Electric power facility safety distance monitoring method, system, equipment and medium

Abstract

The invention relates to the technical field of electric power facility safety distance monitoring, and discloses an electric power facility safety distance monitoring method, system and device and a medium, and the method comprises the steps: obtaining the first historical parameter data of a target electric power facility, carrying out the first preprocessing, cleaning and standardizing the data, guaranteeing the accuracy and consistency, and providing a reliable basis for the subsequent processing; and second preprocessing is carried out on the data after the first preprocessing, through target enhancement and semantic segmentation, the data feature expression ability is improved, and hidden danger identification is more accurate. A hidden danger recognition model based on distance judgment is established and used for judging hidden dangers of power equipment, and the hidden danger recognition model can quickly respond to safe distance changes and give an early warning in time. The model is trained by using the data after the second preprocessing, so that the recognition precision and generalization ability are improved, and the model is suitable for a complex environment. And the first real-time parameter data are combined to monitor the safety distance of the electric power facility, so that the real-time monitoring and early warning of the state of the electric power equipment are realized, and the safety accident risk is reduced.

Description

Technical Field

[0001] This invention relates to the field of power facility safety distance monitoring technology, and in particular to a method, system, equipment and medium for power facility safety distance monitoring. Background Technology

[0002] In the field of power facility safety distance monitoring, traditional monitoring methods mainly rely on manual inspections. Workers carry simple measuring tools, such as rangefinders, to periodically measure the distance between power lines and surrounding objects that may pose safety hazards, such as buildings, trees, and construction machinery, and rely on experience to determine whether they are within a safe range. In addition, some more advanced technologies employ single sensors for monitoring. For example, some systems use lidar for distance detection, calculating the distance to target objects by emitting laser beams and receiving reflected light, and then issuing warnings based on preset safety distance thresholds. Other methods are based on image recognition technology, using cameras to capture images of the environment surrounding power facilities, using image processing algorithms to identify objects that may affect the safety distance, and estimating their relative positions to the power facilities.

[0003] Manual inspections are labor-intensive and inefficient. Due to the wide distribution and complex geographical environment of power facilities, it is difficult for personnel to comprehensively and promptly monitor all facilities. Furthermore, human judgment is easily influenced by subjective factors; differences in experience and judgment standards among different staff members may lead to missed or incorrect assessments of safety hazards.

[0004] Systems relying solely on lidar may experience interference with the laser beam in complex environments (such as severe weather or numerous obstructions), leading to decreased measurement accuracy or even malfunction. Image recognition-based methods suffer from low accuracy in low light conditions, when target features are indistinct, or when images are blurry, making it difficult to accurately determine safe distances. Furthermore, the information acquired by a single sensor is limited and cannot comprehensively reflect the true conditions of the environment surrounding power facilities, nor can it adequately consider the impact of multiple factors on safe distances. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a method, system, equipment and medium for monitoring the safe distance of power facilities, which can solve the shortcomings of manual inspection and single sensor technology, and realize comprehensive, accurate and real-time monitoring of the safe distance of power facilities.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides a method for monitoring the safe distance of power facilities, comprising:

[0009] Acquire the first historical parameter data of the target power facility, and perform a first preprocessing on the first historical parameter data;

[0010] The first preprocessing is used to classify the first historical parameter data;

[0011] Perform a second preprocessing on the data after the first preprocessing;

[0012] The second preprocessing is used to perform target augmentation and semantic segmentation on the first historical parameter data after the first preprocessing;

[0013] A hazard identification model is established, which is used to determine potential hazards in power equipment based on distance.

[0014] The hazard identification model is trained based on the second preprocessed data;

[0015] Based on the hazard identification model after training, and combined with the first real-time parameter data, the safe distance of power facilities is monitored.

[0016] As a preferred embodiment of the power facility safety distance monitoring method of the present invention, the second preprocessing includes:

[0017] The first historical parameter data after the first preprocessing is then Gaussian smoothed.

[0018] The data after Gaussian smoothing is then subjected to noise reduction processing.

[0019] The noise-reduced data is normalized according to the numerical range and format required by the hazard identification model.

[0020] As a preferred embodiment of the power facility safety distance monitoring method of the present invention, the second preprocessing further includes:

[0021] Feature extraction is performed on the normalized data, and upsampling is performed to restore the data to the same size as the first historical parameter data.

