Airport bird repelling unmanned aerial vehicle device and safety system based on intelligent sensing

Abstract

The invention relates to the technical field of unmanned aerial vehicles, in particular to an airport bird repelling unmanned aerial vehicle device based on intelligent sensing and a safety system.The airport bird repelling unmanned aerial vehicle device comprises a multi-rotor unmanned aerial vehicle and a multi-rotor flying platform, and the multi-rotor unmanned aerial vehicle is provided with a laser-sound wave composite deterrent system; an AI decision core processor, a millimeter wave radar and an acquisition module are arranged on the three-mode perception driving cabin; and the acquisition module comprises an infrared-visible light fusion camera and a voiceprint acquisition array module. According to the three-mode sensing system, through all-weather coverage, multi-dimensional verification and self-adaptive weight distribution, the bird repelling target recognition omission ratio is reduced to be smaller than 2% and is far higher than that of a single sensor scheme, the typical omission ratio is 15-30%, the sensitivity of 532 nm green laser to the retina of a bird to infrared / red light is far higher than that of the retina of the bird to the wavelength, the deterrent efficiency is improved by 50% or above, and the deterrent effect is good. And the human eye safety wave band needs to be matched with power control, so that the pain points of slow response, easy failure and large interference of the traditional bird repelling means are solved.

Description

TECHNICAL FIELD

[0001] The application relates to an airport bird repelling unmanned aerial vehicle device and a safety system, in particular to an airport bird repelling unmanned aerial vehicle device and a safety system based on intelligent sensing, and belongs to the technical field of unmanned aerial vehicles. BACKGROUND

[0002] At present, airport bird repelling mainly relies on fixed equipment such as bird repelling guns, ultrasonic devices and vehicle-mounted mobile devices, and the following problems exist: The fixed equipment has a monitoring blind area and cannot cover the runway end, construction area and other dynamic areas; The vehicle-mounted device has poor mobility and is difficult to deal with sudden bird invasion; The traditional unmanned aerial vehicle bird repelling function is single, and lacks the closed-loop capability of intelligent sensing-decision-making-attack.

[0003] Therefore, it is urgent to improve the airport bird repelling unmanned aerial vehicle device to solve the above problems. SUMMARY

[0004] The application aims to provide an airport bird repelling unmanned aerial vehicle device and a safety system based on intelligent sensing, a three-mode sensing system, through all-weather coverage, multi-dimensional verification and adaptive weight distribution, can reduce the bird repelling target recognition missing detection rate to <2%, far exceeding the single sensor scheme, the typical missing detection rate is 15-30%, the 532nm green laser is sensitive to the wavelength of the bird retina, far exceeding the infrared / red light, the deterrent efficiency is improved by more than 50%, and the human eye safety band needs to be matched with power control, solving the pain points of slow response, easy failure and large interference of traditional bird repelling methods.

[0005] In order to achieve the above purpose, the main technical scheme adopted by the application comprises: An airport bird repelling unmanned aerial vehicle device based on intelligent sensing, comprising: A multi-rotor unmanned aerial vehicle and a multi-rotor flight platform used in cooperation with the multi-rotor unmanned aerial vehicle, a laser-sound wave composite deterrent system is arranged on the multi-rotor unmanned aerial vehicle, and the laser-sound wave composite deterrent system comprises a 532nm semiconductor laser and a frequency band adjustable sound wave transmitter; A three-mode sensing driving cabin, the three-mode sensing driving cabin is rotationally arranged on the multi-rotor flight platform, an AI decision-making core processor, a millimeter wave radar and a collection module are arranged on the three-mode sensing driving cabin, and the collection module comprises an infrared-visible light fusion camera and a voiceprint collection array module.

[0006] Preferably, the upper end of the multi-rotor flight platform is connected with a support table, and the multi-rotor unmanned aerial vehicle is arranged on the upper side of the support table. A drive motor is fixedly connected to the upper end of the trimodal sensing drive cabin. A drive gear is provided at the output end of the drive motor. A drive groove is opened on the upper side of the support platform. A drive tooth is fixedly provided inside the drive groove. The drive tooth meshes with the drive gear. The trimodal sensing drive cabin is rotatably mounted at the edge of the support platform.

