Power line inspection road risk assessment method and apparatus, computer device, storage medium, and computer program product
By combining gradient boosting tree machine learning algorithm with multi-source data, a landslide hazard assessment model was constructed, which solved the problems of insufficient real-time performance and accuracy in the risk assessment of power line patrol roads, and realized rapid response and efficient prediction of geological disasters.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- ELECTRIC POWER RES INST CHINA SOUTHERN POWER GRID CO LTD
- Filing Date
- 2025-04-09
- Publication Date
- 2026-06-18
AI Technical Summary
Current technologies rely on manual inspections for power line patrol risk assessment, which makes it difficult to meet the real-time and accuracy requirements of patrol work. In particular, in geological disaster prediction, there are problems such as delayed early warning, insufficient comprehensive data analysis, and limited capacity for processing unbalanced samples.
A gradient boosting tree machine learning algorithm was adopted, combined with multi-source environmental factors, to construct a landslide hazard assessment model. Landslide influencing factors were obtained through remote sensing image data, a total landslide sample set was constructed and trained, and imbalanced data was processed using methods such as weighted loss function, ensemble learning, regularization, sample resampling and multi-task learning to improve prediction performance.
It improves the real-time performance and accuracy of risk assessment for power line patrol roads, enabling rapid identification of landslide hazards, providing high-quality risk warning information, and ensuring the safe and stable operation of the power system.
Smart Images
Figure CN2025087911_18062026_PF_FP_ABST
Abstract
Description
A method, apparatus, computer equipment, storage medium, and computer program product for assessing road risks during power line inspections.
[0001] Related applications
[0002] This application claims priority to Chinese patent application No. 2024117961368, filed on December 9, 2024, entitled "A method, apparatus, computer equipment, storage medium and computer program product for assessing road risks of power line inspection", the entire contents of which are incorporated herein by reference. Technical Field
[0003] This application relates to the field of geological disaster risk assessment technology, and in particular to a method, apparatus, computer equipment, storage medium and computer program product for assessing the risk of power line patrol roads. Background Technology
[0004] Technological progress and social development are inseparable from energy, and electricity is currently one of the main energy sources. Therefore, ensuring the safety and stability of the power system is of great significance. During power line inspections, geological disasters are a risk factor that cannot be ignored. In order to cope with geological disasters, it is necessary to conduct risk assessments on power line inspection routes to identify and predict risks.
[0005] In related technologies, road risk assessment for power line patrols mainly relies on manual inspections, which makes it difficult to meet the real-time and accuracy requirements of patrol work. Summary of the Invention
[0006] Therefore, it is necessary to provide a method, device, computer equipment, computer-readable storage medium, and computer program product for power line inspection road risk assessment that can meet the real-time and accuracy requirements of power line inspection work, in response to the above-mentioned technical problems.
[0007] Firstly, this application provides a method for risk assessment of power line inspection routes. The method includes:
[0008] Based on remote sensing images of the road area where the power line patrol route is located, landslide disaster data for the road area is obtained, and a landslide database is established based on the landslide disaster data; the landslide database includes meteorological, geological and remote sensing information of the road area;
[0009] Based on the landslide database, multiple landslide influencing factors were obtained, and a landslide influencing factor dataset was established based on these factors.
[0010] Construct a total landslide sample set based on landslide and non-landslide sample points within the road area;
[0011] Based on the landslide impact factor dataset and the total landslide sample set, target landslide impact factors are selected from multiple landslide impact factors, and a landslide sample set is established based on the target landslide impact factors.
[0012] Based on the training and testing sample sets in the landslide sample set, the gradient boosting tree model is trained to obtain the landslide hazard assessment model.
[0013] Risk assessment of power line patrol roads was conducted based on the landslide hazard assessment model.
[0014] In one embodiment, constructing a total landslide sample set based on landslide sample points and non-landslide sample points within the road area includes:
[0015] Obtain landslide sample points within the road area that meet the first preset requirements;
[0016] Based on the causes of landslides and landslide sample points, non-landslide sample points are obtained;
[0017] Based on the landslide sample points and non-landslide sample points, construct the total landslide sample set.
[0018] In one embodiment, obtaining non-landslide sample points based on the landslide triggering cause and landslide sample points includes:
[0019] When the landslide is triggered by landslide remains, select the same number of non-landslide sample points as the number of landslide sample points.
[0020] When the landslide is triggered by rainfall, non-landslide sample points are selected based on the landslide area and the number of landslide sample points; the ratio of the number of landslide sample points to the number of non-landslide sample points is equal to the ratio of the landslide area to the non-landslide area.
[0021] In one embodiment, based on the landslide impact factor dataset and the total landslide sample set, a target landslide impact factor is selected from multiple landslide impact factors, and a landslide sample set is established based on the target landslide impact factor, including:
[0022] Based on the landslide impact factor dataset, landslide sample points and non-landslide sample points, the multicollinearity and importance of each landslide impact factor were determined;
[0023] Select the landslide impact factor that meets the first preset requirement as the target landslide impact factor;
[0024] Based on the target landslide influencing factors, a landslide sample set is established.
[0025] In one embodiment, the gradient boosting tree model is trained based on the training sample set and the test sample set in the landslide sample set to obtain a landslide hazard assessment model, including:
[0026] According to the preset partitioning method, the landslide sample set is divided into a training sample set and a test sample set; the preset partitioning method includes a preset order and a preset ratio;
[0027] A landslide hazard assessment model is obtained by training a gradient boosting tree model using at least one gradient boosting tree algorithm or a combination of at least two gradient boosting tree algorithms, a training sample set, and a test sample set.
