A target situation awareness method and system based on low-altitude intelligent networking

Abstract

The application discloses a target situation awareness method and system based on low-altitude intelligent networking, and particularly relates to the technical field of low-altitude target awareness, and is used for solving the problem of track situation distortion; the application collects target observation data output by radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes and positioning nodes, generates a low-altitude observation frame sequence through unified time reference correction and unified space coordinate system mapping; constructs a node observation credible domain according to node coverage boundaries, shielding areas, link time delays, historical observation deviations and target height intervals, screens credible observation frames, generates target connection relationships based on the credible observation frames, constructs a track hypothesis tree when target observation is missing, generates a continuous target track chain, generates situation events in combination with identities, air routes, airspace and link states, constructs a situation event graph and outputs target situation awareness results, and can improve the accuracy of continuous tracking, identity verification and risk early warning of low-altitude targets.

Description

Technical Field

[0001] This invention relates to the field of low-altitude target perception technology, and more specifically, to a target situational perception method and system based on low-altitude intelligent networks. Background Technology

[0002] In the field of low-altitude intelligent network target perception, drone operation monitoring technologies have been widely applied for scenarios such as bridge inspection, park security, low-altitude logistics, emergency detection, airport perimeter protection, airport airspace protection zone supervision, runway low-altitude intrusion early warning, terminal area drone patrol and key area protection. Such systems typically coordinate radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes and positioning nodes through computer programs to collect and analyze the azimuth, distance, altitude, speed, heading, electronic identification and communication link status of low-altitude drone targets. Combined with declared flight routes, authorized airspace, airport airspace restrictions, no-fly zones and mission time windows, the system monitors, identifies and alarms the target's operational status.

[0003] However, low-altitude UAVs flying in cross-river passages, near industrial parks and buildings, around bridge towers, near take-off and landing points, airport perimeters, runway clearance areas, near terminals, and in areas with weak communication coverage are easily affected by building obstruction, ground reflection, multipath interference, video false detection, node latency, and link fluctuations. The target observation data output by different sensing nodes differs in time, space, and reliability. Directly fusing multi-source observation results or matching tracks based on nearest neighbor relationships lacks dynamic constraints on the current reliability range of node observations and also lacks information on what happens after the target disappears briefly. The lack of a reliable connection mechanism makes it easy for drones to experience track breaks, duplicate filings, or incorrect connection after crossing obstructed areas, airport perimeter blind spots, or node coverage blind spots. At the same time, there is a lack of continuous reverse verification between electronic identification, declared routes, authorized airspace, airport airspace restrictions, and actual physical tracks. Events such as identity anomalies, boundary approach, airspace intrusion, track conflicts, and communication interruptions are mostly output in the form of isolated alarms, which are difficult to express the evolution of risks from observation instability and track deviation to situational escalation. This affects the accuracy of low-altitude drone supervision, the timeliness of early warnings, and the targeted nature of response. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, the present invention provides a target situational awareness method and system based on low-altitude intelligent networks to solve the problem of trajectory situational distortion in the aforementioned background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A target situational awareness method based on low-altitude intelligent networks includes:

[0007] S1. Collect target observation data output from radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes and positioning nodes within the low-altitude intelligent network, and generate a low-altitude observation frame sequence after unified time reference correction and unified spatial coordinate system mapping.

[0008] S2. Based on the coverage boundary, obstruction area, link delay, historical observation deviation and target altitude range of the sensing node, construct the node observation credibility domain, perform credibility marking and mismatch removal on the low-altitude observation frame sequence, and form a credibility observation frame set.

[0009] S3. Based on the set of trusted observation frames, generate target succession relationships according to time reachability, spatial reachability, motion continuity, identity continuity and overlapping relationship of trusted observation domains, and construct trajectory hypothesis tree when target observation is missing, match subsequent trusted observation frames to reachable envelope, and generate continuous target track chains and trajectory succession events.

[0010] S4. Perform a reverse consistency check between the continuous target track chain and the electronic identity, declared route, authorized airspace and mission time window, and infer the future reachable situation subdomain based on the target motion state, maneuverable boundary and link state, and generate identity anomaly events, track conflict events, boundary risk events and communication interruption events.

[0011] S5. Using identity anomaly events, track conflict events, boundary risk events, communication interruption events, and track continuity events as graph nodes, and using temporal sequence relationships, spatial proximity relationships, target homogeneity relationships, and risk progression relationships as graph edges, construct a situation event graph, identify the target situation evolution path, and output the target situation awareness results.

[0012] In a preferred embodiment, the target observation data includes target azimuth, target distance, target altitude, target speed, target heading, electronic identification, communication link status, node observation time, node observation delay, and observation source type;

[0013] Each low-altitude observation frame in the low-altitude observation frame sequence includes at least the node identifier, target candidate identifier, observation time, spatial coordinates, motion status, identity status, link status, and observation source type.

[0014] In a preferred embodiment, constructing the node observation trust domain includes:

[0015] The basic coverage area of ​​the sensing node is generated based on the installation coordinates, detection distance boundary, azimuth coverage angle and elevation coverage angle of the sensing node;

[0016] Based on the building occlusion area, terrain occlusion area and target height range, the node basic coverage area is occluded and corrected to generate a spatially reliable boundary.

[0017] Generate a reliable time window and a tolerance range based on the link delay fluctuation range and the historical observation deviation range;

[0018] The node observation credibility domain is formed by the spatial credibility boundary, the temporal credibility window, and the deviation allowance range.

[0019] In a preferred embodiment, the low-altitude observation frame sequence is subjected to reliability labeling and mismatch removal, including:

[0020] Determine whether the target candidate point in the low-altitude observation frame is located within the corresponding node's observation confidence domain. If the target candidate point is located within the node's observation confidence domain and the node's observation delay is within the time confidence window, then mark it as a confidence observation frame.

[0021] If a target candidate point exceeds the node observation confidence domain but can be verified by adjacent sensing nodes within an adjacent time window, it is marked as a frame to be verified.

[0022] If a target candidate point exceeds the node observation confidence domain and cannot form a successor verification, it is marked as a mismatched observation frame and removed from the low-altitude observation frame sequence.

[0023] In a preferred embodiment, generating the target continuation relationship includes:

[0024] Within adjacent time windows, candidate observation frame pairs corresponding to different sensing nodes are extracted, and the time interval, spatial displacement, velocity change, altitude change, heading change, electronic identity continuation status, and observation confidence domain overlap status between the candidate observation frame pairs are obtained respectively.

[0025] When the time interval is within the continuity time range, the spatial displacement is within the target reach range, the velocity change and altitude change are within the target maneuverability range, the heading change is within the heading permission range, and the electronic identity continuation state or the observation credibility domain overlap state meets the continuity conditions, the same target continuity relationship is established between candidate observation frame pairs.

[0026] In a preferred embodiment, constructing a trajectory hypothesis tree when target observations are missing includes:

[0027] Using the target's spatial position, altitude, speed, heading, and timestamp before disappearance as the root node, multiple hidden trajectory branches are generated based on the low-altitude channel boundary, node coverage blind zone, maximum speed variation range, maximum altitude variation range, and maximum heading variation range.

[0028] For each hidden state trajectory branch, predictive trajectory nodes are generated in the order of time progression, and an reachable envelope is formed based on the reachable time window, reachable spatial range, speed allowable range, and height allowable range corresponding to the predictive trajectory nodes.

[0029] When a reliable observation frame falls into the reachable envelope and satisfies the motion state continuity condition, the reliable observation frame is continued to the corresponding hidden state trajectory branch, generating a continuous target track chain and trajectory continuation event.

[0030] In a preferred embodiment, the consistency reverse verification includes:

[0031] Match the continuous target track chain with the declared starting point, declared route, authorized airspace and mission time window corresponding to the electronic identity;

[0032] When a continuous target track chain cannot be reached from the declared starting point within the mission time window, or deviates from the declared route and exceeds the authorized airspace boundary, an identity track inconsistency event is generated.

[0033] When the same electronic identity corresponds to multiple spatially inaccessible consecutive target track chains, a suspected identity reuse event is generated;

[0034] When a continuous target track chain remains continuous while an abnormal switch of electronic identity occurs, an identity switch event is generated;

[0035] Events involving inconsistent identity trajectories, suspected identity reuse, and identity switching are grouped into identity anomaly events.

[0036] In a preferred embodiment, extrapolating the future reachable state subdomain includes:

[0037] Based on the target's current position, velocity, heading, altitude, and maneuverable boundaries, generate the target's reachable space range within consecutive time slices;

[0038] Determine whether the reachable space ranges of different targets overlap within the same time slice; if overlap occurs, generate a track conflict event.

[0039] Determine whether the reachable space is close to the authorized airspace boundary, no-fly zone, take-off and landing area, or communication blind spot. If the proximity condition is met, generate a boundary risk event.

[0040] Communication interruption events are generated based on continuous abnormal conditions in the link status.

[0041] In a preferred embodiment, constructing a situation event graph includes:

[0042] The events of identity anomaly, flight path conflict, boundary risk, communication interruption, and trajectory continuity are grouped according to target identification and time sequence, and the time sequence edge, spatial proximity edge, target common origin edge, and risk progression edge between the events are established.

