Power line warning method, device, computer equipment and storage medium
By identifying the power line parameters and dividing the monitoring range and calling the power line warning model, analyzing personnel's behavior and generating warning signals, the problem of insufficient power line safety monitoring in the existing technology is solved, and real-time monitoring and safety of the behavior of personnel around the power line is achieved. Early warning improves the safety of power lines.
Patent Information
- Application Number
- CN202311456691.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-03
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2043-11-03
AI Technical Summary
The existing power line warning signs fail to fully eliminate safety hazards through sound and light warnings, especially in areas with scarce buildings that cannot effectively monitor and prevent people from approaching high-voltage power lines, resulting in high safety risks.
By identifying power line parameters, dividing monitoring ranges, collecting image data in real time, calling pre-trained power line warning models to analyze personnel's behavior, generating warning signals, including identification and alarm of pedestrian staying, gathering and wandering behaviors, improving the sensitivity of warnings and accuracy.
Real-time monitoring and safety warning of the behavior of personnel around the power lines is achieved, the risk of electric shock accidents is reduced, and the safe operation and personnel protection effect of the power lines are improved.
Smart Images

Figure CN117456459B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of power line safety monitoring, and in particular to a power line warning method, device, computer equipment, storage medium and computer program product. Background Art
[0002] With the development of high-voltage power technology, the layout of cross-regional transmission and distribution overhead lines is becoming more and more dense. Transmission and distribution overhead lines are widely distributed in urban and rural areas, often crossing construction sites, rivers and other places, and there are very large safety risks. According to national standards, power lines need to maintain a certain distance from other buildings according to certain distance requirements, so as to avoid safety risks caused by construction sites and other reasons. However, in the wild and in areas with few buildings, it is impossible to conduct comprehensive monitoring of transmission and distribution overhead lines, and only warning signs can be used to remind people who are close.
[0003] Traditional power warning signs generally use infrared detection to identify people approaching, and then use sound and light warnings to warn people to stay away from dangerous areas. However, the warning effectiveness of existing power warning signs is relatively low, and sound and light warnings alone cannot fully eliminate safety hazards. Some people still ignore the sound and light warnings and approach dangerous areas. Summary of the invention
[0004] Based on this, it is necessary to provide a power line warning method, device, computer equipment, computer-readable storage medium and computer program product that can identify and record people close to power transmission and distribution lines and issue warnings based on the behavioral characteristics of the people in response to the above technical problems.
[0005] In a first aspect, the present application provides a power line warning method. The method comprises:
[0006] Divide the monitoring range according to the power line parameters;
[0007] Real-time collection of image data within the monitoring range;
[0008] Identify the image data, and when a person is identified within the monitoring range, call the pre-trained power line warning model to analyze the person and obtain the person type;
[0009] Generates a warning signal if a person type presents a safety risk.
[0010] In one embodiment, defining the monitoring range according to the power line parameters includes:
[0011] Determine the warning position in the power line where a safety warning is required, and determine the power line parameters of the warning position;
[0012] The monitoring range is defined according to the power line parameters; the monitoring range includes a first monitoring range and a second monitoring range, the first monitoring range is a dangerous area, and the second monitoring range is larger than the first monitoring range.
[0013] In one embodiment, when a person type presents a safety risk, generating a warning signal includes:
[0014] Perform target detection on the image data, where the target is the person appearing in the monitoring range;
[0015] When a person is identified within the monitoring range, the power line warning model is called to analyze the person and obtain the person type;
[0016] When the person type is a pedestrian and is within the first monitoring range, a warning signal is generated;
[0017] When the person type is a fisherman and is located in the second monitoring range, a warning signal is generated.
[0018] In one embodiment, when the person type is a pedestrian and is within the first monitoring range, generating a warning signal further includes:
[0019] Detect pedestrian behavior patterns;
[0020] generating a pedestrian staying warning when it is detected that the pedestrian has been in the first monitoring range for a time exceeding a preset time threshold;
[0021] When it is detected that the number of pedestrians in the first monitoring range exceeds a preset number threshold, a pedestrian gathering warning is generated;
[0022] When it is detected that a pedestrian is in the second monitoring range for a long time and enters and exits the first monitoring range multiple times, a pedestrian wandering warning is generated; the second monitoring range includes the first monitoring range.
[0023] In a second aspect, the present application further provides a power line warning model training method, which is used to train any power line warning model of the first aspect, comprising:
[0024] Acquire image data of a person, mark the features of the person, and generate a marked image; the image data includes images of the person in different scenes;
[0025] Preprocess the labeled image to generate a feature arrangement array;
[0026] The feature array is input into the neural network model as training data, and learnable weights are introduced to generate the weights of each feature;
[0027] The weights are adjusted iteratively according to the training data, and the neural network model is trained until convergence or the preset number of training times is reached to obtain the power line warning model.
[0028] In a third aspect, the present application also provides a power line warning device. The device comprises:
[0029] Positioning module, used to define the monitoring range according to power line parameters;
[0030] An acquisition module is used to acquire image data within the monitoring range in real time;
[0031] The analysis module is used to identify the image data. When a person is identified within the monitoring range, the pre-trained power line warning model is called to analyze the person and obtain the person type.
