Self-adaptive bird repelling method and bird repelling rod mechanism for electric power tower area

Abstract

The invention belongs to the technical field of power transmission equipment maintenance, and particularly discloses a bird self-adaptive expelling method and a bird expelling rod mechanism for an electric power tower area, and the method comprises the steps: arranging terminal devices with image collection and expelling execution functions at a plurality of electric power towers, and recognizing the types and body types of birds through the image data collected by the terminals; the identification result is sent to the upper computer; the upper computer calls an expelling parameter combination according to the identification information, generates a control instruction and issues the control instruction to a corresponding terminal; the terminal adopts at least one preset expelling mode to execute expelling operation; after expelling is completed, recording parameters and expelling data and returning the parameters and the expelling data to the upper computer; and the upper computer disturbs at least one feature in the expelling parameters based on the recorded data to generate a new parameter combination, and the new parameter combination is applied when birds of the same type and body form appear again. According to the scheme, dynamic updating and continuous optimization of the repelling strategy can be realized, and the method is suitable for bird repelling management in a multi-tower scene.

Description

Technical Field

[0001] This invention belongs to the field of power transmission equipment maintenance technology, specifically relating to an adaptive bird deterrence method and bird deterrent stick mechanism for power pole areas. Background Technology

[0002] In the existing operating environment of power transmission lines, the behavior of birds perching, nesting, and moving on power poles remains one of the main external factors affecting the safe and stable operation of the lines. In particular, large birds of prey, due to their large size and wide range of activity, are more likely to cause phase-to-phase short circuits, tripping, and other faults by spreading their wings, defecating, or carrying nesting materials, seriously threatening the operational reliability of the power system.

[0003] To address the aforementioned problems, various bird-repelling devices and control methods have been proposed, including physical structures such as bird-proof windmills, bird-proof covers, and bird-proof cutting boards, as well as terminal devices employing sound, light, and electricity to deter birds. These methods often incorporate bird image recognition algorithms to trigger corresponding repelling actions upon identifying a specific bird species. Some solutions also incorporate databases and control strategy libraries to match bird-repelling strategies based on bird species or the time of their appearance. However, analysis reveals the following key issues prevalent in existing technologies: In most current bird control systems, physical structures provide some protection against birds, but bird-proof windmills, bird covers, and bird-proof cutting boards still experience circuit breaker tripping, making it impossible to completely prevent bird-related tripping. For terminal devices using sound, light, or electricity to deter birds, the repulsion methods set for different bird species are preset and fixed, not dynamically changing with actual results. Although the system can record identification results and execute commands, there is a lack of effective feedback mechanisms between bird control parameters and repulsion effectiveness. Furthermore, when multiple terminal devices are deployed in an area, the system has not yet established a globally unified scheduling-based linkage response strategy, preventing the sharing or migration of bird control strategies and resulting in isolated operation of each terminal.

[0004] The aforementioned defects lead to a core problem: when the same species of birds repeatedly perch on the same pole or adjacent poles, repeatedly using the same deterrence parameters will cause the birds to develop an adaptive response to bird deterrence behavior, resulting in a decrease in deterrence efficiency, and in severe cases, even complete failure. Summary of the Invention

[0005] This invention addresses the problems in the prior art by providing an adaptive bird deterrence method and bird deterrence stick mechanism for power pole areas. It solves the problem that when the same species of birds repeatedly stay on the same pole or adjacent poles, the repeated use of the same deterrence parameters will cause the birds to develop an adaptive response to the bird deterrence behavior, resulting in a decrease in deterrence efficiency, or even complete failure in severe cases.

[0006] The technical solution adopted in this invention is as follows: In a first aspect, this application provides an adaptive bird repulsion method for power pole areas, the method comprising the following steps: Step S1: Install terminal devices with image acquisition and expulsion execution functions at multiple power poles; Step S2: Collect image data through the terminal device, identify the species and size of the birds, and send the identification results to the host computer; Step S3: The host computer retrieves the corresponding deportation parameter combination based on the identified bird species and size, generates control commands, and sends them to the corresponding terminal devices. Step S4: The corresponding terminal device, according to the control command, uses at least one of the preset expulsion methods to expel the target birds; Step S5: After the expulsion is completed, record the expulsion parameters and expulsion data, and send the data back to the host computer; Step S6: The host computer analyzes the expulsion parameters and expulsion data recorded in step S5, and perturbs them based on at least one feature of the expulsion parameters to generate a new combination of expulsion parameters. Step S7: When a bird of the same species and size as described above is identified again, a new combination of expulsion parameters is used to expel it, and steps S5 and S6 are repeated.

