Low-altitude unmanned aerial vehicle flight statistical system

Abstract

The invention discloses a low-altitude unmanned aerial vehicle flight statistical system, and relates to the technical field of air traffic management. A flight statistical system of a low-altitude unmanned aerial vehicle comprises a multi-source sensing fusion module, a twin airspace modeling module, a behavior portrait modeling module, a non-cooperative target recognition module, a flight statistical analysis module and a self-adaptive scheduling module. According to the invention, a three-dimensional twin airspace model of a monitoring area is constructed based on high-precision map data, a building three-dimensional model and flight information, and space mapping and collision risk analysis of a flight path are carried out, so that real restoration of the flight path and environment collision prediction are realized; the behavior feature vector is generated and the random forest algorithm is adopted to carry out classification modeling, and the behavior portrait of the unmanned aerial vehicle is constructed, so that accurate identification and classification of the behavior pattern of the unmanned aerial vehicle are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of air traffic management, and in particular to a low-altitude unmanned aerial vehicle (UAV) flight statistics system. Background Art

[0002] With the wide deployment of UAVs in fields such as urban logistics, emergency rescue, traffic inspection, and security monitoring, the management of low-altitude airspace has become an important research direction for smart cities and airspace governance. Currently, a variety of technical means have been used for the monitoring and management of UAVs, including perception systems based on the fusion of multi-source information such as radar, video surveillance, radio detection, and ADS-B (Automatic Dependent Surveillance - Broadcast). Some systems have also introduced flight data analysis and visualization platforms, realizing the trajectory recording of UAVs, the generation of activity heat maps, and the basic classification of flight behaviors.

[0003] In terms of behavior modeling and risk identification, existing research has gradually explored the use of machine learning techniques to classify and analyze the flight characteristics of UAVs, such as behavior portraits based on trajectory pattern recognition and non-cooperative target detection based on electromagnetic characteristics. At the same time, some airspace management platforms have introduced the concept of digital twins and tried to build three-dimensional airspace models to support flight trajectory visualization and safety assessment.

[0004] However, although existing multi-source perception systems can obtain some UAV information, they are still imperfect in terms of non-cooperative target recognition, trajectory parsing, and risk prediction capabilities in complex urban airspaces. Existing behavior recognition models mostly focus on the classification of the trajectory itself and lack the fusion modeling of multi-dimensional behavior characteristics such as speed patterns and hovering behaviors. Airspace strategies are mostly preset static areas and lack an adaptive adjustment mechanism based on real-time statistical data and behavior portraits, making it difficult to achieve dynamic response to flight risks and optimization of flight restriction strategies. Summary of the Invention

[0005] The present invention proposes a low-altitude UAV flight statistics system, which has functions of multi-source fusion perception, behavior feature modeling, non-cooperative recognition, three-dimensional airspace mapping, and risk level assessment, realizing high-precision modeling of urban low-altitude airspace and dynamic flight control.

[0006] A low-altitude UAV flight statistics system includes: A multi-source perception fusion module for collecting flight information of low-altitude UAVs, including the status information, position information, track information, electromagnetic feature information, and image information of the UAVs; A twin airspace modeling module for constructing a three-dimensional twin airspace model of the monitoring area based on high-precision map data, three-dimensional building models, and flight information, and performing spatial mapping of flight trajectories and collision risk analysis; The behavior portrait modeling module is used to extract behavior features based on flight information, including flight trajectory features obtained based on track information, speed pattern features obtained based on status information, and hovering behavior features obtained based on status information and track information; generate corresponding behavior feature vectors and perform classification modeling to construct the behavior portrait of the UAV; the classification modeling adopts a behavior recognition algorithm based on a random forest model; The non-cooperative target recognition module is used to extract the electromagnetic fingerprint of the UAV based on electromagnetic feature information, and identify unregistered or non-cooperative targets without communication devices through cooperative target matching; The flight statistical analysis module is used to obtain the flight duration information, takeoff and landing information, airspace coverage path and density distribution of the UAV, generate multi-dimensional flight statistical data, and output a visualization report; The adaptive scheduling module is used to divide the flight risk level of the UAV according to flight statistical data and behavior portraits, dynamically adjust the flight restriction strategy and warning range of the airspace in combination with the three-dimensional twin airspace model, and link the response system to perform flight scheduling control.

