Low-altitude radar and high-point monitoring combined bird repelling method

Abstract

The invention relates to the technical field of aviation safety, in particular to a low-altitude radar combined high-point monitoring bird repelling method, which comprises the following steps of: acquiring air reflection echo original data in a reconnaissance mode through a low-altitude radar deployed at an airport boundary matrix node; constructing a three-dimensional sensing space point set comprising a detection moment, a target slope distance, an azimuth angle, a pitching angle and echo intensity; and performing intelligent identification on the spatial point set in the step S1 by adopting an improved DeepSORT technology. According to the invention, a three-dimensional sensing network is constructed through an airport boundary low-altitude radar, secondary verification is carried out by using an improved DeepSORT technology, bird identification accuracy is greater than or equal to 98%, positioning refresh is less than 1 second, video monitoring coverage is limited, the radar constructs the three-dimensional sensing network to guide camera focusing, and an EMA module enhances small target feature extraction. And the unmanned aerial vehicle is linked to dynamically plan a path, multi-modal equipment is carried for repelling, a learning strategy is reinforced, and the repeated intrusion rate of birds is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of aviation safety, and specifically to a method for driving birds away by combining low-altitude radar and high-point monitoring. Background Art

[0002] Airport bird strike prevention is an important part of the aviation safety system. Bird strike incidents may cause serious consequences such as damage to aircraft engines and windshield breakage, which not only seriously threaten the lives of passengers and crew, but also cause significant economic losses. Therefore, it is particularly important to develop efficient airport bird-driving technologies and establish a real-time bird flock prevention and control system. Modern bird-driving technologies have evolved from traditional passive defenses to intelligent integrated prevention and control systems that include low-altitude radar detection, optical recognition, and acoustic interference. The core of the technology lies in breaking the adaptive laws of bird behavior to achieve sustainable ecological safety protection. Therefore, it is necessary to establish a three-dimensional intelligent protection network for the airport clearance protection area, and use various technical means to prevent birds from entering the airport flight area, effectively reducing the risk coefficient of collisions between aircraft and birds.

[0003] For example, the patent with the publication number CN118411027A discloses an active bird strike prevention method and device for airports. The bird strike prevention method includes the following steps: collecting bird situation information; comparing and analyzing the collected bird situation information with the bird situation work management database; and sending bird-driving information to the corresponding bird-driving equipment based on the analysis results for bird driving. The present invention sets up an integrated detection and driving system that combines a low-altitude radar for detecting birds and bird-driving linkage. When deploying the bird-driving equipment, a combined and decentralized deployment method is adopted to cover the surrounding area of the airport in all directions. Relying on the intelligent bird prevention platform, the deployment position of the bird-driving equipment can be flexibly adjusted according to the activity rules of birds near the airport to prevent the birds from having an adaptive impact on the equipment. Therefore, based on the existing core area safety objectives and performance indicators, with the help of various existing bird-driving means, the technical means of integrated detection and driving are used to ensure that the birds are immediately dispersed once they approach the core area.