[0043] Identify the target situation evolution path based on continuous graph edges, and output the target's continuous trajectory, identity consistency status, conflict trend status, boundary risk status, node observation reliability status, and situation evolution level.

[0044] A target situation awareness system based on low-altitude intelligent network, used to implement the aforementioned target situation awareness method based on low-altitude intelligent network, includes:

[0045] The observation access module collects target observation data output by radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes and positioning nodes within the low-altitude intelligent network, and generates a low-altitude observation frame sequence after unified time reference correction and unified spatial coordinate system mapping.

[0046] The trusted filtering module constructs a trusted domain for node observation based on the coverage boundary, occlusion area, link delay, historical observation deviation, and target altitude range of the sensing node. It then performs trusted labeling and mismatch removal on the low-altitude observation frame sequence to form a set of trusted observation frames.

[0047] The track continuity module generates target continuity relationships based on a set of reliable observation frames, according to time reachability, spatial reachability, motion continuity, identity continuity and observation reliability domain overlap. When target observations are missing, it constructs a trajectory hypothesis tree, matches subsequent reliable observation frames to the reachable envelope, and generates continuous target track chains and trajectory continuity events.

[0048] The situation simulation module performs a reverse consistency check between the continuous target track chain and electronic identity, declared route, authorized airspace and mission time window, and infers the future reachable situation subdomain based on the target motion state, maneuverable boundary and link state, generating identity anomaly events, track conflict events, boundary risk events and communication interruption events.

[0049] The event graph module uses identity anomaly events, track conflict events, boundary risk events, communication interruption events, and track continuity events as graph nodes, and uses temporal sequence, spatial proximity, target homology, and risk progression as graph edges to construct a situation event graph, identify the target situation evolution path, and output the target situation awareness results.

[0050] The technical effects and advantages of this invention are as follows:

[0051] This invention achieves continuous identification, reliable tracking, and risk assessment of UAV targets in low-altitude intelligent network scenarios by constructing a progressive perception mechanism that includes low-altitude observation frame generation, node observation credibility domain filtering, track continuation reconstruction, situation event generation, and event graph analysis. By fusing data output from radar nodes, communication measurement nodes, video perception nodes, radio frequency monitoring nodes, and positioning nodes, the invention first performs unified time correction and spatial mapping on observation results from different sources, and then constructs an observation credibility domain based on node coverage boundaries, obstruction areas, link delays, and historical deviations. This effectively reduces erroneous observations caused by low-altitude obstruction, multipath reflection, video false detection, and link lag.

[0052] Then, by matching the reachability envelope of short-term disappearing targets through target continuity relationships and trajectory hypothesis trees, the system avoids track breaks and duplicate filings after UAVs cross bridge towers, buildings, or communication blind spots. At the same time, it performs consistency reverse verification by combining electronic identification, declared flight routes, authorized airspace, and mission time windows, and extrapolates future reachability subdomains. This enables timely detection of risk events such as identity anomalies, track conflicts, boundary approach, and communication interruptions. Finally, the system identifies the risk progression path through situational event maps and outputs interpretable target situational awareness results, improving the continuity, accuracy, and targeted early warning and response of low-altitude UAV supervision. Attached Figure Description

[0053] Figure 1 This is a flowchart of a target situational awareness method based on low-altitude intelligent networking according to the present invention.

[0054] Figure 2 This is a schematic diagram of the structure of a target situational awareness system based on low-altitude intelligent networking according to the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1: As Figure 1 As shown, a target situational awareness method based on low-altitude intelligent networks includes the following steps:

[0057] In one specific embodiment, the low-altitude intelligent network is deployed in the low-altitude monitoring area of ​​urban cross-river passages and the waterfront parks on both sides. This area includes the main span of the bridge, bridge ramps, park buildings, low-altitude logistics take-off and landing points, and key protection boundaries. Monitored targets include inspection drones, logistics drones, emergency detection aircraft, and other low-altitude moving targets temporarily entering the area.

[0058] To achieve continuous perception of low-altitude targets, millimeter-wave radar nodes are deployed on the rooftops of buildings at both ends of the bridge, communication measurement nodes are deployed on communication towers in the park, video sensing nodes are deployed at take-off and landing points and open areas at the bridgehead, and radio frequency monitoring nodes are deployed at key protection boundaries. Data output from BeiDou or other positioning nodes carried by low-altitude targets is also connected. Each node is connected to edge processing equipment through the low-altitude intelligent network, and the edge processing equipment uniformly completes observation access, time correction, and spatial coordinate mapping.

[0059] S1. Collect target observation data output from radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes, and positioning nodes within the low-altitude intelligent network. After unified time reference correction and unified spatial coordinate system mapping, generate a low-altitude observation frame sequence. Specific implementation steps include:

[0060] During the data acquisition process, radar nodes output target azimuth, target range, target altitude, and target velocity. Target azimuth represents the horizontal angle of the target relative to the radar node; target range represents the slant range or horizontal distance of the target relative to the radar node; target altitude represents the target's flight altitude relative to a unified altitude reference; and target velocity can be calculated from continuous radar echo position changes or radar radial velocity. Video sensing nodes output target image location, target category, target heading estimation results, and target candidate identifiers. The video sensing nodes first detect low-altitude moving targets in the image, and then, combining camera installation parameters, field-of-view parameters, and geocalibration parameters, convert the image coordinates into estimated spatial locations within the monitored area.

[0061] The communication measurement node is used to output the target's communication link status, communication identifier, node observation time, node observation delay, and signal angle of arrival or time difference of arrival data. The radio frequency monitoring node is used to output the target's electronic identification, radio frequency signal strength, frequency band characteristics, identity broadcast status, and observation source type. The positioning node is used to output the target's positioning coordinates, altitude, speed, heading, and positioning timestamp.

[0062] After receiving data from each node, the edge processing device adds a node identifier and observation source type to each target observation data point, including:

[0063] Node identifiers are used to distinguish specific device numbers for radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes, and positioning nodes;

[0064] The observation source type is used to indicate that the observation data comes from radar ranging, communication measurement, video recognition, radio frequency monitoring, or location reporting;

[0065] For the same low-altitude target, if it carries an electronic identification tag, the edge processing device writes the electronic identification tag into the corresponding observation data;

[0066] If an observation source cannot directly obtain an electronic identity, the edge processing device generates a temporary target candidate identity for the target and associates it with time, space and motion state in subsequent processing.

[0067] The edge processing device uses a unified time reference to correct the node observation time, which solves the problems of inconsistent sampling periods of different sensing nodes, different transmission link delays and deviations of local clocks. Specifically, the edge processing device uses the unified timing signal of the low-altitude intelligent network or the edge server clock as the time reference, reads the node observation time and node observation delay in the observation data of each target, and obtains the actual observation time of the target by subtracting the node observation delay from the node observation time.

[0068] When a node has a calibrated local clock offset, the actual observation time is corrected based on the local clock offset. After time correction, the target observation data uploaded by different nodes in different sampling periods are converted to the same time axis, which facilitates subsequent target continuation, trajectory reconstruction and event merging in adjacent time windows.

[0069] In terms of spatial processing, the edge processing device uses a unified spatial coordinate system pre-established in the monitored area as the mapping reference. This unified spatial coordinate system can be a three-dimensional coordinate system with the center of the low-altitude take-off and landing point, the midpoint of the bridge, or the geographical reference point of the monitored area as the origin. For the azimuth, distance, and altitude data output by the radar node, the edge processing device converts the radar polar coordinate observation results into three-dimensional spatial coordinates under the unified spatial coordinate system according to the radar node's installation coordinates, installation height, installation orientation, and pitch angle.

[0070] For the image coordinates output by the video sensing node, the edge processing device converts the image planar position into spatial coordinates within the monitored area based on the camera's intrinsic and extrinsic parameters, installation position, shooting direction, and target height estimation.

[0071] For the angle of arrival, time difference of arrival, or signal strength variation data output by the communication measurement node, the edge processing device combines the installation coordinates of the communication measurement node and the measurement model to generate an estimated value of the target spatial location. For the latitude, longitude, and altitude data output by the positioning node, the edge processing device converts the geographic coordinates into spatial coordinates under a unified spatial coordinate system.

[0072] After completing time correction and spatial coordinate mapping, the edge processing device generates a low-altitude observation frame sequence according to a preset framing period, including:

[0073] Each low-altitude observation frame corresponds to an observation time or a short time window, used to record the target observation results obtained by one or more sensing nodes within that time window. The low-altitude observation frame contains the node identifier, target candidate identifier, observation time, spatial coordinates, motion state, identity state, link state, and observation source type. Among them, the spatial coordinates are used to represent the position of the target in a unified spatial coordinate system; the motion state includes the target speed, target heading, and target altitude change status.

[0074] Identity status includes whether the electronic identity exists, whether the electronic identity is continuous, and whether the electronic identity has changed.

[0075] Link status includes whether the communication link is normal, whether the node observation delay exceeds the limit, and whether the identity backhaul is interrupted;

[0076] The observation source type indicates that the data in this frame originates from one or more of the following: radar, communications, video, radio frequency, or positioning.