[0007] Furthermore, step S1 also includes: constructing a spatial coordinate system in the host computer based on the latitude and longitude position and installation height information of each power pole, and collecting the position parameters and number information of each terminal device to generate a three-dimensional environment model containing the pole number, terminal position information and image monitoring coverage. Image recognition data and expulsion record data from the terminal device are overlaid in the three-dimensional environment model to form a bird species distribution layer and an expulsion parameter history layer arranged in chronological order.

[0008] Furthermore, in step S2, identifying the species and size of the bird includes the following steps: Step 2-1: Perform resolution equalization and noise filtering on the acquired images; Step 2-2: Extract the contour, edge curvature, symmetry, and color distribution features of the suspected bird region in the image based on the convolutional neural network model; Steps 2-3: Calculate the similarity between the extracted feature vectors and the feature templates contained in the preset bird species database; Steps 2-4: Select the corresponding bird species and size labels based on the similarity score, and output the recognition results with confidence values.

[0009] Furthermore, the bird contour features extracted in step S2-2 include the rate of change of the curvature of the boundary shape in the wing-spreading state, the geometric ratio between the beak and the trunk, the bifurcation angle and length ratio of the tail feather region in a single frame image, and the deformation dynamic trajectory vector group of multiple consecutive frames within the sampling period.

[0010] Furthermore, in step S3, the deportation parameter combination is determined by the host computer from multiple deportation strategy tables, which are indexed according to four dimensions: bird species, body size, historical response behavior, and environmental conditions. Each set of expulsion parameters includes: expulsion method type, action trigger time, action execution duration, execution frequency, action intensity level, and action direction control parameters, and is uniquely identified by multidimensional index encoding.

[0011] Furthermore, in step S5, the recorded expulsion data includes: The currently executed expulsion parameter combination number and the value of each parameter; The time delay between identifying the target bird and executing the expulsion action; The time taken from the execution of the expulsion action until the target birds leave; Changes in the coordinates of the birds' resting positions in the image frame sequence during the expulsion process; The data is associated with and encoded using timestamps and tower numbers, and then stored in the host computer database.

[0012] Furthermore, in step S6, a perturbation step size Δx is applied to one of the parameters in the executed expulsion parameter combination to form a new parameter combination; The perturbation step size Δx is an item in the perturbation sequence of the preset single-point perturbation sequence table, and the perturbation direction is determined by the trend of the expulsion time recorded after the previous round of perturbation. The perturbation execution adopts a rotating perturbation strategy, that is, only one parameter is selected for perturbation in each round among multiple parameters, and the perturbation process follows a single-point perturbation sequence list; The methods for determining the direction of the disturbance include: Calculate the difference between the time taken for the current round of expulsion and the time taken for the previous round of expulsion; When the difference is negative and greater than the set threshold θ, the perturbation direction remains unchanged; When the difference is positive and greater than the set threshold θ, switch the perturbation direction; When the absolute value of the difference is less than the set threshold θ, the current parameter remains unchanged and the disturbance is rotated to the next parameter.

[0013] Secondly, this application provides a bird deterrent mechanism, including a drive unit, a bird deterrent stick fixedly connected at one end to the drive unit, a solar panel attached to the surface of the bird deterrent stick, and a battery fixedly mounted on the drive unit, the battery being electrically connected to the drive unit and the solar panel respectively.

[0014] Furthermore, the bird deterrent stick is made of carbon fiber.