[0022] Perform classification prediction on each pixel of the upsampled data;

[0023] Obtain the distribution feature vectors of different semantic categories in the feature space, and determine the semantic category to which each pixel belongs by classifying and judging the feature vectors of each pixel.

[0024] This preferred approach improves the accuracy and robustness of the hazard identification model. Feature extraction extracts more critical and useful features from the data, while upsampling restores the original data size, making subsequent classification predictions more accurate. Classification prediction at each pixel fully considers the feature information of each pixel, thus more accurately determining its semantic category. Obtaining the distribution feature vectors of different semantic categories in the feature space further improves the accuracy and reliability of hazard identification, providing more accurate and reliable data support for the safe distance monitoring of power facilities.

[0025] As a preferred embodiment of the power facility safety distance monitoring method described in this invention, the establishment of the hazard identification model includes:

[0026] Establish a lidar coordinate system, and combine the known lidar installation position and attitude parameters to calculate the depth information of the target object by using the time difference between the lidar emitting the laser beam and receiving the reflected light.

[0027] Convert the measured depth data to the 84 coordinate system;

[0028] Based on the emission angle of the laser beam and the measured distance, the coordinate position of the laser beam in three-dimensional space is calculated using trigonometric functions to determine the position of each target object.

[0029] For different types of target objects, corresponding safe distance thresholds are set according to the safety regulations and standards of the power industry.

[0030] As a preferred embodiment of the power facility safety distance monitoring method of the present invention, the first historical parameter data of the target power facility includes lidar point cloud data, infrared image data, and ordinary camera image data.

[0031] In the first preprocessing, statistical filtering is used on the lidar point cloud data to remove noise points and outliers, and a spherical region with radius r is selected as the center of each point.

[0032] Calculate the distances from all points in the neighborhood to the center point, and then calculate the mean and standard deviation of these distances.

[0033] Set a distance threshold. If the distance from a point to other points in its neighborhood is greater than the distance threshold, then mark the point as a noise point and remove it.

[0034] As a preferred embodiment of the power facility safety distance monitoring method of the present invention, the first preprocessing further includes:

[0035] For infrared image data, firstly, the grayscale histogram of the infrared image is calculated to obtain the minimum and maximum grayscale values ​​in the image;

[0036] For each pixel in the image, obtain the original grayscale value of each pixel;

[0037] Gray-scale stretching is performed on the original gray-scale value of each pixel.

[0038] As a preferred embodiment of the power facility safety distance monitoring method of the present invention, the step of monitoring the power facility safety distance based on the trained hazard identification model and the first real-time parameter data includes:

[0039] When the actual distance between the target object and the power facility is less than the preset safety threshold, an early warning command is automatically generated. The early warning command is sent to the monitoring center and the on-site alarm device of the power facility via a wireless network for broadcast warning.

[0040] Secondly, the present invention provides a power facility safety distance monitoring system, comprising:

[0041] The data acquisition and processing module is used to acquire the first historical parameter data of the target power facility and perform a first preprocessing on the first historical parameter data;

[0042] The first preprocessing is used to classify the first historical parameter data;

[0043] The preprocessing module is used to perform a second preprocessing on the data after the first preprocessing.

[0044] The second preprocessing is used to perform target augmentation and semantic segmentation on the first historical parameter data after the first preprocessing;

[0045] The model building module is used to build a hazard identification model, which is used to judge the hazards of power equipment based on distance.

[0046] The training module is used to train the hazard identification model based on the second preprocessed data;

[0047] The monitoring module is used to monitor the safe distance of power facilities based on the hazard identification model after training and the first real-time parameter data.

[0048] Thirdly, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0049] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0050] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes a method for monitoring the safe distance of power facilities. The method involves acquiring first historical parameter data of the target power facility and performing a first preprocessing on the first historical parameter data; performing a second preprocessing on the data after the first preprocessing; establishing a hazard identification model, which is used to judge potential hazards in power equipment based on distance; training the hazard identification model based on the data after the second preprocessing; and monitoring the safe distance of power facilities based on the trained hazard identification model in conjunction with the first real-time parameter data. The first preprocessing cleans and standardizes the data, ensuring its accuracy and consistency, and providing a reliable foundation for subsequent processing. The second preprocessing, through target enhancement and semantic segmentation, further improves the feature representation capability of the data, enabling the hazard identification model to more accurately identify potential safety hazards. The hazard identification model is based on distance judgment, a characteristic that allows the system to quickly respond to changes in the safe distance between power equipment and issue timely warnings. Training the hazard identification model with the data after the second preprocessing not only improves the model's recognition accuracy but also enhances its generalization ability, making it applicable to various complex power facility environments. By combining real-time parameter data with power facility safety distance monitoring, real-time monitoring and early warning of power equipment status are achieved, effectively reducing the risk of safety accidents. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart of a method for monitoring the safe distance of power facilities, provided as an embodiment of the present invention.

