Ground-air linkage unmanned aerial vehicle bird impact risk intelligent early warning and repelling method

Abstract

The invention discloses a ground-air linkage unmanned aerial vehicle bird collision risk intelligent early warning and repelling method, and the method comprises the steps: carrying out the real-time monitoring and collection of bird activity data through the deployment of a ground mother station, a high-rise substation and an airborne substation three-in-one distributed monitoring and repelling node network; according to the method, a double-model risk assessment algorithm of bird individuals with attack tendency and common bird groups is constructed, single and indistinct monitoring and repelling are avoided, multi-level early warning and multi-level repelling mechanisms are established corresponding to different risk assessment levels, accurate risk levels and early warning levels are generated, and accurate repelling strategies are executed. Safe, efficient and intelligent bird impact risk early warning and repelling are achieved, safety guarantee is provided for large-scale application of the unmanned aerial vehicle in the urban low-altitude environment, and the unmanned aerial vehicle has great practical value and industrialization prospects.

Description

Technical Field

[0001] The present disclosure relates to the technical field of low-altitude flight safety and ecological environmental protection, and in particular to a ground-to-air linked intelligent early warning and avoidance method for bird strike risks of unmanned aerial vehicles (UAVs). Background Art

[0002] As the low-altitude economy becomes a national strategic emerging industry, Shenzhen, as a core city in the Guangdong-Hong Kong-Macao Greater Bay Area and a national hub for the drone industry, is at the forefront of the industry's development. By the end of 2023, the number of drone companies in Shenzhen will exceed 1,200, with an industry value of 65.8 billion yuan. Shenzhen boasts over 120,000 drones, including 65,000 commercial and professional drones, with an average daily takeoff and landing frequency of over 15,000. By 2035, the number of drone takeoff and landing points will increase from 483 in 2024 to approximately 1,500. Shenzhen is also a key wintering and stopover site for migratory birds along the East Asia-Australasian Flyway, with over 280 bird species recorded annually and over 100,000 seasonal migratory birds passing through. Compared to 2016, the total number of waterbirds recorded in Guangdong's Nei Lingding Futian National Nature Reserve in 2023 increased from 27,392 to 41,217 (a 50.5% increase), and the number of waterbird species increased from 53 to 78 (a 47.2% increase). With the significant increase in the frequency and density of low-altitude flights, conflicts between low-altitude flights and bird activity areas have intensified, and the impact of bird activities on flight safety has become increasingly significant. Low-altitude flight safety faces multiple challenges.

[0003] In high-density cities with densely populated super-tall buildings, bird flight activities are frequent and complex. Among low-altitude aircraft, drones (UAVs) are used in scenarios close to the natural environment. Their flight altitudes overlap significantly with the migratory altitudes of small songbirds (within 300 meters). They also overlap with the primary activity areas of shorebirds (10-50 meters), waders (20-80 meters), gulls (30-120 meters), and birds of prey (50-300 meters). Furthermore, unlike the occasional bird strikes caused by airplanes and large aircraft, small and medium-sized UAVs are smaller and more susceptible to aggressive interference from birds such as magpies, crows, pigeons, and birds of prey. Furthermore, their relatively fragile structures, light weight, and exposed components pose a serious threat to low-altitude flight safety. In the worst case, a UAV's flight path may be deviated, while in the worst case, it may result in a loss of control and a crash. Reducing the incidence of bird strikes is crucial for ensuring the safety of low-altitude aircraft and is crucial for achieving a win-win situation between the low-altitude economy and ecological protection.

