A method for inspecting airport navigation lights based on UAV technology

Abstract

This invention discloses an airport navigation light inspection method based on UAV technology, comprising the following steps: S1, acquiring a historical detection coordinate database of navigation lights; relating to the field of navigation light inspection technology, this application constructs a motion feature vector for each navigation light, processes the motion feature vector of the navigation light using a Kalman filter algorithm, and uses an environmental adaptive noise matrix instead of the fixed noise matrix built into the Kalman filter algorithm. The environmental adaptive noise matrix can be dynamically adjusted according to atmospheric humidity and light intensity, thereby better adapting to changes in sensor noise under complex environments and generating more accurate denoised current navigation light detection coordinates; calculating the mean and standard deviation of latitude and longitude through the historical detection coordinate database of navigation lights, defining the normal fluctuation range, and combining the spatiotemporal consistency score to distinguish between short-term noise interference and true offset, avoiding misjudgment caused by noise in a single frame image, or missed detection due to ignoring small offsets because of a fixed preset route.

Description

Technical Field

[0001] This invention relates to the field of navigation light inspection technology, and in particular to a method for inspecting airport navigation lights based on unmanned aerial vehicle (UAV) technology. Background Technology

[0002] Airport navigation light inspections involve the regular comprehensive inspection, functional testing, and maintenance of all types of lighting equipment used to guide aircraft takeoffs, landings, and taxiing. This ensures the lighting systems are in normal working order and functioning properly, serving as a preventative measure to guarantee safe aircraft takeoffs and landings. The positional accuracy of airport navigation lights is one of the core elements ensuring safe aircraft takeoffs and landings. Even minor deviations, such as those caused by ground subsidence or equipment vibration, can lead to navigational errors and even flight accidents.

[0003] Publication No. CN114051093A discloses a portable on-site inspection system for navigation lights based on image processing technology; Publication No. CN119512181A discloses a method for inspecting airport navigation lights based on UAV technology;

[0004] The navigation light tracking system in Publication No. CN114051093A relies on a gimbal and laser rangefinder to achieve automatic tracking and focusing of the navigation light. When the navigation light experiences dynamic or slight displacement due to foundation settlement or noise interference, the relative distance measured by laser rangefinder cannot directly reflect the foundation settlement displacement. The system has not established a historical detection coordinate database and cannot identify whether the gradual displacement exceeds the normal range through calculation. This can easily lead to slight settlement being regarded as random fluctuations and missed. Furthermore, when the navigation light experiences slight displacement due to settlement, a single frame of laser data may be misjudged as normal due to noise. Therefore, Publication No. CN114051093A has the problem of missing the real displacement or falsely reporting noise interference.

[0005] When using drones for inspection, such as the drone inspection method in publication number CN119512181A, which operates based on a preset flight path, if the navigation lights experience dynamic or slight shifts due to ground subsidence or noise interference, the drone cannot detect and adjust its flight path and inspection strategy in a timely manner due to the fixed nature of the preset flight path, which can easily lead to missed or repeated inspections. Summary of the Invention

[0006] To address the technical problems existing in the background art, this invention proposes an airport navigation light inspection method based on unmanned aerial vehicle (UAV) technology.

[0007] The present invention proposes a method for inspecting airport navigation lights based on unmanned aerial vehicle (UAV) technology, comprising the following steps:

[0008] S1. Obtain the historical detection coordinate database of navigation lights;

[0009] The drone takes off from the ground control station and conducts drone patrols over the airport navigation lights area according to a preset basic route.

[0010] During drone patrol, image data of navigation lights, three-dimensional distance change rate, drone rotation speed, atmospheric humidity and light intensity are collected in real time;

[0011] S2. Based on the image data of the navigation lights, the three-dimensional distance change rate, and the UAV rotation rate, construct the motion feature vector M for each navigation light;

[0012] S3. The motion feature vector M of the navigation light is processed by the Kalman filter algorithm. The environmental adaptive noise matrix is ​​used to replace the fixed noise matrix of the Kalman filter algorithm to generate the current navigation light detection coordinates after denoising.