[0053] Figure 2 This is an internal structural diagram of an electronic device for a method of monitoring the safe distance of power facilities, provided in one embodiment of the present invention. Detailed Implementation

[0054] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0055] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for monitoring the safe distance of power facilities, including:

[0056] Existing technologies have several drawbacks. For example, manual inspections are labor-intensive, inefficient, and susceptible to subjective factors. Systems relying solely on lidar may not function properly in complex environments. Image recognition methods have low accuracy in low light or when target features are indistinct. These issues can lead to missed or misjudged safety hazards, posing risks to the safe operation of power facilities.

[0057] This invention provides a method that can effectively solve the problems mentioned above. The following will describe in detail how to implement the method for monitoring the safe distance of power facilities in conjunction with several embodiments.

[0058] Figure 1 A flowchart of a method for monitoring the safe distance of power facilities is shown, including:

[0059] S101, Obtain the first historical parameter data of the target power facility, and perform the first preprocessing on the first historical parameter data;

[0060] It should be noted that in the power distribution network environment, many complex factors affect the accuracy and reliability of safe distance monitoring. For example, in terms of the natural environment, severe weather (such as heavy rain, heavy snow, dense fog, etc.) can interfere with the propagation of the laser beam of lidar, reducing its measurement accuracy, and can also reduce the clarity of images acquired by cameras, affecting image recognition. In terms of the geographical environment, dense trees and undulating terrain in mountainous areas and forests can easily cause signal obstruction and reflection interference, increasing the difficulty of safe distance judgment.

[0061] Therefore, a method is needed that can process different data and determine safe distances.

[0062] In some specific implementations, the first historical parameter data of the target power facility is obtained as preliminary data for subsequent safety distance determination. The first historical parameter data is obtained through the following steps:

[0063] Different sensor devices are selected based on the application. For lidar: a pulsed lidar with high resolution and a long ranging range is chosen. Its wavelength is specially optimized to maintain good penetration under complex weather conditions (such as light rain and fog), reducing the impact of environmental interference on measurement results. For infrared cameras: a high-sensitivity infrared thermal imaging camera with automatic gain control and digital noise reduction is used. It can clearly image objects by capturing their thermal radiation signals in low light, at night, and in the presence of partially obstructed objects.

[0064] The layout of power facilities should be based on their distribution characteristics and areas with potential hazards, such as installing them near sections of lines where trees are likely to grow too close or buildings are near them, to ensure effective monitoring of potentially dangerous areas. Standard cameras: Equipped with high-definition, wide dynamic range optical cameras, supporting autofocus and optical zoom. Capable of acquiring clear, detailed color images under normal lighting conditions, providing rich visual information for target recognition and environmental detail analysis.

[0065] A unified clock synchronization system is established using NTP (Network Time Protocol) to ensure that LiDAR, infrared cameras, and ordinary cameras can simultaneously initiate data acquisition. When the system receives a monitoring command or initiates the acquisition program according to a preset time interval, a synchronization signal is simultaneously sent to each sensor, causing them to start capturing data at the same time. This ensures the temporal consistency of data from different modalities, facilitating subsequent data fusion and analysis.

[0066] Data collected by the sensors is transmitted to the data processing center in real time via a 5G network. During transmission, CRC (Cyclic Redundancy Check) is used for preliminary verification and error correction to ensure data integrity.

[0067] In this embodiment of the invention, the first preprocessing is used to classify the first historical parameter data.

[0068] It should be noted that because there are many types of first-historical parameter data, they cannot be directly used to train the hazard identification model, so classification processing is required. The classification process involves categorizing different types of data according to their characteristics. For example, LiDAR point cloud data, infrared image data, and ordinary camera image data are classified separately to facilitate targeted preprocessing and feature extraction in the subsequent stages.