[0015] As can be seen from the above technical solutions, the advantages of the present invention are: By introducing a parameter perturbation feedback mechanism, this invention achieves dynamic adjustment and gradual optimization of the bird control strategy, overcoming the problems of increased bird adaptability and gradually declining control effectiveness caused by fixed control methods and lack of feedback adjustment mechanisms in existing technologies. Unlike traditional systems that use static parameters for bird control, this method records control parameters and corresponding control duration data after each control operation. It uses the trend of control time changes to determine the direction of perturbation and applies perturbations to each parameter in a rotating manner, thereby exploring better control combinations under different feature dimensions and forming a continuously iterative and self-updating bird control parameter evolution path. This perturbation strategy, through precise control of control parameters such as sound intensity, frequency, laser intensity, and action trigger delay, combined with a unified management linkage mechanism by a host computer within the region, enables rapid strategy selection and iterative optimization for specific bird species and sizes in multi-tower deployment scenarios. Furthermore, this invention constructs a three-dimensional environment model in the host computer for spatial archiving of terminal device deployment information and historical identification of bird control behavior, which helps improve the overall data integration, analysis, and strategy tracing capabilities of the system, providing a highly efficient and scalable intelligent bird control solution for power systems. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating an adaptive bird repulsion method for power pole areas, as described in an embodiment of the present invention. Figure 2 This is a schematic diagram of the bird deterrent mechanism in an embodiment of the present invention.

[0018] In the picture: 1. Servo motor; 2. Bird deterrent stick; 3. Battery. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, the present invention provides an adaptive bird repulsion method for power pole areas, comprising the following steps: Step S1: Install terminal devices with image acquisition and expulsion execution functions at multiple power poles; The terminal devices are installed on the top of the power pole or at the crossarm location, ensuring that the image acquisition field of view covers the area surrounding the suspended insulator strings and metal components. Each terminal device includes a visible light camera, an infrared sensor, a bird deterrent mechanism, a laser module, a voice controller, and a micro-spray assembly. The terminal devices access the public network via LTE or NB-IoT communication modules and periodically upload their operating status and image data to the host computer. Each terminal device has a unique identification number, marked on the pole structure, for easy maintenance and data tracking in the future.

[0021] A spatial coordinate system is constructed within the host computer based on the latitude and longitude positions and installation height information of each power pole, and the position parameters and number information of each terminal device are collected to generate a three-dimensional environment model containing pole number, terminal position information and image monitoring coverage. The 3D environment model employs a GIS coordinate projection-based modeling method, using the location of the tower as an anchor point. It maps the structural height and terminal installation height to an image monitoring cone-shaped coverage area, establishing the relative spatial relationships between tower groups. In the model, each terminal device is presented as a 3D node, accompanied by real-time status parameters and coverage view boundaries. The system supports heatmap overlay functionality to display the distribution of bird occurrence frequency and expulsion trigger counts in various areas over a recent period.

[0022] Image recognition data and expulsion record data from the terminal device are overlaid in the three-dimensional environment model to form a bird species distribution layer and an expulsion parameter history layer arranged in chronological order. The bird species distribution layer indexes the bird species and size annotations identified at different time points in a timeline manner, forming a spatial-temporal two-dimensional annotation matrix combined with terminal location information. The expulsion parameter history layer records the parameter combinations, expulsion duration, bird reaction time, and other information corresponding to each expulsion operation, and allows for the retrospective analysis of the historical response behavior of any terminal. Users can select the time window and area range through the host computer interface to access the layers for interactive analysis and visualization, providing a basis for strategy evolution and system debugging.

[0023] Step S2: Collect image data through the terminal device, identify the species and size of the birds, and send the identification results to the host computer; The terminal device has a built-in image acquisition unit and an embedded edge computing module. After local processing, the image data generates structured recognition results, including the bird's genus and species number, size classification label (e.g., small, medium, large), confidence score, and timestamp. The recognition results are sent to the host computer via the MQTT protocol and stored according to the pole number and image sequence index for use in dispersal strategy scheduling and subsequent feature perturbation analysis.

[0024] Identifying bird species and size involves the following steps: Step 2-1: Perform resolution equalization and noise filtering on the acquired images; Specifically, this includes applying bilateral filtering and histogram equalization techniques to the image to preserve edge details while suppressing background noise, thereby improving the accuracy of subsequent model recognition. The image preprocessing module is implemented based on the OpenCV library and includes gamma correction for low-light images at night to enhance the clarity of target contours.