[0004] However, when the above technologies are actually used, there are problems such as a high false alarm rate of the low-altitude radar, limited coverage of video monitoring, and lack of coordinated linkage between monitoring and driving equipment. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for driving birds away by combining low-altitude radar and high-point monitoring, so as to solve the problems of high false alarm rate of the low-altitude radar, limited coverage of video monitoring, and lack of coordinated linkage between monitoring and driving equipment.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A method for driving birds away by combining low-altitude radar and high-point monitoring, including the following steps: S1. Initial dataset of low-altitude radar scanning: The original data of air echo is obtained by the low-altitude radar deployed at the matrix nodes on the airport boundary in the detection mode, and a three-dimensional perception space point set containing the detection time, target slant range, azimuth angle, pitch angle and echo intensity is constructed. S2. Initial bird flock identification: The improved DeepSORT technology is used to intelligently identify the space point set in step S1 to form the tracking trajectory of the movable targets in the air. The improvement includes: adding the echo intensity dimension s and height parameter w to the Kalman filter state quantity, and using the weighted metric of Mahalanobis distance and cosine distance for trajectory association to realize the tracking and prediction of the target trajectory. S3. Verification of bird flock identification: For the tracking trajectory identified in step S2, the WGS-84 coordinates are calculated in combination with the low-altitude radar coordinates, and the infrared dual-band intelligent camera is guided to verify the target. The camera verification algorithm is based on Yolov8s. An efficient multi-scale attention module EMA is added after the second convolutional layer of the C2f module in the Backbone, and it forms the C2f-EMA module with C2f, replacing the third and fourth level C2f modules of the original Backbone to improve the feature extraction ability of small targets and fuzzy targets. S4. Start of tracking mode: After the low-altitude radar and the camera in S3 confirm the trajectory, the low-altitude radar switches to the tracking mode, and the update interval of the bird trajectory is shortened from 3 seconds to 0.1 second. S5. Collaborative decision-making and task dispatch: According to the bird flock position, flight direction and airport protection area data in step S4, combined with the course angle change rate and speed vector of the bird flock historical trajectory, the short-term movement trajectory of the bird flock within the next 30 seconds is predicted. According to the predicted trajectory in step S2, the minimum interval between the bird flock and the protection area is predicted. If it is less than the threshold, a bird repelling task is generated and the unmanned aerial vehicle is dispatched. S6. Dynamic path planning and bird repelling execution: During the bird repelling task, the low-altitude radar provides real-time guidance for the unmanned aerial vehicle. The unmanned aerial vehicle plans the path based on the real-time position of the bird flock and carries the multi-modal bird repelling equipment to execute the repelling task; at the same time, the minimum interval between the bird flock and the protection area is calculated until the interval meets the threshold and the predicted trajectory no longer enters the protection area, and the low-altitude radar turns back to the detection mode.

[0007] Preferably, in step S2, the echo intensity dimension s and height parameter w are added to the Kalman filter state quantity, and their calculation formulas are: ; Where A: State transition matrix; B: Control matrix; U: Control input; X: State, and its parameters are ; Where u, v, w: Target center coordinates; r: Aspect ratio; h: target height; s: echo intensity dimension; : velocity corresponding to each parameter.

[0008] Preferably, the formula for weighted measurement of Mahalanobis distance and cosine distance in step S2 is: ; where D is the comprehensive measurement result; is the weighting factor; is the Mahalanobis distance measurement result; is the cosine distance measurement result.

[0009] Preferably, in step S2, the DeepSORT technology divides the trajectories into deterministic trajectories and non-deterministic trajectories. The deterministic trajectories are associated by combining Mahalanobis distance and cosine distance measurements, and the non-deterministic trajectories are associated by the IOU method.

[0010] Preferably, the position update formula of the bird flock in step S4 can be described as: ; where is the bird flock velocity vector, obtained from DeepSORT; is the small increment of time; is the position vector at the current moment; is the position vector at the initial moment.

[0011] Preferably, the UAV path planning formula in step S6 is: ; where is the UAV velocity; is the distance between the UAV and the bird flock ; is the UAV position vector; is the time difference between adjacent moments.

[0012] Preferably, the multi-modal bird repellent device in step S6 includes the switching of sound wave, laser, and air flow disturbance modes.