[0004] Currently, research on bird activity warning and repellent technology is primarily focused on three traditional areas: power grids, buildings, and airports. These technologies rely on fixed monitoring and repelling devices installed at monitoring and repelling sites. Drone-mounted repellent devices also activate acoustic and flashing light repellents upon detecting a bird entering a preset range. However, some birds, such as sparrows and barn swallows, are not known for their aggressive nature. A single repellent strategy can cause excessive interference to less aggressive birds, disrupting the ecological balance. Therefore, there is an urgent need to develop intelligent risk warning and repellent solutions tailored to both aggressive and non-aggressive birds. Summary of the Invention

[0005] In view of this, the present invention discloses a ground-to-air linked intelligent early warning and avoidance method for UAV bird strike risks. By deploying a distributed monitoring and avoidance node network consisting of ground mother stations, high-rise substations and airborne substations, bird activity data at the take-off and landing points and flight paths of UAVs are monitored and collected in real time. A dual-model risk assessment algorithm is constructed for individual birds with attack tendencies and ordinary bird groups. Corresponding to different risk assessment levels, a multi-level early warning and multi-level avoidance mechanism is established. The generated accurate risk level, early warning level and the implementation of accurate avoidance strategies provide safety guarantees for the large-scale application of UAVs in urban low-altitude environments.

[0006] The technical solution of the present invention is a ground-to-air intelligent warning and avoidance method for UAV bird strike risks, the method comprising the following steps: Step 1: Obtain bird information data entering the monitoring area based on a distributed monitoring and repelling node network system deployed on the ground mother station, high-rise building substations, and drone airborne substations; Step 2: Based on the bird database, the bird in step 1 is matched and identified as an aggressive bird analysis target or a normal bird analysis target; Step 3: Construct a dual-model risk assessment algorithm for aggressive bird and ordinary bird. When an aggressive bird is identified as an analysis target, the aggressive bird algorithm model is run to calculate the output risk value Ra; when an ordinary bird is identified as an analysis target, the ordinary bird algorithm model is run to calculate the output risk value Rb. Step 4: Output the risk level based on the risk value Ra and the risk value Rb; output the warning level based on the risk level and the obtained distance between the bird and the drone, execute the warning action corresponding to the warning level; output the avoidance level based on the warning level, and execute the avoidance action corresponding to the avoidance level; Among them, if more than one-third of the birds entering the monitoring area are identified as birds with aggressive tendencies, they are determined to be analysis targets for birds with aggressive tendencies; otherwise, they are determined to be analysis targets for ordinary birds.

[0007] Furthermore, the calculation formula of the aggressive tendency flying bird algorithm model is: Ra=[w1·S+w2·(1 / D)²+w3·V_rel+w4·A_p+w5·T_min-1]×M×H_a; Where: S is the threat coefficient of the bird species, D is the straight-line distance between the bird and the drone (meters), V_rel is the approach speed of the bird relative to the drone (meters / second), A_p is the attack probability estimate based on the historical data of the bird, T_min is the minimum available decision time for the drone to avoid collision (seconds), M is the bird number multiplier, H_a is the weighting factor of the historical attack frequency of the bird species, w1 is the attack tendency influencing parameter, with a value of (0,0.4), which represents the weight value of the threat coefficient S of the bird species, w2 is the straight-line distance influencing parameter, with a value of ( w1+w2+w3+w4+w5=1 is the parameter affecting the attack probability. w2+w3+w4+w5=1 is the parameter affecting the attack probability of the bird. w4+w5=0.2 is the parameter affecting the decision time. w5+w1+w2+w3+w4+w5=1 is the parameter affecting the decision time.

[0008] Furthermore, the calculation formula of the ordinary flying bird algorithm model is: Rb=[α·(D_scale)+β·(V_rel_scale)+γ·(Θ_scale)+δ·(T_avail_scale)]×F_size×F_density; Among them, D_scale is the distance scale factor between the bird and the drone, V_rel_scale is the relative speed scale factor between the bird and the drone, Θ_scale is the angle scale factor between the bird and the drone, T_avail_scale is the available reaction time scale factor of the drone, F_size is the bird size coefficient, F_density is the group density factor, α is the distance influence parameter, which takes values ​​from (0, 0.5) and represents the weight value of the distance scale factor D_scale between the bird and the drone, β is the relative speed influence parameter, which takes values ​​from (0, 0.4) and represents the weight value of the relative speed scale factor V_rel_scale between the bird and the drone, γ is the flight angle influence parameter, which takes values ​​from (0, 0.3) and represents the weight value of the flight angle scale factor Θ_scale between the bird and the drone, and δ is the reaction time influence parameter, which takes values ​​from (0, 0.2) and represents the weight value of the available reaction time scale factor T_avail_scale of the drone, and it must satisfy α + β + γ + δ = 1.

