Linkage control method of airport bird monitoring and early warning bird repelling equipment

Abstract

The invention relates to the technical field of airport safety guarantee, and particularly discloses an airport bird situation monitoring and early warning bird repelling equipment linkage control method comprising the following steps: obtaining bird situation data vectors collected by sensors, calculating fused bird situation data vectors, and obtaining bird flock information; constructing a bird situation risk assessment model based on machine learning, combining the fused bird situation data with an aircraft take-off and landing plan, a runway use state and airport surrounding environment information to form an input feature vector, calculating a risk value through a risk assessment function, and dividing risk levels according to the risk value; according to a risk grade result of the risk assessment module, corresponding bird repelling equipment is controlled to work according to a hierarchical linkage bird repelling strategy; and in the bird repelling process, the risk level is re-evaluated according to the new bird situation data, the parameters of the bird repelling equipment are adjusted, and the bird repelling strategy is optimized by using a reinforcement learning algorithm. According to the invention, the defects of monitoring range and precision of a single sensor are effectively overcome, and the reliability of bird monitoring is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of airport security, and in particular to a linkage control method for airport bird monitoring, early warning and bird-repelling equipment. Background Art

[0002] With the rapid development of the aviation industry, airport safety and operations have become a top priority. Bird strikes, considered an invisible threat to aircraft safety, have always been a focus of industry attention. Currently, widely used bird-repelling methods at airports, such as fixed sonic and visual bird-repelling devices, have provided some protection, but their drawbacks are becoming increasingly prominent.

[0003] During airport operations, bird strikes pose a serious threat to the safe takeoff and landing of aircraft. Common bird-repelling methods currently used at airports, such as fixed acoustic and visual bird-repellent devices, have numerous drawbacks. Firstly, these fixed devices have limited monitoring and bird-repelling ranges, making it difficult to cover the vast expanses of an airport. Secondly, existing bird-repelling devices are independent of each other and lack effective linkage mechanisms. When birds gather in large numbers or suddenly approach the runway, they are unable to function in a timely and coordinated manner, resulting in ineffective bird-repelling effects. This puts aircraft at high risk of bird strikes, seriously impacting flight safety. Summary of the Invention

[0004] The purpose of the present invention is to provide a linkage control method for airport bird monitoring and early warning bird repellent equipment to solve the problems that existing bird repellent equipment has limited coverage, lacks a linkage mechanism, and is difficult to effectively deal with bird strike risks.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] The method for controlling the linkage of bird monitoring, early warning and bird repelling equipment at an airport includes the following steps:

[0007] S1: Deploy various types of sensors around the airport and in the flight area to obtain bird data vectors collected by each sensor. A fused bird data vector is obtained through a multi-sensor data fusion algorithm, and the fused bird data is used to obtain bird flock information.

[0008] S2: Build a bird risk assessment model based on machine learning. This model uses the fused bird data combined with aircraft takeoff and landing plans, runway usage status, and airport surrounding environment information to form an input feature vector. The risk value is calculated using a risk assessment function, and the risk level is determined based on the risk value.

[0009] S3: Based on the risk level results of the risk assessment module, control the corresponding bird-repelling equipment to work according to the hierarchical linkage bird-repelling strategy;

[0010] S4: During the bird-repelling process, reassess the risk level based on the new bird data, adjust the parameters of the bird-repelling equipment, and use reinforcement learning algorithms to optimize the bird-repelling strategy.

[0011] As a further solution of the present invention: in said S1, specifically including:

[0012] Assume that the bird data vector collected by each sensor is where d in represents the n-th dimension data collected by the i-th sensor;

[0013] The fused bird data vector obtained by multi-sensor data fusion algorithm is: m is the number of sensors, w i is the weight of the i-th sensor and

[0014] The fused bird data is used to obtain bird flock information, including the number, location distribution and flight trajectory.

