Intelligent bird accidental collision prevention device and method for overhead line

Abstract

The invention belongs to the technical field of power transmission line inspection, and discloses an intelligent bird accidental collision prevention device and method for an overhead line. The method comprises the following steps: collecting a video stream of bird activities around a power transmission line in a visual range; intercepting the video stream into a plurality of to-be-recognized pictures according to frames, and recognizing whether bird images exist in the to-be-recognized pictures or not; when the bird image is recognized, a first control signal is sent out, otherwise, a second control signal is sent out; when the first control signal is received, the bird blocking module is started to emit ultraviolet light beams to block birds; and in the second control signal, the bird blocking module is closed to save power. In the prior art, an anti-bird-mistaken-collision warning mark or warning ball is installed on a ground wire or a ground wire, power failure is needed to install the anti-bird-mistaken-collision warning mark or warning ball, the anti-bird-mistaken-collision warning mark or warning ball is limited by the fact that a power failure plan is difficult to arrange, and compared with an installation device on an iron tower, the risk coefficient of the installation device on the wire is greatly increased, and the problem can be effectively solved by adopting the anti-bird-mistaken-collision warning mark or warning ball. The effects of protecting rare birds, improving operation safety and providing message results are achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of power transmission line inspection, and in particular relates to an overhead line intelligent bird accidental collision prevention device and method. Background Art

[0002] In recent years, areas with rich natural ecology such as Tibet and Qinghai have frequently encountered casualties caused by rare wild birds such as black-necked cranes, swans, and grey cranes accidentally colliding with power transmission lines, posing a challenge to local biodiversity conservation.

[0003] The current bird deterrent measure, namely installing warning signs on overhead power lines to prevent bird collisions, has a certain effect, but there are significant limitations in the implementation process. This method relies on power outages, and power outage plans for key lines are often difficult to arrange quickly due to a variety of factors, resulting in protective measures that cannot be put in place in a timely manner. Manual installation of warning balls is not only time-consuming and labor-intensive, but also needs to be carried out under power outage conditions, which increases the complexity of the work and safety risks. Electric workers need to climb high to work, which is not only labor-intensive, but also faces high risks of high-altitude operations. In addition, if helicopters are used to assist in installation, although it can reduce manual risks, it is costly and requires extremely high flight skills, which is not feasible under normal circumstances. In view of the above difficulties, there is an urgent need to explore more efficient, safe and economical bird deterrent solutions. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent bird-proofing device and method for overhead lines, which solves the problem in the prior art that power outages are required to install bird-proofing warning signs or warning balls, and power outage plans are difficult to arrange.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: In a first aspect, the present invention provides an overhead line intelligent bird collision prevention device, which is used to prevent birds from accidentally colliding with overhead lines, comprising: A camera module is arranged at a preset position of the transmission tower and is used to collect a video stream of bird activities around the transmission line within a visible range; The bird recognition module is used to obtain a video stream of bird activities around the power transmission line within the visible range from the camera module, intercept the video stream into a number of pictures to be recognized by frame, and recognize whether there is a bird image in the pictures to be recognized; when a bird image is recognized, a first control signal is sent to the control module; when no bird image is recognized, a second control signal is sent to the control module; A control module, used to receive a control signal from the bird identification module, and when the control signal is a first control signal, control the bird prevention module to enter an on state; when the control signal is a second control signal, control the bird prevention module to enter an off state; Bird blocking module, used to emit a UV beam to block birds when turned on.

[0006] The present invention is further improved in that: the camera module includes a night vision camera and a gun, the night vision camera is arranged in the gun, the gun is fixedly arranged on the tower head cross arm or tower body position of the transmission tower, the night vision camera shoots toward the adjacent tower, and the night vision camera is powered by a solar panel.

[0007] A further improvement of the present invention is that the bird identification module includes an AI edge computing device, the AI ​​edge computing device is electrically connected to the control module, and the AI ​​edge computing device is used to identify whether there is a bird image in the picture to be identified.