[0013] In S3, an environment-adaptive noise matrix is ​​used instead of the fixed noise matrix provided by the Kalman filter algorithm. The environment-adaptive noise matrix is ​​constructed as follows:

[0014] S4. Based on the historical detection coordinate database of navigation lights in S1, calculate the mean μ and standard deviation of latitude and longitude of the historical detection coordinates of each navigation light. The normal fluctuation range is set to μ±3. ;

[0015] μ±3 That is, the confidence interval for fluctuation;

[0016] When the current navigation light detection coordinates exceed the normal fluctuation range, a low-altitude circumferential re-inspection is triggered.

[0017] When the current navigation light detection coordinates are within the normal fluctuation range, calculate the spatiotemporal consistency score between the current navigation light detection coordinates and μ. If the spatiotemporal consistency score is greater than the set threshold, trigger a low-altitude orbital re-inspection.

[0018] S5. During low-altitude orbital re-inspection, multi-angle consistency scores are calculated through multi-angle imaging. If the drone patrol triggers low-altitude orbital re-inspection in three consecutive inspections and the multi-angle consistency scores are all greater than the set threshold, it is determined to be a real offset caused by foundation settlement.

[0019] If the displacement is determined to be a genuine result of foundation settlement, an alarm will be issued and a foundation reinforcement and maintenance work order will be generated.

[0020] Preferably, the equipment carried by the UAV includes an atmospheric humidity sensor, a light intensity sensor, an optical flow sensor, a lidar, an attitude sensor, and an image acquisition device.

[0021] Preferably, in S2, the motion feature vector M of the navigation light is constructed as follows:

[0022] The two-dimensional displacement vector of the navigation light in the image plane of the image data was calculated using the optical flow method. and To detect pixel-level offsets, the rate of change of the three-dimensional distance measured in real time by the LiDAR is assumed to be... The drone's rotation speed is Construct motion feature vectors for navigation lights ,in, Let Δt be the change in laser ranging of the lidar, and let Δt represent the time interval between two consecutive lidar measurements. The attitude angular velocity is obtained from the attitude sensor;

[0023] The motion feature vector M of a navigation light specifically refers to the motion feature vector of a single navigation light.

[0024] Preferably, the fixed noise matrix inherent in the Kalman filter algorithm is set as follows: ;

[0025] Let H be the real-time atmospheric humidity obtained in S1 and L be the light intensity;

[0026] Environmental Adaptive Noise Matrix The calculation formula is:

[0027] In the formula, α and β are both environmental sensitivity coefficients, which solves the problem of fixed noise parameters in traditional Kalman filtering and enables target trajectory prediction in scenarios with small offsets; thus improving the prediction accuracy in scenarios with small offsets.

[0028] Preferably, in S4, the spatiotemporal consistency score between the current navigation light detection coordinates and μ is calculated as follows:

[0029] Calculate the Euclidean distance D between the current navigation light detection coordinates and μ, and assume the spatiotemporal consistency score is... ;

[0030] The calculation formula is: In the formula, Δt is the detection time interval.

[0031] Preferably, if the Euclidean distance D between the current detection coordinates and μ is greater than 0.1 meters, this is considered a serious offset. The spatiotemporal consistency score calculation is skipped directly, and a low-altitude orbital re-inspection is forcibly triggered to shorten the anomaly response time.

[0032] Preferably, in S5, the low-altitude orbital re-inspection is as follows:

[0033] When a low-altitude surround re-inspection is triggered, the UAV will take the current navigation light detection coordinates as the center and collect visual images at 8 equally spaced angles within the set altitude and radius area.

[0034] The latitude and longitude coordinates of the current navigation light at the eight angles are obtained by analyzing the view images acquired from eight angles. ,in, Represents 8 equally spaced angles. =1, 2, ..., 8.

[0035] Preferably, in S5, a multi-angle consistency score is calculated. as follows:

[0036] ;in , The latitude and longitude average of the eight angular coordinates of the current navigation light. For preset accuracy parameters, ≠0.

[0037] Preferably, when the atmospheric humidity H > 80% and the light intensity L < 300 lux, when the low-altitude surround re-inspection is triggered, the number of viewing angles for low-altitude surround re-inspection multi-angle imaging is increased to twice the original number, and the UAV altitude is reduced to G times the set altitude, 0 < G < 1, to reduce atmospheric refraction error and improve the reliability of multi-view data.