[0077] For example, in this embodiment, when an inspection drone takes off from the bridgehead landing point and flies along the direction of the bridge, the bridgehead video sensing node first detects the target and generates a target candidate identifier from the video source. Then, the building top radar node outputs the target's azimuth, distance, altitude and speed. The communication measurement node records the communication link status between the target and the low-altitude intelligent network. The radio frequency monitoring node reads the target's electronic identification. The positioning node uploads the target's positioning coordinates and heading.

[0078] The edge processing device uniformly corrects the data from different nodes to the same observation time and maps them to a unified spatial coordinate system, forming a low-altitude observation frame of the same target within the same time window. If a node fails to provide an electronic identity, the edge processing device still retains the spatial coordinates, motion status and observation source type output by that node, and enables the observation frame to participate in subsequent trusted domain screening and track continuation through the target candidate identifier.

[0079] Through the above processing, the low-altitude observation frame sequence is not just a set of target location points, but a data sequence that simultaneously contains the target's physical location, motion state, identity status, link status, and observation source status. This low-altitude observation frame sequence provides the node observation delay, observation source type, and historical observation basis for the subsequent construction of the node observation trusted domain, provides target candidate points, spatial coordinates, and link status for the subsequent generation of trusted observation frame sets, provides the calculation basis for time interval, spatial displacement, velocity change, altitude change, and heading change for the subsequent establishment of target continuity relationships, and also provides the electronic identity, identity status, and link status basis for subsequent identity reverse verification, communication interruption event generation, and situation event map construction.

[0080] After generating the low-altitude observation frame sequence, the edge processing device does not directly use all the low-altitude observation frames for track continuation. Instead, it first limits the observation reliability of each sensing node in the current low-altitude environment. In this embodiment, since the monitored area includes the main span of the bridge, buildings at the bridgehead, park buildings, communication towers, and low-altitude take-off and landing points, the observable range of different nodes is not consistent at different heights and in different directions. For example, radar nodes on the top of buildings have better detection conditions in the direction of the open river surface, but are easily blocked by buildings in the area behind the buildings.

[0081] The video sensing node can reliably identify the target outline near the take-off and landing point, but it will produce positional deviations in the case of backlight, long distance, or occlusion by bridge structure.

[0082] Communication measurement nodes can continuously obtain link status, but observation delay fluctuations can occur under multipath reflection or link congestion. Therefore, edge processing devices need to first construct a node observation trust domain for each sensing node, and then perform trust filtering on the low-altitude observation frame sequence.

[0083] S2. Based on the coverage boundary, obstruction area, link delay, historical observation deviation, and target altitude range of the sensing nodes, a node observation credibility domain is constructed. The low-altitude observation frame sequence is then labeled with credibility and mismatched frames are removed to form a set of credibility observation frames. Specific implementation steps include:

[0084] The edge processing device reads the installation coordinates, installation height, installation orientation, detection distance boundary, azimuth coverage angle, and pitch coverage angle of each sensing node. The installation coordinates are used to determine the fixed position of the node in a unified spatial coordinate system. The detection distance boundary is used to limit the farthest and closest distances at which the node can output effective observation results. The azimuth coverage angle is used to limit the observable direction range of the node in the horizontal plane. The pitch coverage angle is used to limit the observable height range of the node in the vertical direction. Starting from the installation coordinates of the sensing node, the edge processing device expands outward according to the detection distance boundary, azimuth coverage angle, and pitch coverage angle to obtain the three-dimensional coverage space corresponding to the node under unobstructed conditions. This three-dimensional coverage space is the basic coverage domain of the node.

[0085] After obtaining the basic coverage area of ​​the node, the edge processing device combines the building occlusion area, terrain occlusion area, bridge structure occlusion area and target height range of the monitored area to perform occlusion correction on the basic coverage area of ​​the node.

[0086] The area obscured by a building can be determined by the building's outline, height, facade location, and the location of bridge ancillary structures.

[0087] The area obscured by terrain can be determined by the difference in elevation between the riverbanks, the slope, densely wooded areas, or the location of fixed facilities;

[0088] The target altitude range is used to characterize the range of possible flight altitudes of low-altitude targets within the current regulatory area.

[0089] The edge processing device determines whether the observation path of each spatial location is blocked by buildings, terrain or fixed facilities in the node's basic coverage area. If a spatial location is located in an area where the node's line of sight is blocked, then the spatial location is deducted from the node's basic coverage area.

[0090] If the height corresponding to a certain spatial location exceeds the height range that the node's pitch coverage angle can stably observe, then that location will also be excluded from the stable and reliable range.

[0091] After occlusion correction, the edge processing device obtains a spatially reliable boundary, which represents the spatial range within which the corresponding sensing node can reliably output the target location observation results in the current monitored area.

[0092] Regarding link latency, the edge processing device records the node observation latency of each sensing node uploading observation data within multiple consecutive acquisition cycles, and calculates the latency variation range of the node under stable communication conditions. When the observation latency of a certain node is in a state of small fluctuation, the edge processing device takes the fluctuation range as the link latency fluctuation range of the node.

[0093] When a node experiences temporary congestion, backhaul delay, or link switching, the edge processing device readjusts the link latency fluctuation range based on the latency changes of the most recent few frames. The link latency fluctuation range yields a time reliability window, which is used to determine whether the observation time corresponding to a given low-altitude observation frame still represents the true state of the target within the current time window.

[0094] If the node observation delay significantly exceeds the time reliability window, even if the target candidate point falls within the spatial reliability boundary of the node, it may not be suitable to directly participate in the current frame's track continuation due to data backhaul delay.

[0095] Regarding historical observation bias, the edge processing device calculates the deviation changes of each node in the target orientation, target distance, target height, spatial coordinates and motion state based on the historical observation results of multiple sensing nodes in the same area for the same target. For example, when the radar node and the positioning node both observe the same inspection drone in multiple consecutive time windows, the edge processing device can compare the difference between the radar mapping coordinates and the positioning coordinates to obtain the historical observation bias of the radar node in that direction and distance range.

[0096] When video sensing nodes and radar nodes continuously observe the same target, they can compare the differences between the video mapped coordinates and the radar coordinates to obtain the spatial deviation of the video node within the current field of view. The edge processing device generates a permissible deviation range based on these historical deviation changes. This permissible deviation range allows node observations to participate in subsequent judgments within a reasonable error range, preventing the erroneous rejection of real target observations due to small measurement errors.

[0097] The spatial trust boundary, the temporal trust window, and the deviation allowable range together form the node observation trust domain. This node observation trust domain is updated according to the link status of the sensing node, historical observation deviation, and target height range. When the node observation delay of the same node exceeds the temporal trust window, the historical observation deviation increases continuously, the number of mismatched observation frames exceeds the preset number, or the target height range changes across layers within multiple consecutive acquisition cycles, the edge processing device triggers the node observation trust domain update and redetermines the spatial trust boundary, the temporal trust window, and the deviation allowable range. For example, when a communication measurement node experiences an increase in delay within a continuous time window, the edge processing device reduces the temporal trust window corresponding to that node.

[0098] When the positional deviation of a video sensing node increases due to backlighting, the edge processing device expands its target area but reduces its direct trusted labeling ratio.

[0099] When the target height increases from near the bridge deck to above the top of the building, some observation directions that were originally affected by building obstruction can re-enter the spatial credibility boundary;

[0100] Through the above update method, the node observation confidence domain can reflect the actual usable observation capabilities of each node in the current low-altitude intelligent network.

[0101] After constructing the node observation trust domain, the edge processing device performs trust labeling on each frame of the low-altitude observation frame sequence, specifically including:

[0102] For any low-altitude observation frame, the edge processing device first reads the node identifier, target candidate identifier, observation time, spatial coordinates, link status and node observation delay in the frame, and then calls the node observation confidence domain corresponding to the node for judgment.

[0103] When the spatial coordinates of a target candidate point are within the spatial credibility boundary of the node, and the node observation delay is within the temporal credibility window, and the target candidate point is compatible with the permissible deviation range of the node, the edge processing device marks the low-altitude observation frame as a credibility observation frame. A credibility observation frame indicates that the frame can be directly used as the basis data for subsequent target continuity calculation and track chain generation.

[0104] If a candidate target point exceeds the observation confidence domain of its corresponding node but is not obviously abnormal, the edge processing device will not immediately delete it. Instead, it will further determine whether it can be validated by adjacent sensing nodes. Adjacent sensing nodes can be spatially adjacent, have overlapping coverage areas, or be connected sequentially in the direction of low-altitude target movement. The validation logic is as follows:

[0105] The edge processing device searches for target candidate points output by other sensing nodes within adjacent time windows and determines whether the time difference, spatial distance, velocity direction and altitude changes between these target candidate points and the current target candidate point meet the low-altitude target reachability conditions.

[0106] If the observation results of adjacent nodes can indicate that the target candidate point is continuous with the target state of the previous or next frame, then the low-altitude observation frame is marked as a target observation frame to be verified. Target observation frames to be verified are not used as directly reliable data for the time being, but are retained in the cache for subsequent target continuity judgment, trajectory hypothesis tree reconnection, and target reproduction analysis in occluded areas;

[0107] If the target candidate point exceeds the node observation confidence domain and cannot be verified by adjacent sensing nodes within the adjacent time window, the edge processing device marks the low-altitude observation frame as a mismatched observation frame.