[0007] As a preferred technical solution of the present invention, the construction of the three-dimensional twin airspace model of the monitoring area includes: Obtain high-precision map data of the monitoring area, and construct a two-dimensional geographic base layer of the three-dimensional twin airspace model for providing basic geographic information; obtain the three-dimensional model of buildings in the monitoring area, and construct an obstacle modeling layer of the three-dimensional twin airspace model for restoring the three-dimensional building structure, contour and height information in the monitoring area; construct a dynamic track mapping layer based on the flight information of the UAV, and map the position information and track information of the UAV into the three-dimensional twin airspace model to realize real-time visualization of the flight state and path restoration.

[0008] As a preferred technical solution of the present invention, the spatial mapping and collision risk analysis includes: Perform spatial registration of the track information and the three-dimensional building model; based on the current status information and position information of the UAV, use a trajectory extension algorithm to generate a short-term flight prediction trajectory; calculate the minimum spatial distance between the flight prediction trajectory and the building model; set a stepped collision warning threshold, and trigger a collision risk warning when the minimum spatial distance is less than the corresponding threshold; analyze the speed change trend and track deviation of the UAV, and comprehensively evaluate it with the collision risk warning result to obtain the collision risk level, and provide the evaluation result to the adaptive scheduling module.

[0009] As a preferred technical solution of the present invention, the trajectory extension algorithm includes: Construct a flight state vector based on the current position information and status information of the UAV; use the Kalman filtering algorithm to dynamically estimate the flight state vector and predict the motion trajectory of the UAV in a short period of time; project the predicted trajectory into the 3D twin airspace model as the flight prediction trajectory for subsequent collision risk analysis.

[0010] As a preferred technical solution of the present invention, the extracting behavior features according to flight information includes: Conduct a temporal analysis on the track information of the UAV to extract its flight trajectory features, including turning frequency, track curvature, and path repeatability; perform a speed change modeling on the status information of the UAV to extract speed pattern features, including average speed, speed fluctuation amplitude, and acceleration change trend; identify the hovering behavior features that occur during the flight based on the track information and status information, including hovering duration, hovering position frequency, and its correlation with specific geographical areas; vectorize the flight trajectory features, speed pattern features, and hovering behavior features to obtain a behavior feature vector for classification modeling.

[0011] As a preferred technical solution of the present invention, the conducting classification modeling and constructing the behavior portrait of the UAV includes: Construct a training sample set based on the extracted behavior feature vector. The training sample set includes the flight information and behavior category labels of multiple known types of UAVs; use the random forest algorithm to model the training sample set to generate a behavior classification model for identifying and classifying the flight behavior patterns of unknown UAVs; input the behavior feature vector of the UAV to be analyzed into the behavior classification model and output the behavior category label; associate the behavior category label with the behavior feature vector to obtain the behavior portrait of the UAV.

[0012] As a preferred technical solution of the present invention, the identifying unregistered or non - cooperative targets without communication devices includes: Based on the electromagnetic feature information, extract the electromagnetic radiation features of the aircraft at specific frequency bands to construct an electromagnetic fingerprint; match the electromagnetic fingerprint with the electromagnetic fingerprint database of registered aircraft preset in the system for cooperative targets; when no legal registration record can be matched and the target aircraft does not detect communication device signals or identity broadcast information, determine it as a non - cooperative target; send the determination result to the adaptive scheduling module and the response system.

[0013] As a preferred technical solution of the present invention, the generating multi - dimensional flight statistical data includes: Flight duration information: Record the start time and end time of each UAV's flight, and calculate the flight duration and time distribution characteristics; Take - off and landing information: Extract the take - off and landing frequencies, take - off and landing positions, and their affiliated area labels, and generate a spatial distribution map of the take - off and landing positions; Airspace coverage path: Divide the flight trajectory into geographical grids, count the flight track coverage frequency in each grid unit, and construct an airspace coverage path map; Density distribution: Generate a time-space combined flight density heat map based on the density distribution of flight activities in different time periods and different regions; Structurally store the flight statistical data and output the corresponding visualization report for management and scheduling reference.

[0014] As a preferred technical solution of the present invention, the division of the flight risk level of the unmanned aerial vehicle includes: Combined with the density distribution, takeoff and landing information, and airspace coverage path of the flight statistical data, evaluate the spatial activity of the unmanned aerial vehicle activities; combined with the behavior categories, speed mode characteristics, and hovering behavior characteristics of the behavior portrait, evaluate its behavior risk coefficient; combined with the recognition results of the non-cooperative target recognition module, mark whether it is a non-cooperative target or a communication-lost aircraft, and obtain the corresponding cooperation status; set up a multi-dimensional weighted evaluation model, and divide the flight behavior of the unmanned aerial vehicle into flight risk levels according to the spatial activity, behavior risk coefficient, and cooperation status.