[0008] The present invention further improves in that: an improved DeepLabV3+ model is adopted in the AI ​​edge computing device; the improved DeepLabV3+ model is a model that adds a first convolutional block attention module and a second convolutional block attention module to the DeepLabV3+ model; the input of the first convolutional block attention module is connected to the output of the atrous spatial convolution pooling pyramid ASPP in the encoder, and the output of the first convolutional block attention module is connected to the 1×1 convolution module output in the encoder; the input of the second convolutional block attention module is connected to the output of the first feature fusion module in the decoder, the input of the output of the first feature fusion module is connected to the output of the backbone network in the encoder, and the output of the second convolutional block attention module is connected to the input of the interpolation upsampling module.

[0009] A further improvement of the present invention is that the control module includes a relay, which is electrically connected to the GPIO pin on the AI ​​edge computing device. When the bird identification module recognizes that there is a bird image in the picture to be identified, the AI ​​edge computing device drives the GPIO pin level to rise, drives the relay to turn on, and starts the bird blocking module.

[0010] The present invention is further improved in that: the bird-blocking module comprises a mounting shell and an ultraviolet laser, the number of the ultraviolet lasers is two, and the two ultraviolet lasers are arranged in the mounting shell. A further improvement of the present invention is that the wavelength of the laser emitted by the ultraviolet laser is 380 mm-395 mm.

[0011] In a second aspect, the present invention provides an overhead line intelligent bird collision prevention method, comprising: Collect video streams of bird activities around power lines within visual range; Obtain a video stream of bird activities around a power transmission line within a visible range, intercept the video stream into a number of pictures to be identified by frame, and identify whether there is a bird image in the pictures to be identified; send a first control signal when a bird image is identified; send a second control signal when no bird image is identified; When a first control signal is received, the bird-blocking module is turned on; when a second control signal is received, the bird-blocking module is turned off; wherein the bird-blocking module is used to emit an ultraviolet light beam to block birds when in the turned-on state. The present invention is further improved in that: the step of obtaining a video stream of bird activities around a power transmission line within a visible range, intercepting the video stream into a plurality of pictures to be identified by frame, and identifying whether there is a bird image in the pictures to be identified; sending a first control signal when a bird image is identified; and sending a second control signal when no bird image is identified is implemented by a bird identification module; The bird recognition module includes an AI edge computing device, which is electrically connected to the control module and is used to identify whether there is a bird image in the picture to be identified; an improved DeepLabV3+ model is adopted in the AI ​​edge computing device; the improved DeepLabV3+ model is a model in which a first convolutional block attention module and a second convolutional block attention module are added to the DeepLabV3+ model; the input of the first convolutional block attention module is connected to the output of the atrous spatial convolution pooling pyramid ASPP in the encoder, and the output of the first convolutional block attention module is connected to the 1×1 convolution module output in the encoder; the input of the second convolutional block attention module is connected to the output of the first feature fusion module in the decoder, the input of the output of the first feature fusion module is connected to the output of the backbone network in the encoder, and the output of the second convolutional block attention module is connected to the input of the interpolation upsampling module.

[0012] The present invention is further improved in that: when the first control signal is received, the bird blocking module is turned on; when the second control signal is received, the step of turning off the bird blocking module is realized by the control module; The control module includes a relay, which is electrically connected to the GPIO pin on the AI ​​edge computing device. When the bird identification module recognizes that there is a bird image in the picture to be identified, the AI ​​edge computing device drives the GPIO pin level to rise, drives the relay to turn on, and starts the bird blocking module.