[0025] Step 2-2: Extract the contour, edge curvature, symmetry, and color distribution features of the suspected bird region in the image based on the convolutional neural network model; The convolutional neural network employs transfer learning, performing targeted training on top of ImageNet, and outputting multi-scale feature maps. During the feature extraction stage, suspected bird-like regions are located and candidate regions of interest (ROIs) are formed, while multi-dimensional feature vectors are generated for subsequent matching and recognition.

[0026] The extracted bird contour features include the rate of change of the curvature of the boundary shape in the wing-spreading state, the geometric ratio between the beak and the trunk, the bifurcation angle and length ratio of the tail feather region in a single frame image, and the deformation dynamic trajectory vector group of multiple consecutive frames within the sampling period. The rate of change of boundary curvature is calculated by the difference method to determine the change value of the tangent angle of the edge curve. The geometric proportion is measured by principal component analysis to determine the Euclidean distance ratio between key points of bird features. The tail feather bifurcation angle is extracted by morphological segmentation. The deformation dynamic trajectory vector group is combined with optical flow to model the change of target contour in continuous frames and stored as a time series tensor structure to enhance the stability of the recognition of dynamic bird state.

[0027] Steps 2-3: Calculate the similarity between the extracted feature vectors and the feature templates contained in the preset bird species database; The feature template database contains 100 common bird templates, each template including vector forms of contour features, color histograms, and deformation features. Similarity calculation uses a weighted cosine similarity algorithm, with weights for color, geometry, and dynamic features set to 0.3, 0.4, and 0.3 respectively, and a scoring range of 0 to 1.

[0028] Steps 2-4: Select the corresponding bird species and size labels based on the similarity score, and output the recognition results with confidence values; When the similarity score exceeds a set threshold (e.g., 0.85), the category and body type labels are output. The confidence score is obtained by weighted fusion of the model's predicted probability value and the template similarity. The recognition results are packaged into a JSON format structure, containing label number, confidence score, timestamp, and image index, for use in expulsion strategy generation and data statistical analysis.

[0029] Step S3: The host computer retrieves the corresponding deportation parameter combination based on the identified bird species and size, generates control commands, and sends them to the corresponding terminal devices. The host computer is equipped with a strategy scheduling module, which has functions for parameter combination query, strategy generation, instruction encapsulation, and task distribution. Once the identification result is received, the system automatically reads the species code and size level, constructs a four-dimensional index key as the strategy retrieval condition, calls the expulsion parameter combination matching the current target bird, and parses and encapsulates it into a structured control instruction. In addition to the parameter combination number and the value of each parameter, the instruction also includes a unique terminal number and execution timestamp to ensure the timeliness and accuracy of control scheduling. The instruction is published to the terminal's communication channel via MQTT, and AES encryption is used during transmission to ensure data security.

[0030] The combination of expulsion parameters is determined by the host computer by retrieving from multiple expulsion strategy tables. The strategy tables are indexed according to four dimensions: bird species, body size, historical response behavior, and environmental conditions. The strategy table employs a multi-layered hash structure, with bird species as the primary index and size class as the secondary index. Historical response behavior (such as the success or failure of the last expulsion and its duration) and environmental conditions (such as temperature, light intensity, and wind speed) constitute auxiliary filtering conditions. In the historical behavior dimension, the host computer records the average response time and parameter combination number of the most recent N expulsions, and uses historical feedback coefficients to weight and score the priority of parameter selection. Environmental condition data is synchronized in real time by the regional meteorological service interface and updated every 10 minutes to improve the adaptability of the expulsion strategy.

[0031] Each set of expulsion parameters includes: expulsion method type, action trigger time, action execution duration, execution frequency, action intensity level, and action direction control parameters, and is uniquely identified by multidimensional index encoding; The expulsion method type is represented by an enumerated encoding, covering laser irradiation, audio playback, spraying chemicals, mechanical oscillation, etc. The action trigger time represents the delay from the completion of recognition to the initiation of the expulsion action, in milliseconds, used to coordinate with the image frame detection rhythm to set the execution pace. The action execution duration defines the maximum execution time limit for a single expulsion action; the execution frequency applies to periodically driven devices (such as sound waves or laser flashes); the action intensity level is refined to 3-5 levels (e.g., sound pressure level divided into 50 / 60 / 70 / 80dB); the action direction control parameter is the angle or vector information used to control the oscillation, laser scanning, or spray nozzle rotation angle. Each parameter combination is assigned a unique combination code ID, facilitating rapid indexing in the database, version management, and historical tracing during strategy evolution.

