A real-time monitoring method and system for flight behavior of a non-cooperative unmanned aerial vehicle

Abstract

The application discloses a kind of real-time monitoring method and system of non-cooperative unmanned aerial vehicle flight behavior, it is related to unmanned aerial vehicle behavior supervision technical field, to the distributed monitoring array deployment of unmanned aerial vehicle control airspace, execute acoustic sensing under the harmonic monitoring of rotor noise, wake up positioning cluster, execute piezoelectric sensing under the low-frequency pressure wave monitoring, according to edge gateway, if it is non-cooperative unmanned aerial vehicle, return to monitoring central control and carry out kinematic behavior model matching, determine unmanned aerial vehicle behavior data, carry out terminal interface display and abnormal behavior alarm and control, for solving the technical problems that exist in prior art, it is difficult to achieve live adaptive dynamic monitoring and identification, leading to identification precision is limited and alarm exists hysteresis, it is difficult to meet from monitoring to control quick flexible response, give consideration to front-end dynamic monitoring and back-end efficient processing, realize the adaptive flexible supervision and analysis of unmanned aerial vehicle flight behavior, improve identification precision and response immediacy.

Description

Technical Field

[0001] This invention relates to the field of drone behavior monitoring technology, specifically to a real-time monitoring method and system for the flight behavior of non-cooperative drones. Background Technology

[0002] With the popularization of drone technology, non-cooperative drones, i.e., unauthorized or unreported flights, pose a serious threat to airspace security, and existing monitoring methods have obvious limitations.

[0003] Currently, most mainstream solutions rely on single-sensor monitoring. For example, pure acoustic sensing is susceptible to environmental noise interference, leading to false alarms, while pure radar monitoring suffers from high costs and low-altitude blind spots. Monitoring modes lack specificity, and activating all sensor components throughout the process wastes resources. Furthermore, the efficiency in distinguishing between cooperative and non-cooperative drones is low, making it difficult to meet real-time monitoring needs.

[0004] In summary, existing technologies have limitations in front-end monitoring and back-end identification, making it difficult to achieve real-time adaptive dynamic monitoring and identification. This results in limited identification accuracy and delayed alarms, making it difficult to meet the needs of rapid and flexible response from monitoring to control. Summary of the Invention

[0005] This application provides a real-time monitoring method and system for the flight behavior of non-cooperative drones, which addresses the technical problem in the prior art that it is difficult to achieve adaptive dynamic monitoring and identification in a real-time manner, resulting in limited identification accuracy and delayed alarms, making it difficult to meet the requirements for rapid and flexible response from monitoring to control.

[0006] In view of the above problems, this application provides a method and system for real-time monitoring of the flight behavior of non-cooperative unmanned aerial vehicles (UAVs).

[0007] Firstly, this application provides a real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles (UAVs). The method includes: deploying a distributed monitoring array in the UAV-controlled airspace, wherein each monitoring point includes a cascaded deployment of acoustic sensors and piezoelectric sensors; driving the monitoring array to perform rotor noise harmonic monitoring under acoustic sensing, waking up a positioning cluster to perform low-frequency pressure wave monitoring under piezoelectric sensing, and determining the monitoring data, wherein the positioning cluster is a piezoelectric sensor cluster in the vicinity of the acoustic sensor location; judging the monitoring data according to the edge gateway, and if it is a non-cooperative UAV, transmitting it back to the monitoring control center, and determining the UAV behavior data by performing kinematic behavior model matching; displaying the UAV behavior data on a terminal interface, and performing alarms and control of abnormal behavior.

[0008] Secondly, this application provides a real-time monitoring system for the flight behavior of non-cooperative unmanned aerial vehicles (UAVs). The system includes: a monitoring deployment unit for deploying a distributed monitoring array over UAV-controlled airspace, wherein each monitoring point comprises a cascaded deployment of acoustic sensors and piezoelectric sensors; a cascaded monitoring unit for driving the monitoring array to perform rotor noise harmonic monitoring under acoustic sensing, waking up a positioning cluster to perform low-frequency pressure wave monitoring under piezoelectric sensing, and determining the monitoring data, wherein the positioning cluster is a piezoelectric sensor cluster in the vicinity of the acoustic sensor location; a behavior analysis unit for judging the monitoring data based on an edge gateway, and if it is a non-cooperative UAV, transmitting it back to the monitoring control center, and determining the UAV behavior data by performing kinematic behavior model matching; and a behavior management unit for displaying the UAV behavior data on a terminal interface and performing alarms and control of abnormal behavior.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application provides a real-time monitoring method for the flight behavior of non-cooperative drones. A distributed monitoring array is deployed over the drone-controlled airspace. By driving the monitoring array, rotor noise harmonic monitoring is performed using acoustic sensing. A positioning cluster is activated to perform low-frequency pressure wave monitoring using piezoelectric sensing. Monitoring data is determined, and the data is evaluated using an edge gateway. If the drone is non-cooperative, the data is transmitted back to the monitoring control center. Kinematic behavior model matching is performed to determine the drone's behavior data. The drone behavior data is displayed on a terminal interface, and alarms and control measures are implemented for abnormal behavior. This method addresses the technical problems in existing technologies, such as difficulty in achieving real-time adaptive dynamic monitoring and identification, resulting in limited identification accuracy and delayed alarms, hindering rapid and flexible response from monitoring to control. It balances front-end dynamic monitoring with efficient back-end processing, achieving adaptive and flexible monitoring and analysis of drone flight behavior, improving identification accuracy and response immediacy. Attached Figure Description

[0011] Figure 1 This application provides a schematic diagram of a real-time monitoring method for the flight behavior of non-cooperative drones;

[0012] Figure 2 This application provides a schematic diagram of a real-time monitoring system for the flight behavior of non-cooperative unmanned aerial vehicles.

[0013] Explanation of reference numerals in the attached diagram: Monitoring deployment unit 11, cascade monitoring unit 12, behavior analysis unit 13, behavior management unit 14. Detailed Implementation

[0014] This application provides a real-time monitoring method and system for the flight behavior of non-cooperative drones to solve the technical problems in the prior art, which are difficult to achieve real-time adaptive dynamic monitoring and identification, resulting in limited identification accuracy and delayed alarms, making it difficult to meet the requirements of rapid and flexible response from monitoring to control.