[0108] Mismatched observation frames typically correspond to reflection interference, false video detection, communication measurement drift, location lag caused by excessive time delay, or abnormal node output.

[0109] For mismatched observation frames, the edge processing device removes them from the low-altitude observation frame sequence and no longer participates in the calculation of subsequent target succession relationships; at the same time, the mismatch record can be written into the node operation status record for subsequent updates of historical observation deviation intervals and node observation confidence domains.

[0110] In the cross-river passage scenario of this embodiment, when the inspection drone enters the vicinity of the main span of the bridge from the bridgehead landing point, the target candidate point output by the radar node on the top of the building falls into its spatial credibility boundary, and the node observation delay is within the time credibility window. Therefore, the radar observation frame is marked as a credibility observation frame.

[0111] If the target image position is mapped outside the spatial credibility boundary due to the bridge railing obstruction of the bridge-head video sensing node, but the radar node and the positioning node can prove that the target moves continuously along the bridge direction within the adjacent time window, then the video observation frame is marked as the observation frame to be verified.

[0112] If a radio frequency monitoring node generates a target candidate point whose location is significantly off from the main span of the bridge and cannot be verified by adjacent nodes due to multipath reflection, then the radio frequency observation frame is marked as a mismatched observation frame and discarded.

[0113] Through the above processing, the edge processing device writes low-altitude observation frames marked as reliable observation frames into the reliable observation frame set, writes observation frames to be verified into the verification buffer, and removes mismatched observation frames from the low-altitude observation frame sequence. After being verified by adjacent sensing nodes in the subsequent target continuity judgment or trajectory hypothesis tree reconnection process, the observation frames to be verified are transferred to the reliable observation frame set to participate in the generation of continuous target track chains. This enables the direct use of data affected by occlusion, time delay drift, or false detection when calculating time accessibility, spatial accessibility, motion continuity, identity continuity, and the overlap relationship of the observation reliability domain, thereby improving the accuracy of continuous target track chain and trajectory hypothesis tree construction.

[0114] It should be noted that, in this embodiment, the rendezvous time range, target reachability range, target maneuverability permitted range, heading permitted range, preset missing tolerance window, safety interval, and preset proximity range can be determined based on the flight management rules of the low-altitude monitoring area, UAV type, mission route, sensing node sampling cycle, communication feedback cycle, and historical operational data. For inspection UAVs, the speed permitted range, altitude permitted range, and heading permitted range can be determined based on their rated maximum speed, maximum climb speed, maximum descent speed, maximum turning angular velocity, and the permitted flight altitude of the monitoring area. For logistics UAVs, the target reachability range and rendezvous time range can be determined based on their declared route, take-off and landing point location, and mission time window. The above parameters can be pre-configured or corrected based on the node observation confidence domain update results and historical rendezvous success records.

[0115] S3. Based on a set of reliable observation frames, generate target continuity relationships according to temporal reachability, spatial reachability, motion continuity, identity continuity, and overlap of the reliable observation domain. When target observations are missing, construct a trajectory hypothesis tree, match subsequent reliable observation frames to the reachable envelope, and generate continuous target track chains and trajectory continuity events. Specific implementation steps include:

[0116] The edge processing device extracts candidate observation frame pairs from the set of trusted observation frames, using a preset continuation time window as a unit.

[0117] Candidate observation frame pairs can come from adjacent observation frames of the same sensing node, or from observation frames output by different sensing nodes within adjacent time windows. For example, if a bridgehead video sensing node outputs the spatial coordinates and heading of a target in a previous time window, and a building-top radar node outputs the distance, altitude, and velocity of a target at a similar spatial location in a subsequent time window, then these two observation frames can be used as candidate observation frame pairs for continuation judgment.

[0118] For each candidate observation frame pair, the edge processing device first calculates the time interval between the two. The time interval is obtained by subtracting the observation time of the previous observation frame from the observation time of the later observation frame, and is used to determine whether the two frames are within the contiguous time range.

[0119] If the time interval is too large, even if the spatial locations are close, it may not be possible to prove that the two frames correspond to the same target.

[0120] If the time interval is within the range of consecutive time intervals, then continue with the spatial reachability assessment.

[0121] Spatial reachability is determined by the spatial displacement and target mobility between candidate observation frames, specifically including:

[0122] The edge processing device determines the displacement distance of the target in a unified spatial coordinate system based on the spatial coordinates in the two frames, and combines the drone type, low-altitude channel restrictions, the allowable flight speed range in the monitored area and the target speed in the previous observation frame to determine whether the displacement can actually be reached within the corresponding time interval.

[0123] If the spatial displacement between candidate observation frames exceeds the maximum reachable distance of the target within that time interval, then the two are determined not to satisfy spatial reachability.

[0124] If the spatial displacement is within the reach of the target, then continue to assess the continuity of motion.

[0125] Motion continuity includes the continuous judgment of changes in velocity, altitude, and heading. The edge processing device reads the target velocity, target altitude, and target heading from the candidate observation frames and compares the changes in velocity, altitude, and heading between two consecutive frames.

[0126] If the change in speed is within the target's permissible maneuver range, it means that the target can complete the corresponding acceleration or deceleration within that time interval;

[0127] If the change in altitude is within the target's maneuverability range, it means that the target is capable of completing the corresponding climb or descent;

[0128] If the change in heading is within the heading permission range, it means that the target's turning amplitude conforms to the normal maneuvering characteristics of a low-altitude aircraft. When speed, altitude, and heading all meet the continuity condition, the candidate observation frame pair has the motion basis to form a succession relationship of the same target.

[0129] Identity continuity is used to help determine whether candidate observation frame pairs belong to the same target. That is, if both frames contain electronic identity identifiers and the electronic identity identifiers are consistent, the edge processing device recognizes it as an identity continuity state that meets the continuation conditions.

[0130] If one frame lacks an electronic identity, but the spatial location, motion state, and observation source of another frame can correspond continuously with the preceding and following frames, the edge processing device will not directly deny the continuation. Instead, it will continue to judge based on the observation trust domain overlap state. The observation trust domain overlap state refers to whether the two sensing nodes corresponding to the candidate observation frame have coverage connection or trust domain overlap in the target area.

[0131] If the node observation confidence domain of the previous node overlaps or continuously covers the node observation confidence domain of the next node in the direction of target movement, it indicates that the target's movement from one node coverage area to another node coverage area is interpretable.

[0132] When a candidate observation frame pair simultaneously meets the following conditions: the time interval is within the continuity time range, the spatial displacement is within the target reach range, the velocity and altitude changes are within the target maneuverability range, the heading change is within the heading permission range, and the electronic identity continuation status or the observation credibility domain overlap status meets the continuity conditions, the edge processing device establishes the same target continuity relationship between the candidate observation frame pair. After multiple consecutive same target continuity relationships are connected in the order of observation time, a continuous target track chain is formed for the target. This continuous target track chain records the changes in the target's spatial coordinates, velocity, altitude, heading, identity status, and observation source within multiple time windows.

[0133] In actual low-altitude scenarios, the lack of target observation is quite common. For example, when an inspection drone flies along the main span of a bridge, it may not be recognized by the video sensing node for a short period of time when it passes through bridge towers, cable stays, building obstruction areas or communication blind spots. It may also be due to weak radar reflection or communication backhaul delay that no reliable observation frame is generated in the current time window. If the target is directly judged to have disappeared at this time, it will cause the track to break.

[0134] If the target is re-filed when it reappears, the same target will be recorded repeatedly. Therefore, this embodiment constructs a trajectory hypothesis tree when the target observation is missing, in order to maintain the hidden state of the target during the invisible time period.

[0135] Specifically, when a continuous target track chain fails to match a new credible observation frame within one or more consecutive time windows, the edge processing device uses the spatial position, altitude, speed, heading, and timestamp of the last credible observation frame before the target disappears as the root node. The root node represents the known state of the target before it enters the observation missing state.

[0136] Subsequently, the edge processing device combines the low-altitude channel boundary, node coverage blind spot, maximum speed change range, maximum altitude change range, and maximum heading change range to generate multiple hidden state trajectory branches.

[0137] Low-altitude passage boundaries are used to limit the spatial range in which a target may continue to fly, such as inspection passages above bridges, logistics routes in parks, access passages to and from take-off and landing points, or authorized airspace boundaries. Node coverage blind spots are used to indicate areas that sensing nodes cannot stably observe within the current time period, such as areas behind bridge towers, building obstruction areas, video field of view edges, low radar reflection areas, or weak communication coverage areas. Maximum speed variation range, maximum altitude variation range, and maximum heading variation range are used to limit possible maneuvering changes of the target within adjacent time slices to prevent the infinite spread of trajectory branches.

[0138] When generating hidden trajectory branches, the edge processing device uses the speed and heading of the root node as a basis to generate several candidate branches according to the possible movement modes of the target, such as maintaining heading, making a slight left turn, making a slight right turn, decelerating, climbing, descending, or detouring along the boundary of the low-altitude channel. Each branch must be within the allowable range of the boundary of the low-altitude channel and must not cross the area of ​​inaccessible fixed obstacles.