[0013] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides an overhead line intelligent bird-prevention device for preventing birds from accidentally colliding with overhead lines, comprising: a camera module, arranged at a preset position of the transmission tower, for collecting a video stream of bird activities around the transmission line within a visible range; a bird identification module, for obtaining a video stream of bird activities around the transmission line within a visible range from the camera module, intercepting the video stream into a plurality of pictures to be identified by frame, and identifying whether there is a bird image in the picture to be identified; sending a first control signal to a control module when a bird image is identified; sending a second control signal to the control module when no bird image is identified; a control module, for receiving a control signal of the bird identification module, when the control signal is the first control signal, controlling the bird-prevention module to enter an on state; when the control signal is the second control signal, controlling the bird-prevention module to enter an off state; and a bird-prevention module, for emitting an ultraviolet light beam to prevent birds when in an on state. In the present invention, the camera module is arranged at a preset position of the transmission tower, and can collect a video stream of the surrounding environment in real time, ensuring timely monitoring of potential bird approaches, providing rich visual data for the bird identification module, and enhancing the accuracy of identification. The bird recognition module can effectively identify bird images by intercepting pictures in the video stream at preset intervals and performing intelligent analysis. When a bird is identified, the system will send a first control signal and respond quickly to activate the bird blocking module. The control module enables the entire system to have self-regulation capabilities. When no birds are approaching, the control module will enter a closed state to save energy and avoid unnecessary interference. When birds are detected, the bird blocking module will actively work to block birds in a non-invasive manner, reduce damage to birds, and protect the safety of overhead lines. Through data feedback, users can analyze the patterns of bird activities to further optimize the anti-collision strategy. Intelligent real-time monitoring of bird activities around overhead lines effectively reduces the risk of bird accidental collisions. With the cooperation of the camera module and the bird recognition module, the system can quickly identify birds and respond, activate the bird blocking module, and take protective measures when necessary. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0015] Figure 1 A schematic diagram of an embodiment of an intelligent bird collision prevention device for overhead lines according to the present invention; Figure 2 A logic diagram of an embodiment of an intelligent method for preventing accidental bird collisions on overhead lines according to the present invention; Figure 3It is a flow chart of an embodiment of an intelligent method for preventing accidental bird collisions on overhead lines of the present invention; Figure 4 This is a method diagram of an embodiment of an intelligent method for preventing accidental bird collisions on overhead lines of the present invention.

[0016] Description of reference numerals: 100-mounting shell; 200-ultraviolet laser.

[0017] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0020] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0021] An intelligent bird collision prevention device for overhead lines, used to prevent birds from accidentally colliding with overhead lines, comprising: A camera module, which is arranged at a preset position of a transmission tower, and is used to collect video streams within a preset range; in a specific embodiment, the camera module is generally installed on the transmission tower and faces an adjacent transmission tower, and is used to collect video streams of bird activities around the transmission line within a visible range; A bird recognition module, the bird recognition module is electrically connected to the camera module, and the bird recognition module is used to intercept the video stream into a number of pictures to be recognized at a preset interval, and to identify whether there is a bird image in the picture to be recognized; when the image acquisition module recognizes a bird image, it sends a first control signal to the control module; when the image acquisition module does not recognize a bird image, it sends a second control signal to the control module; in a specific embodiment, the bird recognition module intercepts the video stream into a number of pictures to be recognized by frame; A control module, the control module is connected to the bird recognition module, and the control module is used to receive a control signal of the bird recognition module. When the control signal is a first control signal, it indicates that the bird recognition module finds that there are bird activities around the line, and the bird blocking module is controlled to enter an on state; when the control signal is a second control signal, it indicates that there are no flying birds in the field of view, and the bird blocking module is controlled to enter an off state, and the bird blocking module is turned off to save power; The bird-blocking module is connected to the control module and is used to emit ultraviolet light beams to block birds when it is turned on. In a specific embodiment, the bird-blocking module emits ultraviolet light beams to indicate the existence of obstacles in the flying bird space. After the birds find the obstacles, they generally adopt an ascending flight strategy, thereby preventing the flying birds from hitting the overhead wires (generally hitting the ground wires, and in the case where there is no ground wire, they will also hit the wires).

[0022] When the control module enters the on state, the bird blocking module is started; when the control module enters the off state, the bird blocking module is turned off.