[0015] Example 1: As Figure 1 As shown, this application provides a real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles (UAVs), the method comprising:

[0016] S1: Deploy a distributed monitoring array for the airspace controlled by drones, wherein each monitoring point includes a cascaded deployment of acoustic sensors and piezoelectric sensors.

[0017] In this embodiment of the application, a distributed monitoring array is deployed in the airspace controlled by unmanned aerial vehicles. Through multi-point deployment in space and the functional synergy of sensing components, a monitoring coverage without blind spots is formed.

[0018] Specifically, the distributed monitoring array in this application is deployed within the airspace controlled by unmanned aerial vehicles (UAVs) at preset spatial intervals, for example, one monitoring point every 100 meters in a square control area with a side length of 500 meters. Multiple monitoring points are evenly distributed in the three-dimensional space to be controlled, and data interaction between the monitoring points is achieved through wired or wireless communication links to ensure that UAV signals at different locations in the airspace can be effectively captured.

[0019] Meanwhile, each monitoring point includes a cascaded deployment of acoustic and piezoelectric sensors, meaning that the two sensing components have a sequential triggering and data correlation relationship in terms of function, rather than a simple superposition of physical locations.

[0020] In one specific implementation, the acoustic sensing component, as a primary sensing unit, mainly collects sound signals in the airspace to initially identify whether there are acoustic features related to the UAV; the piezoelectric sensing component, as a secondary sensing unit, such as a piezoelectric thin film sensor, is in a low-power standby state when the acoustic sensing component does not detect a target, and is only woken up and started after the acoustic sensing component detects a suspected target.

[0021] For example, when the acoustic sensing component at a certain monitoring point detects rotor noise with a frequency of 200-500Hz, it immediately sends a wake-up signal to the piezoelectric sensing components at the same monitoring point and neighboring monitoring points, enabling them to enter the working state. This achieves the coordinated work of the two sensing components, ensuring the timeliness of monitoring while reducing unnecessary energy consumption.

[0022] S2: By driving the monitoring array, rotor noise harmonic monitoring under acoustic sensing is performed, the positioning cluster is woken up, low-frequency pressure wave monitoring under piezoelectric sensing is performed, and the monitoring data is determined, wherein the positioning cluster is a piezoelectric sensing cluster in the neighborhood of the acoustic sensing position.

[0023] In this embodiment, a start command is sent to the distributed monitoring array to drive the array, causing the sensing components at each monitoring point to enter a working state according to a preset program. In a feasible driving process, an electrical signal from the monitoring control center triggers the operation, ensuring that all monitoring points begin executing monitoring tasks at a unified timing.

[0024] In a specific embodiment, the periodic acoustic signals generated by the rotation of the UAV rotor are used as the acoustic monitoring standard to perform rotor noise harmonic monitoring under acoustic sensing. The rotor noise harmonics refer to signal waves whose frequency characteristics are directly related to the rotor speed and the number of blades. For example, quadcopter UAVs typically generate characteristic harmonics in the 150-800Hz range. The acoustic sensing component collects acoustic signals within the UAV's controlled airspace, uses filtering to separate environmental harmonics from rotor noise harmonics, extracts the rotor noise harmonic components, and completes directional monitoring of rotor noise.

[0025] Subsequently, when the acoustic sensor detects a suspected rotor noise harmonic, it triggers the activation mechanism of the positioning cluster. In this embodiment, the positioning cluster is a piezoelectric sensor cluster that monitors the location of the acoustic sensor and neighboring monitoring points.

[0026] Specifically, if the acoustic sensing component at a certain monitoring point detects the target signal, the piezoelectric sensing components contained in that monitoring point and other monitoring points within a 50-meter radius of it will form a positioning cluster. This cluster can improve positioning accuracy through multi-point collaboration while avoiding resource consumption caused by the activation of piezoelectric sensing components across the entire area.

[0027] Subsequently, low-frequency pressure wave monitoring under piezoelectric sensing is performed. In the drone flight scenario, the drone generates periodic pressure disturbances to the surrounding air, forming low-frequency pressure waves with a frequency below 20Hz. The piezoelectric sensing component can obtain wavefront data by sensing this pressure change to reflect the propagation direction and intensity of the pressure wave. In this application, the spatial positioning of the drone can be achieved by combining the monitoring results of multiple sensing components.

[0028] In summary, by fusing the rotor noise harmonic data acquired by acoustic sensing with the low-frequency pressure wavefront data acquired by piezoelectric sensing, monitoring data containing target acoustic characteristics and spatial location information is formed, providing complete input for subsequent target determination.

[0029] Furthermore, before driving the monitoring array, step S2 of this application includes:

[0030] The rotor noise spectrum of each UAV model is collected, and rotor-specific harmonics are screened based on environmental noise to determine a rotor harmonic feature set, wherein each UAV model corresponds to at least one rotor harmonic feature; a first sensing rule is determined by directional picking of the rotor harmonic feature set; and the acoustic sensing part in the monitoring array is initialized based on the first sensing rule.

[0031] In this embodiment, the rotor noise spectrum of various drone models is collected. Specifically, for various common drones on the market, including consumer and industrial models, the noise signal generated by the rotor rotation during flight is collected, and the noise signal is converted into spectrum data with frequency as the horizontal axis and sound pressure level as the vertical axis through spectrum analysis. For example, for a quadcopter consumer drone, its noise spectrum in typical states such as hovering and cruising can be collected to obtain the noise energy distribution at different frequencies.

[0032] Furthermore, by performing rotor-specific harmonic screening under environmental noise, harmonic characteristics that are distinct from environmental interference are identified as rotor harmonic feature sets to meet the needs of efficient and accurate front-end monitoring and discrimination.

[0033] The rotor-specific harmonics refer to specific frequency components in the UAV rotor noise that differ from ambient noise. Ambient noise typically includes randomly distributed broadband noise, such as wind noise and vehicle noise. The rotor, due to its periodic rotation, generates harmonics of fixed frequencies. For example, at a rotor speed of 600 revolutions per minute, the fundamental frequency is 10Hz, and its second harmonic (20Hz), third harmonic (30Hz), etc., are all harmonic components. By comparing the UAV rotor noise spectrum with the ambient noise spectrum, harmonic components that are not present in the ambient noise or whose intensity is significantly higher than the ambient noise are identified, thus obtaining the rotor-specific harmonics.