[0028] In one embodiment, risk assessment of the road area based on a landslide hazard assessment model includes:
[0029] Based on the landslide hazard assessment model, the landslide hazard index of the road area is predicted, the landslide hazard index of the road area is obtained, and the road area is divided into hazard zones based on the landslide hazard index of the road area.
[0030] To obtain the exposure of power line patrol roads in the road area and the vulnerability indicators of power line patrol roads after a disaster;
[0031] Based on the exposure of power line patrol roads in the road area, the vulnerability indicators of power line patrol roads after disasters, and the landslide hazard index of the target area, the risk index of the target area is determined, and the risk assessment of the target area is completed.
[0032] Secondly, this application also provides a power line inspection road risk assessment device. The device includes:
[0033] The data acquisition module is used to acquire landslide disaster data of the road area based on remote sensing images of the road area where the power line inspection road is located, and to establish a landslide database based on the landslide disaster data; the landslide database includes meteorological, geological and remote sensing information of the road area;
[0034] The data processing module is used to obtain multiple landslide influencing factors from the landslide database and to build a landslide influencing factor dataset based on these factors.
[0035] The sample construction module is used to construct a total landslide sample set based on landslide sample points and non-landslide sample points within the road area; it is also used to select a target landslide influencing factor from multiple landslide influencing factors based on the landslide influencing factor dataset and the total landslide sample set, and to establish a landslide sample set based on the target landslide influencing factor.
[0036] The model training module is used to train the gradient boosting tree model based on the training sample set and the test sample set in the landslide sample set, so as to obtain the landslide hazard assessment model.
[0037] The risk assessment module is used to conduct risk assessments on power line patrol roads based on landslide hazard assessment models.
[0038] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the methods described above.
[0039] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the methods described above.
[0040] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the methods described above.
[0041] The aforementioned methods, devices, computer equipment, storage media, and computer program products for assessing the risks of power line patrol roads capture the dynamic changes of disaster factors by extracting and fusing features from multi-source data such as geological, meteorological, and remote sensing data. This comprehensively identifies landslide influencing factors and improves the comprehensiveness and real-time nature of the assessment. Furthermore, this application employs a gradient boosting tree model suitable for imbalanced disaster sample data, addressing the problem of insufficient sample numbers for model training in minority disaster categories. This significantly improves the predictive ability for imbalanced sample data and effectively avoids the neglect of minority samples. In addition, training the landslide hazard assessment model based on the gradient boosting tree model allows for continuous optimization through training, continuously improving the accuracy and reliability of predictions and providing high-quality risk warning information and safety assurance for power line patrol personnel. Attached Figure Description
[0042] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the published drawings without creative effort.
[0043] Figure 1 shows the application environment of the power line inspection road risk assessment method in one embodiment;
[0044] Figure 2 is a flowchart illustrating the risk assessment method for power line patrol roads in one embodiment;
[0045] Figure 3 is a schematic diagram of the process of constructing a total landslide sample set based on landslide sample points and non-landslide sample points within a road area in one embodiment.
[0046] Figure 4 is a flowchart illustrating the process of obtaining non-landslide sample points based on the landslide triggering cause and landslide sample points in one embodiment.
[0047] Figure 5 is a flowchart illustrating the process of selecting a target landslide impact factor from multiple landslide impact factors and establishing a landslide sample set based on the landslide impact factor dataset and the total landslide sample set in one embodiment.
[0048] Figure 6 is a schematic diagram of the process of training a gradient boosting tree model based on the training sample set and the test sample set in a landslide sample set in one embodiment to obtain a landslide hazard assessment model.
[0049] Figure 7 is a flowchart illustrating the process of risk assessment of power line patrol roads based on a landslide hazard assessment model in one embodiment.
[0050] Figure 8 is a flowchart illustrating the power line inspection road risk assessment method in another embodiment;
[0051] Figure 9 is a schematic diagram of the design concept of a power line inspection road risk assessment method in one embodiment;
[0052] Figure 10 is a structural block diagram of a power line inspection road risk assessment device in one embodiment;
[0053] Figure 11 is an internal structure diagram of a computer device in one embodiment. Detailed Implementation
[0054] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0056] The power system is one of the fundamental infrastructures of modern society, playing an indispensable role in energy supply, economic development, social life, national security, environmental protection, and technological innovation. Therefore, ensuring the safe and stable operation of the power system is crucial for the normal functioning and sustainable development of society. Power lines are a vital component of the power system, and their safe and stable operation directly affects the continuity and reliability of power supply. Thus, risk assessments of power line patrol routes are necessary to promptly identify and address potential safety hazards, prevent accidents, and ensure the safe and stable operation of the power system.
[0057] In related technologies, power line patrol road risk assessment mainly relies on manual inspections, analysis of historical disaster records, and preliminary assessments of terrain and meteorological conditions. Common methods include on-site surveys, terrain feature extraction, and disaster frequency statistics to assess potential geological hazard risks along the patrol roads. For data collection, remote sensing imagery, geological survey data, and meteorological information are typically used to assist in identifying potential disaster threat factors. Some technologies also incorporate GIS (Geographic Information System) systems, utilizing geographic information technology to spatially analyze environmental parameters such as terrain slope and vegetation cover along the road to identify risk areas. For specific high-risk areas, regular on-site surveys or sensor monitoring are also conducted.