[0007] Preferably, the three-modal sensing drive cabin is provided with a limiting guide roller, which abuts against the lower side of the support platform; The bottom of the trimodal sensing drive cabin is connected to an L-shaped connecting block, and the acquisition module is rotatably mounted on the L-shaped connecting block.

[0008] Preferably, each of the multi-rotor drones is fixedly provided with a drone metal support foot, and an electromagnet assembly is fixedly provided inside the support platform. The electromagnet assembly includes an iron core and a conductor coil, and the conductor coil is wound around the iron core.

[0009] Preferably, a gong is fixedly installed in the middle of the multi-rotor UAV, and several evenly distributed striking mechanisms are arranged around the periphery of the gong. Each striking mechanism includes a striking rod, and a striking ball is fixedly installed at one end of the striking rod. The striking ball is used to strike the gong.

[0010] Preferably, the striking rod is rotatably mounted on the multi-rotor drone, and a connecting block is rotatably connected to the lower side of the striking rod via a fixed block. The connecting block is connected to an electric telescopic rod via a spring, and the electric telescopic rod is fixedly mounted on the multi-rotor drone.

[0011] Preferably, the AI ​​decision core processor includes: Bird classification model based on ResNet-LSTM fusion network; A bird flock trajectory prediction algorithm combining fluid dynamics and reinforcement learning.

[0012] According to claim 1, the airport bird-repelling drone device and safety system based on intelligent perception, the fluid dynamics model formula is as follows: ; in, This represents the predicted speed at which individual i is affected by the flock of birds; Let i represent the set of neighbors of individual i; W Represents the kernel function. For smooth length; Indicates external force; The reinforcement learning optimizes the individual obstacle avoidance path through a deep Q-network, and the action value function is: Q ; where Q(s,a) represents the expected long-term cumulative reward that can be obtained after performing action a in state s; E[·]:Expectation, represents the average over all possible future state transitions and rewards; r represents the immediate reward, i.e., the reward obtained directly after performing action a in state s; γ is the discount factor, which weighs the importance of current rewards versus future rewards: γ≈0: only focus on immediate rewards; γ≈1: more emphasis on long-term returns; is the state at the next time step , select the action that maximizes the Q-value; s is the state at the current time t, and a is the action; The final output fusion prediction formula is: ; The adaptive weight is: = ; represents the maximum allowed neighbor density; w represents a trainable parameter, and the initial value is 2.0.

[0013] A safety system of a search and strike integrated airport bird repelling unmanned aerial vehicle based on intelligent sensing, comprising: At least 3 unmanned aerial vehicles as claimed in any one of claims 1-8; A ground control station comprising a digital twin operation interface and an Internet of Things security module, the Internet of Things security module communicates encrypted through TLS / SSL protocol to prevent counterfeit devices from accessing; An airport ADS-B data fusion module for airspace collaborative obstacle avoidance.

[0014] Preferably, when multiple machines work cooperatively, a Voronoi diagram is used to divide the monitoring area; The bird repelling target is dynamically allocated through an auction algorithm; A laser beam energy field intelligent superposition control module.

[0015] The present application has at least the following beneficial effects: 1. The three-mode perception system reduces the bird target identification missed detection rate to <2% through all-weather coverage, multi-dimensional verification and adaptive weight distribution, far exceeding the single sensor solution, typical missed detection rate 15~30%, 532nm green laser is sensitive to the wavelength of bird retina, far exceeding infrared / red light, and the deterrent efficiency is improved by more than 50%, and the human eye safety band needs to cooperate with power control, solving the pain points of traditional bird repelling methods such as slow response, easy failure and large interference.

[0016] 2. The three-mode perception driving cabin is arranged at the edge of the support table, and the driving motor on the three-mode perception driving cabin is started to drive the driving gear on the driving motor to rotate, and the driving gear is engaged with the driving teeth when rotating in the driving groove, so that the support table can be rotated at the edge of the support table, and the detection accuracy can be improved.