[0032] Step S4: The corresponding terminal device, according to the control command, uses at least one of the preset expulsion methods to expel the target birds; After receiving control commands from the host computer, the terminal device parses the deportation parameter combination number and action execution parameters, activates the corresponding components according to the deportation method specified in the command, and begins the action execution process. If the command specifies a combination method, the terminal activates multiple devices sequentially or in parallel according to a set order to execute the deportation task. During the deportation action, the terminal simultaneously activates the image acquisition function to record the dynamic response trajectory of the target bird. All executed actions and response times are packaged into an execution record and reported to the host computer database.

[0033] Methods of expulsion include: The mechanical swinging motion is performed by the bird deterrent mechanism, such as... Figure 2 As shown, the bird deterrent mechanism includes a drive unit, which is a servo motor or a servo motor 1. In this embodiment, a servo motor 1 is used. The bird deterrent mechanism also includes a bird deterrent stick 2, one end of which is fixedly connected to the output shaft of the servo motor 1. A solar panel is attached to the surface of the bird deterrent stick 2. The bird deterrent mechanism also includes a battery 3 fixedly mounted on the servo motor 1. The battery 3 is electrically connected to the servo motor 1 and the solar panel respectively. The bird deterrent stick 2 is made of carbon fiber.

[0034] The expulsion audio signal is played through a speaker. The frequency range of the audio signal is 100Hz~800Hz, and the sound pressure level range is 50dB~85dB. The laser emits a visible laser beam with a wavelength of 520nm~660nm and a laser intensity of 50mW~300mW. Bird repellent is sprayed using a spraying device, which includes a drug storage chamber, a micro electric pump, and a nozzle assembly. The spraying pressure ranges from 0.05 MPa to 0.2 MPa, and the volume of a single spray is 10 mL to 50 mL. Step S5: After the expulsion is completed, record the expulsion parameters and expulsion data, and send the data back to the host computer; The terminal device has a built-in log recording module that archives all data from the expulsion operation immediately after it concludes. All data is cached locally in structured data packets and synchronized to the host computer via a wireless network. The data transmission process employs data compression and verification mechanisms to ensure the integrity and real-time performance of remote transmission, preventing missed reports or retransmissions. Upon receiving the data, the host computer parses and archives it, then correlates it with previously identified records at a timeline level to support subsequent parameter perturbation analysis and strategy optimization.

[0035] The recorded eviction data includes: The currently executed expulsion parameter combination number and the value of each parameter; This number serves as a unique key pointing to a specific combination of parameters in the expulsion strategy table. Parameter values ​​include expulsion method type, action trigger time (in milliseconds), action duration (in milliseconds), action intensity level (e.g., laser intensity 200mW, sound pressure level 65dB, etc.), and directional control angle. Before execution begins, the terminal device automatically formats and records the parameters within the instruction in a structured manner.

[0036] The time delay between identifying the target bird and executing the expulsion action; This delay parameter reflects the terminal's response efficiency and the stability of the synchronization timing. It is calculated as: the expulsion start timestamp minus the target recognition completion timestamp, in milliseconds (ms). This data can be used to subsequently infer the coupling relationship between image recognition processing capabilities, communication latency, and execution mechanisms.

[0037] The time taken from the execution of the expulsion action until the target birds leave; This data is obtained by time-annotating the image frame sequence, specifically the time span from the frame initiating the action to the last frame where the target bird no longer appears in the monitored image. It is used to quantify the expulsion efficiency and is a key indicator for subsequent disturbance feedback evaluation. If the detected object re-enters the field of view in the image, it will be considered a new event to ensure data consistency.

