertinence and effectiveness of data processing.

[0137] Specifically, first, for the first identified tree, the first probability of the tree's lodging tendency is predicted based on the soil characteristics of the area where it is located, as well as the predicted wind direction, predicted wind speed and predicted rainfall in the preset time zone. This probability reflects the possibility of the tree lodging in the preset time zone and affecting the safety of the line.

[0138] Furthermore, the ratio of the first lodging tendency probability to the average historical lodging tendency probability of the tree is set as the risk weight adjustment coefficient, and the initial risk weight is optimized using this coefficient to obtain the first adaptive risk weight, and the adaptive risk weight is not greater than 0.7.

[0139] At the same time, the first adaptation risk weight is subtracted from 1 to obtain the first adaptation distance weight, so as to realize the dynamic adjustment of the weights of risk coefficient and compensation distance in importance assessment.

[0140] Furthermore, based on the first adapted risk weight and the first adapted distance weight, combined with the first compensated tree-line distance and the first risk coefficient, the first data importance is evaluated and determined.

[0141] Finally, repeat the above steps to obtain the data importance of all identified trees in turn, and then perform differentiated processing on the line inspection data of the corresponding coverage area based on these importances.

[0142] Through this step, a differentiated processing mechanism with data importance as the core was constructed, which can carry out targeted processing of line inspection data according to the actual threat level of trees to transmission lines.

[0143] For example, for areas of high importance, resources can be concentrated for detailed processing and key monitoring; for areas of low importance, the processing flow can be appropriately simplified to reduce the interference of redundant information and invalid data, improve the accuracy and efficiency of data processing, and provide more targeted technical support for ensuring the safety of transmission lines.

[0144] Step S140 in the method provided in the embodiment of the present application includes:

[0145] Predicting and obtaining a first lodging tendency probability of the first identified tree based on soil characteristics of the area where the first identified tree is located, as well as the predicted wind direction, predicted wind speed, and predicted rainfall;

[0146] The ratio of the first lodging tendency probability to the mean historical lodging tendency probability of the first identified tree is set as the risk weight adjustment coefficient, and the initial risk weight is optimized to obtain a first adapted risk weight, wherein the initial risk weight is 0.2 and the adapted risk weight is not greater than 0.7;

[0147] Subtract the first adaptation risk weight from 1 to obtain a first adaptation distance weight;

[0148] Based on the first adapted risk weight and the first adapted distance weight, a first data importance is determined according to the first compensation tree-line distance and the first risk coefficient evaluation, and is added to the plurality of data importances.

[0149] In the embodiment of the present application, in order to scientifically evaluate the importance of data and realize differentiated processing of line inspection data, it is necessary to dynamically adjust the weight ratio of the hazard coefficient and the compensation distance in combination with the probability of the tree's tendency to fall, and finally form a data importance index that can accurately reflect the safety priority.

[0150] Specifically, first, for the first identified tree, a lodging risk assessment model is constructed based on the soil characteristics of the area where it is located, as well as the predicted wind direction, predicted wind speed and predicted rainfall in the preset time zone.

[0151] Among them, the lodging risk assessment model is based on the gradient boosting decision tree framework, with soil characteristics such as soil moisture and soil firmness, as well as parameters such as slope, predicted wind direction, predicted wind speed, and predicted rainfall as input, and outputs the first lodging tendency probability of the first identified tree. It is constructed by training historical data samples.

[0152] Specifically, first, based on historical inspection data, a large number of records of tree lodging under different soil characteristics and meteorological conditions were collected to form a sample data set. Each sample contains soil moisture (such as 20%, 35%, etc.), soil firmness (such as 50kPa, 80kPa, etc.), slope (such as 5°, 15°, etc.), predicted wind direction (such as east wind, northwest wind, etc., which need to be converted into numerical codes), predicted wind speed (such as 5m / s, 12m / s, etc.), predicted rainfall (such as 10mm, 50mm, etc.) and the corresponding actual lodging status of the tree (lodging is recorded as 1, and non-lodging is recorded as 0).

[0153] Among them, the numerical encoding of the predicted wind direction means that when processing data related to wind direction, since wind directions such as east wind and northwest wind belong to the category information of text description, they cannot be directly used for calculations in data processing processes such as machine learning models. Therefore, it is necessary to convert these text-based wind directions into corresponding numerical forms.