[0034] Subsequently, the aforementioned specific harmonics were classified and organized according to UAV models to form a rotor harmonic feature set. Each UAV model corresponds to at least one rotor harmonic feature. For example, a six-rotor UAV can correspond to the harmonic periodicity at the fundamental frequency of 15Hz and the third harmonic frequency of 45Hz.

[0035] Furthermore, using the directional pickup of the rotor harmonic feature set as the first sensing rule, the directional pickup in this application refers to the acoustic sensing component only collecting specific frequency components in the rotor harmonic feature set, rather than receiving sound signals of all frequencies indiscriminately.

[0036] Based on this, the first sensing rule is specifically the setting specification of the working parameters of the acoustic sensing component, including the target monitoring frequency range, such as the concentrated frequency band of the focused rotor harmonics and the frequency band trend; signal strength threshold, such as triggering acquisition when the signal strength of a certain frequency exceeds 3 times the average environmental noise, etc.

[0037] For example, for drone models that include 15Hz and 45Hz features, the first sensing rule can be set to prioritize monitoring the frequency ranges of 15-18Hz and 43-47Hz, and mark signals with a signal strength ≥60dB in these ranges as valid signals.

[0038] Furthermore, the acoustic sensing portion of the monitoring array is initialized according to the first sensing rule, that is, the hardware parameters and software logic of each acoustic sensing component in the monitoring array are configured according to the first sensing rule.

[0039] In one specific implementation, on the hardware side, the filtering module of the acoustic sensor is adjusted to allow only frequency signals corresponding to the rotor harmonic characteristic set to pass through. For example, a bandpass filter is used to retain the target frequency band signal of 10-200Hz and filter out high-frequency environmental noise. On the software side, a preset signal acquisition trigger condition is set. When a signal that meets the first sensing rule is detected, data recording is automatically started; otherwise, a low-power standby state is maintained. Subsequently, the recorded data is periodically analyzed to determine whether it meets the motion characteristics of the UAV rotor.

[0040] In summary, the acoustic sensing component can accurately focus on the target signal and perform orientation determination, reducing invalid data collection and improving subsequent monitoring efficiency.

[0041] Furthermore, performing rotor noise harmonic monitoring via acoustic sensing, step S2 of this application includes:

[0042] Acoustic harmonic data is determined by driving the acoustic sensors in the monitoring array; the acoustic harmonic data is verified using the rotor harmonic feature set as the directional picking target; if the verification result is empty, the subsequent monitoring thread is terminated.

[0043] In this embodiment of the application, by monitoring the control signal sent by the central control unit, all acoustic sensing components in the monitoring array are switched from the initialization state to the real-time acquisition state. Each acoustic sensing component acquires the acoustic signal in the airspace controlled by the UAV according to the first sensing rule, and converts the acquired time-domain sound signal into frequency-domain acoustic harmonic data through the built-in spectrum analysis module, such as by using Fourier transform.

[0044] Subsequently, using the rotor harmonic feature set as the directional picking target, the acoustic harmonic data is verified, that is, the real-time acquired acoustic harmonic data is compared and matched with the pre-constructed rotor harmonic feature set. Specifically, the sound pressure level of each frequency point is extracted from the acoustic harmonic data, and frequency components with sound pressure levels exceeding a preset threshold, such as 6dB higher than the ambient noise reference value, are screened out. Then, the screened frequency components are compared one by one with the specific harmonics corresponding to each UAV model in the rotor harmonic feature set to determine whether there is a match. For example, if the deviation of a certain frequency component from the fundamental frequency or harmonic of a certain model in the rotor harmonic feature set does not exceed ±0.5Hz, it is determined to be a match.

[0045] If the test result is empty, it indicates that no frequency components matching the rotor harmonic feature set were found in the acoustic harmonic data, meaning no suspected UAV rotor noise signal was detected within the current monitoring range. In this case, to avoid unnecessary resource consumption, the monitoring process termination mechanism is triggered.

[0046] One specific termination process includes: controlling the acoustic sensing component to switch from real-time acquisition state back to low-power standby state, without sending a wake-up signal to the piezoelectric sensing component, and stopping subsequent data transmission and processing threads related to this monitoring, for example, terminating the further analysis process of acoustic data for this period until the monitoring process is re-executed when the next periodic start or external trigger signal arrives.

[0047] Furthermore, step S2 of this application includes:

[0048] If the verification result is not empty, the target acoustic sensing component is locked, wherein the target acoustic sensing component is the component verified in the non-empty test; the target acoustic sensing component, along with piezoelectric sensing components at the same monitoring point and neighboring monitoring points, are used as a positioning cluster to wake up the positioning cluster and drive the low-frequency pressure wave monitoring under piezoelectric sensing to determine the wavefront data; the verification result and the wavefront data are integrated as the monitoring data and transmitted to the target edge gateway to which the coverage area belongs.

[0049] In this embodiment of the application, if the detection result is not empty, it indicates that there is a signal in the acoustic harmonic data that matches the rotor harmonic feature set, that is, a suspected UAV target appears within the monitoring range. At this time, it is necessary to lock the target's acoustic sensing components.

[0050] Specifically, this refers to recording the physical location identifier of the component, such as number S-012, and maintaining its continuous acquisition state without entering a low-power mode. The target acoustic sensing component is a non-empty verification component; for example, when acoustic sensing components numbered S-012 and S-013 both detect matched harmonics, both components are identified as the target acoustic sensing component.

[0051] Subsequently, using the target acoustic sensing component as a reference, piezoelectric sensing components at the same monitoring point and neighboring monitoring points are selected as a positioning cluster. The same monitoring point refers to a piezoelectric sensing component deployed at the same physical location as the target acoustic sensing component, and the neighboring monitoring points refer to other monitoring points within a preset radius centered on the monitoring point where the target acoustic sensing component is located. The piezoelectric sensing components at these monitoring points are collectively organized into a positioning cluster.

[0052] This then wakes up the positioning cluster, which is achieved by sending an electrical signal, such as a 3.3V trigger pulse, through the target acoustic sensing component. This switches the piezoelectric sensing component, which is in standby mode, to the working mode, driving it to perform low-frequency pressure wave monitoring under piezoelectric sensing.