[0015] As a preferred technical solution of the present invention, the dynamic adjustment of the airspace includes: Based on the divided flight risk levels, dynamically mark the spatial range of different risk areas in the three-dimensional twin airspace model; set flight restrictions, altitude limits, or warning flight strategies for high-risk areas, and update the airspace usage rights in real time; automatically adjust the warning range according to the flight density heat map and the behavior concentration area; combined with the real-time trajectory information of the unmanned aerial vehicle, when it is about to enter a high-risk area, link the response system to issue a warning or execute a flight avoidance instruction.

[0016] The present invention has the following advantages: The present invention collects the flight information of low-altitude unmanned aerial vehicles, including status information, position information, flight track information, electromagnetic feature information, and image information, and fuses multi-source perception data, realizing all-round and three-dimensional perception of low-altitude unmanned aerial vehicle flight activities, and significantly improving the system's target detection and behavior monitoring capabilities.

[0017] By constructing a three-dimensional twin airspace model of the monitoring area based on high-precision map data, three-dimensional building models, and flight information, and performing spatial mapping and collision risk analysis of flight trajectories, the true restoration of flight paths and environmental collision prediction are realized, effectively ensuring the safety of low-altitude flights, and especially suitable for the supervision of unmanned aerial vehicle activities in complex urban airspaces; by extracting flight trajectory characteristics, speed mode characteristics, and hovering behavior characteristics, generating behavior feature vectors and using the random forest algorithm for classification modeling, constructing the behavior portrait of the unmanned aerial vehicle, realizing the accurate recognition and classification of the behavior patterns of the unmanned aerial vehicle, and improving the judgment ability of abnormal flight behaviors and potential threats.

[0018] The present invention extracts the electromagnetic fingerprint of an unmanned aerial vehicle (UAV) based on electromagnetic feature information and matches it with the electromagnetic feature database of registered aircraft, enabling the identification of unregistered or non-cooperative targets without communication devices. This breaks through the traditional identification method that solely relies on communication signals and enhances the system's ability to identify and prevent illegal and irregular flight targets.

[0019] By recording the flight duration information, takeoff and landing information, airspace coverage path, and density distribution of the UAV, the present invention generates multi-dimensional flight statistical data and outputs a visual report, facilitating supervisors to intuitively grasp the usage status of the low-altitude airspace and providing strong data support for management decision-making, trend research, and judgment. Further, by combining the flight statistical data with the behavior portrait, the risk level of UAV flight activities is classified, and the flight restriction strategy and warning range are dynamically adjusted based on the three-dimensional twin airspace model, realizing the adaptive dynamic scheduling management of the airspace and significantly enhancing the intelligent level of the UAV supervision system and the airspace safety guarantee ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only schematic diagrams of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on the provided drawings without creative efforts. Figure 1 It is a schematic structural diagram of a low-altitude UAV flight statistics system adopted in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings. Obviously, the described embodiments are only partial embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0022] Embodiment 1, a low-altitude UAV flight statistics system, specifically includes: 1. A multi-source perception fusion module for collecting the flight information of low-altitude UAVs, including the status information, position information, flight track information, electromagnetic feature information, and image information of the UAV; The status information at least includes the flight attitude, flight speed, acceleration, heading angle, pitch angle, roll angle, flight mode, battery power, and communication status of the UAV; The position information is obtained by a GNSS, RTK module, or inertial navigation system (INS), and at least includes longitude and latitude, altitude, velocity vector, and timestamp; The flight track information is a continuous record of the historical flight path of the UAV in the airspace, obtained by fusing position information and status data, and has the characteristics of a time series; The electromagnetic characteristic information is extracted by passively receiving the radio signals emitted by the UAV, and at least includes spectral distribution, signal strength, modulation characteristics, and frequency hopping patterns to form a unique electromagnetic "fingerprint"; The image information is obtained by ground optoelectronic observation equipment, radar imaging systems, or the UAV's own camera, and includes real-time images or video streams for target recognition and behavior correlation analysis.

[0023] After the above multi-source flight information is collected, it undergoes time synchronization, format standardization, and missing value filling through data preprocessing, and is unifiedly modeled using a feature fusion algorithm to provide unified data support for subsequent modules; this module supports parallel perception and processing of multiple flight targets, and has the characteristics of strong real-time performance, rich information dimensions, and adaptability to complex environments.