[0023] In the prior art, warning signs or warning balls to prevent bird accidental collisions are installed on ground wires or access wires. A power outage is required to install the warning signs or warning balls to prevent bird accidental collisions. Due to the difficulty in arranging power outage plans, the risk factor of installing devices online is much higher than that of installing devices on towers. The present invention can effectively avoid this problem, thereby protecting rare birds and improving operational safety.

[0024] The present invention can effectively reduce safety accidents caused by accidental bird collisions and protect birds and power facilities by real-time monitoring of bird activities around overhead lines. The camera module and the bird recognition module are used to realize automatic identification of birds without human intervention. When a bird image is identified, a control signal is quickly sent to start the bird-blocking module, which can effectively prevent birds from approaching overhead lines and reduce the risk of accidental collisions. The control module can dynamically adjust the on and off states of the bird-blocking module according to the identification of birds, avoid unnecessary energy consumption, and achieve energy-saving effects. The camera module continuously collects video streams, which can be used for subsequent data analysis and bird behavior research, and provide valuable information for bird protection. Compared with traditional physical barriers or bird-blocking devices, the non-contact method of the intelligent bird-blocking device reduces interference with the bird habitat and protects the ecological balance.

[0025] In some embodiments, the camera module includes a night vision camera and a gun, the night vision camera is arranged in the gun, the gun is fixedly arranged at the tower head cross arm or tower body of the transmission tower, the night vision camera shoots toward the adjacent tower, and the night vision camera is powered by solar panels. The night vision camera is precisely installed inside the gun to ensure its stability and protection performance, while avoiding the influence of external environmental factors on the camera. The gun is fixed at the tower head cross arm or tower body to ensure its wide field of view and wide coverage. The night vision camera shoots toward the adjacent tower to achieve comprehensive monitoring of the surrounding environment. The use of solar panels for power supply enables the camera module to operate independently and reduce dependence on external power supply. The night vision camera is equipped with infrared technology, which can be used for clear shooting in low light or completely dark environments, improving the effectiveness of night monitoring. The camera module can be connected to the AI ​​edge computing device to transmit the captured video data in real time for real-time analysis and recognition. At the same time, the processed data can be sent to the control center via a wireless network to achieve remote monitoring and management. The night vision camera can work effectively in low light environments, ensuring that bird activities can be monitored day and night to improve safety. The gun camera is fixed on the tower head or tower body, which reduces the shaking of the equipment caused by external factors such as wind and rain, and ensures the stability of the shooting picture. The use of solar panels for power supply reduces the dependence on traditional power supplies and realizes the use of green energy. The camera shoots towards the adjacent tower, which can effectively cover a larger area, improve the monitoring effect, and timely detect potential bird approaches. The solar power supply system can operate sustainably, extend the service life of the equipment, and reduce the maintenance frequency. The centralized installation on the tower body simplifies the wiring and installation, and facilitates subsequent maintenance and inspection. In the cross arm of the overhead transmission line conductor, the camera can also be installed on the tower body below the cross arm, and shoot towards the adjacent tower to capture whether there are bird flight activities in the file. The camera that can be used has a pixel of 16 million. The camera does not need the pan-tilt rotation function, and only needs to shoot towards the opposite tower, so the gun camera can be installed. The camera has night vision function, and the night vision pixel is 2 million. The camera is equipped with a solar panel and installed on the angle steel of the tower. It can use universal solar cells and installation accessories.

[0026] In some embodiments, the bird recognition module is connected to the camera module, and the bird activity around the power transmission line within the visible range is obtained from the camera module; the bird recognition module includes an AI edge computing device, the AI ​​edge computing device is electrically connected to the control module module, and the AI ​​edge computing device is used to identify whether there is a bird image in the picture to be identified. The AI ​​edge computing device uses advanced deep learning algorithms to analyze the bird images captured by the camera module, identify different types of birds, and improve the accuracy and speed of recognition. The advantage of edge computing is that it can process data locally on the device, reduce latency, and ensure that bird recognition is almost real-time, thereby quickly responding to environmental changes. Through AI edge computing, only processed and filtered data will be sent to the control module, reducing bandwidth requirements and improving the overall efficiency of the system. The AI ​​device can continuously learn and update the recognition model based on the newly captured images, enhance the system's ability to recognize new types of birds, and maintain efficient adaptability. The recognition results can be fed back to the control module in real time, and the control module automatically adjusts the strategy of the bird-blocking module based on the recognition information to form a closed-loop system and improve the repelling effect.