[0032] In a fourth aspect, the present application further provides a power line warning model training device. Used to train any power line warning model of the first aspect, the device comprises:
[0033] A data acquisition module is used to acquire image data of a person, mark the features of the person, and generate a marked image; the image data includes images of the person in different scenes;
[0034] A preprocessing module is used to preprocess the labeled image and generate a feature arrangement array;
[0035] The initialization module is used to input the feature array as training data into the neural network model and introduce learnable weights to generate the weights of each feature;
[0036] The training module is used to iteratively adjust the weights according to the training data, train the neural network until convergence or reaching a preset number of training times, and obtain a power line warning model.
[0037] In a fifth aspect, the present application also provides a power line warning device. The device includes:
[0038] The shell comprises a first side surface, a second side surface, a first panel and a second panel; the first side surface of the shell is connected to the first bracket, the second side surface is connected to the second bracket, the first panel is connected to the solar panel, and the second panel is connected to the infrared detection module, the sound and light alarm module and the video monitoring module;
[0039] A warning bar, fixed between the first bracket, the second bracket and the second panel of the housing;
[0040] As in the power line warning device of the third aspect, the power line warning device is disposed in a housing.
[0041] In a sixth aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:
[0042] Divide the monitoring range according to the power line parameters;
[0043] Real-time collection of image data within the monitoring range;
[0044] Identify the image data, and when a person is identified within the monitoring range, call the pre-trained power line warning model to analyze the person and obtain the person type;
[0045] Generates a warning signal if a person type presents a safety risk.
[0046] In a seventh aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:
[0047] Divide the monitoring range according to the power line parameters;
[0048] Real-time collection of image data within the monitoring range;
[0049] Identify the image data, and when a person is identified within the monitoring range, call the pre-trained power line warning model to analyze the person and obtain the person type;
[0050] Generates a warning signal if a person type presents a safety risk.
[0051] The above-mentioned power line warning method, device, computer equipment, storage medium and computer program product define the monitoring range by analyzing the power line parameters. When the power line parameters are different, an adjustable monitoring range is adopted instead of a fixed monitoring range, so that the monitoring range is more comprehensive and reasonable, and the sensitivity of the warning is improved. By acquiring the image data within the monitoring range and uploading it to the power line warning model for processing and analysis, real-time monitoring of the monitoring area is achieved; combined with the calculation and processing of the power line warning model, it is possible to identify whether there is a detection target in the image data, so that potential safety risks can be quickly discovered, the safe operation of the power line can be fully guaranteed, and potential personnel risks can be responded to in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 A diagram showing an application environment of a power line warning method in an embodiment;
[0053] Figure 2A schematic diagram of a flow chart of a power line warning method in one embodiment;
[0054] Figure 3 A schematic diagram of a flow chart of a power line warning model training method in one embodiment;
[0055] Figure 4 A schematic diagram of a flow chart of a power line warning method in another embodiment;
[0056] Figure 5 is a structural block diagram of a power line warning device in one embodiment;
[0057] Figure 6 It is a structural block diagram of a power line warning model training device in one embodiment;
[0058] Figure 7 It is a structural block diagram of a power line warning device in one embodiment;
[0059] Figure 8 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0061] The power line warning method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Among them, the terminal 102 can be but is not limited to various personal computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.
[0062] In one embodiment, Figure 2 As shown, a power line warning method is provided, which is applied to Figure 1 The server in the example is used to illustrate the following steps:
[0063] Step 202, dividing the monitoring range according to the power line parameters.
[0064] Among them, power line parameters are parameters used to describe the characteristics of power lines, including voltage level, current, line length, capacitance, inductance and resistance. These parameters are important parameters in the operation and safety management of power lines and need to be monitored and managed to ensure the operation efficiency and safety of the lines.
[0065] Step 204: collect image data within the monitoring range in real time.
[0066] Step 206, the image data is identified, and when a person is identified within the monitoring range, a pre-trained power line warning model is called to analyze the person and obtain the person type.
[0067] Step 208: If there is a safety risk for the personnel type, a warning signal is generated.