[0024] 2. The twin airspace modeling module is used to construct a three-dimensional twin airspace model of the monitoring area based on high-precision map data, three-dimensional building models, and flight information, and perform spatial mapping and collision risk analysis of flight trajectories; thus realizing digital twin modeling and dynamic safety assessment of the low-altitude airspace.

[0025] The process of constructing the three-dimensional twin airspace model includes: Obtain high-precision map data of the monitoring area and construct a two-dimensional geographic base layer of the three-dimensional twin airspace model. This layer provides basic geographical location information, including static geographical elements such as roads, water bodies, administrative boundaries, and functional area divisions, to support the accurate projection of the UAV trajectory in the geographical space; Obtain three-dimensional model information of buildings in the monitoring area, including building outline lines, height data, volume shape, number of floors, and spatial coordinates, etc., and construct an obstacle modeling layer in the twin airspace model to restore the entity structures and obstacle distributions existing in the actual airspace and achieve accurate modeling of the three-dimensional space environment; Based on the position information and flight track information collected by the UAV, construct a dynamic flight track mapping layer. This layer maps the real-time or historical flight tracks of the UAV to the above three-dimensional twin airspace, realizing path reconstruction and state visualization of flight behaviors in the digital space. This mapping layer supports the simultaneous display of multiple UAVs and has functions such as trajectory tracking, status query, and time backtracking.

[0026] When conducting spatial mapping and collision risk analysis, the drone trajectory information in the dynamic flight path mapping layer is spatially registered with the building model in the obstacle modeling layer to ensure coordinate system consistency and spatial accuracy. Subsequently, based on the current state information and position information of the drone, a trajectory extension algorithm is called to generate a short-term flight prediction trajectory. The minimum spatial distance between the prediction trajectory and the surrounding building models is calculated in real time, and a multi-level stepped collision warning threshold is set. If the distance is less than a certain threshold, a corresponding level of collision risk warning is triggered.

[0027] Combined with dynamic parameters such as the speed change trend, acceleration direction, and flight path deviation of the drone, further analysis is carried out, and the above analysis results and the predicted distance are used as inputs to calculate the collision risk level of the current flight state. The collision risk level is uploaded to the adaptive scheduling module as the basis for airspace dynamic adjustment and flight behavior response.

[0028] The trajectory extension algorithm specifically includes: constructing a flight state vector based on the current position information and state information (speed, acceleration, heading angle) of the drone, and using the Kalman filter algorithm to dynamically estimate it and predict its continuous motion trajectory in the short term in the future. This predicted trajectory is then projected into the three-dimensional twin airspace model and compared with the obstacle model to achieve early perception of flight conflicts.

[0029] 3. The behavior portrait modeling module is used to extract behavior characteristics according to flight information, including flight trajectory characteristics obtained based on flight path information, speed pattern characteristics obtained based on state information, and hovering behavior characteristics obtained based on state information and flight path information; generate corresponding behavior feature vectors and perform classification modeling to construct the behavior portrait of the drone; the classification modeling uses a behavior recognition algorithm based on a random forest model. The extraction of the behavior characteristics includes: Perform temporal analysis on the flight path information of the drone to extract its flight trajectory characteristics, including turning frequency (the number of heading changes per unit time), flight path curvature (describing the degree of bending of the flight path), path repeatability (the proportion of repeated flight paths in a specific area), etc., which are used to reflect the motion law and spatial distribution characteristics of the drone in the airspace.

[0030] Perform speed modeling on the state information of the drone to extract its speed pattern characteristics, including parameters such as average speed, speed fluctuation amplitude, and acceleration change trend, which reflect the flight stability and action intention of the drone. For example, high-fluctuation speed and sudden acceleration indicate behavior patterns such as task execution, avoidance, and tracking.

[0031] Jointly identify the hovering behavior characteristics that occur during flight based on track information and status information, such as hovering duration, hovering position frequency, and the spatial correlation between these positions and specific geographical regions (such as no-fly zones, sensitive targets, densely populated areas). This feature is particularly crucial for identifying illegal reconnaissance, suspicious surveillance, and other behaviors.

[0032] Vectorize the above three types of behavior characteristics to form a behavior feature vector in a unified format, which is used as the input of the classification model.