[0139] For branches entering the node coverage blind zone, the edge processing device allows them to remain in the blind zone for a certain period of time, but the duration must not exceed the preset missing tolerance window;

[0140] For branches that clearly deviate from the authorized path and cannot be explained by the target's maneuverability, the edge processing device will not retain them;

[0141] When the duration of a hidden state trajectory branch exceeds a preset missing tolerance window, the predicted trajectory node enters an inaccessible obstacle region, the predicted trajectory node exceeds the authorized airspace boundary and is not verified by subsequent trusted observation frames, or the branch fails to match any trusted observation frame within multiple consecutive time slices, the edge processing device terminates the hidden state trajectory branch.

[0142] For each hidden trajectory branch, the edge processing device generates predicted trajectory nodes in chronological order. These predicted trajectory nodes represent the possible position, altitude, velocity, and heading of the target during the period of missing observations. Each predicted trajectory node corresponds to an reachable time window, reachable spatial range, velocity permissible range, and altitude permissible range. The reachable time window is determined by the root node timestamp and the continuously advancing time slices.

[0143] The reachable space is jointly defined by the position, velocity, heading of the previous predicted trajectory node and the target's maneuverability range;

[0144] The permissible speed range is determined by the target's current speed, the permissible acceleration range, and the permissible deceleration range.

[0145] The altitude permissible range is determined by the target's current altitude, the permissible climb range, the permissible descent range, and the low-altitude passage altitude limit. These ranges together form the reachable envelope.

[0146] The reachability envelope is used to determine whether a subsequently reappearing reliable observation frame can be continued to a certain hidden state trajectory branch, specifically including:

[0147] When a new trusted observation frame appears, the edge processing device reads the observation time, spatial coordinates, velocity, altitude, heading, and identity status of the trusted observation frame, and matches them with the reachable envelopes of each hidden state trajectory branch in the trajectory hypothesis tree. If the observation time of the trusted observation frame falls within the reachable time window of a certain branch, its spatial coordinates fall within the reachable space range corresponding to that branch, and its velocity, altitude, and heading are continuous with the motion state of the predicted trajectory node of that branch, then it is determined that the trusted observation frame can be continued to that hidden state trajectory branch.

[0148] If multiple hidden state trajectory branches can be connected to the same trusted observation frame, the edge processing device prioritizes the hidden state trajectory branch with the smoothest motion change, the most consistent low-altitude channel constraints, the most complete connection relationship of the observation trusted domain, and the most stable electronic identity continuation state. The smoothest motion change means that the branch has small and continuous speed, altitude, and heading changes from the root node to the subsequent trusted observation frame.

[0149] The most consistent low-altitude corridor constraints mean that the branch is always located within the authorized corridor or reasonable flight area;

[0150] The most complete connection relationship of the observation credibility domain means that there is spatial continuity between the blind zone, the occlusion zone and the coverage area of ​​the re-emerging node traversed by the branch;

[0151] The most stable electronic identity continuation state means that the reappearing credible observation frame is consistent with or does not conflict with the electronic identity identifier before the target disappeared.

[0152] In this embodiment, when the inspection drone flies along the main span of the bridge, before entering the area blocked by the bridge tower, the last reliable observation frame output by the radar node shows that the drone is located on the south side of the bridge tower, with its heading pointing to the middle of the bridge and its altitude within the height range of the inspection channel. In the following two time windows, neither the video sensing node nor the radar node outputs a reliable observation frame. Based on this, the edge processing device determines that the target observation is missing and uses the last reliable observation frame as the root node to construct a trajectory hypothesis tree.

[0153] Based on the boundary of the inspection channel above the bridge, the occlusion range of the bridge tower, and the maximum turning range of the drone, the edge processing equipment generates hidden trajectory branches such as maintaining the heading through the occlusion area of ​​the bridge tower, slightly detouring along the outside of the bridge tower, and slightly climbing and then passing through the upper edge of the occlusion area.

[0154] Subsequently, when the radar node on the north side of the bridge tower outputs a reliable observation frame again, the edge processing device determines that the observation time, spatial coordinates, altitude, and heading of the reliable observation frame all fall within the reachable envelope of the branch that maintains the heading through the bridge tower's obstruction area. Therefore, the reliable observation frame is continued to the hidden state trajectory branch to maintain the continuous target track chain of the inspection UAV.

[0155] When a reliable observation frame is connected to the corresponding hidden state trajectory branch, the edge processing device merges the root node, the predicted trajectory node in the hidden state trajectory branch and the subsequent reliable observation frame in chronological order to generate a continuous target trajectory chain. This continuous target trajectory chain not only includes the actually observed reliable observation frames, but also records the hidden state trajectory branches and corresponding reachable envelope information during the observation missing period.

[0156] At the same time, the edge processing device generates a trajectory continuation event, which includes at least the target candidate identifier, the continuation occurrence time, the last reliable observation frame before continuation, the reliable observation frame after continuation, the selected hidden state trajectory branch, the reachable envelope matching result, and the source of the continuation node.

[0157] Through the above processing, this embodiment can maintain the continuity of target tracks in the case of low-altitude target cross-node movement, short-term occlusion, link anomaly, or coverage blind spot. The generated continuous target track chain provides the basis for the actual movement process of the target for subsequent consistency reverse verification. The track continuation event is used as a graph node in the subsequent situation event graph to characterize the target's occlusion continuation, blind spot recurrence, or cross-node continuation process, thereby supporting the comprehensive judgment of subsequent identity anomaly events, track conflict events, and boundary risk events.

[0158] After obtaining continuous target track chains and trajectory continuity events, the edge processing device further comprehensively judges the target's identity credibility, track compliance, future conflict trends, and communication stability. In this embodiment, the low-altitude intelligent network not only needs to determine the target's current location, but also needs to determine whether the target matches its declared identity, whether it is still within the authorized airspace, whether it may intersect with other targets, and whether there is a continuous communication link anomaly. Therefore, the edge processing device uses continuous target track chains as the basis for physical movement, and uses electronic identity, declared flight routes, authorized airspace, and mission time windows as the basis for mission compliance. It generates corresponding situation events through consistency reverse verification and future reachable situation subdomain deduction.

[0159] S4. Perform a reverse consistency check between the continuous target track chain and the electronic identity, declared route, authorized airspace, and mission time window. Based on the target's motion state, maneuverable boundaries, and link status, infer the future reachable situation subdomain and generate identity anomaly events, track conflict events, boundary risk events, and communication interruption events. Specific implementation steps include:

[0160] The edge processing device first reads the target candidate identifier, spatial coordinate sequence, altitude sequence, speed sequence, heading sequence, observation time sequence, electronic identity change record and trajectory continuity event record in the continuous target track chain. For targets with electronic identity identifiers, the edge processing device retrieves the corresponding declaration start point, declaration route, authorized airspace and mission time window from the low-altitude mission declaration database or the regulatory mission table. The declaration start point is used to indicate the take-off position or the position where the target is allowed to start the mission or enter the regulatory area.

[0161] The declared flight path is used to indicate the low-altitude passage or flight corridor that the target should pass through;

[0162] Authorized airspace is used to define the horizontal and vertical boundaries of the target's permitted activities;

[0163] The task time window is used to limit the start time, end time, and allowable deviation range of the target task.

[0164] The processing logic for the consistency reverse verification is as follows:

[0165] The edge processing device does not simply rely on whether the target carries an electronic identification to determine the legality of the target. Instead, it substitutes the continuous target track chain into the declaration information corresponding to the electronic identification to determine whether the physical movement process of the target can be reasonably explained by the declaration starting point, declaration route, authorized airspace and mission time window. Specifically, the edge processing device first determines whether the track chain can be reached from the declaration starting point within the mission time window based on the first valid position, the first valid observation time and the target's movement capability.

[0166] If the starting position of the track chain is too far from the declared starting point, and the target cannot be reached within the corresponding time according to the allowed speed and mission start time, it is considered that although the target carries an electronic identification, its physical trajectory does not match the declared starting point.

[0167] Subsequently, the edge processing device spatially matches each trajectory point in the continuous target trajectory chain with the declared route to determine whether the trajectory point runs along the declared route or the allowable deviation zone of the declared route.

[0168] At the same time, the horizontal position and altitude corresponding to the trajectory point are compared with the authorized airspace boundary to determine whether the target exceeds the authorized airspace.

[0169] When a continuous target track chain cannot be reached from the declared starting point within the mission time window, or when the continuous target track chain deviates from the declared route and exceeds the authorized airspace boundary, the edge processing device generates an identity trajectory inconsistency event. The identity trajectory inconsistency event records the electronic identity identifier, target candidate identifier, time of anomaly occurrence, abnormal trajectory segment, declared starting point, declared route, authorized airspace boundary, and reason for inconsistency.

[0170] For example, in a cross-river passage scenario, the electronic identity of a target broadcast corresponds to the task of "flying along the inspection channel on the south side of the bridge from the bridgehead take-off and landing point". However, the continuous target track chain shows that the target first appears above the buildings on the north side of the bridge. According to its speed, it is impossible to reach the location from the bridgehead take-off and landing point in a short time after the task starts. In this case, the edge processing device generates an identity trajectory inconsistency event.

[0171] When the same electronic identity corresponds to multiple consecutive target track chains within the same task time window, the edge processing device further determines whether these track chains have spatial reachability. Spatial reachability can be determined by the end position of the track chain, the starting position of another track chain, the time difference between the two, the target's maximum permissible speed, altitude change capability, and low-altitude channel connectivity.