[0068] Specifically, according to the power line parameters and the actual location of the power line, the safety distance can be determined according to the voltage level and spacing of the line. Generally speaking, the higher the voltage level and the closer the spacing, the larger the safety distance should be, that is, the monitoring range should also increase with the safety distance. If there are public facilities such as roads and ditches on both sides of the line, the safety area needs to be expanded accordingly. The image data within the monitoring range is collected by the image acquisition device, and the collected image data is uploaded to the power line warning model. After calculation and processing by the power line warning model, it is identified whether there are detection targets such as personnel, animals, etc. in the image data. When a person is identified, the distance of the currently detected person is measured by the infrared detection module. Since the image acquisition module can detect the position of the person in the image screen, combined with the distance detected by the infrared module, the position of the currently detected person can be determined, and it can be judged whether the person is in the monitoring range. When the detected person is in the monitoring range, it is determined that the person has a safety risk, a warning signal is generated, and the detection results and real-time monitoring video are uploaded to the inspection cloud platform and the backend server. After receiving the warning signal, the backend personnel will conduct remote monitoring and early warning, and the local personnel will be expelled through the local alarm. Among them, the local alarm includes sound and light alarm, through warning pictures and intelligent sound and light alarm system, and according to the warning requirements of different environments, built-in multiple voice alarm sounds, which can emit harsh high-frequency and high-volume voice to drive people away, and the volume can be adjusted as needed. At the same time, the warning sign alarm mode and voice can be modified by remote control. Set flash warning, such as high-brightness LED flash warning, with high flashing brightness and strong penetration, and the warning distance can reach more than 300 meters at night.
[0069] Optionally, it detects whether there are animals approaching the monitoring range. When an animal is detected to have entered the monitoring range, in addition to sound and light alarms, the ultrasonic module can also emit ultrasonic waves to drive the animal away to avoid the risk of electric shock.
[0070] In the above-mentioned power line warning method, the monitoring range is defined by analyzing the power line parameters. When the power line parameters are different, an adjustable monitoring range is used instead of a fixed monitoring range, so that the monitoring range is more comprehensive and reasonable, and the sensitivity of the warning is improved. By acquiring the image data within the monitoring range and uploading it to the power line warning model for processing and analysis, real-time monitoring of the monitoring area is achieved; combined with the calculation and processing of the power line warning model, it is possible to identify whether there is a detection target in the image data, so that potential safety risks can be quickly discovered, the safe operation of the power line can be fully guaranteed, and potential personnel risks can be responded to in a timely manner.
[0071] In one embodiment, based on the power line parameters, defining the monitoring range includes: determining the warning position in the power line where a safety warning is required, and determining the power line parameters of the warning position; defining the monitoring range based on the power line parameters; the monitoring range includes a first monitoring range and a second monitoring range, the first monitoring range is a dangerous area, and the second monitoring range is larger than the first monitoring range.
[0072] Specifically, according to the power line parameters and the actual location of the power line, the safety distance can be determined according to the voltage level and spacing of the line. Generally speaking, the higher the voltage level and the closer the spacing, the larger the safety distance should be, that is, the monitoring range should also increase with the safety distance. The positioning module locates the warning position in the current power line through GPS, demarcates the monitoring range with the warning position as the center, and sets the first monitoring range and the second monitoring range. The second monitoring range is mainly used to monitor fishing behavior. Since many fishing lines are made of conductive materials, such as carbon steel materials, woolen lines, etc., and the basic action of fishing is to swing the rod, the force will be relatively large when swinging the rod, and the fishing line is relatively long. Therefore, when fishing in areas with safety risks such as overhead power transmission and distribution lines, it is easy for the fishing line to hang on the high-voltage wire, resulting in electric shock accidents and affecting personal safety. Moreover, in this case, even if the person is not particularly close to the dangerous area, if the fishing line may touch the transmission line when swinging the rod, there may be safety hazards. In order to more accurately monitor and alarm the fishing behavior, the second monitoring range is set to be larger than the first monitoring range. The first monitoring scope mainly monitors human behaviors such as wandering, staying and gathering.
[0073] In this embodiment, the monitoring range is further refined, and a first monitoring range and a second monitoring range are delineated. The delineated first monitoring range ensures the safety of people who mistakenly enter dangerous areas; the delineation of the second monitoring range expands the monitoring range of fishing areas such as rivers, thereby preventing anglers from accidentally touching high-voltage wires with tools such as fishing lines, causing accidents such as electric shock, and improving the warning capability and warning range of power lines.
[0074] In one embodiment, when there is a safety risk in the type of person, a warning signal is generated; when a person is identified to be present within the monitoring range, a power line warning model is called to analyze the person to obtain the type of person; when the type of person is a pedestrian and is within the first monitoring range, a warning signal is generated; when the type of person is a fisherman and is within the second monitoring range, a warning signal is generated.
[0075] Specifically, when an identification target appears in the image data, it is identified and judged. When the target in the image is identified as a person, the pre-trained power line warning model is called to analyze the identified person and judge the type of person. When the person type is a pedestrian and it is judged that it is located in the first monitoring range, it is considered that the person has a safety risk, and a warning signal is generated and uploaded to the background server, so that the background personnel can grasp the pedestrian's movements in time and take countermeasures to avoid the risk of electric shock. When the person type is a pedestrian and it is judged that it is located in the second monitoring range, no warning signal is generated. When the person type is a fishing person and it is located in the second monitoring range, it is judged that the fishing person has a safety risk, and a warning signal is generated and uploaded to the background server, so that the background personnel can grasp the pedestrian's movements in time and take countermeasures to avoid the risk of electric shock. Since the second monitoring range is larger than the first monitoring range, when the fishing person is in the first monitoring range, the warning signal will also be triggered.