[0009] The risk level includes four levels, namely, small risk, low risk, medium risk, and high risk. When the Ra or Rb value is [0, 25), the output is small risk; when the Ra or Rb value is [26, 50), the output is low risk; when the Ra or Rb value is [51, 75), the output is medium risk; when the Ra or Rb value is [76, 100), the output is high risk; The warning level includes four levels, namely, level one warning, level two warning, level three warning, and level four warning. When the risk level is low risk and the bird enters the monitoring range and the closest distance between the bird and the drone is greater than 500 meters, a level one warning is output; when the risk level is low risk and the closest distance between the bird and the drone is less than 300 meters, a level two warning is output; when the risk level is medium risk, the bird's trajectory intersects the drone's route path and the distance between the bird and the drone is less than 150 meters, a level three warning is output; when the risk level is high risk and the bird is identified as an aggressive bird analysis target and the closest distance between the aggressive bird and the drone is less than 100 meters, or is identified as an ordinary bird analysis target and the closest distance between the ordinary bird and the drone is less than 75 meters, a level four warning is output; The repelling level includes three levels, namely, level one repelling, level two repelling, and level three repelling. When the level three warning is on, the level one repelling is output, and when the level four warning is on, the level two repelling is output. When the Ra or Rb value is greater than 100, and the closest distance between the bird and the drone is less than 50 meters,

[0010] The warning action executed by the first-level warning is to maintain the normal monitoring state and record the monitoring events; the warning action executed by the second-level warning is to automatically increase the monitoring frequency, and the ground mother station or the high-rise sub-station automatically enters the avoidance preparation state, and the UAV maintains the original route; the warning action executed by the third-level warning is to trigger the first-level avoidance, and the avoidance action of the first-level avoidance is that the ground mother station or the high-rise sub-station enters the alert state and sends a warning alarm to the UAVs in the nearby area; the warning action executed by the fourth-level warning is to trigger the second-level avoidance, and the avoidance action of the second-level avoidance is that the ground mother station and the high-rise sub-station respond synchronously, the ground mother station takes over, sends avoidance instructions and adjusts the flight route of the UAVs in the nearby area; the action executed by the third-level avoidance is that the ground mother station forces manual takeover, starts the UAV protection mode, and the UAV hovers in place or starts emergency avoidance path planning.

[0011] Compared with the existing technical solutions, the present invention has the following beneficial effects: This proposed method utilizes a ground-to-air intelligent warning and avoidance system for drone bird strikes. Based on a distributed network of monitoring and avoidance nodes deployed on the ground, including a mother station on the ground, substations on high-rise buildings, and drone-mounted substations, it not only implements general monitoring and avoidance but also constructs a dual-model risk assessment algorithm for both individual birds with aggressive tendencies and general bird populations. By identifying birds that enter the monitoring area and calculating risk values ​​using the dual-model algorithm, a more scientific and accurate risk assessment level is generated. This method avoids the problem of single, indiscriminate bird monitoring and avoidance, which can lead to excessive interference. Risk quantification and a multi-level progressive warning and avoidance mechanism are established. The risk assessment results are matched with the multi-level warning and avoidance mechanisms, resulting in a more scientific and accurate warning level and the implementation of a more scientific and accurate avoidance strategy. Therefore, this method provides multi-level risk assessment, warning, and avoidance strategies for safe drone flight in low-altitude urban environments, addressing the limitations of existing single warning and avoidance solutions in complex and changing scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a flowchart of the steps of the ground-to-air linked drone bird strike risk intelligent warning and avoidance method of the present invention.