[0058] However, the relevant technologies still have significant shortcomings in practical applications, making it difficult to meet the real-time and accuracy requirements of pipeline inspection work. Firstly, manual inspection methods are time-consuming and inefficient, failing to enable rapid response to sudden geological disasters, resulting in delayed early warnings and insufficient coverage. Secondly, the technologies have a low level of intelligence, making it difficult to fully utilize multi-source data such as meteorological, geological, and topographical data for comprehensive risk analysis, especially struggling to capture the dynamic changes of disaster factors, leading to inaccurate prediction results. Furthermore, geological disasters are characterized by strong regionality, low incidence but high destructiveness, and the insufficient sample size of a few types of disasters makes model training and evaluation more difficult. The relevant technologies also have limited capacity to process imbalanced data, making it difficult to accurately identify and predict a few types of risks.
[0059] Therefore, this application proposes a method for rapid and accurate risk assessment of power line inspection roads based on gradient boosting tree machine learning algorithm and combined with multiple environmental factors, in order to solve the risk prediction problem of geological disasters on power line inspection roads in complex terrain.
[0060] The gradient boosting tree algorithm used in this application has the following advantages when dealing with imbalanced sample data:
[0061] Weighted Loss Function: Gradient boosting trees allow the use of weighted loss functions, which assign different weights to samples of different classes. In cases of imbalanced data, increasing the weight of minority class samples allows the model to focus more on these samples during training, thereby improving the model's predictive performance.
[0062] Ensemble learning: Gradient boosting trees are an ensemble learning method that progressively improves a model's predictive performance by constructing multiple decision trees. In cases of imbalanced data, each decision tree can focus on a different subset of samples, thereby enhancing the model's generalization ability.
[0063] Regularization: Gradient boosting trees typically include regularization terms to help prevent overfitting. In cases of imbalanced data, regularization can reduce the model's over-reliance on the majority class, thereby improving the model's predictive performance.
[0064] Sample resampling: During the training of gradient boosting trees, sample resampling techniques can be used to balance the sample data. This can generate copies of minority class samples during the training of each decision tree, thereby improving the model's predictive performance.
[0065] Multi-task learning: Gradient boosting trees can be used for multi-task learning, meaning they can handle multiple related tasks simultaneously. In cases of imbalanced data, multi-task learning can improve the model's predictive performance; for example, it can simultaneously predict the class and probability of a sample.
[0066] Feature selection: Gradient boosting trees can automatically perform feature selection during training, which helps improve the model's predictive performance. In cases of imbalanced data, feature selection can help the model identify features that have a greater impact on the predictive performance of minority class samples.
[0067] Model interpretability: Gradient boosting trees offer good model interpretability, allowing users to understand the model's prediction process by analyzing the tree's structure. In cases of imbalanced data, model interpretability helps users understand the model's predictive performance on minority class samples.
[0068] In summary, gradient boosting trees offer advantages such as weighted loss function, ensemble learning, regularization, sample resampling, multi-task learning, feature selection, and model interpretability when dealing with imbalanced sample data. These advantages help improve the predictive performance of the model.
[0069] The power line patrol road risk assessment method provided in this application embodiment can be applied to the application environment shown in Figure 1. The image acquisition device 102 and the terminal 104 communicate via a network. The image acquisition device 102 is used to acquire remote sensing images of the road area where the power line patrol road is located and transmits these images to the terminal 104. The terminal 104 executes the various steps in the power line patrol road risk assessment method of this application based on the remote sensing images of the road area where the power line patrol road is located. The terminal 104 can be, but is not limited to, various personal computers, laptops, and workstations.
[0070] In an exemplary embodiment, as shown in Figure 2, a method for assessing road risks during power line inspection is provided. Taking the application of this method to the terminal in Figure 1 as an example, the method includes the following steps:
[0071] Step S202: Based on remote sensing images of the road area where the power line patrol road is located, acquire landslide disaster data for the road area, and establish a landslide database based on the landslide disaster data; the landslide database includes meteorological, geological and remote sensing information of the road area.
[0072] Among them, the road area refers to the entire power line inspection road area. Furthermore, risk assessments can also be conducted on specific areas within the road area, such as selected target areas.
[0073] Optionally, based on high-resolution, low-cloud-coverage optical remote sensing images of power line patrol roads, data on historical landslide remains and landslide disasters triggered by events such as rainfall can be acquired to construct a landslide database.
[0074] For example, remote sensing images may be selected, but are not limited to, image data with a resolution greater than 6 meters, cloud coverage less than 10%, and no cloud or fog obstruction.
[0075] For example, the landslide database contains historical landslide remnant data and landslide data triggered by rainfall. The landslide data includes spatial and attribute data, such as landslide location, extent boundaries, area, and quantity information.
[0076] For example, the landslide database may also include regional topographic and geomorphological conditions, regional engineering geological conditions, regional meteorological and hydrological conditions, regional surface cover conditions, and regional landslide logging data.
[0077] For example, the landslide database format includes, but is not limited to, “.shp”, “.kml”, and “.kmz”.
[0078] Step S204: Based on the landslide database, obtain multiple landslide influencing factors and establish a landslide influencing factor dataset based on the multiple landslide influencing factors.
[0079] Optionally, based on multi-source data in the landslide database, multiple landslide influencing factors such as geology, topography, and meteorology are selected, and a landslide influencing factor dataset is established after normalization of each landslide influencing factor.