[0172] If two track chains overlap in time, or the spatial distance between them exceeds the maximum reachable range of the target within the corresponding time difference, or the two track chains are located in unconnected authorized airspace areas, it is considered that the same electronic identity cannot correspond to multiple physical targets at the same time. The edge processing device generates an identity reuse suspected event, which is used to characterize that the electronic identity may be reused, impersonated, misbound, or repeatedly broadcast.

[0173] When a continuous target track chain remains continuous in spatial coordinates, speed, altitude and heading, but the electronic identity identifier changes abnormally within adjacent time windows, the edge processing device generates an identity switching event. This situation may occur when the target electronic identity is abnormally transmitted back, the identity broadcasting module fails, the target impersonates the identity and switches, or multiple targets are mistakenly associated in an intersecting area.

[0174] When generating an identity switching event, the edge processing device records the electronic identity identifier before the switch, the electronic identity identifier after the switch, the time of the switch, the trajectory segments before and after the switch, and the source of the observation.

[0175] The aforementioned events of inconsistent identity trajectories, suspected identity reuse, and identity switching are all grouped into identity anomaly events for use as graph nodes when constructing the situational event graph.

[0176] After completing the identity consistency judgment, the edge processing device infers the future reachable situation subdomain based on the target's motion state, maneuverable boundary and link state. The target's motion state includes the target's current position, speed, heading and altitude.

[0177] The maneuverable boundaries include the range of permissible speed changes, permissible altitude changes, permissible heading changes, low-altitude passage boundaries, authorized airspace boundaries, and safety interval requirements. The edge processing equipment extrapolates the possible spatial range that the target may reach based on continuous time slices.

[0178] For example, in the first time slice after the current moment, the reachable space of the target is determined by the current position, current speed, current heading, and the allowable range of heading changes;

[0179] In the second time slice, the edge processing device continues to superimpose the permissible ranges of speed change, altitude change and heading change based on the reachable space range of the previous time slice to obtain the reachable space range of a further time slice. The reachable space ranges in multiple consecutive time slices together constitute the future reachable situation subdomain.

[0180] It should be noted that the future reachable situation subdomain is not a single predicted point, but rather a spatial area that the target may occupy in the short future. This area is simultaneously constrained by the target's mobility and the low-altitude scenario. For a drone flying along a bridge inspection channel, its future reachable situation subdomain mainly extends along the direction of the bridge; for a logistics drone climbing from the park's take-off and landing point, its future reachable situation subdomain includes both the horizontal forward range and the altitude ascent range.

[0181] For targets near the no-fly zone boundary, their future reachability subdomain will be truncated by the authorized airspace boundary, and their extension trend toward the boundary side will be recorded in detail.

[0182] After generating future reachability subdomains for multiple targets, the edge processing device determines whether the reachability space ranges of different targets overlap in the same time slice. If the reachability space ranges of two targets overlap or the safety interval is insufficient in the same future time slice, and their altitude ranges also overlap or are close, the edge processing device generates a track conflict event. The track conflict event records the conflict target identifier, the expected conflict time slice, the conflict space region, the current speed, heading, altitude of the two targets, and the cause of the conflict.

[0183] For example, if an inspection drone flies along the direction of the bridge and another logistics drone crosses the bridgehead area laterally from the park's take-off and landing point, and if the reachable space ranges of the two overlap above the bridgehead within the next three time slices, and the height difference is less than the safe interval, a flight path conflict event will be generated.

[0184] The edge processing device also determines whether the future reachable situation subdomain is close to the authorized airspace boundary, no-fly zone boundary, take-off and landing area or communication blind spot. Among them, the proximity judgment can be determined by comprehensively considering factors such as the minimum distance between the target's future reachable space range and the corresponding boundary, whether the target's heading points to the boundary, whether the target's speed continuously increases in the direction of the boundary, and whether the target's altitude is close to the limit altitude.

[0185] If the reachable situation subdomain extends beyond the boundary of the authorized airspace in the future, or may enter the no-fly zone, take-off and landing safety protection zone or communication blind zone in the future time slice, the edge processing device will generate a boundary risk event.

[0186] In boundary risk events, record the target candidate identifier, the type of approaching boundary, the expected approach time slice, the target's current motion status, and the risk formation path.

[0187] For link status, the edge processing device reads the communication link status, node observation delay, identity feedback status and observation source changes corresponding to each low-altitude observation frame in the continuous target track chain;

[0188] If a target experiences communication link interruptions, persistently excessive node observation latency, abnormal electronic identity verification, missing location data, or can only be passively observed by non-communication sensing nodes within multiple consecutive time slices, the edge processing device determines that the link status is continuously abnormal and generates a communication interruption event. A communication interruption event is not equivalent to a single link jitter, but rather indicates that the link abnormality is persistent in time, potentially leading to delays in issuing subsequent regulatory instructions or the inability to continuously confirm the target's identity.

[0189] For example, if an inspection drone completes its trajectory in the bridge tower obstruction area, its electronic identity remains unchanged, but the continuous flight path shows that it begins to deviate from the declared inspection channel and moves towards the no-fly boundary outside the bridge, the edge processing device first generates an identity trajectory inconsistency event or boundary risk related record through consistency reverse verification.

[0190] Subsequently, based on its current position, speed, heading, and maneuverable boundaries, the future reachable situation subdomain is extrapolated. If it is found that the target may enter the no-fly zone within a certain number of time slices, a boundary risk event is generated. If the target's communication link status also shows continuous abnormal feedback at this time, the edge processing device will also generate a communication interruption event.

[0191] The above events collectively demonstrate that the target not only deviates from its flight path, but also exhibits a tendency to approach the boundary and become unstable via the link.

[0192] For example, near the take-off and landing point in the park, two drones enter the same bridgehead airspace from different channels. One target patrols along the direction of the bridge, while the other target takes off from the logistics take-off and landing point and crosses laterally. The edge processing device generates future reachability subdomains based on the two continuous target track chains. If the reachability ranges of the two overlap within the same future time slice and their altitude ranges are similar, a track conflict event is generated. If the declared route corresponding to the electronic identity of one of the targets does not pass through the bridgehead airspace, the edge processing device will also generate an identity anomaly event based on the deviation between the continuous target track chain and the declared route. Thus, it is possible to simultaneously identify two different types of situational issues: identity inconsistency and future conflict.

[0193] Through the above processing, the S4 phase further transforms the continuous target track chain obtained in the S3 phase from continuous position records into a set of events that can be used for situation assessment. Identity anomaly events are used to reflect the inconsistency between electronic identity and physical track.

[0194] Track conflict events are used to reflect the risk of overlapping future reachable spaces for multiple targets; boundary risk events are used to reflect the trend of targets approaching authorized airspace boundaries, no-fly zones, or communication blind spots.

[0195] Communication interruption events are used to reflect persistent anomalies in the target link status. These events are accompanied by the target identifier, the time of occurrence, the spatial location, the associated track segment, and the cause of the event.

[0196] After generating identity anomaly events, track conflict events, boundary risk events, communication interruption events, and track continuation events, the edge processing device further organizes these events into a situational event diagram that can express the target risk evolution process. This situational event diagram is not only used to store a single alarm result, but also to represent the sequence of events, spatial relationships, target attribution relationships, and risk progression relationships that occur in a continuous time for the same target or multiple related targets, so that the low-altitude target situation can be transformed from discrete events into a structured result with causal order and evolution path.

[0197] S5. Using identity anomaly events, track conflict events, boundary risk events, communication interruption events, and track continuity events as graph nodes, and using temporal sequence, spatial proximity, target homology, and risk progression as graph edges, construct a situational event graph to identify the target situational evolution path and output the target situational awareness results. Specific implementation steps include:

[0198] The edge processing device first performs unified event formatting processing on the trajectory continuation events generated by S3 and the identity anomaly events, track conflict events, boundary risk events and communication interruption events generated by S4.

[0199] Each event is recorded with an event identifier, event type, target identifier, associated track chain identifier, occurrence time, occurrence location, associated sensing node, event source, event triggering reason, and event status. The event type is used to distinguish between identity anomalies, track conflicts, boundary risks, communication interruptions, and track continuity; the target identifier points to the corresponding target candidate identifier or electronic identity identifier.

[0200] The associated track chain identifier is used to point to the continuous target track chain corresponding to this event;

[0201] The occurrence time is used to indicate the time slice in which the event is triggered; the occurrence spatial location is used to indicate the target location, conflict area, boundary proximity area, or trajectory continuation area corresponding to the event.

[0202] The associated sensing nodes are used to record the radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes, or positioning nodes involved in the event.

[0203] The edge processing device then merges the events according to their target identifier and time sequence, including:

[0204] For multiple events corresponding to the same target identifier, the edge processing device arranges them from earliest to latest according to the occurrence time of the events, forming a target event sequence;

[0205] When the same electronic identity corresponds to multiple consecutive target track chains, the edge processing device simultaneously retains the event sequences corresponding to multiple track chains and marks whether there is a spatial unreachability relationship between them;

[0206] In the event of a track conflict between different targets, the edge processing device will associate the event sequences of multiple targets involved in the conflict, so that the track conflict event can be connected to the corresponding multiple target track chains at the same time.