[0038] Changes in the coordinates of the birds' resting positions in the image frame sequence during the expulsion process; The center coordinates (x, y) of the bird in each frame are extracted using an inter-frame target detection algorithm, and its relative displacement trajectory within the terminal's field of view is recorded. This data is used to analyze the bird's dynamic response to the current dispersal method, such as whether it flies away in a straight line, lingers for a short time, or attempts to return. The stopping path can be generated into a two-dimensional heat map and mapped onto a three-dimensional tower environment model for spatial response analysis.

[0039] The data is associated with and encoded using timestamps and tower numbers, and then stored in the host computer database; All expulsion data is indexed using a unique "tower number + timestamp" format for categorized storage and horizontal comparative analysis. The host computer database uses a time-series database structure for archiving, featuring fast query response, strong time-series analysis capabilities, and supports subsequent historical expulsion trajectory backtracking, statistical summarization, and parameter trend analysis.

[0040] Step S6: The host computer analyzes the expulsion parameters and expulsion data recorded in step S5, and perturbs them based on at least one feature of the expulsion parameters to generate a new combination of expulsion parameters. The analysis process is completed by a parameter perturbation analysis module deployed in the host computer. Based on reported historical bird flight data, this module constructs corresponding bird flight parameter response profiles according to the target bird species and size. The analysis module uses the flight time (from the start of the action to the bird's departure) as the primary evaluation indicator, and combines it with the fluctuation characteristics of the stop path coordinates for dynamic scoring. If the flight time does not meet the set response time threshold Tmax, the system determines that the current parameter combination needs optimization and triggers the perturbation strategy process. The new combination is formed by adjusting the original parameters according to the perturbation logic and is registered in the combination version table, forming a complete strategy evolution path.

[0041] Apply a perturbation step size Δx to one of the parameters in the already executed expulsion parameter combination to form a new parameter combination; The parameter perturbation mechanism employs a dimension-by-dimensional perturbation strategy. Each perturbation modifies only one parameter, while the remaining parameters remain unchanged, thus clearly identifying the impact of single variables on the expulsion effect. Δx is the perturbation step size; for example, the perturbation for laser illumination intensity is Δx = ±10mW, for audio sound pressure level it is Δx = ±3dB, and for spray pressure it is Δx = ±0.02MPa. The modified parameter values ​​form new parameter combinations, which the system marks as "derived versions," and records the source combination number as the parent node, forming a tree-like strategy evolution structure.

[0042] The perturbation step size Δx is an item in the perturbation sequence of the preset single-point perturbation sequence table, and the perturbation direction is determined by the trend of the expulsion time recorded after the previous round of perturbation. The single-point perturbation sequence table is defined in array form, with multiple standard perturbation step sizes set for each type of parameter. For example, the perturbation sequence for sound pressure level is [±1dB, ±3dB, ±5dB], and for laser intensity it is [±10mW, ±30mW]. The system automatically retrieves the position of the perturbation sequence to which the parameter belongs in the current round based on the combination identifier. The perturbation direction is recorded using dual-state bits (positive / negative), and whether to reverse it is determined based on the trend of expulsion efficiency change after the previous perturbation.

[0043] The perturbation execution adopts a rotating perturbation strategy, that is, only one parameter is selected for perturbation in each round among multiple parameters, and the perturbation process follows a single-point perturbation sequence list; The parameters are perturbed cyclically in a predetermined order. Each optimization process affects only one parameter in the current round, while the others remain unchanged to achieve independence analysis between variables. After each round of perturbation, the system updates the perturbation pointer, moves to the next parameter dimension, and records the perturbation history to prevent repeated perturbations.

[0044] The methods for determining the direction of the disturbance include: Calculate the difference between the time taken for the current round of expulsion and the time taken for the previous round of expulsion; The system uses t_current to represent the expulsion time of the current round and t_previous to represent the expulsion time of the previous round with the same parameter combination. The difference Δt = t_current - t_previous is calculated, and the effectiveness of the disturbance is judged by combining the direction of change and the absolute value.

[0045] When the difference is negative and greater than the set threshold θ, the perturbation direction remains unchanged; For example, if θ is set to 0.5 seconds, and Δt = -0.8 seconds, it means that the expulsion time is shortened and the disturbance is effective in the positive direction. In this case, the current disturbance direction will continue to act on the current parameter.