[0013] Figure 2 This is a block diagram of the four-level risk level, four-level warning, and four-level avoidance mechanism of the present invention. DETAILED DESCRIPTION

[0014] The technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present disclosure, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present disclosure without making any creative efforts are also within the scope of protection of the present disclosure.

[0015] Please refer to Figure 1 、 Figure 2 , is the ground-to-air linkage UAV bird strike risk intelligent warning and avoidance method disclosed in this solution, the method comprising the following steps: Step 1: Obtain bird information data entering the monitoring area based on a distributed monitoring and repelling node network system deployed on the ground mother station, high-rise building substations, and drone airborne substations; Step 2: Based on the bird database, the birds in step 1 are matched and identified as either aggressive bird analysis targets or normal bird analysis targets. If more than one-third of the birds entering the monitoring area are identified as aggressive birds, they are identified as aggressive bird analysis targets; otherwise, they are identified as normal bird analysis targets. Step 3: Construct a dual-model risk assessment algorithm for aggressive bird and ordinary bird. When an aggressive bird is identified as an analysis target, the aggressive bird algorithm model is run to calculate the output risk value Ra; when an ordinary bird is identified as an analysis target, the ordinary bird algorithm model is run to calculate the output risk value Rb. Step 4: Output the risk level based on the numerical values ​​of risk value Ra and risk value Rb; output the warning level based on the level of risk level and the obtained numerical value of the distance between the bird and the drone, execute the warning action corresponding to the warning level, output the avoidance level based on the level of warning level, and execute the avoidance action corresponding to the avoidance level.

[0016] The calculation formula of the aggressive tendency flying bird algorithm model is: Ra=[w1·S+w2·(1 / D)²+w3·V_rel+w4·A_p+w5·T_min-1]×M×H_a; Where: S is the threat coefficient of the bird species, D is the straight-line distance between the bird and the drone (meters), V_rel is the approach speed of the bird relative to the drone (meters / second), A_p is the attack probability estimate based on the historical data of the bird, T_min is the minimum available decision time for the drone to avoid collision (seconds), M is the bird number multiplier, H_a is the weighting factor of the historical attack frequency of the bird species, w1 is the attack tendency influencing parameter, with a value of (0,0.4), which represents the weight value of the threat coefficient S of the bird species, w2 is the straight-line distance influencing parameter, with a value of ( w1+w2+w3+w4+w5=1 is the parameter affecting the attack probability. w2+w3+w4+w5=1 is the parameter affecting the attack probability of the bird. w4+w5=0.2 is the parameter affecting the decision time. w5+w1+w2+w3+w4+w5=1 is the parameter affecting the decision time.

[0017] The bird species threat coefficient is initially determined based on existing bird species data. The species threat coefficient is matched to the species with the highest identified threat coefficient among multiple birds. The species threat coefficient ranges from 1 to 10. For example, the system initially set the threat coefficients for eagles and goshawks between 9 and 10; crows between 6 and 8; magpies between 4 and 6; and pigeons between 2 and 3. During ongoing monitoring, through continuous monitoring and behavioral pattern mining, the threat coefficient is adjusted based on the probability of bird repellent events in the database. The species threat coefficient is calculated as 1 + probability of repelling the corresponding bird × 9. The probability of repelling the corresponding bird can be expressed as F, F = N / T; where F is the attack frequency, N is the number of attack events in a specified time period, and T is the total number of flights or observations in that time period.

[0018] Definition of the Historical Attack Frequency Weighting Factor: Historical attack frequency refers to the number of attacks on aircraft by a specific bird species within a specific time period. This historical attack data is relative; that is, the accumulated data from the entire system operation also represents historical attack data. F = N / T; where F is the attack frequency, N is the number of attack events within a specified time period, and T is the total number of flights or observations within that time period.

[0019] Historical frequency weighting factor H_a: This factor is normalized based on the historical attack frequency for use in the risk assessment model. This factor can be determined by the following method: Normalization: w=(F−Fmin) / (Fmax−Fmin), where Fmin and Fmax are the minimum and maximum attack frequencies in the historical records, respectively.