[0207] Through the above processing, the state changes of a single target and the interactive risks between multiple targets are both incorporated into the same situation event diagram.

[0208] When constructing graph nodes, the edge processing device uses each formatted event as a graph node in the situation event graph. Among them, the identity anomaly event node is used to indicate that there is an inconsistency between the target identity and the physical track, suspected identity reuse, or identity switching anomaly.

[0209] Track conflict event nodes are used to indicate that the reachable spatial ranges of multiple targets overlap or the safety interval is insufficient in a future time slice;

[0210] Boundary risk event nodes are used to indicate that the target's future reachability subdomain is approaching the authorized airspace boundary, no-fly zone, take-off and landing area, or communication blind spot;

[0211] Communication interruption event nodes are used to indicate that the target link status is interrupted, delay exceeds limits, or identity backhaul is abnormal within a continuous time slice;

[0212] The trajectory continuation event node is used to indicate that the target completes the continuation through the trajectory hypothesis tree after being in an occluded area, a blind spot, or a short-term observation gap.

[0213] When constructing graph edges, the edge processing device establishes temporal edges, spatial proximity edges, target origin edges, and risk progression edges respectively. Among them, temporal edges are used to connect events that occur consecutively in time for the same target or related targets. For example, if a target first experiences a trajectory continuation event, then a communication interruption event, and then a boundary risk event, then temporal edges are established between the above events according to the time of occurrence.

[0214] Spatial proximity edges are used to connect events that occur in close proximity to each other. For example, if a target generates a trajectory continuation event in the area obscured by a bridge tower, and then generates a boundary risk event in the adjacent bridge deck airspace, a spatial proximity edge is established if the spatial distance between the two events is within a preset proximity range. Target origin edges are used to connect events that belong to the same continuous target track chain or the same electronic identity, so that events from different time slices and different node sources can be attributed to the same target. Risk progression edges are used to connect events with risk evolution relationships. For example, communication interruption may lead to the inability to continuously confirm identity status, identity anomalies may be accompanied by track deviation, track deviation may lead to boundary risks, and the overlap of boundary risks with the future reachability subdomains of other targets may lead to track conflicts.

[0215] In a specific implementation process, the risk progression edge can be established according to preset event evolution rules, specifically including:

[0216] If a communication interruption event occurs within a certain number of consecutive time slices after the trajectory continuation event, the edge processing device considers that the target has a progressive risk relationship of "link instability after occlusion continuation";

[0217] If, after a communication interruption event, there is a loss of electronic identity, identity switching, or inconsistent identity trajectory, a risk progression edge is established from the communication interruption event to the identity anomaly event.

[0218] If the track segment corresponding to the identity anomaly event continues to extend outside the authorized airspace, a risk progression edge is established from the identity anomaly event to the boundary risk event.

[0219] If the future reachable situation subdomain corresponding to the boundary risk event overlaps with the future reachable situation subdomain of other targets, then a risk progression edge from the boundary risk event to the track conflict event is established.

[0220] Using the above rules, the situation event diagram can express the continuous evolution of risk from low-confidence observation, trajectory continuation, identity anomaly, boundary approach to conflict formation.

[0221] In a cross-river passage scenario, a certain inspection drone experienced a short-term observation loss in the area obstructed by the bridge tower. The edge processing device completed the continuation through the trajectory hypothesis tree and generated a trajectory continuation event.

[0222] Subsequently, the target reappeared on the north side of the bridge tower, but the communication link experienced excessive backhaul delays over multiple consecutive time slices, triggering a communication interruption event generated by the edge processing device. During continued flight, the target's trajectory gradually deviated from the declared inspection channel and approached the no-fly zone boundary outside the bridge. The edge processing device generated an identity anomaly event and a boundary risk event. At this point, the situational event diagram sequentially displayed trajectory continuity event nodes, communication interruption event nodes, identity anomaly event nodes, and boundary risk event nodes. These nodes were connected by temporal precedence edges and target origin edges. Since these events occurred consecutively in the adjacent space between the bridge tower and the outer boundary of the bridge, spatial proximity edges were also established. Furthermore, because the events conformed to the evolutionary relationship of "link instability after obstruction continuity, identity anomaly after link instability, and boundary approach after identity anomaly," risk progression edges were also established. Based on this, the edge processing device identified the target's situational evolution path as obstruction continuity, communication anomaly, identity anomaly, and finally boundary approach.

[0223] In another embodiment, two low-altitude targets enter the airspace at the bridgehead from the bridge inspection passage and the park logistics take-off and landing point, respectively. The first target generates a boundary risk event during the preceding process, while the second target flies within the normal declared flight path. However, the future reachability situation subdomains of the two targets overlap in the same time slice, and the edge processing device generates a track conflict event. At this time, the boundary risk event node of the first target is connected to the track conflict event node in the situation event graph, and the continuous track chain corresponding to the second target is connected to the track conflict event node. Spatial proximity edges and risk progression edges are also established.

[0224] After completing the construction of the situation event graph, the edge processing device identifies the target situation evolution path based on continuous graph edges. Continuous graph edges refer to the set of graph edges that can form continuous connections in terms of time, space, target homology, or risk progression.

[0225] The edge processing device starts from the initial event node of any target, determines the order of events along the time sequence edge, confirms whether the events belong to the same target or the same track chain along the target same origin edge, judges whether the events occur in a continuous spatial area along the spatial proximity edge, and judges whether the events constitute a risk escalation process along the risk progression edge.

[0226] If multiple event nodes can form a continuous path through the edges of the graph, then the path is determined as the target situation evolution path.

[0227] The target situation evolution path can correspond to different situation types. For example, a path that only has normal track continuity and is not accompanied by communication interruption, identity abnormality, boundary risk or track conflict can be identified as normal operation or obstructed continuity situation.

[0228] A path where communication interruption events occur consecutively after a trajectory continuation event can be identified as an abnormal link situation;

[0229] The path of consecutive occurrences of communication interruption events and identity anomaly events can be identified as identity pending verification or an identity anomaly situation;

[0230] The path of boundary risk events following an identity anomaly event can be identified as a yaw approaching situation;

[0231] The consecutive occurrence of border risk events and flight path conflict events can be identified as a conflict formation situation;

[0232] If communication interruptions, identity irregularities, boundary risks, and flight path conflicts occur consecutively within a short period of time, it can be identified as an abnormal disconnection situation.

[0233] The edge processing device ultimately outputs target situational awareness results, which specifically include:

[0234] The target situation awareness results include the target continuous track, identity consistency status, conflict trend status, boundary risk status, node observation credibility status, and situation evolution level. Among them, the target continuous track is determined by the continuous target track chain generated by S3, including the target's spatial position, altitude, speed, heading, and trajectory continuation segments at each observation time.

[0235] The identity consistency status is determined by identity anomaly events, which can be represented as identity consistency, identity pending verification, inconsistent identity trajectory, suspected identity reuse, or identity switching anomaly.

[0236] The conflict trend status is determined by track conflict events and can be represented as no conflict, potential intersection, expected conflict, or insufficient safety separation;

[0237] The boundary risk status is determined by boundary risk events and can be represented as normal, close to authorized boundary, close to no-fly boundary, close to take-off and landing protection zone, or close to communication blind zone;

[0238] The node observation credibility status is determined by the credible observation frame, the observation frame to be kerneled, the mismatched observation frame and the node observation credibility domain update in S2. It is used to indicate whether the current observation results of the target mainly come from credible nodes, the observation frame to be kerneled, or whether there are node observation anomalies.

[0239] The situation evolution level is determined based on the number of events, event types, number of risk progression edges, and risk path length in the situation event diagram, and is used to characterize the degree of risk of a target going from normal operation to abnormal disconnection.

[0240] It should be noted that the situation evolution level can be divided into normal, attention, warning, and high risk. When there are only normal trajectory continuation events in the situation event map and no risk progression edge, the normal level is output. When there are communication interruption events or identity verification events but no boundary risk events are formed, the attention level is output. When identity anomaly events and boundary risk events are connected through risk progression edge, the warning level is output. When boundary risk events and trajectory conflict events are connected continuously, or when the same target has communication interruption, identity anomaly, and boundary risk at the same time, the high risk level is output.

[0241] Through the above method, this embodiment can unify the set of reliable observation frames, continuous target track chains, trajectory continuation events, and various situational events obtained in the preceding steps into a situational event map, so that the low-altitude target situational awareness results have traceable data sources, interpretable event chains, and outputtable risk levels. This clearly shows what kind of event started the target risk, what spatial and temporal links it went through, and what kind of situational result it ultimately formed. This provides continuous and interpretable target situational awareness data for low-altitude monitoring platforms, edge control nodes, or manually monitored terminals.

[0242] Example 2: A target situational awareness system based on low-altitude intelligent network, such as Figure 2 As shown, it specifically includes:

[0243] The observation access module collects target observation data output by radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes and positioning nodes within the low-altitude intelligent network, and generates a low-altitude observation frame sequence after unified time reference correction and unified spatial coordinate system mapping.

[0244] The trusted filtering module constructs a trusted domain for node observation based on the coverage boundary, occlusion area, link delay, historical observation deviation, and target altitude range of the sensing node. It then performs trusted labeling and mismatch removal on the low-altitude observation frame sequence to form a set of trusted observation frames.