[0069] In this embodiment of the invention, the first historical parameter data of the target power facility includes lidar point cloud data, infrared image data, and ordinary camera image data.

[0070] In the first preprocessing step, statistical filtering is used on the lidar point cloud data to remove noise points and outliers, and a spherical neighborhood with radius r is selected as the center of each point.

[0071] Calculate the distances from all points in the neighborhood to the center point, and then calculate the mean and standard deviation of these distances.

[0072] Set a distance threshold. If the distance from a point to other points in its neighborhood is greater than the distance threshold, then mark the point as a noise point and remove it.

[0073] In this embodiment of the invention, the first preprocessing further includes:

[0074] For infrared image data, firstly, the grayscale histogram of the infrared image is calculated to obtain the minimum and maximum grayscale values ​​in the image;

[0075] For each pixel in the image, obtain the original grayscale value of each pixel;

[0076] Perform grayscale stretching on the original grayscale value of each pixel.

[0077] Specifically, statistical filtering is used for lidar point cloud data to remove noise points and outliers, and a spherical region with radius r is selected as the center of each point.

[0078] Furthermore, calculate the distances from all points in the neighborhood to the center point, and calculate the mean μ and standard deviation σ of these distances;

[0079] Furthermore, a distance threshold T is set, usually T = μ + kσ (k is an empirical coefficient, generally between 1 and 3); if the distance from a point to other points in its neighborhood is greater than T, then the point is marked as a noise point and removed.

[0080] For infrared images, first, the gray-level histogram of the infrared image is calculated to obtain the minimum gray-level value min_gray and the maximum gray-level value max_gray in the image;

[0081] Furthermore, for each pixel P(x,y) in the image, its original grayscale value is gray(x,y), and the grayscale value new_gray(x,y) after grayscale stretching is calculated according to the following formula:

[0082]

[0083] Furthermore, through grayscale stretching, pixels that were originally in darker or brighter areas and had similar grayscale values ​​will have greater differences in grayscale after stretching, thus improving the clarity and recognizability of the image.

[0084] For ordinary camera images, the image resolution is determined, and the nearest neighbor interpolation method is used. For each pixel P(x,y) in the image, the nearest neighbor pixel corresponding to it is found in the original image, and the gray value of that pixel is assigned to P(x,y), thereby improving the image resolution.

[0085] It should be noted that acquiring the first historical parameter data of the target power facility and performing initial preprocessing on this data can improve its accuracy and reliability. Preprocessing removes noise and outliers, making subsequent processing and analysis more accurate. Furthermore, preprocessing formats and standardizes the data, providing a unified data format and standard for subsequent steps, thus facilitating data processing and analysis. In this way, the efficiency and accuracy of the entire power facility safety distance monitoring system can be significantly improved.

[0086] S102, Perform a second preprocessing on the data after the first preprocessing;

[0087] It should be noted that after the first preprocessing, the preprocessed data needs to be further optimized to improve the usability of the data and the accuracy of hazard identification.

[0088] In this embodiment of the invention, the second preprocessing is used to perform target enhancement and semantic segmentation on the first historical parameter data after the first preprocessing;

[0089] Specifically, the second preprocessing includes two steps: target enhancement and semantic segmentation. Target enhancement aims to highlight key targets in the image or point cloud data, such as power facilities, trees, and buildings. By enhancing the features of these targets, the hazard identification model can more easily capture them. Semantic segmentation, on the other hand, performs pixel-level classification on the image or point cloud data, distinguishing different objects or regions and providing more detailed information for subsequent hazard assessment. Through this second preprocessing, the feature representation capability of the data is significantly improved, providing a solid foundation for the training and monitoring of the subsequent hazard identification model.

[0090] In this embodiment of the invention, the second preprocessing includes:

[0091] The first historical parameter data after the first preprocessing is then Gaussian smoothed.

[0092] The data after Gaussian smoothing is then subjected to noise reduction processing.

[0093] The noise-reduced data is normalized according to the numerical range and format required by the hazard identification model.

[0094] In this embodiment of the invention, the second preprocessing further includes:

[0095] Feature extraction is performed on the normalized data, and upsampling is performed to restore the data to the same size as the first historical parameter data.