[0154] For example, east wind can be coded as 0, south wind as 1, west wind as 2, north wind as 3, northeast wind as 4, southeast wind as 5, northwest wind as 6, southwest wind as 7, etc. Through such numerical coding conversion, wind direction information can be used in subsequent data processing links such as model training, calculation analysis, etc. in numerical form.

[0155] Furthermore, the sample data set is divided into a training set and a validation set in a ratio of 7:3. The training set is used to train the gradient boosting decision tree model. Multiple decision trees are generated through iteration. Each tree learns based on the prediction residual of the previous model, and the model parameters are continuously optimized.

[0156] For example, the first decision tree is built based on the original training data, predicts the probability of tree falling, and then calculates the residual; the second decision tree is trained with this residual as the target to correct the previous prediction deviation. This process is repeated for each subsequent tree to gradually reduce the overall prediction error.

[0157] At the same time, a learning rate (such as 0.1) is introduced during the training process to control the contribution weight of each tree to avoid overfitting of the model. The prediction accuracy of the validation set is used as an indicator. When the accuracy improvement of five consecutive iterations is less than 0.01, the training is stopped and the lodging risk assessment model is judged to have converged.

[0158] Finally, the obtained lodging risk assessment model can output the lodging tendency probability of the first identified tree in the preset time zone based on the input soil characteristics and meteorological parameters, providing a basis for subsequent weight adjustment and data importance assessment.

[0159] Furthermore, the lodging risk assessment model comprehensively considers the impact of external environmental factors on the stability of trees, and predicts the first lodging tendency probability of the first identified tree. This probability reflects the possibility that the tree will fall within the preset time zone and affect the safety of the line.

[0160] For example, if the first identified tree grows in a sandy soil area with a slope of 15°, the predicted wind direction in the preset time zone is northwest wind (coded as 6), the predicted wind speed is 12m / s, and the predicted rainfall is 50mm, combining these factors and calculating through the lodging risk model, it is concluded that the first lodging tendency probability of the tree in the preset time zone is 0.35, indicating that it has a high risk of lodging and may pose a sudden threat to the safety of the transmission line.

[0161] Furthermore, the first probability of lodging tendency is divided by the average historical probability of lodging tendency of the tree to obtain the hazard weight adjustment coefficient. The specific calculation formula is "hazard weight adjustment coefficient = first probability of lodging tendency / average historical probability of lodging tendency".

[0162] Among them, the hazard weight adjustment coefficient is used to dynamically reflect the degree of deviation of the currently predicted lodging risk from the historical average level. The larger the hazard weight value, the more significantly the current tree lodging risk is compared with the historical average state, and a higher weight needs to be given to the hazard coefficient in the importance assessment; conversely, the smaller the hazard weight value, the lower the current lodging risk is than the historical average level, and the weight of the hazard coefficient in the assessment can be reduced accordingly.

[0163] Furthermore, when the risk weight adjustment coefficient is used to optimize the initial risk weight (fixed at 0.2), the initial value of the first adaptive risk weight can be obtained through the calculation formula "initial value of the first adaptive risk weight = initial risk weight × risk weight adjustment coefficient".

[0164] Subsequently, the initial value of the first adaptive risk weight is threshold-checked. If it is less than or equal to 0.7, it is directly determined as the first adaptive risk weight; if it is greater than 0.7, the first adaptive risk weight is forcibly set to 0.7.

[0165] Furthermore, the first adaptation distance weight is obtained by subtracting the first adaptation risk weight from 1. The specific calculation formula is "first adaptation distance weight = 1 - first adaptation risk weight" to achieve a dynamic balance between the risk coefficient and the compensation distance in the importance assessment.

[0166] For example, if the first lodging tendency probability of the first identified tree is 0.3, and its historical lodging tendency average probability is 0.15, then the hazard weight adjustment coefficient is 2 (0.3 / 0.15=2), and the initial hazard weight of 0.2 is optimized to obtain the first adapted hazard weight of 0.4 (not exceeding the upper limit of 0.7), and the corresponding first adapted distance weight is 0.6 (1-0.4=0.6).