[0053] Specifically, the low-frequency pressure waves generated by the rotor's air disturbance during drone flight act on the sensitive elements of the piezoelectric sensing components, such as the piezoelectric film, causing a change in charge. This change is collected and converted into an electrical signal, which is then converted from analog to digital to obtain wavefront data. The wavefront data includes the time difference between the arrival of the pressure wave at each sensing component (e.g., the reception time difference between P-012 and P-013 is 0.02s); and the pressure amplitude (e.g., 20mV corresponds to 0.5Pa pressure).

[0054] Furthermore, the test results are integrated with the wavefront data, that is, the harmonic frequencies and signal strengths in the test results are associated with the time difference and pressure amplitude in the wavefront data and stored to form structured data containing the target acoustic characteristics and spatial propagation characteristics.

[0055] For example, the data could be encapsulated in JSON format as {harmonic frequency, intensity, time difference, etc.}. This monitoring data is then transmitted to the target edge gateway whose coverage area is assigned. In this application, "coverage area" refers to the sensor components associated with each edge gateway within its coverage area. Data backhaul from the sensor monitoring end to the edge gateway is then performed based on this association.

[0056] For example, if the target acoustic sensing component S-012 is located in the coverage area of ​​the edge gateway numbered G-03, the data is transmitted to the G-03 gateway to provide complete input for subsequent target determination.

[0057] S3: Based on the edge gateway, the monitoring data is judged. If it is a non-cooperative drone, it is sent back to the monitoring center. By performing kinematic behavior model matching, the drone behavior data is determined.

[0058] In this embodiment of the application, localized analysis and identification operations are performed on the monitoring data transmitted to it by utilizing an edge gateway deployed within the unmanned aerial vehicle (UAV) controlled airspace partition.

[0059] Among them, the edge gateway, as an intermediate node with lightweight data processing and storage capabilities, pre-stores the registration information of all cooperative drones within its coverage area, including model, preset route, authorized flight time, and corresponding characteristic parameters.

[0060] Specifically, the edge gateway first extracts core features from the monitoring data, including the rotor harmonic characteristics of the UAV, the real-time position coordinates obtained from wavefront data inversion, and the preliminary trajectory trend. These extracted features are then compared and matched with a feature library of cooperative UAVs stored locally on the edge gateway.

[0061] Firstly, if the rotor harmonic characteristics, real-time position, and trajectory trend in the monitoring data all fall within the preset parameter range of a certain type of cooperative drone, for example, if the model matches and the position is on its authorized flight path, then it is determined to be a cooperative drone. In this case, the edge gateway only records this information and does not upload subsequent data.

[0062] Secondly, if the features in the monitoring data do not match the preset parameters of all cooperative drones, such as not finding the registration information of the corresponding rotor harmonic features, or the position deviating from all authorized routes by more than the preset threshold, then it is determined to be a non-cooperative drone.

[0063] Specifically, if the determination result is that the drone is a non-cooperative type, the edge gateway will transmit the monitoring data, including complete acoustic harmonic data, wavefront data and preliminary determination results, back to the monitoring and control system through an encrypted communication link, such as a 5G private network.

[0064] Subsequently, after receiving the data, the monitoring control center initiates the kinematic behavior model matching process. In this application, the kinematic behavior model refers to a set of standardized behavior patterns constructed based on the flight characteristics of the UAV. For example, the straight-line cruise model corresponds to a stable speed of 5-10 m / s and a heading angle change rate of ≤1° / s, while the hovering reconnaissance model corresponds to a speed of ≤0.5 m / s and a position change range of ≤3 meters.

[0065] The specific matching process is as follows: The monitoring control center parses the real-time motion parameters of the non-cooperative drone from the transmitted data, such as the current speed of 8 m / s and acceleration of 0.2 m / s². 2 The heading angle changes by 5° every 10 seconds. Similarity analysis is performed on the parameters of each model in the preset kinematic behavior model library. In a preferred embodiment, Euclidean distance or cosine similarity is used as the analysis method to select the behavior model with the highest similarity, such as more than 90%, as the matching result.

[0066] Subsequently, based on the matching results, combined with the drone's real-time location and trajectory trend, the drone's behavior data is determined. For example, if the drone is matched with a straight-line cruise model and its location is moving towards the no-fly zone, the behavior data is recorded as a suspected violation of the rules, with a current speed of 8 m / s and an estimated entry into the no-fly zone in 5 minutes.

[0067] Furthermore, prior to determining the monitoring data based on the edge gateway, the deployment of the edge gateway, in step S3 of this application, includes:

[0068] The flight range information of the cooperating drones is obtained, and the flight range information is spatially segmented according to the coverage of the edge gateway to determine the flight range fragment map. Based on the regional association, the flight range fragment map is distributed and stored to each edge gateway, wherein the flight range fragment map is updated synchronously as the flight range of the cooperating drones changes.

[0069] In this embodiment, the flight path information of the cooperative drone refers to the collection of preset flight path data of all authorized cooperative drones through the cooperative drone registration system or flight plan submission platform. This data includes, but is not limited to, the latitude and longitude coordinates of the takeoff point, waypoints, and landing point, the estimated flight time for each segment, and the effective time period of the route. The above information is summarized in a structured data format, such as GPX format, to form a complete cooperative drone flight path dataset.

[0070] Subsequently, the flight information is spatially segmented based on the coverage area of ​​the edge gateway. Specifically, the coverage area of ​​the edge gateway is a pre-defined geographical region, and each edge gateway corresponds to a local area within the UAV controlled airspace.

[0071] In a feasible embodiment, using Geographic Information System (GIS) tools, the complete flight path of the cooperative UAV is spatially overlaid with the coverage area of ​​each edge gateway. The segments of the flight path that fall within the coverage area of ​​each edge gateway are extracted, forming flight path sub-segments corresponding to that edge gateway. All flight path sub-segments corresponding to all edge gateways are then organized according to geographic coordinates and time series to determine the flight path fragment map.

[0072] For example, during the complete flight of the collaborative drone U1, the segment that falls within the coverage area of ​​the edge gateway G1 is extracted as a record in G1's flight fragment map, containing the start and end coordinates of the segment and the estimated transit time.

[0073] Furthermore, based on regional associations, i.e. the geographical correspondence between each edge gateway and its coverage area, for example, the flight fragment map corresponding to edge gateway G1 is sent to G1's local storage module, such as an embedded database, via an encrypted transmission protocol, such as HTTPS.