[0033] The classification modeling includes: Construct a training sample set based on the flight sample data of multiple known types of unmanned aerial vehicles. Each sample data contains the corresponding behavior feature vector and behavior category label (cruising, tracking, detecting, hovering). Use the random forest algorithm to model the training samples and construct a multi-class flight behavior recognition model.

[0034] Input the behavior feature vector of the unmanned aerial vehicle to be analyzed into the trained behavior classification model, output its behavior category label, and associate it with the original behavior feature vector to form the final behavior portrait. This portrait not only contains the recognition results of flight behaviors but also retains the original behavior data such as trajectory features, speed features, and hovering features, which is convenient for subsequent risk assessment, pattern clustering, and behavior evolution analysis.

[0035] 4. Non-cooperative target recognition module, which is used to extract the electromagnetic fingerprint of the unmanned aerial vehicle based on electromagnetic feature information and identify unregistered or communication device-free non-cooperative targets through cooperative target matching; Specifically, through radio monitoring devices or electromagnetic sensing nodes set in the monitoring area, passively receive the radio signals emitted by low-altitude aircraft during flight, and focus on their electromagnetic radiation characteristics in specific frequency bands (2.4 GHz, 5.8 GHz, 900 MHz). By performing spectrum analysis, time series analysis, and modulation mode recognition on the received signals, extract unique and stable electromagnetic fingerprint features, including at least carrier frequency, bandwidth, power spectral density, modulation method, signal hopping behavior, and emission cycle characteristics.

[0036] The system pre-establishes a database of electromagnetic fingerprints of registered aircraft. The fingerprint data in this database comes from the communication devices, remote control links, or video transmission devices of legally registered unmanned aerial vehicles. The database is classified and managed according to aircraft models, manufacturers, and device types, and is dynamically updated to ensure its timeliness and coverage.

[0037] Match the currently extracted electromagnetic fingerprint with the database for cooperative target matching. If the match is successful, determine that the target is a cooperative aircraft; if no registered record can be matched and no signal or identity broadcast information of a legitimate communication device (such as ADS-B or Wi-Fi identification code) is detected around the target aircraft, further determine that the target is a non-cooperative target. The determination result will be synchronously sent to the adaptive scheduling module and the response system.

[0038] The non-cooperative target recognition module proposed in the present invention does not rely on the active communication behavior of the aircraft and realizes detection when the aircraft does not turn on the identity broadcast or flies in the "silent mode"; the electromagnetic fingerprint has strong device uniqueness and anti-forgery ability, is difficult to counterfeit, enhances the recognition accuracy and robustness of illegal flight behaviors, and is used in coordination with the behavior profiling module to conduct early warning and risk escalation for high-risk and concealed flight targets. It breaks through the limitations of the traditional mechanism relying on identity broadcast or signal registration, strengthens the recognition ability of low-altitude gray flight and black flight behaviors, and improves the overall airspace safety prevention and control level.

[0039] 5. Flight statistical analysis module, which is used to obtain the flight duration information, takeoff and landing information, airspace coverage path and density distribution of the unmanned aircraft, generate multi-dimensional flight statistical data, and output a visual report to provide data support for airspace management, behavior evaluation and scheduling decision-making; The flight statistical analysis module first performs real-time parsing and archiving on the received flight information data, automatically records the start time and end time of each unmanned aircraft's flight, calculates its flight duration, and combines time tags to count its daily activity frequency and time period distribution characteristics, such as behavior patterns during peak hours and night activities.

[0040] Extract the takeoff position and landing position of each flight mission, and match its corresponding area labels (administrative area, functional area, restricted area) and the takeoff and landing frequencies corresponding to the area labels according to the geographical coordinates, and draw a spatial distribution map of the takeoff and landing points of the unmanned aircraft to analyze the flight activity density in hot spots or sensitive areas.

[0041] For the track data generated during the flight, based on the regional grid division algorithm, the monitored airspace is divided into equal-size geographical grid units (such as 50m×50m), the track coverage frequency in each unit is counted, and an airspace coverage path map is constructed to intuitively present the activity intensity and coverage breadth of the unmanned aircraft in different areas.

[0042] Based on the two dimensions of time and space, a joint analysis model is constructed to generate a time-space joint density heat map to show the flight density distribution trend in a specific area during a specific time period. For example, visually display the differences in unmanned aircraft activities between "weekday daytime VS weekend night" to assist in identifying abnormal flight patterns.