[0167] Furthermore, based on the first adaptation hazard weight and the first adaptation distance weight, combined with the first compensating tree-line distance and the first hazard coefficient, the first data importance is determined by weighted summation evaluation. The specific calculation formula is "first data importance = first adaptation hazard weight × first hazard coefficient + first adaptation distance weight × first compensating tree-line distance", and it is added to several data importance sets.

[0168] For example, if the first compensation tree-line distance is 0.5 and the first risk coefficient is 0.7, the first data importance is 0.4×0.7+0.6×0.5=0.58. After calculation, the first data importance is added to several data importance sets.

[0169] Finally, repeat the above steps to obtain the data importance of all identified trees in turn, and then perform differentiated processing on the line inspection data of the corresponding coverage area based on these importances.

[0170] Specifically, four processing strategies are set based on data importance levels. An importance level of less than 0.4 corresponds to no importance and requires minimal processing, which means retaining basic identification records (such as minimal information on "risk-free trees") with the shortest possible data retention period or directly archiving the data as a background measurement sample library, minimizing resource investment.

[0171] In addition, if the importance is 0.4 ≤ < 0.6, it corresponds to a low importance level and requires lightweight processing. This means compressing the data (e.g., JPEG compression, point cloud simplification) and temporarily storing it locally to avoid long-term resource usage. At the same time, tree growth trend values ​​are recorded for periodic comparison to balance data value and storage costs.

[0172] Furthermore, if the importance level is 0.6 ≤ < 0.8, it corresponds to a medium importance level and requires regionalization and keying. This means only cropping data from tree-covered areas is saved, the data is converted into structured storage (e.g., organized by coordinates, tree height, and type), and after noise reduction, it is uploaded to the cloud and added to the operations and maintenance queue to focus on key risk points.

[0173] In addition, importance levels ≥ 0.8 correspond to high importance levels and require the highest level of full processing. This involves preserving the original imagery in full, performing detailed processing with densification of the 3D point cloud data, annotating tree locations, numbers, and morphological labels, generating a risk layer, and uploading it to a dedicated cloud service. This layer is then prioritized for AI model training, providing a basis for in-depth analysis and risk warnings.

[0174] Ultimately, through this differentiated processing method, while ensuring the accurate processing of key data, the interference of redundant information and invalid data is reduced, effectively improving the accuracy and efficiency of overall data processing.

[0175] The embodiments of the present application achieve the following technical effects through the above specific implementation methods:

[0176] This application proposes a method for processing power line inspection data. First, it identifies transmission line images from drone inspections. Based on historical transmission line inspection logs, it collects a set of sample line images. The shortest relative distances between trees and lines are identified to obtain a set of sample relative distances. This set of sample distances is then trained with a convolutional neural network until convergence, resulting in a distance recognition plug-in. This plug-in is then used to identify inspection images of areas covered by trees and predict the tree-line distance. Next, based on predicted meteorological information and historical data for a preset time zone, a tree growth prediction plug-in, derived from a trained machine learning model, analyzes the growth distance. The predicted distance is then inwardly compensated to obtain an initial compensation distance. A simulated swaying space is then constructed, and swaying is simulated using predicted wind direction and speed. The average of the shortest distances from multiple simulations is taken as the compensated tree-line distance. Next, tree overflow height differences are identified based on historical data. This is then processed to obtain a hazard factor, combined with, for example, the flammability score from a tree feature database. Finally, the compensation distance and hazard factor are combined to optimize weights based on the probability of lodging propensity to determine data importance, and the inspection data is then differentiated accordingly.

[0177] The method provided in the embodiment of the present application solves the problems of low processing accuracy and poor efficiency caused by data redundancy and insufficient consideration of dynamic factors in traditional power line detection through the technical solution of "distance identification-dynamic compensation-hazard assessment-importance classification", realizes targeted processing of detection data, reduces redundant information and invalid data, improves the accuracy and efficiency of data processing, and provides reliable technical support for ensuring the safety of transmission lines.