[0076] In this embodiment, the type of person entering the monitoring area is identified and judged through the power line warning model. When pedestrians enter the second monitoring range, the monitoring will not issue a warning, but a warning signal will be generated for the detected fishing personnel to warn them, which reduces the disturbance to pedestrians and ensures the personal safety of fishing personnel. Only when pedestrians enter the first monitoring range and are in a dangerous area will a warning be issued, which protects the personal safety of people of different types and improves the warning capability.
[0077] In one embodiment, when the type of person is a pedestrian and is within the first monitoring range, generating a warning signal also includes: detecting the behavior pattern of the pedestrian; generating a pedestrian stay warning when it is detected that the time the pedestrian is within the first monitoring range exceeds a preset time threshold; generating a pedestrian gathering warning when it is detected that the number of pedestrians within the first monitoring range exceeds a preset number threshold; generating a pedestrian wandering warning when it is detected that the pedestrian is within the second monitoring range for a long time and enters and exits the first monitoring range multiple times; the second monitoring range includes the first monitoring range.
[0078] Specifically, since people often linger and wander in areas where power lines are arranged, being in this area for a long time increases the chances of people coming into contact with power lines, further increasing the risk of electric shock accidents, the type of pedestrians is further judged. When a pedestrian is detected to be in the first monitoring range, an on-site alarm can be issued through the sound and light alarm module, and the detection of pedestrian wandering, lingering, and gathering behaviors can be triggered. When it is detected that a pedestrian is in the first monitoring range for a long time, when the target exceeds the preset time threshold in the first monitoring range, it is considered that pedestrians are currently lingering, a pedestrian lingering warning is generated, and the detection results and real-time monitoring videos are reported to the background. When multiple pedestrians are detected to be in the first monitoring range, and the number of pedestrians exceeds the preset number threshold, it is considered that pedestrians are currently gathering, a pedestrian gathering warning is generated, and the detection results and real-time monitoring videos are reported to the background. If it is detected that the pedestrian's movement trajectory is within the first monitoring range for a long time, and the pedestrian enters and exits the first monitoring range multiple times within the set time period, for example, when the first monitoring range is included in the second monitoring range, and the pedestrian wanders between the first monitoring range and the second monitoring range multiple times, it is considered that the pedestrian is currently wandering, a pedestrian wandering warning is generated, and the detection results and real-time monitoring video are reported to the background.
[0079] On the backend side, according to the received warning signal, the pedestrian wandering, lingering, and gathering behavior signs at the corresponding monitoring location can be marked on the electronic circuit diagram of the power transmission and distribution overhead line, so that the management personnel can respond in time. In addition, the location of pedestrians in the first monitoring range and the second monitoring range can be displayed in real time on the backend interface, and the interface can also display the monitoring video synchronously, so that the backend personnel can grasp the pedestrian's movement in time and take countermeasures to avoid the risk of electric shock.
[0080] In this embodiment, the image recognition, behavior analysis and alarm functions are combined for pedestrians' wandering, lingering, gathering and other behaviors, so as to realize the monitoring of pedestrian behavior and safety warning to ensure the safe operation of power lines. It helps to reduce the risk of electric shock accidents, provide timely remote response and management, and effectively protect the safety of personnel and power facilities.
[0081] In one embodiment, Figure 3 As shown, a method for training a power line warning model is provided, and the method is used to train a power line warning model, comprising the following steps:
[0082] Step 302, acquiring image data of a person, marking features of the person, and generating a marked image; the image data includes images of the person in different scenes;
[0083] Step 304, preprocessing the marked image to generate a feature arrangement array;
[0084] Step 306, inputting the feature arrangement array into the neural network model as training data, and introducing learnable weights to generate weights for each feature;
[0085] Step 308, repeatedly iteratively adjust the weights according to the training data, train the neural network model until convergence or reaching a preset number of training times, and obtain a power line warning model.
[0086] Specifically, images of fishermen in normal weather, outdoor riverbanks and other scenes are collected, with as many image types as possible and as varied backgrounds as possible. Since fishing incidents are not likely to occur in some abnormal weather conditions, the normal weather scenes here refer to other related severe weather scenes except rainy days, foggy days, snowy days, etc. Data annotation uses rectangular boxes to annotate all people and fishing gear. The annotated boxes should contain as few backgrounds and interfering objects as possible other than people and fishing rods to avoid extracting irrelevant information features other than fishing during feature extraction, thereby affecting the detection effect of fishing behavior. Considering that there are limited images of real scenes in water areas, the network has limited features such as spatial invariance and pixel-level invariance learned by the network. Data enhancement methods such as translation, scaling, rotation, and changing hue values are used to allow the model to train various types of data and improve the model's discrimination on test data.