[0043] The flight statistics are stored in a structured manner to form a unified data set, which supports query and comparison by target classification, time screening, spatial area division, etc. Visual reports such as statistical charts, heat maps, and behavior trend curves are generated on demand and provided to the supervision platform or dispatch center for decision-making assistance.

[0044] This module has high timeliness, refinement and multi-dimensional analysis capabilities. It not only realizes the archiving and auditing of historical data, but also can dynamically update it in combination with real-time data. It is suitable for the refined airspace management needs at the city level, park level and key area level.

[0045] 6. Adaptive scheduling module, which is used to classify the flight risk level of drones according to flight statistics and behavior portraits, dynamically adjust the flight restriction strategy and warning range of the airspace in combination with the three-dimensional twin airspace model, and link the response system for flight scheduling control.

[0046] The response system is used to receive the flight risk assessment results and airspace strategy adjustment instructions output by the adaptive scheduling module. When it is detected that the target UAV has high-risk behavior, is about to enter a restricted flight area, or is identified as a non-cooperative target, the corresponding response measures are executed. The response measures include: sending flight warnings or avoidance instructions to UAVs with communication interfaces; implementing forced responses to target linkage intervention devices that do not have communication capabilities, including but not limited to signal interference, warning broadcasts, physical interception, etc.; and feeding back the response results to the system log or the supervision platform for subsequent audits and situation review.

[0047] Specifically, the flight risk level classification process includes: Based on the data provided by the flight statistics analysis module, the flight activity indicators of each drone during the monitoring period are evaluated, including at least flight density (frequency of tracks per unit area), take-off and landing frequency (activity level in hot spots), and airspace coverage path (span and overlap of activity tracks), to reflect its spatial usage intensity and the possible risk of airspace resource occupation.

[0048] Combined with the behavior category labels and behavior feature vectors generated by the behavior profiling module, the behavioral risk factors of the drone are analyzed, including whether there are abnormal trajectory behaviors (frequent hovering, sharp turns, sudden changes in curvature), unstable speed patterns, etc., so as to derive the behavior stability and abnormal probability and form a behavior risk coefficient.

[0049] Based on the recognition results of the non-cooperative target recognition module, it is determined whether the drone is a non-cooperative target or a communication loss target. Due to the high uncontrollability and unknown identity of such aircraft, the risk level weight will be significantly increased.

[0050] Based on the above multi-dimensional indicators, a multi-dimensional weighted evaluation model is constructed. By setting threshold stratification and scoring rules, the flight behaviors of drones are classified into multiple levels (low, medium, high, and extremely high risks). The risk level will serve as the basis for subsequent airspace control, strategy response, and resource allocation.

[0051] Combined with the three-dimensional twin airspace model, dynamic airspace strategy adjustment is carried out, including: Dynamically label the spatial boundaries and attributes of areas with different risk levels in the twin airspace model to support the visualization of risk areas updated over time; automatically set flight restriction, altitude limit, slow flight, and early warning flight strategies for high-risk areas, and actively update airspace usage permissions based on the real-time situation; automatically adjust the early warning range according to the flight density heat map and high-frequency behavior areas. For example, increase the risk response level around densely populated areas and key infrastructure. Real-time monitor the change of the target drone's trajectory. When it is predicted that the drone is about to enter a high-risk area or approach the flight restriction boundary, automatically link the response system to send an early warning signal, or push an avoidance instruction to the aircraft (if it has a communication interface), or link intervention measures (such as lasers, signal jamming devices) for response control.

[0052] Embodiment 2, a low-altitude drone flight statistics system, see Figure 1 As shown, it includes the following modules and units: 1. Multi-source perception fusion module, including: Flight information collection unit, used to collect the status information, position information, flight track information, electromagnetic feature information, and image information of the drone; Data preprocessing unit, used to unify the format, synchronize the time, and filter the noise of the collected multi-source data to ensure the data quality and fusibility; Feature fusion modeling unit, used to fuse and model the perception data from different sources, and construct a unified flight data structure for subsequent analysis modules to use.

[0053] 2. Twin airspace modeling module, including: Geographical environment modeling unit, used to construct a geographical base layer and an obstacle modeling layer based on high-precision map data and three-dimensional building models; Dynamic trajectory mapping unit, used to map the real-time or historical flight trajectories of drones into the three-dimensional twin airspace model to achieve spatial restoration and visualization; Collision risk analysis unit, used to predict the short-term flight path based on the trajectory extension algorithm, calculate the spatial distance from buildings, and judge the collision risk level.