[0038] Preferably, if a navigation light has not triggered an alarm for 6 consecutive months, its normal fluctuation range is adjusted from μ±3. Adjusted to μ±2.5 To improve detection sensitivity;

[0039] The airport navigation light inspection method based on UAV technology proposed in this invention has the following beneficial technical effects:

[0040] 1. This application constructs motion feature vectors for each navigation light and processes them using a Kalman filter algorithm. An environmentally adaptive noise matrix replaces the fixed noise matrix inherent in the Kalman filter algorithm, dynamically adjusting the noise matrix based on atmospheric humidity and light intensity. This better adapts to changes in sensor noise under complex environments, generating more accurate denoised current navigation light detection coordinates. Furthermore, by calculating the latitude and longitude mean and standard deviation from the historical navigation light detection coordinate database, a normal fluctuation range is defined. Combined with spatiotemporal consistency scoring, short-term noise interference is distinguished from true offsets, avoiding misjudgments caused by noise in a single frame image or ignoring minor offsets due to a fixed preset route. This system addresses the issue of missed detections due to ground subsidence. By implementing a low-altitude surround re-inspection mechanism and calculating multi-angle consistency scores through multi-angle imaging, it further eliminates single-view interference. Only after three consecutive re-inspections trigger a re-inspection and the multi-angle consistency scores are all greater than the threshold is the system determined to be a genuine shift caused by ground subsidence, significantly improving the accuracy of judgment. This system also solves the problem of missed detections or misjudgments caused by the fixed preset flight paths, sensor noise interference, and the precise descent or slight shift of navigation lights during dynamic detection when inspecting airport navigation lights using traditional UAVs. It ensures the accuracy of navigation light positions, maintains the normal function of the airport navigation system, provides reliable guidance for aircraft takeoff and landing, reduces the risk of flight accidents caused by navigation light shifts from the source, and ensures aviation safety.

[0041] 2. By using an environment-adaptive noise matrix instead of the fixed noise matrix inherent in the Kalman filter algorithm, and dynamically adjusting the noise matrix of the Kalman filter through real-time environmental perception, the problem of fixed noise parameters in the traditional Kalman filter is solved, thereby enabling target trajectory prediction in scenarios with small offsets and improving prediction accuracy in such scenarios.

[0042] 3. This application constructs motion feature vectors for each navigation light and integrates the image data of the navigation light, the three-dimensional distance change rate, and the rotation rate of the UAV to more comprehensively characterize the motion state of the navigation light. Compared with a single data source that relies solely on laser ranging or preset routes, it can more comprehensively and accurately reflect the actual situation of the navigation light and enhance its adaptability to complex scenarios.

[0043] 4. This application forms a complete automated process from data acquisition, feature vector construction, filtering, offset judgment to triggering re-inspection and final decision-making to generate a foundation reinforcement maintenance work order. It does not require manual intervention throughout the process and can quickly and continuously monitor the status of navigation lights, thereby improving inspection efficiency.

[0044] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0045] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0046] Embodiments of the present invention are described in detail below. Examples of these embodiments are illustrated in the accompanying drawings, wherein the same or similar symbols denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0047] like Figure 1 The method for inspecting airport navigation lights based on UAV technology, as shown, includes the following steps:

[0048] S1. Obtain the historical detection coordinate database of navigation lights;

[0049] The drone takes off from the ground control station and conducts drone patrols over the airport navigation lights area according to a preset basic route.

[0050] The equipment carried by the drone includes an atmospheric humidity sensor, a light intensity sensor, an optical flow sensor, a lidar, an attitude sensor, and an image acquisition device;

[0051] During drone patrol, image data of navigation lights, three-dimensional distance change rate, drone rotation speed, atmospheric humidity and light intensity are collected in real time;

[0052] S2. Based on the image data of the navigation lights, the three-dimensional distance change rate, and the UAV rotation rate, construct the motion feature vector M for each navigation light;

[0053] In an optional embodiment, in S2, the motion feature vector M of the navigation light is constructed as follows:

[0054] The two-dimensional displacement vector of the navigation light in the image plane of the image data was calculated using the optical flow method. and To detect pixel-level offsets, the rate of change of the three-dimensional distance measured in real time by the LiDAR is assumed to be... The drone's rotation speed is Construct motion feature vectors for navigation lights ,in, Let Δt be the change in laser ranging of the lidar, and let Δt represent the time interval between two consecutive lidar measurements. The attitude angular velocity is obtained from the attitude sensor;

[0055] The motion feature vector M of a navigation light specifically refers to the motion feature vector of a single navigation light.