[0087] Optionally, two data augmentation methods can be used in data preprocessing. The first is a combination of a series of atomic linear transformations, mainly including: center, perspective, translation, scale, rotation, and shear. The second is the proposed Crop Mixup feature enhancement, which improves the original Mixup method and performs feature enhancement after processing with a convolutional neural network (CNN). The original image processing method is to use the original Mixup as a data augmentation method, and process the data before the neural network is built through the data preprocessing stage and data augmentation methods such as linear transformation. Since the original confusion enhancement (Mixup) cannot guarantee the semantic information of the synthesized image, the synthesized image information is often not correctly understood by the network, so the mixing ratio parameter (mixrate) is increased to control the ratio of cropping and reorganization. The mixing ratio parameter ranges from 0 to 1. When the mixing ratio parameter approaches 0, the transformation effect of the model training data will be smaller; when the mixing ratio parameter approaches 1, the transformation effect of the model training data will be more significant. At the same time, in order to ensure sufficient mixing of data, a batch of data is divided into three minimum batches (minbatches), so that the network input has more feature arrangement methods.
[0088] In the backbone network, since the Convolutional Neural Network (CNN) is translation invariant and local, and lacks the ability to model globally, the framework (Transformer) in the field of natural language processing is introduced to form the CNN+Transformer architecture. First, after the input image undergoes a 4×4 convolution and layer normalization (Layer Normalization, Layer Norm), the shape of the feature map changes from (H, W, 3) to (H / 4, W / 4, C). Secondly, through the downsampling operation (Patch Merging), the feature map is downsampled by half and the number of channels is doubled. Finally, through the slicing operation, the feature layers are spliced and the pixel channel dimension is normalized. Therefore, after the feature map is processed three times, the feature map sizes are (H / 8, W / 8, 2C), (H / 16, W / 16, 4C) and (H / 32, W / 32, 8C) respectively.
[0089] By using the Weighted Multihead Self-Attention (WMSA) and Shifted Windows Multihead Self-Attention (SW MSA) in the visual self-attention model (Swin Transformer Block), that is, after the previous layer uses W MSA, the next layer uses the SWMSA module to realize communication between windows, thereby achieving the effect of global modeling.
[0090] The feature fusion network (Neck) is a network layer inserted between the Backbone and the detection head to collect different feature maps. Usually, a Neck consists of multiple forward calculations (bottom up) and top-down paths (top down). Considering that the fishing targets in real water scenes are at different distances from the video equipment, the target sizes vary greatly. The feature fusion network (Path Aggregation Network, PANet) in the original image processing method simply adds the features after sampling, without considering the differences in the contribution of features of different resolutions. Therefore, a bidirectional feature pyramid network (Bi-directional Feature Pyramid Network, BIFPN) is used to introduce learnable weights to learn the importance of different input features.
[0091] Based on the above network structure, by labeling information such as fishermen and fishing rods, the labeled images are input into the network for model training until the loss function converges, and the trained network parameters can be used to detect fishing behavior. In addition to positioning loss, target confidence loss and classification loss, the loss function also includes Crop Mixup loss.
[0092] Optionally, the target detection method can select target detection models such as YOLOv3 and YOLOv5. The position corresponding to the target output by the YOLOv3 target detection model corresponds to all the targets detected in an image and the position information of these targets in the fixed monitoring screen. Each target on the corresponding image is cut out by a rectangular frame, and the position of the target is represented by the width, height and center point coordinate information of the rectangular frame. The YOLOv3 target detection algorithm is mainly composed of convolutional layers and pooling layers. The naming rules of the layers in the network are composed of their categories and the number of times they appear in the network. For example, conv8 represents the 8th convolutional layer in the network. The output feature map size of each layer in the network is expressed as "resolution width × resolution height × number of channels". After multiple layers of convolutional pooling layers, the rectangular frames and classifications of each target in the image are finally obtained to complete the detection of the target. The pooling layer is an image downsampling operation. Although it will reduce the parameters of the convolutional feature layer and speed up the model operation speed, it will cause semantic information loss to the convolutional feature map of the previous layer. On the other hand, the receptive field of the shallow convolution feature contains less background noise and has better representation ability for small targets. At the same time, when the target is occluded, only the local features of the target exist. Since the shallow convolution layer is usually more sensitive to the local or small targets of the target, the shallow feature output layer of the network is increased, and the samples of the target local feature learning are increased to enhance the recognition ability of occluded targets and small targets. Due to the problem of computing resources, the basic skeleton parameters of the YOLOv3 target detection network are only 4M, and the volume is small and suitable for landing. In order to train the target detection model of the main figure, it is necessary to collect a large amount of pedestrian image information as a training data set, and then manually mark the rectangular boxes of pedestrians one by one from the image information as annotation data. Further design the neural network structure and loss function of the target detection model, and use the annotation data to train the network parameters of the target detection model. After the model training is completed, save the model structure and parameters for subsequent target detection. According to actual needs, pedestrian detection can also be adaptively increased or refined.
[0093] In this embodiment, the CNN+Transformer architecture in the neural network is used to realize the modeling of the power line warning model, fully combining the advantages of both, and improving the effect of small targets and intensive prediction tasks in natural scenes. The convolutional neural network model has high accuracy and fast calculation speed, and can also be used in combination with other technologies and algorithms, such as genetic algorithms, BP neural networks, etc., to improve the performance and adaptability of personnel behavior detection in power lines. The image data of fishermen and pedestrians collected by the image acquisition device is used as the input of the neural network. It can effectively reduce the operating risk of power lines and improve the warning ability of power lines.