[0015] As a further solution of the present invention: in said S2, specifically including:

[0016] The input feature vector is composed of the fused bird data, aircraft takeoff and landing plan S, runway usage status R, and airport surrounding environment information E. Through the risk assessment function Calculate the risk value R risk ;

[0017] Risk levels are divided according to risk value:

[0018] When R risk When <T1, it is judged as low risk;

[0019] When T1≤R risk When <T2, it is judged as medium risk;

[0020] When T2≤R risk When it is less than T3, it is judged as high risk;

[0021] When R risk When ≥T3, it is judged as an urgent risk;

[0022] Among them, T1, T2 and T3 are preset division thresholds and T1<T2<T3.

[0023] As a further solution of the present invention: in said S4, specifically including:

[0024] During the bird-repelling process, a new fusion bird data vector is obtained after time Δt and the input feature vector Recalculate risk value If the current bird-repellent strategy does not achieve the expected results, the strategy will be readjusted according to the new risk level.

[0025] As a further solution of the present invention: the bird risk assessment model based on machine learning is a neural network model, which is trained using a back propagation algorithm, and the training data includes at least 1,000 sets of historical bird data and corresponding aircraft operation safety data.

[0026] As a further solution of the present invention: in S4, the reinforcement learning algorithm adopts a deep Q network algorithm, the experience replay pool capacity is not less than 5000 sets of data, and the learning rate is 0.001-0.01.

[0027] The beneficial effects of the present invention are as follows: the present invention realizes comprehensive and accurate monitoring of bird conditions at airports through multi-source data acquisition and fusion technology, effectively making up for the shortcomings of the monitoring range and accuracy of a single sensor, and greatly improving the reliability of bird monitoring; the bird risk grading assessment model based on machine learning can comprehensively consider multiple factors and accurately assess the risk level of bird flocks to aircraft takeoff and landing, providing a scientific basis for subsequent graded linkage bird repellent, making bird repellent measures more targeted; the graded linkage bird repellent strategy implements differentiated bird repellent schemes according to different risk levels, realizes the efficient coordination of bird repellent equipment, significantly improves the bird repellent effect, minimizes the risk of aircraft encountering bird strikes, and effectively ensures the flight safety of the airport; the dynamic strategy adjustment and optimization mechanism can timely adjust the bird repellent strategy according to the real-time changes of the bird flock, ensuring the effectiveness of the bird repellent work. At the same time, the application of reinforcement learning algorithm enables the bird repellent strategy to continuously optimize itself and adapt to the complex and changing bird environment; the data recording and analysis function provides data support for airport bird control work, helps to discover bird patterns, thereby continuously optimizing and improving the entire bird monitoring, early warning and bird repellent system, and improving the overall level of airport bird control. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention will be further described below with reference to the accompanying drawings.

[0029] Figure 1 It is a flow chart of the linkage control method of the airport bird situation monitoring, early warning and bird-repelling equipment of the present invention. DETAILED DESCRIPTION

[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0031] See also Figure 1 As shown, the present invention is a method for controlling the linkage of bird monitoring, early warning and bird repelling equipment at an airport, comprising the following steps:

[0032] Multi-source data collection and fusion: Various types of sensors are deployed around the airport and in the flight area, including high-definition cameras, millimeter-wave radars, ultrasonic sensors, and thermal imagers. Let the bird data vector collected by the i-th sensor be Where n is the data dimension collected by the sensor (such as image features collected by high-definition cameras, distance and speed measured by millimeter-wave radar, etc.). Through multi-sensor data fusion algorithms, such as weighted average method, the data collected by different sensors are integrated and processed to obtain the fused bird data vector

[0033] Where m is the number of sensors, w i is the weight of the i-th sensor, and Weight w i It can be determined based on factors such as the reliability and accuracy of the sensor.

[0034] By using the fused bird data, more comprehensive and accurate bird information can be obtained, such as the number of bird flocks N, specific location distribution (x, y, z), flight trajectory wait.