[0096] Perform classification prediction on each pixel of the upsampled data;

[0097] Obtain the distribution feature vectors of different semantic categories in the feature space, and determine the semantic category to which each pixel belongs by classifying and judging the feature vectors of each pixel.

[0098] Specifically, Gaussian smoothing of the original image typically involves convolution operations using a Gaussian kernel function. The Gaussian kernel function has the following form:

[0099]

[0100] In the formula: x and y represent the coordinate positions of image pixels in the horizontal and vertical directions; σ is the standard deviation; e is the natural constant.

[0101] Furthermore, downsampling is performed on the smoothed image. This typically involves taking every single pixel in the horizontal and vertical directions to obtain a new image with half the resolution of the original. Through the Gaussian smoothing and downsampling process described above, a series of images at different resolutions can be obtained, forming a Gaussian pyramid structure, thereby achieving target enhancement.

[0102] Furthermore, the acquired images, after target augmentation, undergo preprocessing operations such as normalization. This ensures that the numerical range and format of the image data meet the input requirements of the FCN (Fully Convolutional Neural Network) model, thus completing data preparation and preprocessing.

[0103] Furthermore, the preprocessed image is input into the FCN model, where it first undergoes a series of convolutional and pooling layers for feature extraction.

[0104] Furthermore, after feature extraction, an upsampling layer is used to gradually restore the feature map to the size of the original image. Deconvolution is used for upsampling to enhance the resolution of the feature map, while classification prediction is performed at each pixel. During training, the FCN model learns the distribution of different semantic categories in the feature space. By classifying the feature vector of each pixel, it determines its semantic category, such as accurately distinguishing between power facilities, buildings, vegetation, and the sky.

[0105] Furthermore, in training the model, a dataset is constructed using pre-annotated power scene images, divided into training, testing, and validation sets in a 7:1.5:1.5 ratio. This allows for the training of the FCN model. The annotation information clearly defines the semantic category of each pixel in the image. The model continuously adjusts its weight parameters by minimizing the loss function between the predicted result and the annotation information. During training, optimization algorithms such as stochastic gradient descent can be used to gradually improve the model's performance, enabling it to more accurately segment different types of power-related objects and their surrounding environment. Even when image features change after target enhancement, high segmentation accuracy can still be maintained.

[0106] It should be noted that performing a second preprocessing step on the data after the first preprocessing significantly improves the accuracy and efficiency of hazard identification. Through target augmentation, the features of key targets are highlighted, making it easier for the hazard identification model to capture these targets, thereby improving identification accuracy. Simultaneously, semantic segmentation classifies image or point cloud data at the pixel level, providing more detailed information for subsequent hazard assessment and further enhancing the system's monitoring capabilities. Furthermore, the second preprocessing includes data normalization, feature extraction, and upsampling. These steps work together to make the data more suitable for the processing requirements of the hazard identification model, thus improving the overall system's processing efficiency and accuracy. Therefore, the second preprocessing is an indispensable part of the power facility safety distance monitoring system of this invention and plays a crucial role in improving the overall performance of the system.

[0107] S103, Establish a hazard identification model. The hazard identification model is used to judge the hazards of power equipment based on distance.

[0108] It should be noted that after the first and second preprocessing operations, a hazard identification model needs to be established based on the processed data. The core function of this model is to determine whether there are any safety hazards based on the distance information between power facilities and surrounding objects.

[0109] To achieve this goal, the hazard identification model needs to be able to accurately understand the characteristics of the data, including the location, shape, and size of power facilities, as well as the type and distance of surrounding objects. This information will serve as an important basis for the model to judge hazards.

[0110] In some specific implementations, the hazard identification model can be built using deep learning algorithms, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs). These algorithms are able to learn feature representations from large amounts of historical parameter data and automatically extract key information related to hazards.

[0111] To train the hazard identification model, a large dataset containing various power facility scenarios and hazard conditions can be constructed. This dataset should cover different weather conditions, geographical environments, and types of power facilities to ensure the model's generalization ability.

[0112] During training, a supervised learning method is used, with preprocessed historical parameter data as input and manually labeled hazard tags as output. The model parameters are optimized by minimizing the prediction error.

[0113] In addition, data augmentation techniques, such as rotation, scaling, and flipping, can be used to increase the diversity of the dataset and further improve the robustness of the model. After training, the hazard identification model can automatically determine whether there are safety hazards in power facilities based on the input distance information and output corresponding early warning information.