[0094] In one embodiment, Figure 4 As shown, a power line warning method is provided, the method comprising the following steps:
[0095] Step 402, acquiring image data of a person, marking features of the person, and generating a marked image; the image data includes images of the person in different scenes;
[0096] Step 404, preprocessing the marked image to generate a feature arrangement array;
[0097] Step 406, inputting the feature arrangement array as training data into the neural network model, and introducing learnable weights to generate weights for each feature;
[0098] Step 408, repeatedly iteratively adjust the weights according to the training data, train the neural network model until convergence or reaching a preset number of training times, and obtain a power line warning model.
[0099] Step 410, determining a warning position in the power line where a safety warning is required, and determining power line parameters at the warning position;
[0100] Step 412, defining a monitoring range according to power line parameters; the monitoring range includes a first monitoring range and a second monitoring range, the first monitoring range is a dangerous area, and the second monitoring range is larger than the first monitoring range.
[0101] Step 414, collecting image data within the monitoring range in real time;
[0102] Step 416, identifying the image data, and when identifying that a person is present within the monitoring range, calling a pre-trained power line warning model, analyzing the person, and obtaining the type of the person;
[0103] Step 418, when it is identified that there is a person in the monitoring range, the power line warning model is called to analyze the person and obtain the type of the person;
[0104] Step 420, detecting the behavior pattern of pedestrians;
[0105] Step 422, generating a pedestrian stay warning when it is detected that the time that the pedestrian is in the first monitoring range exceeds a preset time threshold;
[0106] Step 424, generating a pedestrian gathering warning when it is detected that the number of pedestrians in the first monitoring range exceeds a preset number threshold;
[0107] Step 426, when it is detected that the pedestrian is in the second monitoring range for a long time and enters and exits the first monitoring range multiple times, a pedestrian wandering warning is generated; the second monitoring range includes the first monitoring range.
[0108] Step 428: When the person type is a fisherman and is located in the second monitoring range, a warning signal is generated.
[0109] In this embodiment, the monitoring range is defined by analyzing the power line parameters. When the power line parameters are different, an adjustable monitoring range is used instead of a fixed monitoring range, so that the monitoring range is more comprehensive and reasonable, and the sensitivity of the warning is improved. The monitoring range is further refined, and the first monitoring range and the second monitoring range are defined, and the monitoring range of fishing areas such as the riverside is expanded to prevent fishing tools such as fishing lines from accidentally touching high-voltage wires, causing accidents such as electric shock. The CNN+Transformer architecture is used to model the power line warning model, fully combining the advantages of both, and improving the effects of small targets and intensive prediction tasks in natural scenes. By acquiring image data within the monitoring range and uploading it to the power line warning model for processing and analysis, real-time monitoring of the monitoring area is achieved; combined with the calculation and processing of the power line warning model, it is possible to identify whether there is a detection target in the image data, so that potential safety risks can be quickly discovered, the safe operation of the power line can be fully guaranteed, and potential personnel risks can be responded to in a timely manner.
[0110] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are shown in sequence, these steps are not necessarily executed in sequence in the order indicated. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of the steps or stages in other steps.
[0111] Based on the same inventive concept, the embodiment of the present application also provides a power line warning device for implementing the power line warning method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more power line warning device embodiments provided below can refer to the limitations of the power line warning method above, and will not be repeated here.
[0112] In one embodiment, Figure 5 As shown, a power line warning device is provided, including: a positioning module 502, a collection module 504, an analysis module 506 and a warning module 508, wherein:
[0113] The positioning module 502 is used to define the monitoring range according to the power line parameters.
[0114] The acquisition module 504 is used to acquire image data within the monitoring range in real time.
[0115] The analysis module 506 is used to identify the image data, and when a person is identified within the monitoring range, a pre-trained power line warning model is called to analyze the person and obtain the person type.
[0116] The warning module 508 generates a warning signal when there is a safety risk for the type of personnel.
[0117] In one embodiment, the positioning module 502 further includes:
[0118] The position parameter module is used to determine the warning position in the power line where a safety warning is required, and to determine the power line parameters of the warning position.
[0119] The monitoring module is used to define the monitoring range according to the power line parameters; the monitoring range includes a first monitoring range and a second monitoring range, the first monitoring range is a dangerous area, and the second monitoring range is larger than the first monitoring range.
[0120] In one embodiment, the analysis module 506 further includes:
[0121] The target detection module is used to perform target detection on image data, and the target is the person appearing in the monitoring range.
[0122] The model calling module is used to call the power line warning model to analyze the personnel and obtain the personnel type when a person is identified within the monitoring range.
[0123] The first warning module is used to generate a warning signal when the person type is a pedestrian and is located in a first monitoring range.
[0124] The second warning module is used to generate a warning signal when the person type is a fisherman and is located in the second monitoring range.