[0080] For example, multiple landslide influencing factors include, but are not limited to: historical earthquake parameters, elevation, slope, aspect, curvature, slope position, topographic relief, topographic humidity index, land use type, vegetation cover, stratigraphic lithology, distance from active fault, distance from river, and average annual rainfall over the past five years.
[0081] For example, the spatial resolution of all landslide impact factor data in the landslide impact factor dataset includes, but is not limited to, 12.5m, 30m, and 100m.
[0082] Step S206: Construct a total landslide sample set based on the landslide sample points and non-landslide sample points within the road area.
[0083] Optionally, points in the landslide area within the road area are selected as landslide sample points, and points in the non-landslide area within the road area are selected as non-landslide sample points. A total landslide sample set is constructed based on the landslide sample points and the non-landslide sample points.
[0084] Step S208: Based on the landslide impact factor dataset and the total landslide sample set, select the target landslide impact factor from multiple landslide impact factors, and establish a landslide sample set based on the target landslide impact factor.
[0085] Optionally, based on the landslide impact factor dataset and selected landslide and non-landslide sample points, the multicollinearity and contribution level of each landslide impact factor are analyzed. Landslide impact factors with no multicollinearity and high contribution are selected to establish a landslide training and testing sample set.
[0086] Step S210: Based on the training sample set and test sample set in the landslide sample set, train the gradient boosting tree model to obtain the landslide hazard assessment model.
[0087] Optionally, the established landslide training sample set and test sample set can be input into the gradient boosting tree model to train the gradient boosting tree model until training is completed, thereby constructing a landslide hazard assessment model.
[0088] Step S212: Conduct a risk assessment of the power line patrol road based on the landslide hazard assessment model.
[0089] Optionally, based on the landslide hazard assessment model, and considering the geographical conditions and road design specifications near the power line patrol road, the exposure of the power line patrol road in the target area and the vulnerability index after a disaster are determined, and the risk index of the power line patrol road area is calculated.
[0090] The aforementioned risk assessment method for power line patrol roads fully utilizes multi-source data such as meteorological, geological, and topographical data to comprehensively analyze risks and capture the dynamic changes of disaster factors, thus solving the real-time problem of power line patrol road risk prediction. At the same time, by training a landslide hazard assessment model based on gradient boosting trees, the predictive ability for unbalanced sample data is significantly improved, effectively avoiding the neglect of minority class samples and enhancing the accuracy and reliability of power line patrol road risk prediction.
[0091] To construct a more comprehensive overall sample set of landslides, this application proposes the following scheme for selecting sample points in both landslide and non-landslide areas.
[0092] In an exemplary embodiment, as shown in FIG3, step S206, constructing a total landslide sample set based on landslide sample points and non-landslide sample points within the road area, includes:
[0093] Step S302: Obtain landslide sample points within the road area that meet the first preset requirements.
[0094] The first preset requirement indicates the requirements for the selection method of landslide sample points in the power line patrol area, which can be adaptively adjusted according to different power line patrol conditions.
[0095] For example, the method for selecting landslide sample points in the power line inspection area can be: 1) converting landslide surface features into raster data and extracting each pixel of the converted raster data as a landslide sample point; 2) converting landslide surface features into point features, with one landslide corresponding to one point as a landslide sample.
[0096] Step S304: Based on the landslide triggering cause and landslide sample points, obtain non-landslide sample points.
[0097] Optionally, the cause of landslide triggering, the number of landslide sample points, and the area of landslide sample points can all be used as factors influencing the selection of non-landslide sample points.
[0098] For example, landslides can be triggered by historical landslide remains or rainfall.
[0099] For example, the rules for selecting landslide and non-landslide sample points differ for different types of landslides. When the landslide is triggered by landslide remnants, the same number of non-landslide sample points are selected as the number of landslide sample points. When the landslide is triggered by rainfall, non-landslide sample points are selected based on the landslide area and the number of landslide sample points. The ratio of the number of landslide sample points to the number of non-landslide sample points is equal to the ratio of the landslide area to the non-landslide area.
[0100] Step S306: Construct a total landslide sample set based on landslide sample points and non-landslide sample points.
[0101] This embodiment provides a method for selecting landslide sample points and non-landslide sample points, points out their inherent relationship, and gives different methods for selecting non-landslide sample points according to the causes of landslide triggering, so as to ensure the comprehensiveness of the total landslide sample set.
[0102] The following describes in detail the method for selecting landslide sample points and non-landslide sample points with reference to specific embodiments.
[0103] In an exemplary embodiment, as shown in FIG4, step S304, obtaining non-landslide sample points based on the landslide triggering cause and landslide sample points, includes:
[0104] Step S402: If the landslide is triggered by landslide remains, select the same number of non-landslide sample points as the number of landslide sample points.
[0105] Step S404: When the landslide is triggered by rainfall, select non-landslide sample points based on the landslide area and the number of landslide sample points; the ratio of the number of landslide sample points to the number of non-landslide sample points is equal to the ratio of the landslide area to the non-landslide area.
[0106] For example, the landslide sample point selection method includes, but is not limited to: 1) converting landslide surface features into raster data and extracting each cell of the converted raster data as a landslide sample point; 2) converting landslide surface features into point features, with one landslide corresponding to one point as a landslide sample.
[0107] For example, when the landslide is triggered by landslide remains, the methods for selecting non-landslide sample points include, but are not limited to: 1) extracting non-landslide areas within the road area and randomly selecting points within this range; 2) constructing a net-like point matrix from the non-landslide areas and randomly selecting points from it.