[0020] The historical attack frequency weighting factor H_a = 1+1.5w. Its value range is [1, 2.5].

[0021] Regarding the calculation of A_p attack probability, different attack probabilities are initially set according to the different attack tendencies of birds (as shown in the table below).

[0022] Bird Type Attack Probability Birds of prey (eagles, goshawks) 0.8 crow 0.65 magpie 0.5 Pigeon 0.35 Alternatively, the system can continuously analyze the flight trajectory of the bird, calculate its flight feature vector F = (F1, F2, ..., F_n), and perform similarity matching with the preset attack pattern library to obtain the attack probability A_p, which is calculated as follows: A_p = max_i(similarity(F, Attack_Pattern_i)) Calculation of the minimum available decision time T_min T_min = (D - D_safe) / (V_rel+V_evasion) D_safe is the safety distance threshold, and V_evasion is the maximum evasive speed that the drone can reach.

[0023] Among them, the eigenvector F includes F1: height, in meters, indicating the height of the bird's flight; F2: lateral speed, in meters per second, indicating the speed of the bird's lateral movement; F3: longitudinal speed, in meters per second, indicating the bird's vertical ascent and descent speed; F4: approach angle, in degrees, indicating the angle at which the bird flies toward the drone; F5: flight path stability, in standard deviation, indicating the stability of the flight path calculated by analyzing the trajectory data; F6: distance change rate, in meters per second, indicating the rate of change of the distance between the bird and the drone; F7: behavior state, which is a categorical variable, such as "normal" or "attack" state; F8: group size, in number, indicating the number of birds; F9: migration pattern, which is a categorical variable indicating the normal or abnormal behavior of bird migration; F 10 : Time feature, indicating the time period of bird activity (such as hours). The initial values ​​of w1, w2, w3, w4, and w5 use the set benchmark value w1:w2:w3:w4:w5 = 3:3:2:1:1 (after normalization, w1=0.3).

[0024] The calculation formula of the ordinary flying bird algorithm model is: Rb=[α·(D_scale)+β·(V_rel_scale)+γ·(Θ_scale)+δ·(T_avail_scale)]×F_size×F_density; Where: D_scale is the distance scale factor, calculated as D_scale = (300-D) / 225, where D≤300 meters, otherwise it is 0; V_rel_scale is the relative speed scale factor, calculated as V_rel_scale = V_rel / V_max, where V_max is the maximum detection speed supported by the system; Θ_scale is the angle scale factor, calculated as Θ_scale = (180-θ) / 180, where θ is the flight path angle; T_avail_scale is the available reaction time scale factor, calculated as T_avail_scale = (T_crit - T_avail) / T_crit, when T_avail < T_crit, T_avail_scale is 0 otherwise; F_size is the bird body size coefficient, which is positively correlated with the bird's wingspan size; F_density is the population density factor; α is the distance influence parameter, with a value of (0, 0.5), representing the weight value of the distance scale factor D_scale between the bird and the drone. β is the relative speed influence parameter, with a value of (0, 0.4), representing the weight value of the relative speed scale factor V_rel_scale between the bird and the drone. γ is the flight angle influence parameter, with a value of (0, 0.3), representing the weight value of the flight angle scale factor Θ_scale between the bird and the drone. δ is the reaction time influence parameter, with a value of (0, 0.2), representing the weight value of the drone's available reaction time scale factor T_avail_scale. α + β + γ + δ = 1 must be satisfied. The bird's body size coefficient is positively correlated with the bird's wingspan size, F_size = W / W max , W is the wingspan of the target bird, Wmax is the maximum wingspan set by the system (the largest bird in the area). The population density factor reflects the density of the bird flock in a specific area, F_density=N / N max , N is the number of birds detected, N max is the maximum number of birds for which the system is designed.

[0025] The initial values ​​of α, β, γ, and δ use the set reference value α:β:γ:δ = 4:3:2:1 (α=0.4 after normalization).