[0125] In one embodiment, the first warning module further includes:
[0126] The pattern monitoring module is used to detect the behavior patterns of pedestrians.
[0127] The stay warning module is used to generate a pedestrian stay warning when it is detected that the time a pedestrian stays in the first monitoring range exceeds a preset time threshold.
[0128] The gathering warning module is used to generate a pedestrian gathering warning when it is detected that the number of pedestrians in the first monitoring range exceeds a preset number threshold.
[0129] The wandering warning module is used to generate a pedestrian wandering warning when it is detected that a pedestrian is in the second monitoring range for a long time and enters and exits the first monitoring range many times; the second monitoring range includes the first monitoring range.
[0130] Each module in the above-mentioned power line warning device can be implemented in whole or in part by software, hardware or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.
[0131] Based on the same inventive concept, the embodiment of the present application also provides a power line warning model training device for implementing the power line warning model training method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more power line warning model training device embodiments provided below can refer to the limitations of the power line warning model training method above, and will not be repeated here.
[0132] In one embodiment, Figure 6 As shown, a power line warning model training device is provided, including: a data acquisition module 602, a preprocessing module 604, an initialization module 606 and a training module 608, wherein:
[0133] The data acquisition module 602 is used to acquire image data of a person, mark the features of the person, and generate a marked image; the image data includes images of the person in different scenes.
[0134] The preprocessing module 604 is used to preprocess the marked image to generate a feature arrangement array.
[0135] Initialization module 606 is used to input the feature arrangement array as training data into the neural network model, and introduce learnable weights to generate weights of each feature.
[0136] The training module 608 is used to iteratively adjust the weights according to the training data, train the neural network until convergence or reaching a preset number of training times, and obtain a power line warning model.
[0137] Each module in the above-mentioned power line warning model training device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of the above modules.
[0138] In one embodiment, Figure 7 As shown, a power line warning device is provided, including: a housing 702, a warning bar 704 and a power line warning device. The housing 702 includes a first side 706, a second side 708, a first panel 710 and a second panel 712; the first side 706 of the housing is connected to a first bracket 714, the second side 708 is connected to a second bracket 716, the first panel 710 is connected to a solar panel 718, and the second panel 712 is connected to an infrared detection module 720, an audible and visual alarm module 722 and a video monitoring module 724; the warning bar 704 is fixed between the first bracket 714, the second bracket 716 and the second panel 712 of the housing 702; and the power line warning device is arranged in the housing 702.
[0139] Specifically, the warning image and text are added with an intelligent sound and light alarm system and a speaker is set on the top. According to the warning requirements of different environments, a variety of voice alarm sounds are built-in, which can emit harsh high-frequency and high-volume voices to drive people away. The volume can be adjusted as needed. At the same time, the warning sign alarm mode and voice can be modified by remote control. The high-brightness LED flash warning has high flashing brightness and strong penetration. The warning distance at night can reach more than 300 meters. It is a very conspicuous warning sign and can also have a significant warning effect at night. The high-brightness LED is projected onto the warning board, and the outer layer of the warning board is covered with an acrylic board, which can have good light conductivity. The solar energy collected by the top is used to power the entire system and store the remaining power in the built-in rechargeable lithium battery. The solar panel is set on the top of the warning board. The bracket is designed as a multi-section structure that can be rotated to expand or fold. The system circuit is designed inside the shell or bracket. It has an anti-theft function and an integrated design. The top is designed with an active handle for easy manual carrying. During the day, it charges with solar energy, and when passersby approach, infrared alarms are automatically triggered, which can be transmitted back to the background through wireless communication, or through online video and artificial voice prompts and viewing; at night, the warning light strip automatically lights up to prevent wild fishermen from fishing. It is connected to the hidden danger risk video system of the production command center, so that the background can be monitored online and remotely controlled, and the front end can automatically remind and alarm. It can also automatically retrieve the responsible line patrolman, send positioning and warning information to his mobile phone, dispatch him to the scene to check or give remote voice reminders to prevent accidents.
[0140] In this embodiment, the power line warning device is a foldable and rotatable structure, and a movable handle is designed on the top, which is quick to install and easy to carry. Through the integrated sound and light technology, safety warnings can be given to the monitored area all day long without blind spots. By setting a high-brightness LED flash warning, the warning sign can also be used as a very conspicuous warning sign at night, and can also have a significant warning effect at night. At the same time, the warning device components are all located under the solar panel, which can also slow down the aging speed of the equipment in sunlight and rain.
[0141] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 8As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a power line warning method is implemented.
[0142] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0143] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.
[0144] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.
[0145] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.
[0146] 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, stored data, displayed data, 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 relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.
[0147] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium 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), magnetoresistive 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. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.
[0148] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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.