[0056] This application constructs motion feature vectors for each navigation light and integrates the image data of the navigation light, the three-dimensional distance change rate, and the rotation rate of the UAV to more comprehensively characterize the motion state of the navigation light. Compared with a single data source that relies solely on laser ranging or preset flight paths, it can more comprehensively and accurately reflect the actual situation of the navigation light and enhance its adaptability to complex scenarios.

[0057] S3. The motion feature vector M of the navigation light is processed by the Kalman filter algorithm. The environmental adaptive noise matrix is ​​used to replace the fixed noise matrix of the Kalman filter algorithm to generate the current navigation light detection coordinates after denoising.

[0058] In an optional embodiment, S3 uses an environment-adaptive noise matrix instead of the fixed noise matrix provided by the Kalman filter algorithm. The environment-adaptive noise matrix is ​​constructed as follows:

[0059] Let the fixed noise matrix of the Kalman filter algorithm be... ;

[0060] Let H be the real-time atmospheric humidity obtained in S1 and L be the light intensity;

[0061] Environmental Adaptive Noise Matrix The calculation formula is:

[0062] In the formula, α and β are both environmental sensitivity coefficients;

[0063] The values ​​of α and β are generated through a particle swarm optimization algorithm, which solves the problem of fixed noise parameters in traditional Kalman filtering and enables target trajectory prediction in scenarios with small offsets, thereby improving the prediction accuracy in such scenarios.

[0064] Breaking through the limitations of traditional preset flight paths, it achieves precise tracking of dynamic / micro-deviation navigation lights through real-time environmental perception, target motion prediction, and UAV attitude linkage.

[0065] S4. Based on the historical detection coordinate database of navigation lights in S1, calculate the mean μ and standard deviation of latitude and longitude of the historical detection coordinates of each navigation light. The normal fluctuation range is set to μ±3. ;

[0066] μ±3 That is, the confidence interval for fluctuation;

[0067] When the current navigation light detection coordinates exceed the normal fluctuation range, a low-altitude circumferential re-inspection is triggered.

[0068] When the current navigation light detection coordinates are within the normal fluctuation range, calculate the spatiotemporal consistency score between the current navigation light detection coordinates and μ. If the spatiotemporal consistency score is greater than the set threshold, trigger a low-altitude orbital re-inspection.

[0069] In an optional embodiment, in S4, the spatiotemporal consistency score between the current navigation light detection coordinates and μ is calculated as follows:

[0070] Calculate the Euclidean distance D between the current navigation light detection coordinates and μ, and assume the spatiotemporal consistency score is... ;

[0071] The calculation formula is: In the formula, Δt is the detection time interval;

[0072] S5. During low-altitude orbital re-inspection, multi-angle consistency scores are calculated through multi-angle imaging. If the drone patrol triggers low-altitude orbital re-inspection in three consecutive inspections and the multi-angle consistency scores are all greater than the set threshold, it is determined to be a real offset caused by foundation settlement.

[0073] If the displacement is determined to be a genuine result of foundation settlement, an alarm will be issued and a foundation reinforcement and maintenance work order will be generated.

[0074] The foundation reinforcement maintenance work order includes:

[0075] Work order number, generation time, detection data file, image and video evidence;

[0076] In an optional embodiment, in S5, the low-altitude surround re-check is as follows:

[0077] When a low-altitude surround re-inspection is triggered, the low altitude of the low-altitude surround re-inspection refers to lowering the altitude of the drone. For example, when the drone is cruising, the altitude of the drone is 30m. However, during the low-altitude surround re-inspection, the drone takes the current navigation light detection coordinates as the center and collects visual images at 8 equal intervals within a range of 10m altitude and 5m radius.

[0078] The latitude and longitude coordinates of the current navigation light at the eight angles are obtained by analyzing the view images acquired from eight angles. ,in, Represents 8 equally spaced angles. =1, 2, ..., 8;

[0079] In S5, multi-angle consistency scores are calculated. as follows:

[0080] ;in , The latitude and longitude average of the eight angular coordinates of the current navigation light. For preset accuracy parameters, ≠0;

[0081] If all 3 consecutive inspections trigger > 2 and If the displacement exceeds the set threshold, it is determined to be a real displacement caused by foundation settlement, and foundation settlement is confirmed.