[0046] When the difference is positive and greater than the set threshold θ, switch the perturbation direction; If Δt = +1.0 seconds, the current perturbation direction may lead to a worse expulsion effect, and the parameter direction needs to be adjusted in the opposite direction, that is, from increasing to decreasing, or vice versa.

[0047] When the absolute value of the difference is less than the set threshold θ, keep the current parameter unchanged and rotate to the next parameter for perturbation; For example, if Δt = +0.2 seconds, and this value is lower than θ, the system considers the disturbance insufficient and not a basis for strategy evolution. In this case, the current parameters are frozen, and the disturbance process continues in the next parameter dimension to ensure that the optimization process covers all key expulsion feature variables.

[0048] Step S7: When a bird of the same species and size as described above is identified again, a new combination of expulsion parameters is used to expel it, and steps S5 and S6 are repeated.

[0049] In some embodiments, this application provides an adaptive bird repulsion system for power pole areas, the system comprising: Multiple terminal devices are installed at multiple power poles. Each terminal device includes an image acquisition component, an execution component, a control circuit, and a communication module. The image acquisition component is connected to the control circuit and is used to acquire image data; the execution component is connected to the control circuit and executes the driving action driven by the control circuit; the communication module is connected to the control circuit and is used to send identification data and receive control commands. The host computer establishes communication connections with multiple terminal devices. The host computer includes a processor, communication interfaces, and a database module. The processor is configured as follows: Receive bird identification results and expulsion data from various terminal devices; Retrieve the expulsion parameter combination corresponding to the recognition result; Apply a perturbation step size Δx to a selected parameter in the expulsion parameters based on the expulsion time variation in the expulsion data; A new combination of expulsion parameters is generated and sent to the target terminal device via the communication interface; The database module contains the following information structures: Multiple sets of expulsion parameter strategy tables, each corresponding to different species and sizes of birds; Single-point perturbation sequence table, defining the perturbation order and perturbation step size set for each parameter; Historical eviction record data table, indexed by tower number and timestamp, showing the relationship between eviction parameters and eviction time; The host computer further includes a model structure for establishing a three-dimensional coordinate system based on the spatial location of the power poles. The identification results of the terminal device and the information on the expulsion behavior are superimposed on the coordinate system to form a distribution layer and a recording layer.

[0050] In some embodiments, this application provides a terminal, including: Memory for storing bird adaptive expulsion programs for power pole areas; A processor is configured to implement the steps of the adaptive bird expulsion method for power pole areas when executing the adaptive bird expulsion system for power pole areas.

[0051] In some embodiments, this application provides a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the bird adaptive repulsion method for power pole areas.

[0052] It is understood that the systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer, which can be a personal computer, a laptop computer, a personal digital assistant, a tablet computer, a wearable device, or any combination of these devices.

[0053] In a typical configuration, a computer includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0054] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0055] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined in this embodiment, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0056] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0057] It should be understood that although the terms first, second, third, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of one or more embodiments of this specification, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "in response to a determination," or "when," or "in the event of a determination."

[0058] The above description is merely a preferred embodiment of one or more embodiments of this specification and is not intended to limit the scope of one or more embodiments of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of this specification should be included within the protection scope of one or more embodiments of this specification.

Claims

1. An adaptive bird repulsion method for power pole areas, characterized in that, Includes the following steps: Step S1: Install terminal devices with image acquisition and expulsion execution functions at multiple power poles; Step S2: Collect image data through the terminal device, identify the species and size of the birds, and send the identification results to the host computer; Step S3: The host computer retrieves the corresponding deportation parameter combination based on the identified bird species and size, generates control commands, and sends them to the corresponding terminal devices. Step S4: The corresponding terminal device, according to the control command, uses at least one of the preset expulsion methods to expel the target birds; Step S5: After the expulsion is completed, record the expulsion parameters and expulsion data, and send the data back to the host computer; Step S6: The host computer analyzes the expulsion parameters and expulsion data recorded in step S5, and perturbs them based on at least one feature of the expulsion parameters to generate a new combination of expulsion parameters. Step S7: When a bird of the same species and size as described above is identified again, a new combination of expulsion parameters is used to expel it, and steps S5 and S6 are repeated.