[0013] Compared with the prior art, the beneficial effects of the present invention are: The present invention applies DeepSORT to low-altitude radar for the first time, and adds the echo intensity dimension s and the height parameter w on the basis of the Kalman filter in DeepSORT. Compared with the traditional radar using algorithms such as Kalman filtering, DeepSORT has a better tracking effect on the target, achieving a bird recognition accuracy of ≥98% and a positioning refresh of <1 second. In view of the limited video surveillance coverage, the radar constructs a three-dimensional perception network to guide the camera to focus, the EMA module enhances the feature extraction of small targets, and the drone is linked to dynamically plan the path, equipped with multi-modal equipment to drive away, and the reinforcement learning strategy greatly reduces the repeated invasion rate of birds. At the same time, the improved Yolov8s algorithm is better than the standard Yolov8s algorithm in recognizing small targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic diagram of a low-altitude radar target tracking algorithm flow chart for an improved low-altitude radar combined with high-point monitoring and bird-repelling method of the present invention; Figure 2 A schematic diagram of the improved Yolov8 algorithm model of the low-altitude radar combined with high-point monitoring bird-repelling method of the present invention; Figure 3 It is a schematic diagram of the structure of the EMA attention algorithm for improving the recognition rate of small targets by the low-altitude radar combined with high-point monitoring and bird-repelling method of the present invention; Figure 4 A schematic diagram of a flock of birds captured by a camera in an embodiment of the low-altitude radar combined with high-point monitoring and bird-repelling method of the present invention; Figure 5 This is a schematic diagram of the movement trajectory of a flock of birds tracked by a low-altitude radar in an embodiment of the low-altitude radar combined with high-point monitoring and bird-repelling method of the present invention; Figure 6 This is a schematic diagram of the movement trajectory of a flock of birds and a drone in an embodiment of the low-altitude radar combined with high-point monitoring and bird-repelling method of the present invention; Figure 7 1 is a schematic diagram showing the effect comparison between the improved DeepSORT tracking algorithm and the conventional Kalman filtering method adopted in the present invention; Figure 8 2 is a schematic diagram showing the effect comparison between the improved DeepSORT tracking algorithm and the conventional Kalman filtering method adopted in the present invention; Figure 9 3 is a schematic diagram showing the effect comparison between the improved DeepSORT tracking algorithm and the conventional Kalman filtering method adopted in the present invention; Figure 10 This is a schematic diagram of the improved Yolov8s used in the present invention for identifying small target birds; Figure 11 This is a schematic diagram of the standard Yolov8s for small target bird recognition. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0016] Please refer to Figures 1-11 , the present invention provides a technical solution: a low-altitude radar combined with high-point monitoring bird repelling method, which includes the following steps: S1. Low-altitude radar scanning and bird flock identification: The original point track information of various types of air echo is obtained through the low-altitude radar deployed near the airport in the detection mode. The point track information includes information such as detection time, target slant range, azimuth angle, pitch angle, echo intensity, etc. S2. Use the DeepSORT technology to intelligently identify the spatial point set scanned by the low-altitude radar to form the tracking trajectory of the movable target in the air. As Figure 1 shown, during the low-altitude radar scanning process, the original point track data of the low-altitude radar is sent into DeepSORT as input information. The recognition algorithm divides the trajectory into a deterministic trajectory and a non-deterministic trajectory. For the deterministic trajectory, the method of combining the Mahalanobis distance and the cosine distance metric is used to achieve trajectory association, and the IOU method is used to determine the relevance of the non-deterministic trajectory.

[0017] Improvement of Kalman filter: For the tracking and prediction of the trajectory, the algorithm adds the echo intensity dimension as one of the state quantities on the basis of the original Kalman filter state quantity. At the same time, since the original DeepSORT is used for two-dimensional image tracking, in order to adapt to the three-dimensional target scene, the center coordinate adds the height parameter w as one of the state quantities, that is: ; Where A: state transition matrix; B: control matrix; U: control input; X: state, whose parameters are , u, v, w: target center coordinates; r: aspect ratio; h: target height; s: echo intensity dimension; : the speeds corresponding to each parameter.

[0018] (1) Mahalanobis distance metric method. This method is used to measure the distance between the trajectory predicted by the Kalman filter and the newly input trajectory, which takes into account the uncertainty of state estimation. This method is mainly used to judge the relevance between consecutive frames, and its calculation formula is as follows: ; : The target covariance matrix predicted by the i-th Kalman filter; : The detected position of the j-th target; : The target position predicted by the i-th Kalman filter; The smaller it is, the greater the correlation between the trajectory and the original data.