[0027] The improved DeepLabV3+ model is used in AI edge computing devices; based on the DeepLabV3+ architecture, the CBAM (Convolutional Block Attention Module) attention mechanism is integrated to optimize performance. In the encoding stage, after the dilated convolution is processed, the primary feature layer in the deep convolutional neural network (DCNN) is transformed using 1×1 convolution. Subsequently, through the feature fusion step, the processed features are input into the CBAM attention mechanism for detailed feature weighting operations. This process aims to effectively integrate multi-scale information and extract richer and higher-order feature representations.

[0028] In the decoding stage, in order to further enhance the feature parsing ability, the input image after upsampling is first enlarged by 4 times, and then the CBAM attention mechanism is introduced to process the enlarged low-level features. This strategy aims to filter out redundant background information while emphasizing and highlighting the features of the target area, so that the model can show higher accuracy and robustness in identifying the features of flying birds.

[0029] DeepLabV3+ is used to train a flying bird recognition model. In order to protect birds from harm as much as possible, the model does not need to distinguish the species of birds, because subdividing bird species requires a lot of computing resources and reduces the accuracy. In order to better protect birds from accidentally colliding with the lines, undifferentiated bird prevention should be carried out, which can save computing power.

[0030] Jetson nano is used as the core hardware module of bird recognition, and ONNX technology is used to deploy the bird recognition algorithm model. Due to the limited embedded hardware resources, the lightweight improved DeepLabV3+ algorithm model is deployed.

[0031] After the device is running, the camera video stream is continuously obtained. When the recognition module finds that there are bird activities around the line, the control module controls the bird-blocking module to enter the on state, and the bird-blocking module is started; when there are no flying birds in the field of view, the control module controls the bird-blocking module to enter the off state, and the bird-blocking module is turned off to save electricity; the bird-blocking module emits an ultraviolet light beam, and the ultraviolet light beam is above the overhead line. Birds see the light beam and avoid it by themselves; generally, birds will lift up to avoid obstacles when they encounter obstacles during flight, so the ultraviolet light beam emitted by the bird-blocking module of the present invention is above the overhead line, thus preventing the birds from hitting the overhead line; the present invention prompts the flying birds that there are obstacles in the space by emitting ultraviolet light beams. After discovering them, the birds generally adopt a lifting flight strategy, thereby avoiding the flying birds from hitting the overhead lines (generally hitting the ground wire, and in the case where there is no ground wire, they will also hit the conductor).

[0032] In some embodiments, the AI ​​edge computing device includes a CBAM attention mechanism. In the encoding stage, the AI ​​edge computing device adds the CBAM attention mechanism to perform detailed division of the bird image; in the decoding stage, the AI ​​edge computing device adds the CBAM attention mechanism to further process the bird image. In the encoding stage, CBAM pays attention to the channel and spatial dimensions of the input image, so that the model can focus on the features related to the bird, filter out unnecessary information, and enhance the accuracy of feature representation. The channel attention mechanism of CBAM can effectively identify which features are most important for bird recognition, and automatically adjust the weight of the feature map, thereby improving the efficiency and accuracy of subsequent processing. In the decoding stage, CBAM optimizes the image again in space and channels to improve the quality of the generated image, making the bird features more distinct and facilitating subsequent recognition and classification. CBAM can dynamically adjust the focus area according to the different features of the input image, so that the system can maintain high performance when processing birds of different species or postures. By selectively strengthening the features, CBAM can effectively reduce unnecessary computing resource consumption, optimize the processing speed, and enable the AI ​​edge computing device to operate efficiently under low power consumption. The AI ​​edge computing device can quickly process and recognize images locally, reduce latency, and improve response speed.