[0026] In this solution, the risk level includes four levels, namely, micro risk, low risk, medium risk, and high risk. When the Ra or Rb value is [0, 25), micro risk is output; when the Ra or Rb value is [26, 50), low risk is output; when the Ra or Rb value is [51, 75), medium risk is output; when the Ra or Rb value is [76, 100), high risk is output; the warning level includes four levels, namely, level one warning, level two warning, level three warning, and level four warning. When the risk level is micro risk and a bird enters the monitoring range and the closest distance between the bird and the drone is greater than 500 meters, a level one warning is output; when the risk level is low risk and the closest distance between the bird and the drone is less than 300 meters, a level two warning is output; when the risk level is medium risk, When the bird's trajectory intersects the drone's flight path and the distance between the bird and the drone is less than 150 meters, a level 3 warning is output; when the risk level is high and the bird is identified as an aggressive bird analysis target and the closest distance between the aggressive bird and the drone is less than 100 meters, or when the bird is identified as an ordinary bird analysis target and the closest distance between the ordinary bird and the drone is less than 75 meters, a level 4 warning is output; the avoidance level includes three levels, namely, level 1 avoidance, level 2 avoidance, and level 3 avoidance. When the level 3 warning is issued, a level 1 avoidance is output, and when the level 4 warning is issued, a level 2 avoidance is output. When the Ra or Rb value is greater than 100, and the closest distance between the bird and the drone is less than 50 meters, and the expected collision time is less than the emergency threshold of 5 seconds, a level 3 avoidance is output, where the level 3 avoidance is also a critical avoidance.

[0027] Specifically, the warning action of the first-level warning is to maintain the normal monitoring status and record the monitoring events; the warning action of the second-level warning is to automatically increase the monitoring frequency, and the ground mother station or high-rise sub-station automatically enters the repelling preparation state, and the UAV maintains the original route; the warning action of the third-level warning is to trigger the first-level repelling, and the repelling action of the first-level repelling is that the ground mother station or high-rise sub-station enters the alert state, and sends a warning alarm to the UAVs in the nearby area, activating the repelling function of the ground mother station and high-rise sub-station in the corresponding monitoring area, with an execution intensity of 60%. The UAV maintains the route and monitors the bird's reaction in real time. If the triggering conditions are no longer met, the first-level repelling is lifted; a maximum bird repelling duration threshold is set. If there is no effect after exceeding the preset threshold, it will automatically upgrade to the second-level repelling. The warning action executed by the fourth-level warning is to trigger the second-level repelling. The repelling action of the second-level repelling is that the ground mother station and the high-rise sub-station respond synchronously, the ground mother station takes over, sends an avoidance instruction and adjusts the flight route of the drone in the adjacent area, activates the repelling function of the ground mother station and the high-rise sub-station in the corresponding monitoring area, and the execution intensity is increased to 100%; at the same time, the repelling function on the drone is activated, and the control center sends a deceleration instruction to the drone, decelerating to 50%, and monitoring the bird's reaction in real time. If the triggering conditions are no longer met, the second-level repelling is lifted; a maximum bird repelling duration threshold is set. If there is no effect after exceeding the preset threshold, the system automatically upgrades to the third-level repelling. The action executed by the third-level repelling is that the ground mother station forces manual takeover, starts the drone protection mode, and the drone hovers in place or starts emergency avoidance path planning.