[0114] In this embodiment of the invention, establishing a hazard identification model includes:

[0115] Establish a lidar coordinate system, and combine the known lidar installation position and attitude parameters to calculate the depth information of the target object by using the time difference between the lidar emitting the laser beam and receiving the reflected light.

[0116] Convert the measured depth data to the 84 coordinate system;

[0117] Based on the emission angle of the laser beam and the measured distance, the coordinate position of the laser beam in three-dimensional space is calculated using trigonometric functions to determine the position of each target object.

[0118] For different types of target objects, corresponding safe distance thresholds are set according to the safety regulations and standards of the power industry.

[0119] Specifically, the first step is to establish a lidar coordinate system based on the known installation location and attitude parameters of the lidar. This step ensures that the lidar can accurately emit laser beams and receive reflected light, thereby calculating the depth information of the target object based on the time difference.

[0120] Secondly, the depth data obtained through lidar measurement is converted to the specified WGS84 coordinate system (WGS84, a global coordinate system widely used in geographic information systems). This step helps to unify the coordinate systems of data from different sources, facilitating subsequent processing and analysis.

[0121] Next, based on the laser beam's emission angle and the measured distance, the precise coordinates of the laser point in three-dimensional space are calculated using trigonometric functions. This allows for the determination of the exact location of each detected target object.

[0122] Finally, for different types of target objects, such as buildings, trees, or construction machinery, corresponding safe distance thresholds are set according to the safety regulations and standards of the power industry. These thresholds are important criteria for determining whether there are potential safety hazards. When the detected actual distance is less than the preset safe threshold, the system will automatically generate an early warning command to prevent potential safety accidents.

[0123] It should be noted that the hazard identification model, used to assess potential hazards in power equipment based on distance, can accurately and in real time identify objects too close to power equipment, thus providing early warnings of potential safety risks. Through this model, the system can automatically analyze distance data measured by lidar and compare it with preset safe distance thresholds. Once a distance less than the threshold is detected, an alarm mechanism is immediately triggered, notifying relevant personnel for timely handling. This not only significantly improves the efficiency of hazard identification but also effectively reduces the risk of safety accidents caused by human error or delayed handling. Furthermore, the establishment of the hazard identification model provides data support and decision-making basis for the safety management of power facilities, contributing to an improvement in overall safety management levels.

[0124] S104, Train the hazard identification model based on the second preprocessed data;

[0125] S105, based on the hazard identification model after training, combined with the first real-time parameter data, to monitor the safe distance of power facilities.

[0126] In this embodiment of the invention, monitoring the safe distance of power facilities based on the trained hazard identification model and combined with the first real-time parameter data includes:

[0127] When the actual distance between the target object and the power facility is less than the preset safety threshold, an early warning command is automatically generated. The early warning command is sent to the monitoring center and the on-site alarm device of the power facility via a wireless network for broadcast warning.

[0128] Specifically, the system first acquires initial real-time parameter data of the target power facility. This data may include, but is not limited to, lidar point cloud data, infrared image data, and ordinary camera image data.

[0129] The trained hazard identification model is used to analyze and process this initial real-time parameter data. The model determines the existence of safety hazards based on the actual distance between the target object and the power facilities. In this process, the model uses its learned knowledge and patterns to identify potential dangerous situations.

[0130] The system will automatically generate an early warning command once it detects that the actual distance between a target object (such as a building, tree, or construction machinery) and power facilities is less than a preset safety threshold. The safety threshold mentioned here is set according to different types of target objects and in accordance with safety regulations and standards of the power industry.

[0131] The generated warning commands are then sent via wireless network. This typically involves modern communication technologies, such as 5G networks, to ensure that information can be transmitted quickly and reliably to alarm devices at monitoring centers and power facility sites.

[0132] Finally, the monitoring center and on-site alarm devices that receive the early warning information will react immediately, such as by issuing warnings to relevant personnel through sound, light or other means, to remind them to pay attention and take necessary measures to prevent potential safety accidents.