[0178] Example 2, as shown in the attached Figure 2 As shown, based on the inventive concept of the power line detection data processing method provided in Example 1, the present application also provides a power line detection data processing system, specifically comprising:

[0179] The tree distance prediction module 01 is used to identify the transmission line image inspected by the drone and obtain several predicted tree-line distances for several identified trees;

[0180] Distance fluctuation compensation module 02, for compensating for fluctuations in the predicted tree-line distances based on the predicted meteorological information in a preset time zone and historical inspection data, and outputting a plurality of compensated tree-line distances;

[0181] The risk factor evaluation module 03 is used to evaluate and obtain the risk factors of the plurality of identified trees based on the historical inspection data and preset evaluation indicators;

[0182] The data importance processing module 04 is used to determine a number of data importances based on the several compensation tree-line distances and a number of risk coefficient assessments, and perform differentiated processing on the line inspection data of the coverage areas of the several identification trees according to the several data importances.

[0183] In one embodiment, the tree distance prediction module 01 is further configured to:

[0184] Based on historical transmission line inspection logs, a sample line image set is collected, and the shortest relative distances between trees and transmission lines in different sample line images are identified to obtain a sample relative distance set;

[0185] Using the sample line image set as input and the sample relative distance set as supervision, training the convolutional neural network until convergence, and obtaining a distance recognition plug-in;

[0186] The distance recognition plug-in is used to respectively recognize the line inspection images of the coverage areas of the plurality of identified trees, and output a plurality of predicted tree-line distances.

[0187] In one embodiment, the distance fluctuation compensation module 02 is further configured to:

[0188] Obtaining forecasted weather information within a preset time zone, wherein the preset time zone is the time interval between the current time and the next adjacent inspection time point, and the forecasted weather information includes forecasted wind direction, forecasted wind speed, forecasted mean temperature, forecasted temperature range, and forecasted rainfall. The forecasted wind speed defaults to the maximum forecasted wind speed within the preset time zone, and the forecasted wind direction is the wind direction of the forecasted wind speed;

[0189] Randomly selecting a first identified tree, and obtaining a first predicted tree-line distance and first historical inspection data of the first identified tree;

[0190] Based on the first historical inspection data, as well as the predicted temperature mean, the predicted temperature range, and the predicted rainfall, analyzing to obtain a first predicted tree growth distance for the preset time zone, performing inward fluctuation compensation on the first predicted tree-line distance to obtain a first initial compensated tree-line distance;

[0191] The first initial compensation tree-line distance is simulated for fluctuation compensation according to the predicted wind direction and predicted wind speed, and a first compensation tree-line distance is output, which is added to the plurality of compensation tree-line distances.

[0192] In one embodiment, the risk factor assessment module 03 is further configured to:

[0193] Based on the historical inspection data, identifying a plurality of overflow height differences of the plurality of identified trees;

[0194] Using a tree feature database, obtaining a plurality of flammability scores and a plurality of wind-induced sway intensities of the plurality of identified trees;

[0195] After dimensionless processing of the several overflow height differences, the several flammability scores and the several wind-induced swaying intensities, weighted evaluation is performed to obtain the several hazard coefficients of the several identified trees, wherein the hazard coefficient is positively correlated with the overflow height difference, the flammability score and the wind-induced swaying intensity.

[0196] In one embodiment, the data importance processing module 04 is further configured to:

[0197] Predicting and obtaining a first lodging tendency probability of the first identified tree based on soil characteristics of the area where the first identified tree is located, as well as the predicted wind direction, predicted wind speed, and predicted rainfall;

[0198] The ratio of the first lodging tendency probability to the mean historical lodging tendency probability of the first identified tree is set as the risk weight adjustment coefficient, and the initial risk weight is optimized to obtain a first adapted risk weight, wherein the initial risk weight is 0.2 and the adapted risk weight is not greater than 0.7;

[0199] Subtract the first adaptation risk weight from 1 to obtain a first adaptation distance weight;

[0200] Based on the first adapted risk weight and the first adapted distance weight, a first data importance is determined according to the first compensation tree-line distance and the first risk coefficient evaluation, and is added to the plurality of data importances.

[0201] Example 3, as shown in the attached Figure 3 As shown, based on the inventive concept of the power line detection data processing method provided in the first embodiment, the present application further provides a computer-readable storage medium 300 on which a first computer program 311 is stored.

[0202] When the first computer program 311 is executed by the processor 420 , the power line detection data processing method in the first embodiment is implemented.