[0149] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A power line warning method, It is characterized in that The method comprises: Divide the monitoring range according to the power line parameters; determine the warning position in the power line where a safety warning is required, and determine the power line parameters of the warning position; the monitoring range includes a first monitoring range and a second monitoring range, the first monitoring range is a danger zone, and the second monitoring range is larger than the first monitoring range; the second monitoring range is used to monitor fishing behavior; Collect image data within the monitoring range in real time; The image data is identified, and when a person is identified to be present within the monitoring range, a pre-trained power line warning model is called to analyze the person and obtain the type of person; the power line warning model comprises: obtaining image data of the person, marking the person's features, and generating a marked image; the image data comprises images of the person in different scenes; pre-processing the marked image, enhancing the synthesized image through original confusion, increasing the mixing ratio parameter to control the proportion of clipping and reorganization, and generating a feature arrangement array; the mixing ratio parameter has a value range between 0 and 1; the feature arrangement array is input into the neural network model as training data, and learnable weights are introduced to generate the weights of each feature; the weights are adjusted iteratively according to the training data, and the neural network model is trained until convergence or a preset number of training times is reached, so as to obtain the power line warning model; In the event that there is a safety risk with the type of person, a warning signal is generated; in the event that the type of person is a pedestrian and is within the first monitoring range, generating a warning signal also includes detecting the behavior pattern of the pedestrian; in the event that it is detected that the pedestrian has been within the first monitoring range for a period exceeding a preset time threshold, a pedestrian stay warning is generated; in the event that it is detected that the number of pedestrians within the first monitoring range exceeds a preset number threshold, a pedestrian gathering warning is generated; in the event that it is detected that the pedestrian has been within the second monitoring range for a long time and has entered and exited the first monitoring range multiple times, a pedestrian wandering warning is generated; the second monitoring range includes the first monitoring range.
2. The method according to claim 1, It is characterized in that In the case where the type of personnel presents a safety risk, generating a warning signal includes: When a person is identified to be present within the monitoring range, a power line warning model is called to analyze the person to obtain the type of the person; In the case where the person type is a pedestrian and is located within the first monitoring range, generating a warning signal; When the person type is a fisherman and is located in the second monitoring range, a warning signal is generated.
3. A power line warning device, It is characterized in that The device comprises: A positioning module is used to define a monitoring range according to power line parameters; determine a warning position in the power line where a safety warning is required, and determine the power line parameters of the warning position; the monitoring range includes a first monitoring range and a second monitoring range, the first monitoring range is a dangerous area, and the second monitoring range is larger than the first monitoring range; the second monitoring range is used to monitor fishing behavior; An acquisition module, used for acquiring image data within the monitoring range in real time; An analysis module is used to identify the image data, and when a person is identified to appear within the monitoring range, a pre-trained power line warning model is called to analyze the person to obtain the type of person; the power line warning model includes: obtaining image data of the person, marking the person's features, and generating a marked image; the image data includes images of the person in different scenes; pre-processing the marked image, enhancing the synthesized image through original confusion, increasing the mixing ratio parameter to control the proportion of clipping and reorganization, and generating a feature arrangement array; the mixing ratio parameter has a value range between 0 and 1; inputting the feature arrangement array into the neural network model as training data, and introducing learnable weights to generate the weights of each feature; iteratively adjusting the weights according to the training data, training the neural network model until convergence or reaching a preset number of training times, and obtaining the power line warning model; A warning module is used to generate a warning signal when there is a safety risk in the type of person; when the type of person is a pedestrian and is within the first monitoring range, generating a warning signal also includes detecting the behavior pattern of the pedestrian; when it is detected that the pedestrian has been within the first monitoring range for a time exceeding a preset time threshold, a pedestrian stay warning is generated; when it is detected that the number of pedestrians within the first monitoring range exceeds a preset number threshold, a pedestrian gathering warning is generated; when it is detected that the pedestrian has been within the second monitoring range for a long time and has entered and exited the first monitoring range multiple times, a pedestrian wandering warning is generated; the second monitoring range includes the first monitoring range.
4. A power line warning model training device, used for training the power line warning model as described in any one of claims 1-2, the device include: A data acquisition module is used to acquire image data of a person, mark the features of the person, and generate a marked image; The image data includes images of people in different scenes; A preprocessing module, used for preprocessing the labeled image, enhancing the synthesized image by original confusion, increasing the mixing ratio parameter to control the ratio of cropping and reorganization, and generating a feature arrangement array; The mixing ratio parameter value range is between 0 and 1; An initialization module, used to input the feature arrangement array as training data into the neural network model, and introduce learnable weights to generate weights of each feature; The training module is used to iteratively adjust the weights according to the training data, train the neural network until convergence or reaching a preset number of training times, and obtain the power line warning model.
5. A power line warning device, It is characterized in that The device comprises: A shell, the shell comprising a first side, a second side, a first panel and a second panel; the first side of the shell is connected to the first bracket, the second side is connected to the second bracket, the first panel is connected to the solar panel, and the second panel is connected to the infrared detection module, the sound and light alarm module and the video monitoring module; a warning bar, fixed between the first bracket, the second bracket and the second panel of the housing; The power line warning device as described in claim 3, wherein the power line warning device is arranged in the housing.
6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 2 are implemented.
7. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are implemented.