[0074] The flight path fragment map is updated synchronously with the flight path changes of the cooperating UAV to ensure the real-time nature and effectiveness of the information.

[0075] Specifically, when a cooperative drone adjusts its flight plan, such as temporarily changing waypoints, or when a new cooperative drone is registered, the updated flight information is synchronized to the central server in real time. The central server then re-executes spatial segmentation to generate an updated flight fragment map, and distributes the changed portion to the corresponding edge gateway to ensure that the flight fragment map stored at the edge gateway is consistent with the actual flight information. For example, if drone U1 temporarily cancels a flight segment passing through the coverage area of ​​G1, the flight fragment map of G1 will immediately delete the record of that segment.

[0076] Furthermore, based on the edge gateway, the monitoring data is judged. Step S3 of this application includes:

[0077] The system retrieves the range fragment map stored at the target edge gateway, performs fuzzy inversion of the UAV trajectory based on the wavefront data, and determines the flight data, which includes at least short-term track, heading angle, and flight altitude. Based on the determination result, the UAV model is identified, and the flight data is matched against the range fragment map to determine whether it is a cooperative UAV. If it is, the subsequent monitoring thread is terminated; if not, the UAV model and flight data are encapsulated and sent back to the monitoring center.

[0078] In this embodiment of the application, the range fragment map stored by the target edge gateway is retrieved. That is, the range fragment map that is pre-stored and associated with its own coverage area is read through the local data call interface of the target edge gateway. Specifically, it includes the preset flight segment information of all cooperating UAVs within the coverage area of ​​the edge gateway.

[0079] Subsequently, fuzzy inversion of the UAV trajectory is performed based on the wavefront data. In one specific implementation, low-frequency pressure wavefront data collected by multiple piezoelectric sensing components, including information such as the time difference of pressure wave arrival at each component and pressure amplitude, is used. Combined with the spatial coordinates of each sensing component, the real-time position of the UAV is calculated by acoustic positioning, such as time difference positioning. Based on the position information of multiple consecutive moments, its motion trajectory is fitted to determine features such as short-term track, heading angle, and flight altitude, which are used as the flight data.

[0080] Meanwhile, since environmental interference may cause errors in single positioning, such as ±5 meters, a fuzzy processing method is adopted to allow flight data to contain a certain error range.

[0081] Furthermore, based on the verification results, namely, extracting rotor harmonic features from the acoustic sensing verification results, comparing them with a preset UAV model-harmonic feature correspondence library, and finding the UAV model with the highest feature matching degree, if the matching degree exceeds 95%, it is determined to be that model.

[0082] Subsequently, based on the drone model and the flight data, a match is made in the flight debris map to determine whether it is a cooperative drone. The specific process includes comparing the identified drone model and the retrieved flight data with the cooperative drone models and their preset flight segments recorded in the flight debris map. If a cooperative drone exists whose model matches the identification result, and whose real-time position, direction of movement, and time information in the flight data all fall within the error allowable range of the cooperative drone's preset flight segment (e.g., position deviation ≤ 10 meters, time deviation ≤ 5 minutes), it is determined to be a cooperative drone. Conversely, if no cooperative drone record matching the criteria is found, it is determined to be a non-cooperative drone.

[0083] First, if the drone is determined to be a cooperative drone, in order to avoid unnecessary resource consumption, the subsequent monitoring thread will be terminated, including stopping further piezoelectric sensing data acquisition from the drone and ceasing data transmission and analysis.

[0084] Secondly, if the drone is determined to be a non-cooperative drone, the drone model and flight data are encapsulated. That is, the drone model information and flight data are integrated and packaged according to a preset data format, such as JSON, to form a structured data packet, and then transmitted back to the monitoring center through an encrypted communication link, such as a dedicated wireless network, to provide data support for subsequent behavior analysis and control.

[0085] Furthermore, kinematic behavior model matching is performed to determine the drone's behavior data. Step S3 of this application includes:

[0086] The monitoring and control center receives the UAV model and the flight data, and determines the target kinematic behavior model by identifying behavioral elements and matching based on a behavioral model library, wherein the matching constraint is based on the maximum approximation; for the target kinematic behavior model, the UAV behavior data is determined by multi-threaded evaluation and fusion based on the multi-level approximation of behavioral elements.

[0087] In this embodiment of the application, the monitoring control center receives the UAV model and the flight data, and then performs integrity verification on the data. For example, it verifies whether the data frame is missing sampling points; otherwise, it performs interpolation compensation to ensure the effectiveness of subsequent processing.

[0088] Subsequently, behavioral element identification based on UAV model and flight data is performed, and a behavioral model library is traversed for matching to determine the target kinematic behavior model. The behavioral elements refer to key motion parameters extracted from the flight data, including instantaneous velocity, rate of change of velocity, rate of change of heading angle, and range of position variation; the kinematic behavior model library is a pre-built standardized set of models covering common UAV behavior patterns.

[0089] For example, a straight-line cruise model corresponds to a stable speed of 5-10 m / s and an acceleration ≤0.5 m / s². 2 The hovering model corresponds to a speed ≤0.5m / s and a position change ≤2m; the circumnavigation model corresponds to a heading angle change rate ≥30° / minute and a circular trajectory.

[0090] In one feasible implementation, the aforementioned parameters are extracted from flight data, and then the extracted parameters are compared with the standard parameters of each model in the model library to calculate their approximation. For example, Euclidean distance is used; the smaller the parameter deviation, the higher the approximation. The maximum approximation is used for matching constraints, that is, among the matching results with an approximation greater than a preset approximation, the one with the maximum approximation is determined as the target kinematic behavior model. This avoids matching ambiguity caused by the existence of multiple low-approximation models.

[0091] Subsequently, the target kinematic behavior model is evaluated and fused in separate threads based on a multi-level approximation hierarchy of behavioral elements. Specifically, the multi-level approximation hierarchy refers to classifying behavioral elements according to their importance. For example, first-level elements are core parameters, such as velocity and acceleration, with a weight of 60%; second-level elements are auxiliary parameters, such as the rate of change of heading angle, with a weight of 30%; and third-level elements are secondary parameters, such as the frequency of position changes, with a weight of 10%.