[0028] In this solution, the distributed monitoring and repelling node network system deployed on the ground mother station, high-rise substations in high-rise buildings, and drone airborne substations includes a ground mother station deployed in the drone take-off and landing area, with at least one ground mother station configured at each take-off and landing point, covering a radius of at least 500 meters and a vertical detection range of at least 100 meters. Each ground mother station is equipped with a 24GHz phased array radar and a 1080P high-definition infrared and visible light dual-spectrum camera as a mother station monitoring module; equipped with a directional bird warning sound and natural enemy call sound wave array with adjustable frequency of 2-8kHz, a low-power 532nm band green laser projector, and an electronic bird call generator as a mother station repelling module, responsible for bird monitoring and repelling in the take-off and landing area, providing long-distance early monitoring and bird collision prevention. The high-rise substation is deployed in buildings with a height of ≥100 meters, with the distance between adjacent substations ≤2000 meters, and the coverage overlap rate ≥20%. Each high-rise substation is equipped with a miniaturized 24GHz phased array radar and a 4K high-definition zoom camera system as its monitoring module; a small directional acoustic transmitter, a low-power green laser, and an intelligent strobe light as its repellent module. The airborne substation is equipped with an ultra-lightweight 77GHz millimeter-wave MIMO radar and an edge computing-enhanced vision system as its monitoring module; and a lightweight 20-40kHz ultrasonic transmitter or a high-brightness LED strobe light as its bird repellent module.

[0029] This solution utilizes a dual-model risk assessment algorithm for both individual birds with aggressive tendencies and general bird populations. By identifying birds entering the monitoring area and calculating risk using the dual-model algorithm, a more scientific and accurate risk assessment level can be derived. This avoids the problem of single, indiscriminate monitoring and repelling of birds, which can cause excessive interference. By establishing a risk quantification and multi-level progressive early warning and repelling mechanism, the risk assessment results are matched with the multi-level early warning and repelling mechanisms, resulting in a more scientific and accurate early warning level and the implementation of a more scientific and accurate repelling strategy.

[0030] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A ground-to-air intelligent warning and avoidance method for UAV bird strike risk, characterized by: The method comprises the following steps: Step 1: Obtain bird information data entering the monitoring area based on a distributed monitoring and repelling node network system deployed on the ground mother station, high-rise building substations, and drone airborne substations; Step 2: Based on the bird database, the birds in step 1 are matched and identified as either aggressive bird analysis targets or normal bird analysis targets. If more than one-third of the birds entering the monitoring area are identified as aggressive birds, they are identified as aggressive bird analysis targets; otherwise, they are identified as normal bird analysis targets. Step 3: Construct a dual-model risk assessment algorithm for aggressive bird and ordinary bird. When an aggressive bird is identified as an analysis target, the aggressive bird algorithm model is run to calculate the output risk value Ra; when an ordinary bird is identified as an analysis target, the ordinary bird algorithm model is run to calculate the output risk value Rb. Step 4: Output the risk level based on the numerical values ​​of risk value Ra and risk value Rb; output the warning level based on the level of risk level and the obtained numerical value of the distance between the bird and the drone, execute the warning action corresponding to the warning level, output the avoidance level based on the level of warning level, and execute the avoidance action corresponding to the avoidance level.

2. The ground-to-air coordinated UAV bird strike risk intelligent warning and avoidance method according to claim 1 is characterized in that: The calculation formula of the aggressive tendency flying bird algorithm model is: Ra=[w1·S+w2·(1 / D)²+w3·V_rel+w4·A_p+w5·T_min-1]×M×H_a; Where: S is the threat coefficient of the bird species, D is the straight-line distance between the bird and the drone (meters), V_rel is the approach speed of the bird relative to the drone (meters / second), A_p is the attack probability estimate based on the historical data of the bird, T_min is the minimum available decision time for the drone to avoid collision (seconds), M is the bird number multiplier, H_a is the weighting factor of the historical attack frequency of the bird species, w1 is the attack tendency influencing parameter, with a value of (0,0.4), which represents the weight value of the threat coefficient S of the bird species, w2 is the straight-line distance influencing parameter, with a value of ( w1+w2+w3+w4+w5=1 is the parameter affecting the attack probability. w2+w3+w4+w5=1 is the parameter affecting the attack probability of the bird. w4+w5=0.2 is the parameter affecting the decision time. w5+w1+w2+w3+w4+w5=1 is the parameter affecting the decision time.