[0017] 3. In the process of knocking the copper gong, the connecting block is pulled downward by starting the electric telescopic rod, the fixed block drives the knocking rod to move downward, and then the knocking ball knocks the copper gong. In order to ensure that the copper gong is in a state of vibration, a spring is arranged between the connecting block and the output end of the electric telescopic rod, that is, after knocking, the copper gong can be ensured to be in a state of vibration, so as to achieve the purpose of driving away birds. BRIEF DESCRIPTION OF DRAWINGS

[0018] The drawings described herein are used to provide further understanding of the present application, and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings: Figure 1 is a perspective view of the present application; Figure 2 is a sectional view of the three-mode perception driving cabin of the present application; Figure 3 is a structural view of the three-mode perception driving cabin of the present application; Figure 4 is a structural view of the multi-rotor unmanned aerial vehicle of the present application; Figure 5 is a sectional view of the multi-rotor unmanned aerial vehicle of the present application; Figure 6 is Figure 5 is an enlarged view of A in the middle; Figure 7 is a structural view of the electromagnet assembly of the present application; Figure 8 is a system diagram of the present application.

[0019] In the figure, 1, multi-rotor unmanned aerial vehicle; 2, multi-rotor flight platform; 201, support table; 202, driving groove; 203, driving teeth; 3, electromagnet assembly; 301, core; 302, guide coil; 4, unmanned aerial vehicle metal support foot; 5, copper gong; 6, knocking mechanism; 601, electric telescopic rod; 602, knocking rod; 603, knocking ball; 604, fixed block; 605, connecting block; 606, spring; 7, three-mode sensing driving cabin; 701, driving motor; 702, limiting guide roller; 703, L-shaped connecting block; 704, driving gear; 8, AI decision core processor; 801, millimeter wave radar; 802, acquisition module. DETAILED DESCRIPTION

[0020] The embodiments of the present application will be described in detail below with the accompanying drawings and examples, so that the implementation process of how to apply technical means to solve technical problems and achieve technical effects of the present application can be fully understood and implemented.

[0021] As Figures 1-8 shown, the intelligent sensing-based search and knock integrated airport bird-repelling unmanned aerial vehicle device provided by the embodiment includes: The multi-rotor unmanned aerial vehicle 1 and the multi-rotor flight platform 2 used in cooperation with the multi-rotor unmanned aerial vehicle 1 adopt a six-rotor wind-resistant platform with a wind resistance level ≥ 6 levels, are equipped with a three-axis stabilization gimbal with an accuracy of ± 0.05°, and improve the stability of the multi-rotor flight platform 2 and the multi-rotor unmanned aerial vehicle 1. The multi-rotor unmanned aerial vehicle 1 is provided with a laser-sound wave composite deterrent system, which includes a 532nm semiconductor laser and a frequency band adjustable sound wave emitter. The 532nm green laser is sensitive to the wavelength far more than infrared / red light for the bird retina, and the deterrent efficiency is improved by more than 50%. Moreover, the human eye safety band needs to be matched with power control, which meets the IEC 60825-1 standard. The sound wave emitter covers most of the bird sensitive frequency bands, such as the pigeon sensitive peak of 8kHz. The dynamic switching frequency avoids the adaptability of birds. Therefore, in the process of driving the birds by the multi-rotor unmanned aerial vehicle 1, the purpose of driving can be achieved, and the safety of the staff can be ensured. The traditional bird-repelling means solves the pain points of slow response, easy failure, and large interference, and is especially suitable for high-safety-demand scenes such as airports; In addition, the three-mode perception driving cabin 7 is rotatably arranged on the multi-rotor flight platform 2, and the three-mode perception driving cabin 7 is provided with an AI decision core processor 8, a millimeter wave radar 801 and a collection module 802. The collection module 802 includes an infrared-visible light fusion camera and a voiceprint collection array module. The millimeter wave radar has a normal working wavelength of 3-5 mm in rain, fog and smoke environments, and has a penetration significantly better than that of a laser radar. The detection distance is stably maintained at 500 m, and the performance attenuation is less than 10% when the actual rainfall is less than or equal to 25 mm / h. The FMCW frequency modulation continuous wave technology is adopted to avoid the signal interference of other radars in the airport, and the signal-to-noise ratio is greater than or equal to 20 dB. Moreover, the millimeter wave radar has high-precision dynamic tracking, a speed resolution of 0.1 m / s, and can detect the slight movement of birds, such as the wing vibration frequency of 10 Hz of a pigeon. The infrared-visible light fusion camera can identify the shape of birds in high resolution in the daytime, and can distinguish species with a wingspan greater than 50 cm with an accuracy of 95%. The voiceprint collection array module adopts a three-dimensional sound source positioning 4-microphone ring array beamforming technology, and has a horizontal positioning accuracy of ±3° and a vertical accuracy of ±5°. The sound source tracking is realized through a TDOA time difference algorithm, and a voiceprint feature library covers more than or equal to 200 species of birds, such as the 3-kHz harmonic characteristics of crows and the 6-kHz pulse sound of seagulls. The voiceprint collection array module effectively supplements the optical blind area, such as the monitoring of bird activities under the shelter of tree crowns. Therefore, the three-mode perception system reduces the bird repelling target identification and detection rate to less than 2% through all-weather coverage, multi-dimensional verification and adaptive weight distribution, which is much better than a single sensor solution, and the typical detection rate is 15-30%.