[0082] If the displacement is determined to be a genuine result of foundation settlement, an alarm will be issued and a foundation reinforcement and maintenance work order will be generated.

[0083] Here, multi-angle imaging is set to 8 equally spaced angles for demonstration purposes, but the number of angles is not limited to 8.

[0084] This application forms a complete automated process from data acquisition, feature vector construction, filtering, offset judgment to triggering re-inspection and final decision-making to generate a foundation reinforcement maintenance work order. It requires no manual intervention throughout the process and can quickly and continuously monitor the status of navigation lights, thereby improving inspection efficiency.

[0085] By constructing motion feature vectors for each navigation light and fusing multi-source data, an environment-adaptive noise matrix is ​​used instead of the fixed noise matrix inherent in the Kalman filter algorithm. Combined with a low-altitude surround re-inspection mechanism, the accuracy of identifying true offsets is improved, thereby enhancing the safety of the airport navigation aid system.

[0086] When in use, when the atmospheric humidity H > 80% and the light intensity L < 300 lux, the number of viewing angles for low-altitude surround re-inspection multi-angle imaging increases from 8 to 16, that is, it is set to 16 equally spaced angles, and the drone height is reduced from 10m to 5m to reduce atmospheric refraction error and improve the reliability of multi-view data.

[0087] If the Euclidean distance D between the current detection coordinates and μ is greater than 0.1 meters, this is a serious offset. The spatiotemporal consistency score calculation is skipped directly, and a low-altitude orbital re-inspection is forcibly triggered to shorten the anomaly response time.

[0088] If a navigation light does not trigger an alarm for 6 consecutive months, its normal fluctuation range will be adjusted from μ±3. Adjusted to μ±2.5 To improve detection sensitivity;

[0089] This application constructs a motion feature vector for each navigation light, processes the motion feature vector M using a Kalman filter algorithm, and replaces the fixed noise matrix built into the Kalman filter algorithm with an environment-adaptive noise matrix. This environment-adaptive noise matrix can be dynamically adjusted according to atmospheric humidity and light intensity, thus better adapting to changes in sensor noise under complex environments and generating more accurate denoised current navigation light detection coordinates. Furthermore, by calculating the mean and standard deviation of latitude and longitude from a historical navigation light detection coordinate database, a normal fluctuation range is defined. Combined with a spatiotemporal consistency score, short-term noise interference is distinguished from true offsets, avoiding misjudgments caused by noise in a single frame image or ignoring minor offsets due to a fixed preset route. This system addresses the issue of missed detections due to ground subsidence. By implementing a low-altitude surround re-inspection mechanism and calculating multi-angle consistency scores through multi-angle imaging, it further eliminates single-view interference. Only after three consecutive re-inspections trigger a re-inspection and the multi-angle consistency scores are all greater than the threshold is the system determined to be a genuine shift caused by ground subsidence, significantly improving the accuracy of judgment. This system also solves the problem of missed detections or misjudgments caused by the fixed preset flight paths, sensor noise interference, and the precise descent or slight shift of navigation lights during dynamic detection when inspecting airport navigation lights using traditional UAVs. It ensures the accuracy of navigation light positions, maintains the normal function of the airport navigation system, provides reliable guidance for aircraft takeoff and landing, reduces the risk of flight accidents caused by navigation light shifts from the source, and ensures aviation safety.

[0090] Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0091] In the embodiments provided by this invention, it should be understood that the disclosed system or method can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for instance, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0092] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0093] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in the form of hardware plus software functional modules.

[0094] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the basic characteristics of the present invention.