3. The ground-to-air coordinated UAV bird strike risk intelligent early warning and avoidance method according to claim 1 is characterized in that: The calculation formula of the ordinary flying bird algorithm model is: Rb=[α·(D_scale)+β·(V_rel_scale)+γ·(Θ_scale)+δ·(T_avail_scale)]×F_size×F_density; Among them, D_scale is the distance scale factor between the bird and the drone, V_rel_scale is the relative speed scale factor between the bird and the drone, Θ_scale is the flight angle scale factor between the bird and the drone, T_avail_scale is the available reaction time scale factor of the drone, F_size is the bird size coefficient, F_density is the group density factor, α is the distance influence parameter, which takes values ​​from (0, 0.5), indicating the weight value of the distance scale factor D_scale between the bird and the drone, β is the relative speed influence parameter, which takes values ​​from (0, 0.4), indicating the weight value of the relative speed scale factor V_rel_scale between the bird and the drone, γ is the flight angle influence parameter, which takes values ​​from (0, 0.3), indicating the weight value of the flight angle scale factor Θ_scale between the bird and the drone, and δ is the reaction time influence parameter, which takes values ​​from (0, 0.2), indicating the weight value of the available reaction time scale factor T_avail_scale of the drone, and it must satisfy α + β + γ + δ = 1.

4. The ground-to-air coordinated UAV bird strike risk intelligent warning and avoidance method according to claim 1 is characterized in that: The risk level includes four levels, namely, small risk, low risk, medium risk, and high risk. When the Ra or Rb value is [0,25), small risk is output; when the Ra or Rb value is [26,50), low risk is output; when the Ra or Rb value is [51,75), medium risk is output; when the Ra or Rb value is [76,100), high risk is output.

5. The ground-to-air coordinated UAV bird strike risk intelligent warning and avoidance method according to claim 4 is characterized in that: The warning levels include four levels, namely, level one warning, level two warning, level three warning, and level four warning. When the risk level is low risk and the bird enters the monitoring range, and the closest distance between the bird and the drone is greater than 500 meters, the output is a level one warning; when the risk level is low risk and the closest distance between the bird and the drone is less than 300 meters, the output is a level two warning; when the risk level is medium risk, the bird trajectory intersects the drone route, and the distance between the bird and the drone is less than 150 meters, the output is a level three warning; when the risk level is high risk and it is identified as an aggressive bird analysis target, the closest distance between the aggressive bird and the drone is less than 100 meters, or it is identified as an ordinary bird analysis target, and the closest distance between the ordinary bird and the drone is less than 75 meters, a level four warning is output.

6. The ground-to-air coordinated UAV bird strike risk intelligent warning and avoidance method according to claim 5 is characterized in that: The repellent level includes three levels, namely, level one repellent, level two repellent, and level three repellent. When the level three warning is used, level one repellent is output; when the level four warning is used, level two repellent is output; when the Ra or Rb value is greater than 100, and the closest distance between the bird and the drone is less than 50 meters, and the expected collision time is less than the emergency threshold of 5 seconds, level three repellent is output.

7. The ground-to-air coordinated UAV bird strike risk intelligent warning and avoidance method according to claim 6 is characterized in that: The warning action executed by the first-level warning is to maintain the normal monitoring state and record the monitoring events; The warning action executed by the second-level warning is to automatically increase the monitoring frequency, and the ground mother station or high-rise sub-station automatically enters the evasion preparation state, and the drone maintains its original route; the warning action executed by the third-level warning is to trigger the first-level evasion, and the evasion action of the first-level evasion is that the ground mother station or high-rise sub-station enters the alert state and sends a warning alarm to drones in the nearby area; The warning action executed by the fourth-level warning is to trigger the second-level avoidance. The avoidance action of the second-level avoidance is that the ground mother station and the high-rise sub-station respond synchronously, and the ground mother station takes over, sends an avoidance command and adjusts the flight path of the drone in the nearby area; The actions performed by the third-level avoidance are forced manual takeover by the ground mother station, starting the drone protection mode, and the drone hovering in place or starting emergency avoidance path planning.

Images