[0019] (2) Cosine distance metric method. This method is a metric method based on appearance information. Since in the process of low-altitude radar detection, the echo of birds usually cannot normally reflect the shape characteristics of the birds themselves, and at the same time, due to the existing electromagnetic interference, the shape of the same target's adjacent two echoes may be quite different, so the cosine distance is used to implement.

[0020] The algorithm combines the advantages of Mahalanobis distance and cosine distance to provide a solution for long-term and short-term tracking of targets in target tracking. The Mahalanobis distance measures the gap in the position of the target object in two consecutive frames of images, solving the problem of target tracking between consecutive frames. The cosine distance compares the similarity between the features of newly detected objects and the features of the already tracked objects, solving the problem of target tracking between interval frames. To construct the association problem, DeepSORT combines the two metrics through weighted summation as follows:

[0021] ; where D is the comprehensive metric result, is the Mahalanobis distance metric result, is the cosine distance metric result, is the weighting factor.

[0022] S3. For the trajectories of movable targets identified by the low-altitude radar, combined with the coordinate position of the low-altitude radar itself, calculate the WGS-84 coordinate position of the target. Send the target position to the infrared dual-band intelligent camera for further identification of the target. The camera target recognition algorithm, based on Yolov8s, considers the characteristic that the pixel proportion of bird targets is small, and specifically adds an EMA efficient multi-scale attention module in the Backbone, as Figure 3 shown.

[0023] This module is located after the second convolutional layer of the C2f module, used to receive the data after convolutional upsampling, and together with C2f constitutes the C2f-EMA module, replacing the third-level and fourth-level C2f modules at the original Backbone, as Figure 2 shown, which can effectively improve the model's ability to extract features of small targets and fuzzy targets.

[0024] S4. Tracking Mode Activation: After the low-altitude radar and the camera confirm the bird trajectory simultaneously, the low-altitude radar switches from the detection mode to the tracking mode, continuously detecting the birds, and the update interval of the bird trajectory is increased from 3 seconds to 0.1 second.

[0025] S5. Collaborative Decision-making and Mission Dispatch: Based on the bird position and flight direction in S4, combined with the airport protection area data, and considering the course angle change rate and velocity vector of the bird flock's historical trajectory, a short-term prediction of the bird flock's movement trajectory within the next 30 seconds is made. According to the predicted trajectory in step S2, the minimum distance between the bird flock and the protection area is predicted (the threshold setting rule is: the basic threshold is 500 meters, and it can be adjusted according to user needs: for example, for large bird flocks or high-speed flying bird flocks, the threshold is increased to 800 meters). If it is less than the threshold, a bird repelling mission is generated and the drone is dispatched.

[0026] The longitude and latitude data of the bird flock fed back by the low-altitude radar represents the position of the dynamic target, that is, a point in the three-dimensional coordinate system. The position update formula of the bird flock can be described as: ; The current longitude and latitude position of the bird flock: ; is the tiny increment of time; The initial position of the bird flock: ; The velocity vector of the bird flock: , the direction is the flight direction of the bird flock, and the position and velocity can be obtained by DeepSORT.

[0027] If the bird flock position is updated discretely (through the periodic feedback of the low-altitude radar), it can be approximated by a difference equation: ； The drone adjusts its flight path according to the latest bird flock position, with the goal of approaching the bird flock and performing the repelling mission. The path planning of the drone can be based on the following formula: The current position of the drone is , and the target position is set as the current position of the bird flock . Then the target direction vector of the drone is: ; The speed of the drone:

[0028] The distance between the drone and the bird flock: .

[0029] To avoid direct collision between the drone and the bird flock and ensure the approaching mission at the same time, the direction adjustment can Add obstacle avoidance or proximity restrictions, such as setting a minimum safe distance: ; S6. Dynamic path planning and bird repelling execution: After the bird repelling command is issued in step S5, the low-altitude radar gives real-time guidance to the UAV based on the tracking mode. The UAV plans a path based on the real-time positions of the bird flock and carries a multi-modal bird repelling device to execute the repelling task; simultaneously calculate the minimum interval during the movement of the birds and the flight. Until the minimum interval between the birds and the airport protection area meets the threshold requirements and according to the prediction of the low-altitude radar, the future movement trajectory of the birds no longer enters the airport protection area, this bird repelling task ends, and the low-altitude radar switches to the reconnaissance mode.