[0203] Example 4, as shown in the attached Figure 4As shown, based on the inventive concept of the power line detection data processing method provided in Example 1, the present application further provides an inspection terminal 400, which comprises:

[0204] Memory 410, configured to store a second computer program 411;

[0205] The processor 420 is configured to read and execute the second computer program 411 .

[0206] When the second computer program 411 is executed by the processor 420 , the power line detection data processing method in the first embodiment is implemented.

[0207] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0208] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0209] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A method for processing power line detection data, characterized in that: The method comprises: Identifying transmission line images inspected by drones based on a distance recognition plug-in to obtain a number of predicted tree-line distances corresponding to a number of identified trees, wherein the distance recognition plug-in is trained based on a convolutional neural network and recognizes line inspection images of areas covered by the identified trees to obtain a number of predicted tree-line distances; Based on the predicted meteorological information in a preset time zone and historical inspection data, the fluctuation compensation of the plurality of predicted tree-line distances is performed, and the plurality of compensated tree-line distances are outputted, wherein the preset time zone is the time interval between the current time and the next adjacent inspection time point, and the predicted meteorological information includes the maximum predicted wind speed in the preset time zone, the wind direction of the predicted wind speed, the predicted temperature mean, the predicted temperature range, and the predicted rainfall; Randomly selecting a first identified tree, and obtaining a first predicted tree-line distance and first historical inspection data of the first identified tree; Based on the first historical inspection data, as well as the predicted temperature mean, the predicted temperature range, and the predicted rainfall, a first predicted tree growth distance for the preset time zone is obtained through analysis, and the first predicted tree-line distance is compensated for inward fluctuations to obtain a first initial compensated tree-line distance; the inward fluctuation compensation refers to subtracting the first predicted tree growth distance from the first predicted tree-line distance; Performing simulated fluctuation compensation on the first initial compensation tree-line distance according to the wind direction of the predicted wind speed and the maximum predicted wind speed, outputting a first compensation tree-line distance, and adding the first compensation tree-line distance to the plurality of compensation tree-line distances; Based on the historical inspection data, a plurality of risk factors corresponding to the plurality of identified trees are obtained according to preset evaluation indicators; A number of data importances are determined based on the several compensation tree-line distances and several risk coefficient assessments, and line inspection data of the coverage areas of the several identification trees are differentially processed based on the several data importances.

2. The power line detection data processing method according to claim 1, characterized in that: The training steps of the distance recognition component include: Based on historical transmission line inspection logs, a sample line image set is collected, and the shortest relative distance between trees and transmission lines in different sample line images is identified to obtain a sample relative distance set; The sample line image set is used as input, the sample relative distance set is used as supervision, a convolutional neural network is trained until convergence, and the distance recognition plug-in is obtained.

3. The power line detection data processing method according to claim 1, characterized in that: Based on the first historical inspection data, as well as the predicted temperature mean, the predicted temperature range, and the predicted rainfall, a first predicted tree growth distance in the preset time zone is obtained through analysis, including: Based on the first historical inspection data, a sample temperature mean set, a sample temperature interval set, and a sample rainfall set of the first identified tree are collected, and historical tree growth distances of different sample temperature means, sample temperature intervals, and sample rainfall in historical time zones are obtained to obtain a sample tree growth distance set; Using the sample temperature mean set, the sample temperature interval set, the sample rainfall set, and the sample tree growth distance set, a machine learning model is trained until convergence to obtain a first tree growth prediction plug-in; The first tree growth prediction plug-in is used to obtain the first predicted tree growth distance in the preset time zone based on the predicted temperature mean, predicted temperature range and predicted rainfall analysis.

4. The method for processing power line detection data according to claim 1, wherein: Performing simulated fluctuation compensation on the first initial compensation tree-line distance according to the wind direction of the predicted wind speed and the maximum predicted wind speed, and outputting a first compensation tree-line distance, includes: Simulating and constructing a first simulated swaying space between the first identified tree and the transmission line, and performing parameter identification on the first simulated swaying space according to the first initial compensation tree-line distance; In the first simulated swing space, multiple swing simulations are performed based on the wind direction of the predicted wind speed and the maximum predicted wind speed, and multiple simulated shortest distances between the first identification tree and the transmission line are output, and the average of the multiple simulated shortest distances is set as the first compensation tree-line distance.