[0092] Furthermore, parallel threaded evaluations are performed for elements at each level, meaning independent calculation threads are initiated for each level of element. For example, the first-level thread evaluates the deviation rate between the actual speed and the target model's standard speed, such as a deviation rate of 2.5% between the actual speed of 7.8 m / s and the standard value of 8 m / s; the second-level thread evaluates the consistency of the heading angle change rate. After the evaluation is completed, the results at each level are integrated by weighted summation.

[0093] Finally, the above fusion results are used to determine the drone behavior data, which includes, but is not limited to, behavior type, such as straight-line cruising; behavior characteristic parameters, such as average speed of 7.6 m / s and stable heading; and behavior trend, such as continuous movement in the northeast direction, expected to reach 116.40° east longitude in 5 minutes. At the same time, the confidence level of each parameter is marked to provide a quantitative basis for subsequent abnormal behavior judgment.

[0094] Furthermore, regarding the construction of the behavior model library, step S3 of this application includes:

[0095] The drone flight records are retrieved, and the records are clustered and kinematic behavior features are mined based on behavior categories to establish a behavior model library, which contains multiple kinematic behavior models. The behavior model library is traversed, and behavior weights are adjusted. The adjustment is embedded in the monitoring center, where the behavior weights are adjusted by increasing the weights of specific behaviors and decreasing the weights of non-specific behaviors among the kinematic behavior models.

[0096] In this embodiment of the application, retrieving and retrieving UAV flight records can be achieved by collecting flight data of different models and scenarios of UAVs through UAV management databases, public flight logs, and actual measurement data collection. The UAV flight records include time-series location coordinates, flight parameters at corresponding times, and behavioral tags, such as straight-line cruising, hovering, and circling.

[0097] Subsequently, behavior categories are used for record clustering and kinematic behavior feature mining to establish a behavior model library. Specifically, the behavior categories refer to categories divided according to flight modes, such as straight-line cruise, hovering, circling, and rapid penetration. Based on the behavior categories, the UAV flight records are clustered, grouping records of the same flight mode, i.e., records with similar flight parameters, into one category.

[0098] Further kinematic behavior feature mining is performed on various types of UAV flight records. This involves extracting the core parameter range from each clustered record, such as extracting the feature ranges of speed, acceleration, and heading angle change rate from the straight cruise record. This range and the corresponding behavior labels are then encapsulated into a kinematic behavior model.

[0099] Using this method, behavioral features are mined from various types of drone flight records to generate multiple models corresponding to various flight modes, which are then integrated to establish a behavioral model library. For example, the library includes straight-line cruise models, hovering models, and circling flight models.

[0100] Subsequently, the behavioral model library is traversed to adjust the behavioral weights in order to strengthen the distinguishing features between models, weaken the non-distinguishing features between models, and improve the recognizability.

[0101] Specifically, each kinematic behavior model in the model library is read one by one, and weight values ​​are assigned to the behavioral feature parameters contained in the model. The initial weights are balanced weights. Subsequently, further adjustments are made between kinematic behavior models, with the weights for specific behaviors increasing and the weights for non-specific behaviors decreasing.

[0102] Specifically, the specific behavior refers to the characteristic parameters that are unique to a certain model or significantly different from other models. For example, the heading angle change rate in the flight around model is ≥30° / min. This parameter rarely appears in the straight-line cruise and hovering models, so its weight is increased from the initial value.

[0103] Non-specific behavior refers to common parameters present in multiple models. For example, a speed of 5-8 m / s appears in both straight-line cruising and rapid penetration models, and its weight is reduced from the initial value. This adjustment enhances the discriminative power between models and avoids feature confusion.

[0104] Furthermore, the behavior model library is embedded in the monitoring and control center. For example, the behavior model library with adjusted weights is integrated into the hardware carrier of the monitoring and control center, such as an industrial control computer, as an algorithm module. By compiling the program, it becomes a built-in functional unit of the control system, which can directly call the model library for behavior matching without relying on external storage or computing resources, thus ensuring the real-time performance of behavior recognition.

[0105] S4: Display the drone behavior data on the terminal interface and execute alarms and control for abnormal behavior.

[0106] Furthermore, regarding the early warning and control of abnormal behavior, step S4 of this application includes:

[0107] The drone behavior data is identified, and behavior alarm information and behavior control schemes are generated based on drone management rules. According to the behavior alarm information and behavior control schemes, control terminals are deployed and non-cooperative drones are controlled.

[0108] In this embodiment, the drone behavior data is displayed on a terminal interface, that is, the determined drone behavior data is presented on the display interface of the monitoring terminal in a visual manner. Specific display examples include: a dynamic trajectory graph, which uses real-time updated lines to mark the drone's flight path, such as marking the already flown path with a solid red line on an electronic map, and using dashed lines to predict future trajectories; and a parameter dashboard, which displays key parameters such as current speed and heading angle, as well as behavior labels, in numerical and pointer formats. Simultaneously, the interface supports historical data review, allowing users to view changes in behavior data within past time zones via a timeline selection.

[0109] Subsequently, the behavior type and motion parameters in the behavior data are compared with preset safety thresholds. For example, when the behavior type is flying around and the center of the flight is a no-fly zone, such as an airport runway, or the speed exceeds the typical upper limit of the drone model, it is judged as abnormal behavior.

[0110] In this application, the drone management rules include response standards corresponding to different types and levels of anomalies. Taking abnormal behavior in no-fly zones as an example: Level 1 anomaly is suspected approach to the no-fly zone and corresponds to a warning prompt; Level 2 anomaly is entering the edge of the no-fly zone and corresponds to an audible and visual alarm and a drive-away command; Level 3 anomaly is high-speed intrusion into the core no-fly zone and corresponds to a forced landing command.

[0111] Meanwhile, the behavior alarm information must clearly specify the type of anomaly, such as Level 2 anomaly - intrusion into the edge of the no-fly zone, current location and risk level; the behavior control plan should include specific operation instructions. For example, the plan corresponding to Level 2 anomaly is to activate the directional acoustic drive-away device, issue a 120-decibel warning sound, and at the same time send a no-fly instruction via radio.

[0112] Finally, based on the behavior alarm information and the behavior control scheme, the control terminal is deployed and non-cooperative drones are controlled.