[0030] The present invention combines the sensing capabilities of a low-altitude phased array radar and an infrared dual-band intelligent camera, and constructs a three-dimensional sensing network through the optimization of various algorithms, improving the accuracy and timeliness of bird flock recognition. The system adopts a multi-modal data fusion algorithm based on deep learning, which can synchronously analyze key parameters such as bird species, quantity, and flight trajectory. Through testing, the bird recognition accuracy rate ≥ 98%, and the positioning refresh frequency is less than 1 second.

[0031] The present invention constructs a closed-loop management of intelligent decision-making and adaptive optimization. Through a risk assessment model, it combines real-time bird flock data and flight dynamics to accurately quantify the bird strike risk and realize the automatic dispatch of bird repelling tasks.

[0032] The present invention deploys a low-altitude radar at the airport boundary matrix node, obtains the original data of the air echo in the reconnaissance mode to construct a three-dimensional sensing space point set, and adopts an improved DeepSORT technology (adding the echo intensity dimension and height parameter to the Kalman filter state quantity, and using the weighted metric of Mahalanobis distance and cosine distance for trajectory association) to form the tracking trajectory of the movable target in the air, and traction an infrared dual-band intelligent camera based on Yolov8s and adding an EMA module in the Backbone to verify the target trajectory. After the radar and the camera confirm, it switches to the tracking mode to shorten the trajectory update interval from 3 seconds to 0.1 second. Combining the bird flock position, flight direction and airport protection area data to generate a bird repelling task, the low-altitude radar guides the UAV to plan a path based on the real-time position of the bird flock and carry acoustic, optical, and airflow disturbance multi-modal bird repelling devices to execute the repelling task until the threshold requirements are met; its technical feature is to combine the sensing capabilities of a low-altitude phased array radar and an infrared dual-band intelligent camera, and realize the quantification of bird strike risk and the automatic dispatch of bird repelling tasks through a multi-modal data fusion algorithm (recognition accuracy rate ≥ 98%, positioning refresh frequency < 1 second), a risk assessment model, and use a reinforcement learning model to dynamically adjust the bird repelling strategy. The measured bird repeated intrusion rate is reduced by 83%, and a high-precision and highly collaborative intelligent bird repelling system is constructed, solving the problems of high false alarm rate of radar, insufficient recognition of small targets in video monitoring, and lack of collaborative linkage of equipment in the traditional bird repelling system.

[0033] The present invention was tested and verified in the courtyard of a household in Nanhuaijian Village, Ninghe District, Tianjin on June 3, 2025. Compared with the traditional radar using algorithms such as Kalman filtering, DeepSORT has a better tracking effect on targets. As Figure 7 , Figure 8 , Figure 9 shown, where the red marks are the original traces reported by the low-altitude radar, the blue marks are the bird flight trajectories fused by the improved DeepSORT method adopted by the present invention, and the orange marks are the bird flight trajectories fused directly by the Kalman filtering algorithm. It can be seen from the figure that the improved DeepSORT algorithm is superior to the Kalman filtering algorithm of the traditional radar in terms of track smoothness.

[0034] The recognition effect of the improved Yolov8s of the present invention for small targets is as Figure 10 shown, and the recognition effect of the standard Yolo8s for birds in the same environment is as Figure 11 shown. It can be seen from the figure that the improved Yolov8s recognition algorithm has a better detection effect on small targets than the standard Yolo8s algorithm.