[0095] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for inspecting airport navigation lights based on unmanned aerial vehicle (UAV) technology, characterized in that, Includes the following steps: S1. Obtain the historical detection coordinate database of navigation lights; The drone takes off from the ground control station and conducts drone patrols over the airport navigation lights area according to a preset basic route. During drone patrol, image data of navigation lights, three-dimensional distance change rate, drone rotation speed, atmospheric humidity and light intensity are collected in real time; S2. Based on the image data of the navigation lights, the three-dimensional distance change rate, and the UAV rotation rate, construct the motion feature vector M for each navigation light; S3. The motion feature vector M of the navigation light is processed by the Kalman filter algorithm. The environmental adaptive noise matrix is ​​used to replace the fixed noise matrix of the Kalman filter algorithm to generate the current navigation light detection coordinates after denoising. S4. Based on the historical detection coordinate database of navigation lights in S1, calculate the mean μ and standard deviation of latitude and longitude of the historical detection coordinates of each navigation light. The normal fluctuation range is set to μ±3. ; When the current navigation light detection coordinates exceed the normal fluctuation range, a low-altitude circumferential re-inspection is triggered. When the current navigation light detection coordinates are within the normal fluctuation range, calculate the spatiotemporal consistency score between the current navigation light detection coordinates and μ. If the spatiotemporal consistency score is greater than the set threshold, trigger a low-altitude orbital re-inspection. S5. During low-altitude orbital re-inspection, multi-angle consistency scores are calculated through multi-angle imaging. If the drone patrol triggers low-altitude orbital re-inspection in three consecutive inspections and the multi-angle consistency scores are all greater than the set threshold, it is determined to be a real offset caused by foundation settlement. If the displacement is determined to be a genuine result of foundation settlement, an alarm will be issued and a foundation reinforcement and maintenance work order will be generated.

2. The airport navigation light inspection method based on UAV technology according to claim 1, characterized in that, The equipment carried by the drone includes an atmospheric humidity sensor, a light intensity sensor, an optical flow sensor, a lidar, an attitude sensor, and image acquisition equipment.

3. The airport navigation light inspection method based on UAV technology according to claim 2, characterized in that, In S2, the motion feature vector M of the navigation lights is constructed as follows: The two-dimensional displacement vector of the navigation light in the image plane of the image data was calculated using the optical flow method. and Assume the rate of change of the three-dimensional distance measured in real time by the lidar is... The drone's rotation speed is Construct motion feature vectors for navigation lights ,in, Δt represents the change in laser ranging of the lidar, and Δt represents the time interval between two consecutive lidar measurements.

4. The airport navigation light inspection method based on UAV technology according to claim 3, characterized in that, In S3, an environment-adaptive noise matrix is ​​used instead of the fixed noise matrix provided by the Kalman filter algorithm. The environment-adaptive noise matrix is ​​constructed as follows: Let the fixed noise matrix of the Kalman filter algorithm be... ; Let H be the real-time atmospheric humidity obtained in S1 and L be the light intensity; Environmental Adaptive Noise Matrix The calculation formula is: In the formula, α and β are both environmental sensitivity coefficients.

5. The airport navigation light inspection method based on UAV technology according to claim 1 or 4, characterized in that, In S4, the spatiotemporal consistency score between the current navigation light detection coordinates and μ is calculated as follows: Calculate the Euclidean distance D between the current navigation light detection coordinates and μ, and assume the spatiotemporal consistency score is... ; The calculation formula is: In the formula, Δt is the detection time interval.

6. The airport navigation light inspection method based on UAV technology according to claim 5, characterized in that, In S5, the low-altitude orbital re-inspection is as follows: When a low-altitude surround re-inspection is triggered, the UAV will take the current navigation light detection coordinates as the center and collect visual images at 8 equally spaced angles within the set altitude and radius area. The latitude and longitude coordinates of the current navigation light at the eight angles are obtained by analyzing the view images acquired from eight angles. ,in, Represents 8 equally spaced angles. =1, 2, ..., 8.

7. The airport navigation light inspection method based on UAV technology according to claim 6, characterized in that, In S5, multi-angle consistency scores are calculated. as follows: ;in , The latitude and longitude average of the eight angular coordinates of the current navigation light. For preset accuracy parameters, ≠0.

8. The airport navigation light inspection method based on UAV technology according to claim 6, characterized in that, When the atmospheric humidity H > 80% and the light intensity L < 300 lux, when the low-altitude orbital re-inspection is triggered, the number of viewing angles for low-altitude orbital re-inspection multi-angle imaging increases to twice the original number, and the drone altitude decreases to G times the set altitude, where 0 < G < 1.

9. The airport navigation light inspection method based on UAV technology according to claim 5, characterized in that, If the Euclidean distance D between the current detection coordinates and μ is greater than 0.1 meters, skip the spatiotemporal consistency score calculation and force a low-altitude orbital re-inspection.

Images