[0035] Bird risk grading assessment: A bird risk assessment model based on machine learning is constructed. The model takes the bird data after multi-source fusion as input and combines multi-dimensional data such as aircraft take-off and landing plans, runway usage status, and airport surrounding environment information (such as whether it is close to bird habitats, water bodies, etc.) as features. Suppose the input feature vector is Among them, S represents the parameters related to the aircraft take-off and landing plan (such as the take-off and landing time interval, the number of flights, etc.), R represents the runway usage status (such as whether the runway is in use, the frequency of use, etc.), and E represents the airport surrounding environment information (such as the distance to the bird habitat, the water area, etc.).

[0036] Using a large amount of historical bird data and the corresponding aircraft operation safety conditions for training, a risk assessment function is obtained. This function is used to calculate the risk value of bird flocks to aircraft takeoff and landing

[0037] According to the risk value R risk , the risk level is divided into four levels: low, medium, high, and emergency, and the thresholds are T1, T2, and T3 (T1<T2<T3):

[0038] Risk levels are divided according to risk value:

[0039] When R riskWhen <T1, it is judged as low risk;

[0040] When T1≤R risk When <T2, it is judged as medium risk;

[0041] When T2≤R risk When it is less than T3, it is judged as high risk;

[0042] When R risk When ≥T3, it is judged as an urgent risk.

[0043] Hierarchical linkage bird-repellent strategy: Based on the bird risk assessment results, the central control system automatically triggers the linkage of bird-repellent equipment at the corresponding level.

[0044] Low risk: Start the voice bird repellent system and set the volume of the voice bird repellent system to V voice , the playback time is t voice Turn on some low-intensity optical bird-repelling devices, such as LED lights with a low flashing frequency. Set the flashing frequency of the LED lights to f LED The initial values of these parameters, such as V voice =V1,t voice =t1,f LED =f1.

[0045] Medium risk: Based on the voice and optical bird repellent, activate the ultrasonic bird repellent, and set the frequency of the ultrasonic bird repellent to f ultra , power is P ultra Control some mobile bird-repelling equipment (such as bird-repelling vehicles) to move slowly along the preset route. The speed of the bird-repelling vehicle is v car The initial values of parameters can also be determined based on experience, such as f ultra =f2,P ultra =P1,v car =v1.

[0046] High risk: All types of bird repellent devices are turned on, including high-intensity sonic bird repellents, strong strobe lights, laser bird repellents, etc. Suppose the volume of the sonic bird repellent is V sound , the flashing frequency of the strong strobe light is f strobe , the power of the laser bird repellent is P laser The bird-scaring car quickly drives towards the area where the bird flock is located. Let the speed of the bird-scaring car be v car2 The initial value of the parameter can be set to V sound =V2,f strobe =f3,P laser =P2,v car2 =v2. Furthermore, an early warning message is sent to the airport tower, reminding pilots to pay attention to the bird situation and adjust the aircraft's takeoff and landing attitude or delay takeoff and landing if necessary.

[0047] Emergency Risk: In addition to activating all bird-repelling equipment to repel the flock, immediately suspend aircraft takeoff and landing operations on the relevant runways to ensure aircraft safety. Simultaneously, issue an emergency bird alert to airport staff via the airport public address system, and organize personnel to perform auxiliary manual bird repellent operations, such as using flares and other equipment.

[0048] Dynamic strategy adjustment and optimization: During the bird-repelling process, the dynamic changes of the bird flock are continuously monitored. After a time Δt, the new fused bird data vector is The new input feature vector is Recalculate risk value

[0049] If the current bird repellent strategy is found to be ineffective, the operating parameters of the bird repellent equipment will be dynamically adjusted according to the new risk level. For example, for a sonic bird repellent, the volume adjustment formula is:

[0050]

[0051] Where α is the adjustment coefficient.

[0052] A similar method can be used to adjust the parameters of other bird-repelling equipment.