[0035] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0036] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for driving away birds by combining low-altitude radar and high-point monitoring, characterized in that, It includes the following steps: S1. Scanning the initial data set by low-altitude radar: The low-altitude radar deployed at the airport boundary matrix node acquires the original data of the air reflected echo in the detection mode, and constructs a three-dimensional perception space point set including the detection time, target slant range, azimuth angle, pitch angle, and echo intensity; S2. Initial bird flock recognition: An improved DeepSORT technology is used to intelligently identify the space point set in step S1 to form a tracking trajectory of movable targets in the air. The improvement includes: adding the echo intensity dimension s and the height parameter w to the Kalman filter state quantity, and the trajectory association uses the weighted metric of Mahalanobis distance and cosine distance to achieve the tracking and prediction of the target trajectory; S3. Verification of bird flock recognition: For the tracking trajectory recognized in step S2, the WGS-84 coordinates are calculated in combination with the low-altitude radar coordinates, and an infrared dual-band intelligent camera is towed for target verification. The camera verification algorithm is based on Yolov8s. An efficient multi-scale attention module EMA is added after the second convolutional layer of the C2f module in the Backbone, and it forms a C2f-EMA module with C2f to replace the third and fourth level C2f modules of the original Backbone, improving the feature extraction ability for small targets and fuzzy targets; S4. Starting the tracking mode: After the low-altitude radar and the camera in S3 confirm the trajectory, the low-altitude radar switches to the tracking mode, and the bird trajectory update interval is shortened from 3 seconds to 0.1 second; S5. Collaborative decision-making and task dispatch: According to the bird flock position, flight direction and airport protection area data in step S4, combined with the heading angle change rate and velocity vector of the bird flock historical trajectory, a short-term prediction of the movement trajectory of the bird flock within the next 30 seconds is carried out. According to the predicted trajectory in step S2, the minimum interval between the bird flock and the protection area is predicted. If it is less than the threshold, a bird repelling task is generated and a drone is dispatched; S6. Dynamic path planning and bird repelling execution: During the bird repelling task, the low-altitude radar provides real-time guidance for the drone. The drone plans the path based on the real-time position of the bird flock and carries a multi-modal bird repelling device to execute the repelling task; Synchronously calculate the minimum interval between the bird flock and the protection area until the interval meets the threshold and the predicted trajectory no longer enters the protection area, and the low-altitude radar turns back to the detection mode.

2. The low-altitude radar combined with high-point monitoring bird repelling method according to claim 1, characterized in that: In step S2, the echo intensity dimension s and the height parameter w are added to the Kalman filter state quantity, and its calculation formula is: ； Where A: State transition matrix; B: Control matrix; U: Control input; X: State, whose parameter is ; Where u, v, w: Target center coordinates; r: Aspect ratio; h: Target height; s: Echo intensity dimension; : The speed corresponding to each parameter.

3. The low-altitude radar combined with high-point monitoring bird repelling method according to claim 1, characterized in that: The formula for the weighted metric of Mahalanobis distance and cosine distance in step S2 is: ； Where D is the comprehensive metric result; is a weighting factor; is the Mahalanobis distance metric result; is the cosine distance metric result.

4. The low-altitude radar combined with high-point monitoring bird repelling method according to claim 1, characterized in that: In step S2, the DeepSORT technology divides the trajectory into a deterministic trajectory and a non-deterministic trajectory. The deterministic trajectory uses the combination of Mahalanobis distance and cosine distance for metric and association, and the non-deterministic trajectory uses the IOU method for association.

5. The low-altitude radar combined with high-point monitoring bird repelling method according to claim 1, wherein: The bird flock position update formula in step S4 can be described as: ； Among them is the velocity vector of the bird flock, obtained by DeepSORT; is a small increment of time; is the position vector at the current moment; is the position vector at the initial moment.

6. The low-altitude radar combined with high-point monitoring bird repelling method according to claim 1, characterized in that: The drone path planning formula in step S6 is: ； Among them is the speed of the drone; is the distance between the drone and the flock of birds ; is the UAV position vector; is the time difference between adjacent moments.

7. The low-altitude radar combined with high-point monitoring bird repelling method according to claim 1, characterized in that: The multi-modal bird repelling device in step S6 includes the switching of acoustic wave, laser, and airflow disturbance modes.

Images