[0022] Further, as Figure 1 and Figure 2As shown, the upper end of the multi-rotor flight platform 2 is connected with a support table 201, the multi-rotor unmanned aerial vehicle 1 is arranged on the upper side of the support table 201, and the multi-rotor unmanned aerial vehicle 1 is fixedly arranged on the upper side of the support table 201, so as to facilitate the multi-rotor unmanned aerial vehicle 1 to work and improve the driving efficiency. In order to detect the airport in all directions, the three-mode sensing driving cabin 7 is arranged on the side of the support table 201. In the detection process, the upper end of the three-mode sensing driving cabin 7 is fixedly connected with a driving motor 701, the output end of the driving motor 701 is provided with a driving gear 704, the upper side of the support table 201 is provided with a driving groove 202, the inside of the driving groove 202 is fixedly provided with a driving tooth 203, the driving tooth 203 is in meshing connection with the driving gear 704, and the three-mode sensing driving cabin 7 is rotatably arranged at the edge of the support table 201. By starting the driving motor 701 on the three-mode sensing driving cabin 7, the driving gear 704 on the driving motor 701 is driven to rotate after the driving motor 701 is started. The driving gear 704 is in meshing connection with the driving tooth 203 when rotating in the inside of the driving groove 202. Therefore, the three-mode sensing driving cabin 7 can rotate at the edge of the support table 201. Therefore, the surroundings of the multi-rotor flight platform 2 can be detected in all directions, and the detection accuracy is improved. In order to improve the stability of the movement, as shown in the drawings, Figure 3 As shown, the three-mode sensing driving cabin 7 is provided with a limiting guide roller 702, the limiting guide roller 702 abuts against the lower side of the support table 201, the three-mode sensing driving cabin 7 is provided with the limiting guide roller 702 abutting against the bottom side of the support table 201, which plays a certain supporting and limiting role on the three-mode sensing driving cabin 7, improves the stability of the three-mode sensing driving cabin 7, and the bottom of the three-mode sensing driving cabin 7 is connected with an L-shaped connecting block 703. The collecting module 802 is rotatably arranged on the L-shaped connecting block 703. The structure is simple, and the detection range is improved.

[0023] Further, in order to ensure the stability of the multi-rotor unmanned aerial vehicle 1 on the multi-rotor flight platform 2, as shown in the drawings, Figure 7 As shown, the support legs of the multi-rotor unmanned aerial vehicle 1 are fixedly provided with unmanned aerial vehicle metal support legs 4, the inside of the support table 201 is fixedly provided with an electromagnet assembly 3, the electromagnet assembly 3 comprises an iron core 301 and a guide coil 302, the guide coil 302 is wound on the iron core 301, the inside of the support table 201 is fixedly provided with the electromagnet assembly 3, the electromagnet assembly 3 comprises the iron core 301 and the guide coil 302, and after the guide coil 302 is connected with an external power supply, the iron core 301 generates magnetism and attracts the unmanned aerial vehicle metal support legs 4 at the bottom of the multi-rotor unmanned aerial vehicle 1. Specifically, When driving is needed, the multi-rotor unmanned aerial vehicle 1 needs to be started, so the guide coil 302 is in an open circuit state, the iron core 301 has no magnetism, and the multi-rotor unmanned aerial vehicle 1 can directly fly away from the multi-rotor flight platform 2 and work; When the unmanned aerial vehicle returns to the support table 201 of the multi-rotor flight platform 2, the guide coil 302 is connected to the power supply and attracts the unmanned aerial vehicle metal support foot 4 at the bottom of the multi-rotor unmanned aerial vehicle 1, and the multi-rotor unmanned aerial vehicle 1 is fixed on the support table 201 under the action of magnetism, thereby improving the stability of the multi-rotor unmanned aerial vehicle 1.