[0108] The landslide was triggered by rainfall. The method for selecting the sample size according to the ratio of landslide to non-landslide area includes, but is not limited to: 1) taking a large number of random points in the entire road area, and then counting whether the area where the points are located belongs to the landslide area, and labeling each point as a landslide or non-landslide sample; 2) first randomly selecting an appropriate amount of point data in the landslide area, and then selecting corresponding sample points in the non-landslide area according to the ratio of landslide to non-landslide area.
[0109] This embodiment improves the selection scheme for landslide sample points and non-landslide sample points. Based on this, the total landslide sample set constructed is more comprehensive and accurate, providing a good foundation for the subsequent construction of a landslide hazard assessment model.
[0110] After the overall landslide sample set is constructed, landslide influencing factors need to be selected to form landslide samples for training the landslide hazard assessment model.
[0111] In an exemplary embodiment, as shown in FIG5, step S208, selecting a target landslide impact factor from multiple landslide impact factors based on the landslide impact factor dataset and the total landslide sample set, and establishing a landslide sample set based on the target landslide impact factor includes:
[0112] Step S502: Based on the landslide impact factor dataset, landslide sample points and non-landslide sample points, determine the multicollinearity and importance of each landslide impact factor.
[0113] Optionally, based on the landslide impact factor dataset and the selected landslide sample points and non-landslide sample points, the multicollinearity and contribution of each landslide impact factor can be analyzed.
[0114] For example, multicollinearity analysis methods for landslide influencing factors include, but are not limited to: 1) Variance Inflation Factor (VIF) analysis: By calculating the Variance Inflation Factor (VIF) value of each influencing factor, the degree of collinearity among landslide influencing factors is determined. When the VIF value is higher than a specific threshold (such as 5 or 10), it indicates that collinearity exists among landslide influencing factors; 2) Correlation coefficient matrix analysis: The correlation coefficients among landslide influencing factors are calculated to generate a correlation coefficient matrix. High correlation coefficients (close to 1 or -1) indicate that collinearity may exist among landslide influencing factors; 3) Condition number analysis: By calculating the condition number of the characteristic matrix of landslide influencing factors, the collinearity problem is determined. When the condition number is greater than 30, it indicates that severe collinearity exists.
[0115] For example, the contribution analysis methods for landslide influencing factors include, but are not limited to: 1) Multiple regression analysis: using a multiple linear regression model to analyze the contribution of landslide influencing factors to the probability of landslide occurrence, and using standardized regression coefficients to determine the importance of each factor. The larger the absolute value of the coefficient, the greater the contribution of the factor to the landslide; 2) Random forest feature importance: using the random forest algorithm to calculate the feature importance of each influencing factor. The higher the importance of the factor, the greater its contribution to landslide prediction; 3) Information gain: using the information gain method to evaluate the contribution of each influencing factor to landslide classification information. The larger the information gain, the higher the importance of the factor to the occurrence of landslide.
[0116] Step S504: Select the landslide impact factor that meets the first preset requirement as the target landslide impact factor.
[0117] The first preset requirement represents the performance requirements for selecting landslide influencing factors, such as no multicollinearity and a high ranking in terms of contribution.
[0118] Optionally, a landslide training and testing sample set can be established by selecting target landslide influencing factors that are free from multicollinearity and have a high contribution rate. The number of target landslide influencing factors can be adjusted according to training requirements, and this application does not impose any restrictions.
[0119] Step S506: Establish a landslide sample set based on the target landslide influencing factors.
[0120] In the aforementioned risk assessment method for power line inspection roads, multicollinearity and contribution analysis are performed on landslide influencing factors extracted from multi-source data such as geology, meteorology, and remote sensing to remove redundant information and obtain accurate and comprehensive target landslide influencing factors, thereby capturing the dynamic changes of landslide influencing factors. In addition, by using multiple multicollinearity and contribution analysis methods, the limitations of a single selection method are avoided, ensuring the reliability and comprehensiveness of the landslide sample set and enhancing the prediction accuracy of landslide risk disasters.
[0121] After the sample set is built, it is necessary to train the model based on the landslide sample set. The model training method of this application is described in detail below with reference to the embodiments.
[0122] In an exemplary embodiment, as shown in Figure 6, step S210 involves training the gradient boosting tree model based on the training sample set and the test sample set in the landslide sample set to obtain a landslide hazard assessment model, including:
[0123] Step S602: According to the preset division method, the landslide sample set is divided into a training sample set and a test sample set; the preset division method includes a preset order and a preset ratio.
[0124] The preset division method includes a preset order and a preset ratio. The preset order can be random, sequential, or interval-based. The preset ratio can be 9:1, 8:2, 7:3, 6:4, etc.
[0125] For example, landslide sample points and non-landslide sample points in the landslide sample set are labeled. The labeling methods include: landslide sample points are labeled as "1" and non-landslide sample points are labeled as "2", or a combination of "1" and "0".
[0126] Step S604: Train the gradient boosting tree model according to at least one gradient boosting tree algorithm or a combination of at least two gradient boosting tree algorithms, a training sample set and a test sample set to obtain a landslide hazard assessment model.