[0054] 3. Behavior portrait modeling module, including: Behavior feature extraction unit, used to extract behavior features from flight track information, speed status, and hovering behavior; A behavioral feature vector construction unit, which is used to vectorize multi-dimensional behavioral features and generate a structured feature vector; A behavioral classification modeling unit, which is used to classify the behavioral feature vector by using a random forest model and output behavioral portrait labels.

[0055] 4. The non-cooperative target recognition module includes: An electromagnetic fingerprint extraction unit, which is used to extract electromagnetic characteristic parameters of an unmanned aerial vehicle in a specific frequency band based on radio signal analysis; A target matching and recognition unit, which is used to compare the extracted electromagnetic fingerprint with a registered aircraft database to identify non-cooperative targets; An identification result feedback unit, which is used to send the identification result to a scheduling control system or a response system to trigger a subsequent response process.

[0056] 5. The flight statistical analysis module includes: A data recording unit, which is used to obtain basic statistical information such as the flight duration, takeoff and landing time and location, and trajectory data of the unmanned aerial vehicle; A spatial distribution analysis unit, which is used to construct an airspace coverage path map, a takeoff and landing point distribution map, and a time-space joint heat map; A report generation unit, which is used to generate a visualization chart or a structured report of the statistical results for use by a management platform.

[0057] 6. The adaptive scheduling module includes: A risk assessment unit, which is used to comprehensively analyze flight density, behavioral portrait, and cooperation status and divide flight risk levels; An airspace strategy adjustment unit, which is used to dynamically adjust flight restriction, altitude limit, and early warning strategies for high-risk areas in a twin airspace model; A linkage response control unit, which is used to trigger an early warning by linking systems when a high-risk behavior occurs, or execute avoidance or intervention instructions on a target aircraft.

[0058] The specific implementation manners described above further elaborate the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific implementation manners of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A low-altitude UAV flight statistics system, characterized in that, Including: A multi-source perception fusion module, which is used to collect the flight information of low-altitude unmanned aerial vehicles (UAVs), including the status information, position information, flight path information, electromagnetic feature information, and image information of the UAVs; A twin airspace modeling module, which is used to construct a three-dimensional twin airspace model of the monitoring area based on high-precision map data, three-dimensional building models, and flight information, and conduct spatial mapping of the flight trajectory and collision risk analysis; A behavior portrait modeling module, which is used to extract behavior features according to the flight information, including flight trajectory features obtained based on the flight path information, speed pattern features obtained based on the status information, and hovering behavior features obtained based on the status information and flight path information; Generate corresponding behavior feature vectors and conduct classification modeling to construct the behavior portrait of the UAV; the classification modeling adopts a behavior recognition algorithm based on a random forest model; A non-cooperative target recognition module, which is used to extract the electromagnetic fingerprint of the UAV based on the electromagnetic feature information, and identify unregistered or non-cooperative targets without communication devices through cooperative target matching; A flight statistical analysis module, which is used to obtain the flight duration information, takeoff and landing information, airspace coverage path, and density distribution of the UAV, generate multi-dimensional flight statistical data, and output a visualization report; An adaptive scheduling module, which is used to divide the flight risk level of the UAV according to the flight statistical data and behavior portrait, dynamically adjust the flight restriction strategy and early warning range of the airspace in combination with the three-dimensional twin airspace model, and link the response system to conduct flight scheduling control.

2. The low-altitude UAV flight statistics system according to claim 1, characterized in that, The construction of the three-dimensional twin airspace model of the monitoring area includes: Obtain high-precision map data of the monitoring area, and construct a two-dimensional geographic base layer of the three-dimensional twin airspace model to provide basic geographic information; obtain the three-dimensional building model of the monitoring area, and construct an obstacle modeling layer of the three-dimensional twin airspace model to restore the three-dimensional building structure, contour, and height information in the monitoring area; construct a dynamic flight path mapping layer based on the flight information of the UAV, and map the position information and flight path information of the UAV into the three-dimensional twin airspace model to realize real-time visualization of the flight state and path restoration.

3. The low-altitude UAV flight statistics system according to claim 1, wherein, The spatial mapping and collision risk analysis include: Perform spatial registration of the flight path information and the three-dimensional building model; based on the current status information and position information of the UAV, use a trajectory extension algorithm to generate a short-term flight prediction trajectory; calculate the minimum spatial distance between the flight prediction trajectory and the building model; set a stepped collision warning threshold, and trigger a collision risk warning when the minimum spatial distance is less than the corresponding threshold; analyze the speed change trend and flight path deviation of the UAV, and comprehensively evaluate it with the collision risk warning result to obtain the collision risk level, and provide the evaluation result to the adaptive scheduling module.