[0024] Further, in order to improve the driving efficiency, the noise is also advantageous, such as Figure 5 and Figure 6 As shown in the figure, the middle part of the multi-rotor unmanned aerial vehicle 1 is fixedly provided with a gong 5, and the circumferential side of the gong 5 is provided with a plurality of uniformly distributed knocking mechanisms 6. The knocking mechanism 6 includes a knocking rod 602, one end of the knocking rod 602 is fixedly provided with a knocking ball 603, and the knocking ball 603 is used to knock the gong 5. The gong 5 is fixedly arranged in the multi-rotor unmanned aerial vehicle 1, and the gong 5 is knocked by the knocking mechanism 6. The sound generated by the knocking mechanism 6 drives away the birds. The knocking rod 602 is rotatably arranged on the multi-rotor unmanned aerial vehicle 1. The lower side of the knocking rod 602 is rotatably connected with a connecting block 605 through a fixed block 604. The connecting block 605 is connected with an electric telescopic rod 601 through a spring 606. The electric telescopic rod 601 is fixedly arranged on the multi-rotor unmanned aerial vehicle 1. During the process of knocking the gong 5, the electric telescopic rod 601 is started to pull the connecting block 605 to move downward, the fixed block 604 drives the knocking rod 602 to move downward, and then the knocking ball 603 knocks the gong 5. In order to ensure that the gong 5 is in a vibrating state, a spring 606 is arranged between the connecting block 605 and the output end of the electric telescopic rod 601, that is, after knocking, the gong 5 can be ensured to be in a vibrating state, so as to achieve the purpose of driving away the birds.

[0025] In addition, the AI decision core processor 8 includes: The bird classification model based on the ResNet-LSTM fusion network, the ResNet-18 backbone network processes the infrared-visible light image, extracts the static features such as feather texture and wing span shape, and the Top-3 classification accuracy is 92%. The LSTM network analyzes the voiceprint sequence 20ms / frame and the radar trajectory fragment 30Hz update, captures the flight rhythm features such as pigeon wing vibration frequency 8-10Hz, and after the time-domain-space domain feature fusion, the accuracy of distinguishing similar species such as crow vs. raven is improved by 25%; The bird flock trajectory prediction algorithm combining fluid mechanics and reinforcement learning; The fluid mechanics model formula is: ; Among them, represents the predicted speed of individual i affected by the bird flock; represents the neighbor set of individual i; W The kernel function is represented by a cubic spline function, is a smoothing length; represents an external force; The reinforcement learning optimizes the individual obstacle avoidance path through a deep Q network, and the action value function is: Q ; wherein Q(s,a) represents the long-term cumulative reward expected to be obtained after performing action a in state s; E[·]: expected value, representing the average of all possible future state transitions and rewards; r represents the immediate reward, i.e., the reward obtained directly after performing action a in state s; gamma is 0<=gamma<1, a discount factor, which weighs the importance of current rewards and future rewards: gamma=0: only focus on immediate rewards; gamma=1: more emphasis on long-term returns; is the state at the next state time, and action corresponds to the value of the maximum Q value; , the state at the current time t is s, and the action is a; The final output fusion prediction formula is: ; The adaptive weight is: = ; represents the maximum allowed neighbor density; w represents a trainable parameter; The initial value is 2.0, the prediction length is 5s, which is enough for the UAV to adjust the interception path, the root mean square error RMSE is less than 1.2m, the fluid mechanics layer is updated 100ms / time + the reinforcement learning layer is updated 50ms / time, the total calculation time is less than 30ms, and the two types of models realize the bird behavior "perception-prediction-intervention" closed loop through the combination of physical law embedding and data-driven optimization. Compared with a pure data-driven scheme such as pure LSTM prediction, the comprehensive performance is improved by more than 40%.