[0127] The gradient boosting tree algorithm includes, but is not limited to: 1) Gradient Boosting Decision Tree (GBDT): A classic gradient boosting decision tree algorithm that optimizes the model's residuals by progressively building multiple decision trees, with each tree learning the error predicted in the previous step. GBDT performs well in regression and classification tasks; 2) Extreme Gradient Boosting (XGBoost): An improved gradient boosting tree algorithm with higher efficiency and scalability. XGBoost prevents overfitting through parallel processing and regularization, and allows for custom objective functions; 3) Categorical Boosting (CatBoost): A gradient boosting algorithm optimized for categorical features. It processes categorical features through a special target encoding method, reducing model bias and variance, and preventing overfitting.
[0128] Optionally, for different training tasks and evaluation needs, one gradient boosting tree algorithm or a combination of multiple gradient boosting tree algorithms can be selected to train the gradient boosting tree model, thereby obtaining a landslide hazard assessment model built based on the gradient boosting tree model.
[0129] In the aforementioned power line inspection road risk assessment method, by dividing the training sample set and the test sample set, and utilizing the advantages of the gradient boosting tree algorithm and its combination algorithm in imbalanced sample data scenarios, the landslide hazard assessment model's predictive ability for imbalanced sample data is significantly improved, effectively avoiding the neglect of minority class samples. Simultaneously, by selecting appropriate machine learning algorithms or combinations of multiple algorithms for different training tasks, the landslide hazard assessment model can be continuously optimized through training, enabling it to perform risk assessment tasks with different requirements. This enhances the comprehensiveness and scalability of risk assessment, and improves the accuracy and reliability of predictions.
[0130] In an exemplary embodiment, as shown in FIG7, step S212, conducting a risk assessment of the power line patrol road according to the landslide hazard assessment model, includes:
[0131] Step S702: Based on the landslide hazard assessment model, predict the landslide hazard index of the road area to obtain the landslide hazard index of the road area, and classify the road area into hazard zones based on the landslide hazard index of the road area.
[0132] For example, the range of the landslide hazard index includes 0-1 or other ranges greater than 0 and less than 1; the specific value depends on the calculation results.
[0133] For example, landslide hazard zoning methods include, but are not limited to: the natural discontinuity method, which divides the hazard zone into 5 categories: extremely low hazard zone, low hazard zone, medium hazard zone, high hazard zone and extremely high hazard zone; and the equidistant classification method, which divides the 0-1 interval into 5 segments on average.
[0134] Step S704: Obtain the exposure of power line patrol roads in the road area and the vulnerability indicators of power line patrol roads after a disaster.
[0135] The defining elements of road exposure include, but are not limited to: geographical location, importance, frequency of use, traffic flow, and surrounding environment, etc., and each of these elements is evaluated and assigned a value; the method for calculating exposure (E) is as follows (1):
[0136] Where E represents exposure level, n represents factor type, and w i x represents the weight of the i-th exposure factor. i Let i represent the i-th exposure factor.
[0137] The defining elements of road vulnerability include, but are not limited to, geological conditions, topographical factors, climatic conditions, and degree of aging. Road vulnerability (V) is calculated according to the following formula (2):
[0138] Where V is the road vulnerability value, n represents the number of factor types, and F i For the i-th vulnerability factor, w i Let be the weight of the i-th vulnerability factor.
[0139] Step S706: Based on the exposure of the power line patrol road in the road area, the vulnerability index of the power line patrol road after the disaster, and the landslide hazard index of the target area, determine the risk index of the target area and complete the risk assessment of the target area.
[0140] Optionally, the method for calculating the risk index of power line patrol road areas includes, but is not limited to:
[0141] 1) Classical product formula (3): R=H×EXV (3)
[0142] Where R represents the regional risk index, H represents the regional hazard index or disaster susceptibility index, E represents exposure, and V represents vulnerability.
[0143] 2) Risk level matrix method: This method assigns threat, exposure, and vulnerability values to different levels, and then substitutes these values into a formula for calculation. It is suitable for rapid risk screening of large-scale roads.
[0144] 3) Fuzzy comprehensive evaluation method: This method uses fuzzy logic to process the membership degree of each factor and calculates the comprehensive risk index. It is suitable for complex situations that are difficult to quantify precisely. As shown in formula (4):
[0145] Where, μ i (F i ) represents the fuzzy membership degree of the i-th factor, w i For the corresponding weights.
[0146] Based on the risk index, the target area is divided into 5 levels of geological disaster risk: 1-extremely low risk area, 2-low risk area, 3-medium risk area, 4-high risk area and 5-extremely high risk area.
[0147] The aforementioned risk assessment method for power line patrol roads utilizes a landslide hazard index to delineate the hazard zones of road areas and provides calculation formulas for the exposure and vulnerability indicators of power line patrol roads. Furthermore, it proposes different risk index calculation methods for routine patrol roads, rapid risk screening of large-scale roads, and complex situations where precise quantification is difficult. Faced with complex and diverse landslide risks, this method improves the accuracy and reliability of predictions, providing high-quality risk warning information and safety assurance for power line patrol personnel.
[0148] In another exemplary embodiment, as shown in FIG8, a method for assessing the risk of power line inspection roads is provided, including:
[0149] Step S802: Based on the high-resolution, low-cloud-coverage optical remote sensing images of the power line inspection road area, acquire historical landslide disaster data and rainfall-triggered landslide disaster data for the target area, and construct a landslide database; the landslide database includes meteorological, geological, and remote sensing information of the road area.
[0150] Step S804: Select multiple influencing factors such as geology, topography, and meteorology, and then normalize each landslide influencing factor to establish a landslide influencing factor dataset.