[0133] In summary, this invention proposes a method for monitoring the safe distance of power facilities. The method involves acquiring first historical parameter data of the target power facility and performing a first preprocessing step on this data; performing a second preprocessing step on the preprocessed data; establishing a hazard identification model, which is used to determine potential hazards in power equipment based on distance; training the hazard identification model using the second preprocessed data; and monitoring the safe distance of power facilities using the trained hazard identification model in conjunction with the first real-time parameter data. The first preprocessing cleans and standardizes the data, ensuring its accuracy and consistency, and providing a reliable foundation for subsequent processing. The second preprocessing, through target enhancement and semantic segmentation, further improves the feature representation capability of the data, enabling the hazard identification model to more accurately identify potential safety hazards. The hazard identification model is based on distance judgment, a characteristic that allows the system to quickly respond to changes in the safe distance between power equipment and issue timely warnings. Training the hazard identification model with the second preprocessed data not only improves the model's recognition accuracy but also enhances its generalization ability, making it applicable to various complex power facility environments. By combining the first real-time parameter data with power facility safe distance monitoring, real-time monitoring and early warning of the power equipment status are achieved, effectively reducing the risk of safety accidents.

[0134] Example 2: In a preferred embodiment, when the target object is construction machinery and the actual distance to the power facility is less than 5 meters, an early warning command is automatically generated. The early warning command sends the warning information to the monitoring center and the on-site alarm device of the power facility via a wireless network, and broadcasts the following warning: "Alarm, a large machine is approaching a high-voltage line at the construction site. Please stop work immediately and evacuate!"

[0135] When the target object is a tree and the actual distance between the treetop and the power line is less than 3 meters, an early warning command is automatically generated. The early warning command is sent to the monitoring center and the alarm device at the power facility via a wireless network, and the following warning is broadcast: "Attention, nearby trees are growing too fast and are about to approach the high-voltage line. Please take immediate action!"

[0136] When the target object is a building and the actual distance between the top of the building and the power line is less than 6 meters, an early warning command is automatically generated. The early warning command is sent to the monitoring center and the alarm device at the power facility via a wireless network, and the following warning is broadcast: "Emergency situation, the height of the adjacent building exceeds the safe range, please check for any illegal construction."

[0137] When the target object is a pedestrian or non-motorized vehicle and the actual distance between it and the warning area around the power facility is less than 2 meters, an early warning command will be automatically generated. The early warning command will be sent to the monitoring center and the on-site alarm device of the power facility via wireless network, and will broadcast the following warning: "Warning, someone has entered the dangerous area of ​​the power facility. Please leave immediately to ensure personal safety."

[0138] When the target object is a vehicle (not construction machinery) and the actual distance between the vehicle and the power facility is less than 4 meters, an early warning command will be automatically generated. The early warning command will be sent to the monitoring center and the alarm device at the power facility via a wireless network, and the following warning will be broadcast: "Reminder: The vehicle is too close to the power facility. Please drive carefully and stay away from the area."

[0139] Example 3, referring to Figure 2 This embodiment also provides a power facility safety distance monitoring system, including:

[0140] The data acquisition and processing module is used to acquire the first historical parameter data of the target power facility and perform the first preprocessing on the first historical parameter data;

[0141] The first preprocessing step is used to classify the first historical parameter data.

[0142] The preprocessing module is used to perform a second preprocessing on the data after the first preprocessing.

[0143] The second preprocessing is used to perform target augmentation and semantic segmentation on the first historical parameter data after the first preprocessing.

[0144] The model building module is used to build a hazard identification model, which is used to judge the potential hazards of power equipment based on distance.

[0145] The training module is used to train the hazard identification model based on the second preprocessed data;

[0146] The monitoring module is used to monitor the safe distance of power facilities based on the hazard identification model after training and the first real-time parameter data.

[0147] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0148] This embodiment also provides an electronic device, which can be a terminal, and its internal structure diagram can be as follows. Figure 2 As shown, the electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the 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 communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for monitoring the safe distance of power facilities. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the device's casing, or an external keyboard, touchpad, or mouse.

[0149] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps:

[0150] Acquire the first historical parameter data of the target power facility, and perform a first preprocessing on the first historical parameter data;

[0151] The first preprocessing step is used to classify the first historical parameter data.

[0152] Perform a second preprocessing on the data after the first preprocessing;

[0153] The second preprocessing is used to perform target augmentation and semantic segmentation on the first historical parameter data after the first preprocessing.

[0154] Establish a hazard identification model, which is used to determine potential hazards in power equipment based on distance.