[0113] Specifically, alarm information and control plans are sent to the corresponding control terminals via data transmission links, such as ground control consoles or drone deterrence equipment terminals deployed around the no-fly zone. For example, the command to initiate directional acoustic deterrence is sent to the acoustic device terminal closest to the drone's current location, ensuring that the terminal completes command parsing upon receipt. The control terminal executes the control plan; for example, the acoustic device issues a warning sound as instructed. If the drone does not change its trajectory, radio jamming equipment is further activated to block the communication signal between the drone and the remote controller, forcing the drone to hover or return to base. Preferably, for level three abnormal situations, drone capture equipment, such as net gun launchers, can be linked to perform forced interception, while the equipment status during the control process is transmitted back to the monitoring center in real time, forming a closed-loop control system.

[0114] The real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles (UAVs) provided in this application has the following technical effects:

[0115] 1. Distributed Cascaded Sensor Array Deployment: A distributed monitoring array cascaded with acoustic and piezoelectric sensors is adopted. The acoustic sensors are responsible for the initial detection of rotor noise, while the piezoelectric sensors initiate positioning after acoustic detection triggers. This collaborative mode of detection followed by positioning avoids the blind zone problem of a single sensor and reduces ineffective energy consumption.

[0116] 2. Rotor Harmonic Feature Orientation Identification: A rotor harmonic feature set is pre-constructed, and acoustic sensors pick up signals directionally based on the feature set, initiating subsequent processes only when a matching harmonic is detected. This enables accurate differentiation between the UAV and environmental noise, reducing the false detection rate, and shortening target identification time through directional monitoring.

[0117] 3. Rapid Determination of Cooperative and Non-Cooperative Targets at the Edge: The edge gateway stores fragmented flight path maps of cooperative drones and determines target types locally by combining real-time flight data. Dynamic Behavior Model Matching is Employed: A dynamic model library is built based on kinematic features, and behavior types are determined through maximum approximation matching and multi-level factor evaluation. This can identify multiple typical behaviors, improving behavior recognition accuracy and solving the problem of poor adaptability in traditional fixed feature matching. A closed loop is formed from behavior display to anomaly alarms and control command issuance, supporting hierarchical control and achieving end-to-end response from detection to control, shortening the time for handling abnormal behaviors.

[0118] Example 2: Based on the same inventive concept as the real-time monitoring method for the flight behavior of a non-cooperative drone in the foregoing examples, such as... Figure 2 As shown, this application provides a real-time monitoring system for the flight behavior of non-cooperative unmanned aerial vehicles (UAVs), the system comprising:

[0119] The monitoring deployment unit 11 is used to deploy a distributed monitoring array in the airspace controlled by unmanned aerial vehicles, wherein each monitoring point includes a cascaded deployment of acoustic sensors and piezoelectric sensors.

[0120] The cascaded monitoring unit 12 is used to drive the monitoring array to perform rotor noise harmonic monitoring under acoustic sensing, wake up the positioning cluster, perform low-frequency pressure wave monitoring under piezoelectric sensing, and determine the monitoring data. The positioning cluster is a piezoelectric sensing cluster in the vicinity of the acoustic sensing position.

[0121] The behavior analysis unit 13 is used to determine the monitoring data based on the edge gateway. If it is a non-cooperative drone, it is sent back to the monitoring control center, and the drone behavior data is determined by kinematic behavior model matching.

[0122] The behavior management unit 14 is used to display the drone behavior data on the terminal interface and to perform alarms and control of abnormal behavior.

[0123] Furthermore, the cascaded monitoring unit 12 performs the following steps: collecting the rotor noise spectrum of the UAV model, determining the rotor harmonic feature set based on rotor-specific harmonic screening under environmental noise, wherein each UAV model corresponds to at least one rotor harmonic feature; determining the first sensing rule by directional picking of the rotor harmonic feature set; and initializing the acoustic sensing part in the monitoring array based on the first sensing rule.

[0124] Furthermore, the cascaded monitoring unit 12 performs the following steps: by driving the acoustic sensors in the monitoring array to monitor and determine acoustic harmonic data; using the rotor harmonic feature set as the directional picking target, verifying the acoustic harmonic data; if the verification result is empty, terminating the subsequent monitoring thread.

[0125] Furthermore, the cascaded monitoring unit 12 performs the following steps: if the verification result is not empty, lock the target acoustic sensing component, wherein the target acoustic sensing component is a component verified as not empty; using the target acoustic sensing component, and the piezoelectric sensing components of the same monitoring point and neighboring monitoring points as a positioning cluster, wake up the positioning cluster, drive the low-frequency pressure wave monitoring under piezoelectric sensing, and determine the wavefront data; integrate the verification result and the wavefront data as the monitoring data, and transmit them to the target edge gateway to which the coverage area belongs.

[0126] Furthermore, the behavior analysis unit 13 performs the following steps: acquiring the flight range information of the cooperating UAV, spatially segmenting the flight range information based on the coverage of the edge gateway, and determining the flight range fragment map; and storing the flight range fragment map to each edge gateway according to regional association, wherein the flight range fragment map is updated synchronously with the flight range changes of the cooperating UAV.

[0127] Furthermore, the behavior analysis unit 13 performs the following steps: retrieves the range fragment map stored in the target edge gateway, performs fuzzy inversion of the UAV trajectory based on the wavefront data, and determines the flight data, wherein the flight data includes at least short-term track, heading angle, and flight altitude; based on the verification result, identifies the UAV model, and matches it with the flight data in the range fragment map to determine whether it is a cooperative UAV; if yes, terminates the subsequent monitoring thread; if no, encapsulates the UAV model and flight data and sends them back to the monitoring center.

[0128] Furthermore, the behavior analysis unit 13 performs the following steps: the monitoring center receives the UAV model and the flight data, and determines the target kinematic behavior model by identifying behavioral elements and matching based on the behavior model library, wherein the matching constraint is based on the maximum approximation; for the target kinematic behavior model, the UAV behavior data is determined by performing multi-threaded evaluation and fusion based on the multi-level approximation of behavioral elements.

[0129] Furthermore, the behavior analysis unit 13 performs the following steps: retrieving UAV flight records, clustering the records by behavior class and mining kinematic behavior features, and establishing a behavior model library, wherein the behavior model library contains multiple kinematic behavior models; traversing the behavior model library, performing behavior weight adjustment, and embedding it in the monitoring center, wherein the behavior weight adjustment method is to increase the specific behavior weights and decrease the non-specific behavior weights among the kinematic behavior models.