[0151] Step S806: Select points within the road area as landslide sample points. If the landslide is triggered by landslide remains, select the same number of non-landslide sample points as the number of landslide sample points; if the landslide is triggered by rainfall, select non-landslide sample points based on the landslide area and the number of landslide sample points; the ratio of the number of landslide sample points to the number of non-landslide sample points is equal to the ratio of the landslide area to the non-landslide area.
[0152] Step S808: Based on the landslide impact factor dataset and selected landslide and non-landslide sample points, analyze the multicollinearity and contribution level of each landslide impact factor. Select landslide impact factors with a variance inflation factor (VIF) of no more than 10 to establish a landslide training test sample set.
[0153] Step S810: Based on the established landslide training sample set and test sample set, train a landslide hazard assessment model constructed based on the gradient boosting tree XGBoost model.
[0154] Step S812: Use the trained landslide hazard assessment model to predict the landslide hazard index in the power line patrol road area.
[0155] Step S814: Based on the geographical conditions and road design specifications near the power line inspection road, determine the road's exposure in the study area and its vulnerability index after a disaster.
[0156] Step S816: Calculate the risk index of the power line patrol road area according to the classic product formula of the risk index.
[0157] This embodiment is a feasible assessment process for the risk assessment of power line patrol roads in this application. Based on the design concept shown in Figure 9, firstly, the multi-source landslide data is analyzed and organized to obtain a set of landslide influencing factors. Then, the landslide influencing factors are selected by using the variance inflation factor (VIF). The landslide hazard assessment model based on the gradient boosting tree XGBoost model is trained to conduct landslide susceptibility analysis. Different assessment schemes are adopted for different landslide triggering causes. Finally, the risk index of the power line patrol road area is calculated using the classic risk index product formula to complete the risk assessment of the power line patrol road.
[0158] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.
[0159] Based on the same inventive concept, this application also provides a power line patrol road risk assessment device for implementing the power line patrol road risk assessment method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations of one or more power line patrol road risk assessment device embodiments provided below can be found in the limitations of the power line patrol road risk assessment method described above, and will not be repeated here.
[0160] In one embodiment, as shown in Figure 10, a power line inspection road risk assessment device is provided, comprising: a data acquisition module 1002, a data processing module 1004, a sample construction module 1006, a model training module 1008, and a risk assessment module 1010, wherein:
[0161] The data acquisition module 1002 is used to acquire landslide disaster data of the road area based on remote sensing images of the road area where the power line patrol road is located, and to establish a landslide database based on the landslide disaster data; the landslide database includes meteorological, geological and remote sensing information of the road area.
[0162] The data processing module 1004 is used to obtain multiple landslide influencing factors from the landslide database and to establish a landslide influencing factor dataset based on the multiple landslide influencing factors.
[0163] The sample construction module 1006 is used to construct a total landslide sample set based on landslide sample points and non-landslide sample points within the road area; it is also used to select a target landslide influencing factor from multiple landslide influencing factors based on the landslide influencing factor dataset and the total landslide sample set, and to establish a landslide sample set based on the target landslide influencing factor.
[0164] The model training module 1008 is used to train the gradient boosting tree model based on the training sample set and test sample set in the landslide sample set, so as to obtain the landslide hazard assessment model.
[0165] Risk assessment module 1010 is used to conduct risk assessments on power line patrol roads based on landslide hazard assessment models.
[0166] In an exemplary embodiment, the sample construction module 1006 is further configured to obtain landslide sample points within the road area that meet the first preset requirements; obtain non-landslide sample points based on the landslide triggering cause and the landslide sample points; and construct a total landslide sample set based on the landslide sample points and the non-landslide sample points.
[0167] In an exemplary embodiment, the sample construction module 1006 is further configured to select non-landslide sample points with the same number as landslide sample points when the landslide triggering cause is landslide remnants; and to select non-landslide sample points based on the landslide area and the number of landslide sample points when the landslide triggering cause is rainfall; the ratio of the number of landslide sample points to the number of non-landslide sample points is equal to the ratio of the landslide area to the non-landslide area.
[0168] In an exemplary embodiment, the sample construction module 1006 is further configured to determine the multicollinearity and importance of each landslide influencing factor based on the landslide influencing factor dataset, landslide sample points and non-landslide sample points; select landslide influencing factors that meet the first preset requirements as target landslide influencing factors; and establish a landslide sample set based on the target landslide influencing factors.
[0169] In an exemplary embodiment, the model training module 1008 is further configured to divide the landslide sample set into a training sample set and a test sample set according to a preset division method; the preset division method includes a preset order and a preset ratio; and to train the gradient boosting tree model according to at least one gradient boosting tree algorithm or a combination of at least two gradient boosting tree algorithms, the training sample set and the test sample set to obtain a landslide hazard assessment model.
[0170] In an exemplary embodiment, the risk assessment module 1010 is further configured to predict the landslide hazard index of the road area where the power line patrol road is located based on the landslide hazard assessment model, obtain the landslide hazard index of the road area, and classify the road area into hazard zones based on the landslide hazard index of the road area; obtain the exposure degree of the power line patrol road in the road area and the vulnerability index of the road area after a disaster; determine the risk index of the road area based on the exposure degree of the power line patrol road in the road area, the vulnerability index of the road area after a disaster, and the landslide hazard index of the road area, and complete the risk assessment of the power line patrol road.
[0171] Each module in the aforementioned power line inspection road risk assessment device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.
[0172] In an exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram is shown in Figure 11. The computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a power line inspection risk assessment method.
[0173] Those skilled in the art will understand that the structure shown in Figure 11 is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0174] In one exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method embodiment described above.