[0053] Furthermore, a reinforcement learning algorithm is used to continuously optimize the bird-repelling strategy based on the feedback r of each bird-repelling operation (such as the proportion of birds dispersed and the distance the flock moves away from the runway). Let the strategy be π, and the bird-repelling efficiency is improved by updating the strategy π. The update formula can be based on the policy gradient algorithm of reinforcement learning.

[0054] Data Recording and Analysis: The system automatically records the entire process of each bird incident, including bird data, risk assessment results, the startup and operation of bird repellent equipment, adjustments to bird repellent strategies, and the final bird repellent effect. Regular in-depth analysis of this data uncovers patterns and trends in bird incidents, providing strong data support for optimizing sensor layout, adjusting risk assessment model parameters, and improving bird repellent strategies.

[0055] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for controlling the linkage of bird monitoring and early warning equipment for bird repelling at an airport, characterized in that: The following steps are involved: S1: Deploy various types of sensors around the airport and in the flight area to obtain bird data vectors collected by each sensor. A fused bird data vector is obtained through a multi-sensor data fusion algorithm, and the fused bird data is used to obtain bird flock information. S2: Build a bird risk assessment model based on machine learning. This model uses the fused bird data combined with aircraft takeoff and landing plans, runway usage status, and airport surrounding environment information to form an input feature vector. The risk value is calculated using a risk assessment function, and the risk level is determined based on the risk value. S3: Based on the risk level results of the risk assessment module, control the corresponding bird-repelling equipment to work according to the hierarchical linkage bird-repelling strategy; S4: During the bird-repelling process, reassess the risk level based on the new bird data, adjust the parameters of the bird-repelling equipment, and use reinforcement learning algorithms to optimize the bird-repelling strategy.

2. The method for controlling the linkage of bird monitoring, early warning and bird repelling equipment at an airport according to claim 1, characterized in that: In the above S1, it specifically includes: Assume that the bird data vector collected by each sensor is where d in represents the n-th dimension data collected by the i-th sensor; The fused bird data vector obtained by multi-sensor data fusion algorithm is: m is the number of sensors, w i is the weight of the i-th sensor and The fused bird data is used to obtain bird flock information, including the number, location distribution and flight trajectory.

3. The method for controlling the linkage of airport bird monitoring, early warning and bird repelling equipment according to claim 1, characterized in that: In the above S2, it specifically includes: The input feature vector is composed of the fused bird data, aircraft takeoff and landing plan S, runway usage status R, and airport surrounding environment information E. Through the risk assessment function Calculate the risk value R risk ; Risk levels are divided according to risk value: When R risk When <T1, it is judged as low risk; When T1≤R risk When <T2, it is judged as medium risk; When T2≤R risk When it is less than T3, it is judged as high risk; When R risk When ≥T3, it is judged as an urgent risk; Among them, T1, T2 and T3 are preset division thresholds and T1<T2<T3.

4. The method for controlling the linkage of airport bird monitoring, early warning and bird repelling equipment according to claim 1, characterized in that: In the above S4, it specifically includes: During the bird-repelling process, a new fusion bird data vector is obtained after time Δt and the input feature vector Recalculate risk value If the current bird-repellent strategy does not achieve the expected results, the strategy will be readjusted according to the new risk level.

5. The method for controlling the linkage of bird monitoring, early warning and bird repelling equipment at an airport according to claim 1, characterized in that: The bird risk assessment model based on machine learning is a neural network model, which is trained using a back-propagation algorithm. The training data includes at least 1,000 sets of historical bird data and corresponding aircraft operation safety data.

6. The method for controlling the linkage of airport bird monitoring, early warning and bird repelling equipment according to claim 1, characterized in that: In S4, the reinforcement learning algorithm adopts a deep Q network algorithm, the experience replay pool capacity is not less than 5000 sets of data, and the learning rate is 0.001-0.01.

Images