4. The low-altitude UAV flight statistics system according to claim 3, characterized in that, The trajectory extension algorithm includes: Construct a flight state vector based on the current position information and status information of the UAV; use the Kalman filter algorithm to dynamically estimate the flight state vector and predict the movement trajectory of the UAV in a short time; project the predicted trajectory into the three-dimensional twin airspace model as the flight prediction trajectory for subsequent collision risk analysis.

5. A low-altitude UAV flight statistics system according to claim 1, characterized in that, The extraction of behavior features according to the flight information includes: Perform temporal analysis on the trajectory information of the UAV, extract its flight trajectory features, including turning frequency, trajectory curvature, and path repeatability; perform speed change modeling on the state information of the UAV, extract speed pattern features, including average speed, speed fluctuation amplitude, and acceleration change trend; identify the hovering behavior features that occur during the flight based on the trajectory information and state information, including hovering duration, hovering position frequency, and its correlation with specific geographical regions; vectorize the flight trajectory features, speed pattern features, and hovering behavior features to obtain a behavior feature vector for classification modeling.

6. The low-altitude UAV flight statistics system according to claim 1, characterized in that, The classification modeling is carried out to construct the behavior portrait of the UAV, including: Construct a training sample set based on the extracted behavior feature vectors. The training sample set includes the flight information and behavior category labels of multiple known types of UAVs; use the random forest algorithm to model the training sample set to generate a behavior classification model for identifying and classifying the flight behavior patterns of unknown UAVs; input the behavior feature vector of the UAV to be analyzed into the behavior classification model and output the behavior category label; associate the behavior category label with the behavior feature vector to obtain the behavior portrait of the UAV.

7. The low-altitude UAV flight statistics system according to claim 1, characterized in that The identification of unregistered or non-cooperative targets without communication devices includes: Based on the electromagnetic feature information, extract the electromagnetic radiation characteristics of the aircraft at specific frequency bands to construct an electromagnetic fingerprint; match the electromagnetic fingerprint with the electromagnetic fingerprint database of registered aircraft preset in the system for cooperative targets; when no legal registration record can be matched and the target aircraft does not detect communication device signals or identity broadcast information, determine it as a non-cooperative target; send the determination result to the adaptive scheduling module and the response system.

8. A low-altitude drone flight statistics system according to claim 1, characterized in that, The generation of multi-dimensional flight statistical data includes: Flight duration information: Record the start time and end time of each UAV's flight, and calculate the flight duration and time distribution characteristics. Takeoff and landing information: Extract the takeoff and landing frequencies, takeoff and landing positions, and their affiliated area labels to generate a spatial distribution map of the takeoff and landing positions. Airspace coverage path: Divide the flight trajectory into geographical grids, count the track coverage frequencies in each grid cell, and construct an airspace coverage path map. Density distribution: Based on the density distribution of flight activities in different time periods and different regions, generate a time-space joint flight density heat map. Structurally store the flight statistical data and output the corresponding visualization report for management and scheduling reference.

9. The low-altitude UAV flight statistics system according to claim 1, characterized in that, The division of the flight risk level of the UAV includes: Combine the density distribution, takeoff and landing information, and airspace coverage path of the flight statistical data to evaluate the spatial activity of UAV activities; combine the behavior category, speed pattern features, and hovering behavior features of the behavior portrait to evaluate its behavior risk coefficient; combine the identification results of the non-cooperative target recognition module to mark whether it is a non-cooperative target or a communication-lacking aircraft to obtain the corresponding cooperation status; set up a multi-dimensional weighted evaluation model, and divide the flight behavior of the UAV into flight risk levels according to the spatial activity, behavior risk coefficient, and cooperation status.

10. The low-altitude UAV flight statistics system according to claim 1, wherein The dynamic adjustment of airspace includes: Based on the divided flight risk levels, dynamically mark the spatial ranges of different risk areas in the 3D twin airspace model; set flight restrictions, altitude limits or warning flight strategies for high-risk areas and update airspace usage permissions in real time; automatically adjust the warning range according to the flight density heat map and behavior concentration areas; combine the real-time trajectory information of the UAV, and when it is about to enter a high-risk area, link the response system to issue a warning or execute a flight avoidance instruction.

Images