2. The adaptive bird repulsion method for power pole areas according to claim 1, characterized in that, Step S1 also includes: constructing a spatial coordinate system in the host computer based on the latitude and longitude position and installation height information of each power pole, and collecting the position parameters and number information of each terminal device to generate a three-dimensional environment model containing the pole number, terminal position information and image monitoring coverage. Image recognition data and expulsion record data from the terminal device are overlaid in the three-dimensional environment model to form a bird species distribution layer and an expulsion parameter history layer arranged in chronological order.

3. The adaptive bird repulsion method for power pole areas according to claim 1, characterized in that, In step S2, identifying the species and size of the bird includes the following steps: Step 2-1: Perform resolution equalization and noise filtering on the acquired images; Step 2-2: Extract the contour, edge curvature, symmetry, and color distribution features of the suspected bird region in the image based on the convolutional neural network model; Steps 2-3: Calculate the similarity between the extracted feature vectors and the feature templates contained in the preset bird species database; Steps 2-4: Select the corresponding bird species and size labels based on the similarity score, and output the recognition results with confidence values.

4. The adaptive bird repulsion method for power pole areas according to claim 3, characterized in that, The bird outline features extracted in step S2-2 include the rate of change of the curvature of the boundary shape in the wing-spreading state, the geometric ratio between the beak and the trunk, the bifurcation angle and length ratio of the tail feather region in a single frame image, and the deformation dynamic trajectory vector group of multiple consecutive frames within the sampling period.

5. The adaptive bird repulsion method for power pole areas according to claim 1, characterized in that, In step S3, the expulsion parameter combination is determined by the host computer from multiple expulsion strategy tables. The strategy tables are indexed according to four dimensions: bird species, body size, historical response behavior, and environmental conditions. Each set of expulsion parameters includes: expulsion method type, action trigger time, action execution duration, execution frequency, action intensity level, and action direction control parameters, and is uniquely identified by multidimensional index encoding.

6. The adaptive bird repulsion method for power pole areas according to claim 1, characterized in that, In step S5, the recorded expulsion data includes: The currently executed expulsion parameter combination number and the value of each parameter; The time delay between identifying the target bird and executing the expulsion action; The time taken from the execution of the expulsion action until the target birds leave; Changes in the coordinates of the birds' resting positions in the image frame sequence during the expulsion process; The data is associated with and encoded using timestamps and tower numbers, and then stored in the host computer database.

7. The adaptive bird repulsion method for power pole areas according to claim 1, characterized in that, In step S6, a perturbation step size Δx is applied to one of the parameters in the executed expulsion parameter combination to form a new parameter combination; The perturbation step size Δx is an item in the perturbation sequence of the preset single-point perturbation sequence table, and the perturbation direction is determined by the trend of the expulsion time recorded after the previous round of perturbation. The perturbation execution adopts a rotating perturbation strategy, that is, only one parameter is selected for perturbation in each round among multiple parameters, and the perturbation process follows a single-point perturbation sequence list; The methods for determining the direction of the disturbance include: Calculate the difference between the time taken for the current round of expulsion and the time taken for the previous round of expulsion; When the difference is negative and greater than the set threshold θ, the perturbation direction remains unchanged; When the difference is positive and greater than the set threshold θ, switch the perturbation direction; When the absolute value of the difference is less than the set threshold θ, the current parameter remains unchanged and the disturbance is rotated to the next parameter.

8. A bird deterrent mechanism, characterized in that, The bird deterrent mechanism includes a drive unit and a bird deterrent stick mechanism. One end of the bird deterrent stick (2) is fixedly connected to the drive unit. A solar panel is attached to the surface of the bird deterrent stick (2). The bird deterrent stick mechanism also includes a battery (3) fixedly connected to the drive unit. The battery (3) is electrically connected to the drive unit and the solar panel respectively.

9. The bird deterrent mechanism according to claim 8, characterized in that, The bird deterrent stick (2) is made of carbon fiber.

Images