5. The power line detection data processing method according to claim 1, characterized in that: Based on the historical inspection data, a number of risk factors corresponding to the number of identified trees are obtained according to preset evaluation indicators, including: Based on the historical inspection data, a plurality of overflow height differences corresponding to the plurality of identified trees are identified; Using a tree feature database, obtaining a number of flammability scores and a number of wind-induced sway intensities corresponding to the number of identified trees; After dimensionless processing of the several overflow height differences, the several flammability scores and the several wind-induced sway intensities, weighted evaluation is performed to obtain the several hazard coefficients corresponding to the several identified trees, wherein the hazard coefficient is positively correlated with the overflow height difference, the flammability score and the wind-induced sway intensity.

6. The method for processing power line detection data according to claim 1, wherein: The importance of a number of data is determined based on the distances between the compensation trees and the line and a number of risk factors, including: Predicting and obtaining a first lodging tendency probability of the first identified tree based on soil characteristics of the area where the first identified tree is located, as well as the wind direction of the predicted wind speed, the maximum predicted wind speed, and the predicted rainfall; The ratio of the first lodging tendency probability to the mean historical lodging tendency probability of the first identified tree is set as the risk weight adjustment coefficient, and the initial risk weight is optimized to obtain a first adapted risk weight, wherein the initial risk weight is 0.2 and the adapted risk weight is not greater than 0.7; Subtract the first adaptation risk weight from 1 to obtain a first adaptation distance weight; Based on the first adapted risk weight and the first adapted distance weight, a first data importance is determined according to the first compensation tree-line distance and the first risk coefficient evaluation, and is added to the plurality of data importances.

7. A power line detection data processing system, characterized in that: The system is used to execute the power line detection data processing method according to any one of claims 1 to 6, and the system includes: A tree distance prediction module is used to identify transmission line images inspected by drones based on a distance recognition plug-in, and obtain a number of predicted tree-line distances corresponding to a number of identified trees. The distance recognition plug-in is trained based on a convolutional neural network and recognizes line inspection images of the areas covered by the identified trees to obtain a number of predicted tree-line distances. a distance fluctuation compensation module, configured to compensate for fluctuations in the plurality of predicted tree-line distances based on predicted meteorological information within a preset time zone and historical inspection data, and output a plurality of compensated tree-line distances, wherein the preset time zone is the time interval between the current time and the next adjacent inspection time point, and the predicted meteorological information includes the maximum predicted wind speed within the preset time zone, the wind direction of the predicted wind speed, the predicted mean temperature, the predicted temperature range, and the predicted rainfall; The distance fluctuation compensation module is further configured to randomly select a first identified tree and obtain a first predicted tree-line distance and first historical inspection data of the first identified tree; Based on the first historical inspection data, as well as the predicted temperature mean, the predicted temperature range, and the predicted rainfall, a first predicted tree growth distance for the preset time zone is obtained through analysis, and the first predicted tree-line distance is compensated for inward fluctuations to obtain a first initial compensated tree-line distance; the inward fluctuation compensation refers to subtracting the first predicted tree growth distance from the first predicted tree-line distance; Performing simulated fluctuation compensation on the first initial compensation tree-line distance according to the wind direction of the predicted wind speed and the maximum predicted wind speed, outputting a first compensation tree-line distance, and adding the first compensation tree-line distance to the plurality of compensation tree-line distances; A risk factor evaluation module is used to evaluate and obtain a number of risk factors corresponding to the number of identified trees based on the historical inspection data and preset evaluation indicators; The data importance processing module is used to determine a number of data importances based on the several compensation tree-line distances and a number of risk coefficient assessments, and to perform differentiated processing on the line inspection data of the coverage areas of the several identification trees based on the several data importances.

8. A non-transitory computer-readable storage medium, characterized in that The storage medium stores a first computer program, which, when executed by a processor, implements the power line detection data processing method according to any one of claims 1 to 6.

9. A patrol inspection terminal, characterized in that: The inspection terminal includes a processor and a memory: a memory for storing a second computer program; A processor is configured to read and execute the second computer program, thereby implementing the power line detection data processing method according to any one of claims 1 to 6.

Images