[0033] In some embodiments, the control module includes a relay, which is electrically connected to a GPIO pin on an AI edge computing device. When the bird recognition module identifies a bird image in a night vision camera, the AI ​​edge computing device drives the GPIO pin level to rise, drives the relay to turn on, and starts the bird blocking module. The AI ​​edge computing device analyzes the image data instantly through the bird recognition module. Once a bird is identified, it quickly sends a signal to the relay through the GPIO pin to achieve an immediate response and ensure that the bird blocking measures can be started in time. The use of GPIO pins simplifies circuit connections, reduces the need for additional controllers, makes the overall design more compact, reduces power consumption and improves system reliability. The control module can achieve different activation modes through software updates according to different identification situations and bird blocking strategies to meet the needs of different scenarios. The design allows linkage with other monitoring and bird blocking technologies, such as sound devices or flashlights, to form multiple bird blocking means and improve the repelling effect. The control module can integrate a self-test function to monitor the status of the relay and GPIO pins to ensure that the system is promptly alerted when a fault occurs to avoid potential risks. The design can be combined with a user-friendly control interface to allow users to monitor and adjust the working status of the bird blocking module and improve the operability of the system. When the bird recognition module detects a bird image, the system can automatically start the bird blocking module, improve the reaction speed, and drive away potential threats in time. Through relay control, the bird blocking module can be accurately operated, avoiding unnecessary interference and reducing the impact on the environment. The relay consumes less energy when in standby mode, which can effectively extend the overall operation time of the system and is suitable for long-term monitoring and use. Using GPIO pins to directly drive the relay reduces the complexity of circuit design and makes the system easier to integrate and maintain.

[0034] In some embodiments, the bird-blocking module includes a mounting shell 100 and an ultraviolet laser 200, the number of ultraviolet lasers 200 is two, and the two ultraviolet lasers 200 are arranged in the mounting shell 100. The ultraviolet laser 200 can produce light that is not easily adapted to by birds, thereby effectively deterring birds from approaching overhead lines and reducing the risk of accidental collision. Unlike traditional noise or physical barriers, the ultraviolet laser 200 provides a non-invasive method that reduces interference with the surrounding environment and other animals. The two ultraviolet lasers 200 can form cross-irradiation, provide a wider coverage range, and ensure that birds in multiple directions can be deterred. The output intensity and working mode of the laser can be adjusted according to the environment and bird behavior to improve the flexibility of the effect. The ultraviolet laser 200 usually has a long service life and low maintenance requirements, reducing subsequent maintenance costs. The mounting shell 100 can integrate the lasers together for easy installation and maintenance, while providing certain protection to ensure the reliability of the equipment. The two ultraviolet lasers 200 can be designed to emit beams in different directions to form a more comprehensive protective net and increase the coverage of surrounding birds. Combined with sensors and AI technology, the UV laser 200 can intelligently adjust its working state according to real-time monitoring data, such as automatically increasing the laser intensity when birds approach, providing a stronger deterrent effect. The UV laser 200 does not generate noise or chemicals, reduces the impact on the ecological environment, and conforms to modern environmental protection concepts. The mounting shell 100 can prevent the laser beam from accidentally irradiating people or other animals, ensure the safety of the equipment, and reduce the risk of accidental injury.

[0035] An intelligent method for preventing accidental bird collisions on overhead lines, the method for preventing accidental bird collisions is applicable to an anti-accidental bird collision device, and the specific steps of the method for preventing accidental bird collisions include: S1. Install the camera module at a preset position of the transmission tower to collect video streams of bird activities around the transmission line within the visible range; S2, the bird recognition module obtains a video stream of bird activities around the power transmission line within the visible range, intercepts the video stream into a number of pictures to be recognized by frame, and recognizes whether there is a bird image in the pictures to be recognized; when a bird image is recognized, a first control signal is sent; when no bird image is recognized, a second control signal is sent; S3, the control module receives the control signal of the bird identification module, and when the control signal is the first control signal, the bird blocking module is turned on; when the control signal is the second control signal, the bird blocking module is turned off; the bird blocking module is used to emit an ultraviolet light beam to block birds when it is turned on.