[0175] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method embodiments described above.
[0176] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method embodiments described above.
[0177] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.
[0178] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0179] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0180] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A method for risk assessment of power line inspection roads, wherein, The method includes: Based on remote sensing images of the road area where the power line patrol route is located, landslide disaster data for the road area is obtained, and a landslide database is established based on the landslide disaster data; the landslide database includes meteorological, geological, and remote sensing information of the road area; Based on the landslide database, multiple landslide influencing factors are obtained, and a landslide influencing factor dataset is established based on these multiple landslide influencing factors. Based on the landslide sample points and non-landslide sample points within the road area, a total landslide sample set is constructed, including: obtaining landslide sample points within the road area that meet a first preset requirement; obtaining non-landslide sample points based on the landslide triggering cause and the landslide sample points; and constructing a total landslide sample set based on the landslide sample points and the non-landslide sample points. Based on the landslide impact factor dataset and the total landslide sample set, a target landslide impact factor is selected from multiple landslide impact factors, and a landslide sample set is established based on the target landslide impact factor. Based on the training and testing sample sets in the landslide sample set, the gradient boosting tree model is trained to obtain the landslide hazard assessment model. A risk assessment was conducted on the power line patrol road based on the landslide hazard assessment model.
2. The method for risk assessment of power line inspection roads according to claim 1, wherein, The step of obtaining non-landslide sample points based on the landslide triggering cause and the landslide sample points includes: If the landslide is triggered by landslide remains, select the same number of non-landslide sample points as the landslide sample points; When the landslide is triggered by rainfall, non-landslide sample points are selected based on the landslide area and the number of landslide sample points; the ratio of the number of landslide sample points to the number of non-landslide sample points is equal to the ratio of the landslide area to the non-landslide area.
3. The method for risk assessment of power line inspection roads according to claim 1, wherein, The step of selecting a target landslide impact factor from multiple landslide impact factors based on the landslide impact factor dataset and the total landslide sample set, and establishing a landslide sample set based on the target landslide impact factor, includes: Based on the landslide impact factor dataset, the landslide sample points, and the non-landslide sample points, the multicollinearity and importance of each landslide impact factor are determined. Select the landslide impact factor that meets the first preset requirement as the target landslide impact factor; Based on the target landslide influencing factors, a landslide sample set is established.
4. The method for risk assessment of power line inspection roads according to claim 1, wherein, The gradient boosting tree model is trained based on the training and testing sample sets in the landslide sample set to obtain the landslide hazard assessment model, which includes: According to a preset partitioning method, the landslide sample set is divided into a training sample set and a test sample set; the preset partitioning method includes a preset order and a preset ratio; The gradient boosting tree model is trained using at least one gradient boosting tree algorithm or a combination of at least two gradient boosting tree algorithms, the training sample set, and the test sample set to obtain a landslide hazard assessment model.
5. The method for risk assessment of power line inspection roads according to claim 1, wherein, The risk assessment of the power line patrol road based on the landslide hazard assessment model includes: Based on the landslide hazard assessment model, the landslide hazard index of the road area where the power line patrol road is located is predicted to obtain the landslide hazard index of the road area, and the road area is classified into hazard zones based on the landslide hazard index of the road area. The exposure of the power line patrol road in the road area and the vulnerability index of the road area after a disaster are obtained. Based on the exposure of the power line patrol road in the road area, the vulnerability index of the road area after a disaster, and the landslide hazard index of the road area, the risk index of the road area is determined, and the risk assessment of the power line patrol road is completed.
6. A power line inspection road risk assessment device, wherein, The device includes: The data acquisition module is used to acquire landslide disaster data of the road area where the power line inspection road is located based on remote sensing images of the road area, and to establish a landslide database based on the landslide disaster data; the landslide database includes meteorological, geological and remote sensing information of the road area; The data processing module is used to obtain multiple landslide influencing factors based on the landslide database, and to establish a landslide influencing factor dataset based on the multiple landslide influencing factors. The sample construction module is used to construct a total landslide sample set based on landslide sample points and non-landslide sample points within the road area, including: obtaining landslide sample points within the road area that meet a first preset requirement; obtaining non-landslide sample points based on the landslide triggering cause and the landslide sample points; and constructing a total landslide sample set based on the landslide sample points and the non-landslide sample points. It is also used to select a target landslide impact factor from multiple landslide impact factors based on the landslide impact factor dataset and the total landslide sample set, and to establish a landslide sample set based on the target landslide impact factor; The model training module is used to train the gradient boosting tree model based on the training sample set and the test sample set in the landslide sample set, so as to obtain the landslide hazard assessment model. The risk assessment module is used to conduct a risk assessment of the power line patrol road based on the landslide hazard assessment model.
7. A power line inspection road risk assessment device according to claim 6, wherein, The risk assessment module is also used to predict the landslide hazard index of the road area where the power line patrol road is located based on the landslide hazard assessment model, obtain the landslide hazard index of the road area, and classify the road area into hazard zones based on the landslide hazard index of the road area. The exposure level of the power line patrol road in the road area and the vulnerability index of the road area after a disaster are obtained; based on the exposure level of the power line patrol road in the road area, the vulnerability index of the road area after a disaster, and the landslide hazard index of the road area, the risk index of the road area is determined, and the risk assessment of the power line patrol road is completed.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein... When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.
9. A computer-readable storage medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.
10. A computer program product comprising a computer program, wherein, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.