[0155] The hazard identification model is trained based on the second preprocessed data;

[0156] Based on the hazard identification model after training, and combined with the first real-time parameter data, the safe distance of power facilities is monitored.

[0157] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0158] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0159] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring the safe distance of power facilities, characterized in that, include: Acquire the first historical parameter data of the target power facility, and perform a first preprocessing on the first historical parameter data; The first preprocessing is used to classify the first historical parameter data; Perform a second preprocessing on the data after the first preprocessing; The second preprocessing is used to perform target augmentation and semantic segmentation on the first historical parameter data after the first preprocessing; A hazard identification model is established, which is used to determine potential hazards in power equipment based on distance. The hazard identification model is trained based on the second preprocessed data; Based on the hazard identification model after training, and combined with the first real-time parameter data, the safe distance of power facilities is monitored.

2. The method for monitoring the safe distance of power facilities as described in claim 1, characterized in that, The second preprocessing includes: The first historical parameter data after the first preprocessing is then Gaussian smoothed. The data after Gaussian smoothing is then subjected to noise reduction processing. The noise-reduced data is normalized according to the numerical range and format required by the hazard identification model.

3. The method for monitoring the safe distance of power facilities as described in claim 2, characterized in that, The second preprocessing also includes: Feature extraction is performed on the normalized data, and upsampling is performed to restore the data to the same size as the first historical parameter data. Perform classification prediction on each pixel of the upsampled data; Obtain the distribution feature vectors of different semantic categories in the feature space, and determine the semantic category to which each pixel belongs by classifying and judging the feature vectors of each pixel.

4. The method for monitoring the safe distance of power facilities as described in claim 3, characterized in that, The establishment of the hazard identification model includes: Establish a lidar coordinate system, and combine the known lidar installation position and attitude parameters to calculate the depth information of the target object by using the time difference between the lidar emitting the laser beam and receiving the reflected light. Convert the measured depth data to the 84 coordinate system; Based on the emission angle of the laser beam and the measured distance, the coordinate position of the laser beam in three-dimensional space is calculated using trigonometric functions to determine the position of each target object. For different types of target objects, corresponding safe distance thresholds are set according to the safety regulations and standards of the power industry.

5. The method for monitoring the safe distance of power facilities as described in claim 4, characterized in that, The first historical parameter data of the target power facility includes lidar point cloud data, infrared image data, and ordinary camera image data. In the first preprocessing, statistical filtering is used on the lidar point cloud data to remove noise points and outliers, and a spherical region with radius r is selected as the center of each point. Calculate the distances from all points in the neighborhood to the center point, and then calculate the mean and standard deviation of these distances. Set a distance threshold. If the distance from a point to other points in its neighborhood is greater than the distance threshold, then mark the point as a noise point and remove it.

6. The method for monitoring the safe distance of power facilities as described in claim 5, characterized in that, The first preprocessing also includes: For infrared image data, firstly, the grayscale histogram of the infrared image is calculated to obtain the minimum and maximum grayscale values ​​in the image; For each pixel in the image, obtain the original grayscale value of each pixel; Gray-scale stretching is performed on the original gray-scale value of each pixel.

7. The method for monitoring the safe distance of power facilities as described in claim 6, characterized in that, The step of monitoring the safe distance of power facilities based on the trained hazard identification model and the first real-time parameter data includes: When the actual distance between the target object and the power facility is less than the preset safety threshold, an early warning command is automatically generated. The early warning command is sent to the monitoring center and the on-site alarm device of the power facility via a wireless network for broadcast warning.

8. A power facility safety distance monitoring system, using the method described in any one of claims 1 to 7, characterized in that, include: The data acquisition and processing module is used to acquire the first historical parameter data of the target power facility and perform a first preprocessing on the first historical parameter data; The first preprocessing is used to classify the first historical parameter data; The preprocessing module is used to perform a second preprocessing on the data after the first preprocessing. The second preprocessing is used to perform target augmentation and semantic segmentation on the first historical parameter data after the first preprocessing; The model building module is used to build a hazard identification model, which is used to judge the hazards of power equipment based on distance. The training module is used to train the hazard identification model based on the second preprocessed data; The monitoring module is used to monitor the safe distance of power facilities based on the hazard identification model after training and the first real-time parameter data.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the power facility safety distance monitoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the power facility safety distance monitoring method according to any one of claims 1 to 7.

Images