[0130] Furthermore, the behavior management unit 14 performs the following steps: identifying the drone behavior data, generating behavior alarm information and behavior control schemes based on drone control rules; and controlling the deployment of control terminals and non-cooperative drones according to the behavior alarm information and the behavior control schemes.

[0131] Through the foregoing detailed description of a real-time monitoring method for the flight behavior of a non-cooperative drone, those skilled in the art can clearly understand the real-time monitoring method and system for the flight behavior of a non-cooperative drone in this embodiment. As for the apparatus disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section description.

[0132] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for real-time monitoring of the flight behavior of non-cooperative unmanned aerial vehicles (UAVs), characterized in that, The method includes: A distributed monitoring array is deployed in the airspace controlled by drones, wherein each monitoring point includes a cascaded deployment of acoustic sensors and piezoelectric sensors; By driving the monitoring array, rotor noise harmonic monitoring under acoustic sensing is performed, the positioning cluster is woken up, low-frequency pressure wave monitoring under piezoelectric sensing is performed, and the monitoring data is determined. The positioning cluster is a piezoelectric sensing cluster in the neighborhood of the acoustic sensing position. The monitoring data is judged by the edge gateway. If it is a non-cooperative drone, it is sent back to the monitoring center. The drone behavior data is determined by kinematic behavior model matching. The drone behavior data is displayed on a terminal interface, and alarms and control are triggered for abnormal behavior. Among them, the monitoring of rotor noise harmonics under acoustic sensing includes: Acoustic harmonic data is determined by driving the acoustic sensors in the monitoring array. Using the rotor harmonic feature set as the directional picking target, the acoustic harmonic data is verified. If the verification result is empty, the subsequent monitoring thread is terminated. If the test result is not empty, lock the target acoustic sensing component, wherein the target acoustic sensing component is the component tested in the non-empty test. Using the target acoustic sensing component, and the piezoelectric sensing components at the same monitoring point and neighboring monitoring points as a positioning cluster, the positioning cluster is woken up and driven to perform low-frequency pressure wave monitoring under piezoelectric sensing to determine wavefront data. The verification results and the wavefront data are integrated and used as the monitoring data, which is then transmitted to the target edge gateway to which the coverage area belongs.

2. The real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles as described in claim 1, characterized in that, Before driving the monitoring array, the following steps are included: The rotor noise spectrum of UAV models is collected to screen rotor-specific harmonics under environmental noise and determine the rotor harmonic feature set, wherein each UAV model corresponds to at least one rotor harmonic feature. The first sensing rule is determined by directional picking of the rotor harmonic feature set; Based on the first sensing rule, the acoustic sensing part in the monitoring array is initialized.

3. The real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles as described in claim 1, characterized in that, Before making a judgment based on the monitoring data, the deployment of the edge gateway includes: The flight range information of the cooperative drones is obtained, and the flight range information is spatially segmented based on the coverage area of ​​the edge gateway to determine the flight range fragmentation map; Based on regional association, the range fragment map is distributed and stored to each edge gateway, wherein the range fragment map is updated synchronously with the range changes of the cooperating UAVs.

4. The real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles as described in claim 3, characterized in that, The monitoring data is judged based on the edge gateway, including: Retrieve the range fragment map stored in the target edge gateway, perform fuzzy inversion of the UAV trajectory based on the wavefront data, and determine the flight data, wherein the flight data includes at least short-term track, heading angle and flight altitude; Based on the verification results, the drone model is identified, and in conjunction with the flight data, it is matched in the flight debris map to determine whether it is a cooperative drone. If so, terminate subsequent monitoring threads; If not, the drone model and flight data are encapsulated and transmitted back to the monitoring center.

5. A real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles as described in claim 4, characterized in that, Perform kinematic behavior model matching to determine UAV behavior data, including: The monitoring control center receives the UAV model and the flight data, and determines the target kinematic behavior model by identifying behavioral elements and matching based on a behavioral model library, wherein the matching constraint is based on the maximum approximation. For the target kinematic behavior model, the behavior data of the UAV is determined by performing multi-threaded evaluation and fusion based on the multi-level approximation level of the behavior elements.

6. The real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles as described in claim 5, characterized in that, The construction of the behavioral model library includes: Retrieve drone flight records, cluster the records by behavior category and mine kinematic behavior features to establish a behavior model library, wherein the behavior model library contains multiple kinematic behavior models; The behavior model library is traversed, and behavior weight adjustments are performed. The adjustments are then embedded in the monitoring and control center. Specifically, the behavior weight adjustments are performed by increasing the weights of specific behaviors and decreasing the weights of non-specific behaviors among the kinematic behavior models.

7. The real-time monitoring method for the flight behavior of non-cooperative unmanned aerial vehicles as described in claim 1, characterized in that, Early warning and control of abnormal behavior, including: Identify the drone behavior data and generate behavior alarm information and behavior control schemes based on drone management rules; Based on the behavior alarm information and the behavior control scheme, control terminals are deployed and non-cooperative drones are controlled.

8. A real-time monitoring system for the flight behavior of non-cooperative unmanned aerial vehicles (UAVs), characterized in that, A system for performing a real-time monitoring method for the flight behavior of a non-cooperative unmanned aerial vehicle (UAV) as described in any one of claims 1-7, the system comprising: The monitoring deployment unit is used to deploy a distributed monitoring array in the airspace controlled by unmanned aerial vehicles. Each monitoring point includes a cascaded deployment of acoustic sensors and piezoelectric sensors. A cascaded monitoring unit is used to drive the monitoring array to perform rotor noise harmonic monitoring under acoustic sensing, wake up the positioning cluster, perform low-frequency pressure wave monitoring under piezoelectric sensing, and determine the monitoring data. The positioning cluster is a piezoelectric sensing cluster in the vicinity of the acoustic sensing position. The behavior analysis unit is used to determine the monitoring data based on the edge gateway. If it is a non-cooperative drone, it is sent back to the monitoring control center. The drone behavior data is determined by kinematic behavior model matching. The behavior management unit is used to display the drone behavior data on the terminal interface and to perform alarms and control of abnormal behavior.

Images