[0036] Specifically, during the peak period of accidental bird collisions, the system device is turned on at dawn or dusk. First, the bird identification module takes pictures of the birds in the transmission tower. When a bird flies over, the relay is driven by the GPIO port of the board to turn on the bird blocking module, emitting an ultraviolet light column, and building a bird blocking light column between adjacent towers. The light column is level with the upper ground wire bracket, and the height is the position of the ground wire. When the birds see the light column, they will avoid it by themselves, thereby preventing the birds from hitting the line. During the operation of the device, the situation of birds in the transmission tower is continuously identified. When it is found that there are no birds in the transmission tower, the relay is driven to close, and the bird blocking module is automatically closed to save power. The power supply of the device is composed of solar energy, battery management module and large-capacity battery, which provide a stable power supply for the device.

[0037] Based on the DeepLabV3+ architecture, the CBAM (Convolutional Block Attention Module) attention mechanism is integrated to optimize performance. In the encoding stage, after the dilated convolution is processed, the primary feature layer in the deep convolutional neural network (DCNN) is transformed using 1×1 convolution. Subsequently, through the feature fusion step, the processed features are input into the CBAM attention mechanism for detailed feature weighting operations. This process aims to effectively integrate multi-scale information and extract richer and higher-order feature representations. By integrating the CBAM attention mechanism, the recognition accuracy of flying birds is improved.

[0038] Jetson nano is used as the core module of bird recognition hardware, and ONNX technology is used to deploy the bird recognition algorithm model. Due to limited embedded hardware resources, the lightweight improved DeepLabV3+ algorithm model is used for deployment. After the device is running, the camera video stream is continuously obtained to identify in real time whether there are birds in the picture. When a bird is found, the bird blocking module is turned on to form a flash column to block the bird. The bird blocking module emits a UV beam from the installation tower to the opposite tower through a UV searchlight. The 380nm-395nm wavelength UV light, which birds are most sensitive to, can improve the birds' ability to identify the light beam.

[0039] The bird-blocking module adopts a simple structure, with two ultraviolet laser light sources installed inside the mounting shell 100, and the overall waterproof level is IP65. The internal relay is connected to the GPIO pin on the Jetson nano of the camera recognition module. When the camera finds a flying bird, the Jetson nano drives the GPIO pin level to pull up, driving the relay to turn on, powering the bird-blocking module and emitting a beam of light. Because subdividing bird species requires a lot of computing resources and the accuracy is reduced, in order to better protect birds from accidentally hitting the line, non-differentiated bird prevention should be carried out, which can save computing power.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An intelligent bird collision prevention device for overhead lines, used to prevent birds from accidentally colliding with overhead lines, characterized in that: include: A camera module is arranged at a preset position of the transmission tower and is used to collect a video stream of bird activities around the transmission line within a visible range; The bird recognition module is used to obtain a video stream of bird activities around the power transmission line within the visible range from the camera module, intercept the video stream into a number of pictures to be recognized by frame, and recognize whether there is a bird image in the pictures to be recognized; when a bird image is recognized, a first control signal is sent to the control module; when no bird image is recognized, a second control signal is sent to the control module; A control module, used to receive a control signal from the bird identification module, and when the control signal is a first control signal, control the bird prevention module to enter an on state; when the control signal is a second control signal, control the bird prevention module to enter an off state; Bird blocking module, used to emit a UV beam to block birds when turned on.

2. The intelligent bird collision prevention device for overhead lines according to claim 1, characterized in that: The camera module includes a night vision camera and a gun, wherein the night vision camera is arranged inside the gun, and the gun is fixedly arranged at the tower head cross arm or tower body position of the transmission tower. The night vision camera shoots toward the adjacent tower, and the night vision camera is powered by a solar panel.