[0245] The track continuity module generates target continuity relationships based on a set of reliable observation frames, according to time reachability, spatial reachability, motion continuity, identity continuity and observation reliability domain overlap. When target observations are missing, it constructs a trajectory hypothesis tree, matches subsequent reliable observation frames to the reachable envelope, and generates continuous target track chains and trajectory continuity events.

[0246] The situation simulation module performs a reverse consistency check between the continuous target track chain and electronic identity, declared route, authorized airspace and mission time window, and infers the future reachable situation subdomain based on the target motion state, maneuverable boundary and link state, generating identity anomaly events, track conflict events, boundary risk events and communication interruption events.

[0247] The event graph module uses identity anomaly events, track conflict events, boundary risk events, communication interruption events, and track continuity events as graph nodes, and uses temporal sequence, spatial proximity, target homology, and risk progression as graph edges to construct a situation event graph, identify the target situation evolution path, and output the target situation awareness results.

[0248] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, ATA hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state ATA hard disk.

[0249] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0250] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0251] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0252] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0253] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0254] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A target situational awareness method based on low-altitude intelligent networks, characterized in that, include: S1. Collect target observation data output from radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes and positioning nodes within the low-altitude intelligent network, and generate a low-altitude observation frame sequence after unified time reference correction and unified spatial coordinate system mapping. S2. Based on the coverage boundary, obstruction area, link delay, historical observation deviation and target altitude range of the sensing node, construct the node observation credibility domain, perform credibility marking and mismatch removal on the low-altitude observation frame sequence, and form a credibility observation frame set. S3. Based on the set of trusted observation frames, generate target succession relationships according to time reachability, spatial reachability, motion continuity, identity continuity and overlapping relationship of trusted observation domains, and construct trajectory hypothesis tree when target observation is missing, match subsequent trusted observation frames to reachable envelope, and generate continuous target track chains and trajectory succession events. S4. Perform a reverse consistency check between the continuous target track chain and the electronic identity, declared route, authorized airspace and mission time window, and infer the future reachable situation subdomain based on the target motion state, maneuverable boundary and link state, and generate identity anomaly events, track conflict events, boundary risk events and communication interruption events. S5. Using identity anomaly events, track conflict events, boundary risk events, communication interruption events, and track continuity events as graph nodes, and using temporal sequence relationships, spatial proximity relationships, target homogeneity relationships, and risk progression relationships as graph edges, construct a situation event graph, identify the target situation evolution path, and output the target situation awareness results.

2. The target situational awareness method based on low-altitude intelligent network according to claim 1, characterized in that: Target observation data includes target azimuth, target distance, target altitude, target speed, target heading, electronic identification, communication link status, node observation time, node observation delay, and observation source type; Each low-altitude observation frame in the low-altitude observation frame sequence includes at least the node identifier, target candidate identifier, observation time, spatial coordinates, motion status, identity status, link status, and observation source type.

3. The target situational awareness method based on low-altitude intelligent network according to claim 1, characterized in that: The construction of the node observation trusted domain includes: The basic coverage area of ​​the sensing node is generated based on the installation coordinates, detection distance boundary, azimuth coverage angle and elevation coverage angle of the sensing node; Based on the building occlusion area, terrain occlusion area and target height range, the node basic coverage area is occluded and corrected to generate a spatially reliable boundary. Generate a reliable time window and a tolerance range based on the link delay fluctuation range and the historical observation deviation range; The node observation credibility domain is formed by the spatial credibility boundary, the temporal credibility window, and the deviation allowance range.

4. The target situational awareness method based on low-altitude intelligent network according to claim 3, characterized in that: The low-altitude observation frame sequence is subjected to reliability labeling and mismatch removal, including: Determine whether the target candidate point in the low-altitude observation frame is located within the corresponding node's observation confidence domain. If the target candidate point is located within the node's observation confidence domain and the node's observation delay is within the time confidence window, then mark it as a confidence observation frame. If a target candidate point exceeds the node observation confidence domain but can be verified by adjacent sensing nodes within an adjacent time window, it is marked as a frame to be verified. If a target candidate point exceeds the node observation confidence domain and cannot form a successor verification, it is marked as a mismatched observation frame and removed from the low-altitude observation frame sequence.

5. A target situational awareness method based on low-altitude intelligent networking according to claim 1, characterized in that: Generate target continuation relationships, including: Within adjacent time windows, candidate observation frame pairs corresponding to different sensing nodes are extracted, and the time interval, spatial displacement, velocity change, altitude change, heading change, electronic identity continuation status, and observation confidence domain overlap status between the candidate observation frame pairs are obtained respectively. When the time interval is within the continuity time range, the spatial displacement is within the target reach range, the velocity change and altitude change are within the target maneuverability range, the heading change is within the heading permission range, and the electronic identity continuation state or the observation credibility domain overlap state meets the continuity conditions, the same target continuity relationship is established between candidate observation frame pairs.

6. A target situational awareness method based on low-altitude intelligent network according to claim 5, characterized in that: Constructing a trajectory hypothesis tree when target observations are missing includes: Using the target's spatial position, altitude, speed, heading, and timestamp before disappearance as the root node, multiple hidden trajectory branches are generated based on the low-altitude channel boundary, node coverage blind zone, maximum speed variation range, maximum altitude variation range, and maximum heading variation range. For each hidden state trajectory branch, predictive trajectory nodes are generated in the order of time progression, and an reachable envelope is formed based on the reachable time window, reachable spatial range, speed allowable range, and height allowable range corresponding to the predictive trajectory nodes. When a reliable observation frame falls into the reachable envelope and satisfies the motion state continuity condition, the reliable observation frame is continued to the corresponding hidden state trajectory branch, generating a continuous target track chain and trajectory continuation event.

7. A target situational awareness method based on low-altitude intelligent networking according to claim 1, characterized in that: Consistency reverse verification includes: Match the continuous target track chain with the declared starting point, declared route, authorized airspace and mission time window corresponding to the electronic identity; When a continuous target track chain cannot be reached from the declared starting point within the mission time window, or deviates from the declared route and exceeds the authorized airspace boundary, an identity track inconsistency event is generated. When the same electronic identity corresponds to multiple spatially inaccessible consecutive target track chains, a suspected identity reuse event is generated; When a continuous target track chain remains continuous while an abnormal switch of electronic identity occurs, an identity switch event is generated; Events involving inconsistent identity trajectories, suspected identity reuse, and identity switching are grouped into identity anomaly events.

8. A target situational awareness method based on low-altitude intelligent networking according to claim 1, characterized in that: The projected future reachable subdomains include: Based on the target's current position, velocity, heading, altitude, and maneuverable boundaries, generate the target's reachable space range within consecutive time slices; Determine whether the reachable space ranges of different targets overlap within the same time slice; if overlap occurs, generate a track conflict event. Determine whether the reachable space is close to the authorized airspace boundary, no-fly zone, take-off and landing area, or communication blind spot. If the proximity condition is met, generate a boundary risk event. Communication interruption events are generated based on continuous abnormal conditions in the link status.

9. A target situational awareness method based on low-altitude intelligent networking according to claim 1, characterized in that: Constructing a situational event graph includes: The events of identity anomaly, flight path conflict, boundary risk, communication interruption, and trajectory continuity are grouped according to target identification and time sequence, and the time sequence edge, spatial proximity edge, target common origin edge, and risk progression edge between the events are established. Identify the target situation evolution path based on continuous graph edges, and output the target's continuous trajectory, identity consistency status, conflict trend status, boundary risk status, node observation reliability status, and situation evolution level.

10. A target situational awareness system based on a low-altitude intelligent network, used to implement the target situational awareness method based on a low-altitude intelligent network as described in any one of claims 1-9, characterized in that, include: The observation access module collects target observation data output by radar nodes, communication measurement nodes, video sensing nodes, radio frequency monitoring nodes and positioning nodes within the low-altitude intelligent network, and generates a low-altitude observation frame sequence after unified time reference correction and unified spatial coordinate system mapping. The trusted filtering module constructs a trusted domain for node observation based on the coverage boundary, occlusion area, link delay, historical observation deviation, and target altitude range of the sensing node. It then performs trusted labeling and mismatch removal on the low-altitude observation frame sequence to form a set of trusted observation frames. The track continuity module generates target continuity relationships based on a set of reliable observation frames, according to time reachability, spatial reachability, motion continuity, identity continuity and observation reliability domain overlap. When target observations are missing, it constructs a trajectory hypothesis tree, matches subsequent reliable observation frames to the reachable envelope, and generates continuous target track chains and trajectory continuity events. The situation simulation module performs a reverse consistency check between the continuous target track chain and electronic identity, declared route, authorized airspace and mission time window, and infers the future reachable situation subdomain based on the target motion state, maneuverable boundary and link state, generating identity anomaly events, track conflict events, boundary risk events and communication interruption events. The event graph module uses identity anomaly events, track conflict events, boundary risk events, communication interruption events, and track continuity events as graph nodes, and uses temporal sequence, spatial proximity, target homology, and risk progression as graph edges to construct a situation event graph, identify the target situation evolution path, and output the target situation awareness results.

Images