[0026] As shown in Figure 8 , the safety system of the search and strike integrated airport bird-repelling UAV based on intelligent perception provided by the embodiment comprises: At least 3 unmanned aerial vehicles as claimed in any one of claims 1-8, a ground control station, including a digital twin operation interface, and an Internet of Things security module, the Internet of Things security module is encrypted for communication through the TLS / SSL protocol to prevent counterfeit devices from accessing and improve the stability of the system, an airport ADS-B data fusion module for airspace collaborative obstacle avoidance, full-area no-dead-angle coverage, each unmanned aerial vehicle autonomously manages a polygonal region, the side length is dynamically adjusted, the minimum is 50m, the coverage rate is improved by 300% compared with a single machine, and the boundary overlap area is about 5m. The monitoring right is allocated through an auction algorithm to avoid repeated driving, a multi-rotor unmanned aerial vehicle 1 is used for a bee swarm strategy, high-altitude mapping is 200m high, the FOV is 120°, a low-altitude surrounding three-dimensional deterrent circle is formed, when multiple machines work cooperatively, a Voronoi diagram is used to divide the monitoring area, a driving bird target is dynamically allocated through an auction algorithm, and a laser beam energy field intelligent superposition control module. The collaborative scheme realizes the leap from "single machine mechanical execution" to "emergence of swarm intelligence" through three technical innovations of airspace topology optimization, game theory decision and physical field cooperation, and is especially suitable for coping with high-density and high-dynamic bird group threats.

[0027] As used in the specification and claims, certain terminology is used to refer to specific components. Those of ordinary skill in the art will appreciate that hardware manufacturers can refer to the same component using different names. The specification and claims do not serve as a means to distinguish components based on differences in names, but rather serve as a means to distinguish components based on differences in functionality. As used throughout the specification and claims, "comprising" is a broad term intended to be open-ended. It specifically can be interpreted as "including, but not limited to." "Consisting essentially of" means including the elements as a set, but excluding other not essential elements of the specific set. "Consisting of" means including the elements as a set, excluding additional elements of any type.

[0028] It should be noted that the terms "comprising," "including," and "having" or any other variation thereof, are intended to cover a non-exclusive inclusion. For example, a process, product, or method that comprises a list of elements is not necessarily limited to only those elements but can include other elements not expressly listed or inherent to such process, product, or method. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, product, or method that comprises the element.

[0029] The foregoing description illustrates and describes the only preferred embodiments of the present application. However, it is to be understood that the application is not limited to the disclosed embodiments, and that numerous modifications are possible without departing from the spirit and scope of the present application. These modifications can be made only by persons skilled in the art, and the present application is limited only by the claims.

Claims

1. A bird-scaring drone device for airports based on intelligent sensing, characterized in that, include: A multi-rotor unmanned aerial vehicle (1) and a multi-rotor flight platform (2) used in conjunction with the multi-rotor unmanned aerial vehicle (1), wherein the multi-rotor unmanned aerial vehicle (1) is equipped with a laser-sonic composite deterrence system, the laser-sonic composite deterrence system comprising a 532nm semiconductor laser and a frequency band adjustable acoustic wave transmitter; The three-modal perception drive cabin (7) is rotatably mounted on the multi-rotor flight platform (2). The three-modal perception drive cabin (7) is equipped with an AI decision core processor (8), a millimeter-wave radar (801), and a data acquisition module (802). The data acquisition module (802) includes an infrared-visible light fusion camera and an acoustic signature acquisition array module.