3. The intelligent bird collision prevention device for overhead lines according to claim 1, characterized in that: The bird identification module includes an AI edge computing device, which is electrically connected to the control module and is used to identify whether there is a bird image in the picture to be identified.

4. The intelligent bird collision prevention device for overhead lines according to claim 3, characterized in that: An improved DeepLabV3+ model is adopted in the AI ​​edge computing device; the improved DeepLabV3+ model is a model in which a first convolutional block attention module and a second convolutional block attention module are added to the DeepLabV3+ model; the input of the first convolutional block attention module is connected to the output of the atrous spatial convolution pooling pyramid ASPP in the encoder, and the output of the first convolutional block attention module is connected to the 1×1 convolution module output in the encoder; the input of the second convolutional block attention module is connected to the output of the first feature fusion module in the decoder, the input of the output of the first feature fusion module is connected to the output of the backbone network in the encoder, and the output of the second convolutional block attention module is connected to the input of the interpolation upsampling module.

5. The intelligent bird collision prevention device for overhead lines according to claim 4, characterized in that: The control module includes a relay, which is electrically connected to the GPIO pin on the AI ​​edge computing device. When the bird identification module recognizes that there is a bird image in the picture to be identified, the AI ​​edge computing device drives the GPIO pin level to rise, drives the relay to turn on, and starts the bird blocking module.

6. The intelligent bird collision prevention device for overhead lines according to claim 1, characterized in that: The bird-blocking module comprises a mounting shell and an ultraviolet laser. The number of the ultraviolet lasers is two, and the two ultraviolet lasers are arranged in the mounting shell.

7. The intelligent bird collision prevention device for overhead lines according to claim 6, characterized in that: The wavelength of the laser emitted by the ultraviolet laser is 380mm-395mm.

8. An intelligent method for preventing accidental bird collisions on overhead lines, characterized in that: include: Collect video streams of bird activities around power lines within visual range; Obtain a video stream of bird activities around a power transmission line within a visible range, intercept the video stream into a number of pictures to be identified by frame, and identify whether there is a bird image in the pictures to be identified; send a first control signal when a bird image is identified; send a second control signal when no bird image is identified; When a first control signal is received, the bird-blocking module is turned on; when a second control signal is received, the bird-blocking module is turned off; wherein the bird-blocking module is used to emit an ultraviolet light beam to block birds when in the turned-on state.

9. The intelligent method for preventing accidental bird collisions on overhead lines according to claim 8, characterized in that: The steps of obtaining a video stream of bird activities around a power transmission line within a visible range, intercepting the video stream into a plurality of pictures to be identified by frame, and identifying whether there are bird images in the pictures to be identified; issuing a first control signal when a bird image is identified; and issuing a second control signal when no bird image is identified are implemented by a bird identification module; The bird recognition module includes an AI edge computing device, which is electrically connected to the control module and is used to identify whether there is a bird image in the picture to be identified; an improved DeepLabV3+ model is adopted in the AI ​​edge computing device; the improved DeepLabV3+ model is a model in which a first convolutional block attention module and a second convolutional block attention module are added to the DeepLabV3+ model; the input of the first convolutional block attention module is connected to the output of the atrous spatial convolution pooling pyramid ASPP in the encoder, and the output of the first convolutional block attention module is connected to the 1×1 convolution module output in the encoder; the input of the second convolutional block attention module is connected to the output of the first feature fusion module in the decoder, the input of the output of the first feature fusion module is connected to the output of the backbone network in the encoder, and the output of the second convolutional block attention module is connected to the input of the interpolation upsampling module.

10. The intelligent method for preventing accidental bird collisions on overhead lines according to claim 9, characterized in that: When the first control signal is received, the bird blocking module is turned on; when the second control signal is received, the step of turning off the bird blocking module is implemented by the control module; The control module includes a relay, which is electrically connected to the GPIO pin on the AI ​​edge computing device. When the bird identification module recognizes that there is a bird image in the picture to be identified, the AI ​​edge computing device drives the GPIO pin level to rise, drives the relay to turn on, and starts the bird blocking module.

Images