2. The airport bird-scaring drone device based on intelligent sensing as described in claim 1, characterized in that: The upper end of the multi-rotor flight platform (2) is connected to a support platform (201), and the multi-rotor UAV (1) is disposed on the upper side of the support platform (201); The upper end of the trimodal sensing drive cabin (7) is fixedly connected to a drive motor (701), and the output end of the drive motor (701) is provided with a drive gear (704). The upper side of the support platform (201) is provided with a drive groove (202), and a drive tooth (203) is fixedly provided inside the drive groove (202). The drive tooth (203) meshes with the drive gear (704), and the trimodal sensing drive cabin (7) is rotatably disposed at the edge of the support platform (201).

3. The airport bird-scaring drone device based on intelligent sensing as described in claim 2, characterized in that: The three-modal sensing drive cabin (7) is provided with a limiting guide roller (702), which abuts against the lower side of the support platform (201); The bottom of the trimodal sensing drive cabin (7) is connected to an L-shaped connecting block (703), and the acquisition module (802) is rotatably mounted on the L-shaped connecting block (703).

4. The airport bird-scaring drone device based on intelligent sensing as described in claim 2, characterized in that: The multi-rotor UAV (1) is fixedly provided with UAV metal support feet (4) on its support feet. The support platform (201) is fixedly provided with an electromagnet assembly (3). The electromagnet assembly (3) includes an iron core (301) and a coil (302). The coil (302) is wound on the iron core (301).

5. The airport bird-scaring drone device based on intelligent sensing as described in claim 1, characterized in that: A gong (5) is fixedly installed in the middle of the multi-rotor UAV (1). Several evenly distributed striking mechanisms (6) are arranged on the periphery of the gong (5). The striking mechanism (6) includes a striking rod (602). A striking ball (603) is fixedly installed at one end of the striking rod (602). The striking ball (603) is used to strike the gong (5).

6. The airport bird-scaring drone device based on intelligent sensing as described in claim 5, characterized in that: The striking rod (602) is rotatably mounted on the multi-rotor drone (1). The lower side of the striking rod (602) is rotatably connected to a connecting block (605) via a fixing block (604). The connecting block (605) is connected to an electric telescopic rod (601) via a spring (606). The electric telescopic rod (601) is fixedly mounted on the multi-rotor drone (1).

7. The airport bird-scaring drone device based on intelligent sensing as described in claim 1, characterized in that: The AI ​​decision core processor (8) includes: Bird classification model based on ResNet-LSTM fusion network; A bird flock trajectory prediction algorithm combining fluid dynamics and reinforcement learning.

8. The airport bird-scaring drone device based on intelligent sensing as described in claim 1, characterized in that: The fluid dynamics model formula is as follows: ； in, This represents the predicted speed at which individual i is affected by the flock of birds; Let i represent the set of neighbors of individual i; W Represents the kernel function (a cubic spline function). For smooth length; Indicates external force; Reinforcement learning optimizes an individual's obstacle avoidance path using a deep Q-network, with the action-value function being: Q ； Where Q(s,a) represents the expected long-term cumulative reward after performing action a in state s; E[·]: Expected value, representing the average of all possible future state transitions and rewards; r represents the immediate reward, which is the reward obtained immediately after performing action a in state s; γ is a discount factor (0 ≤ γ < 1), which weighs the importance of current rewards against future rewards. γ≈0: Focus only on immediate rewards; γ≈1: Greater emphasis is placed on long-term returns; For the next state At that time, choose the action that maximizes the Q value. Corresponding value; The current state at time t is s, and the action is a; The final output fusion prediction formula, with dynamic weights fusing fluid dynamics and reinforcement learning results, is as follows: ; Adaptive weights for: = ； Indicates the maximum allowed neighbor density; w represents the trainable parameters, with an initial value of 2.

0.

9. A bird-scaring drone system for airports based on intelligent sensing, characterized in that, include: At least three drones as described in any one of claims 1 to 8; The ground control station includes a digital twin operating interface and an IoT security module. The IoT security module uses TLS / SSL protocol encryption for communication to prevent counterfeit devices from accessing the site. Airport ADS-B data fusion module is used for airspace collaborative obstacle avoidance.

10. A security system for airport bird-scaring drones based on intelligent perception, as described in claim 9, characterized in that: When multiple machines work together, Voronoi diagrams are used to divide the monitoring area; Bird deterrence targets are dynamically allocated using an auction algorithm; Intelligent superposition control module for laser beam energy field.

Images