Airplane number abnormality early warning method based on call sign reverse lookup and track space-time verification

Abstract

The application discloses an aircraft number abnormality early warning method based on call sign reverse search and track space-time verification, and belongs to the technical field of civil aviation operation data processing. The method obtains the flight number, the departure airport, the actual departure business time and the to-be-verified aircraft number of a target flight, generates a standard call sign according to the flight number, and forms a first candidate aircraft set; carries out time effectiveness screening and time and space proximity merging on ADS-B observation data of the candidate aircraft, forms independent evidence points and observation sequences; determines a departure time interval according to the actual departure business time and the observation sequences, and carries out time constraint fusion; identifies aircraft number abnormality through set matching of candidate aircraft numbers and the to-be-verified aircraft number, and stops generating early warning when evidence is insufficient. The application can reduce the influence of multi-source time deviation, same-call-number multiple aircrafts and repeated observation on aircraft number abnormality identification, and improve early warning accuracy and reliability.

Description

Technical Field

[0001] This invention relates to the field of aviation anomaly early warning technology, specifically to an aircraft registration anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification. Background Technology

[0002] Civil aviation operation data platforms typically maintain flight dynamic data such as flight number, departure and arrival airports, actual departure time, and aircraft registration number, and combine this data with ADS-B data to obtain aircraft call sign, ICAO address, location, and time information. The aircraft registration number is a crucial identifier linking flight operation data and real-time trajectory data, and its accuracy directly impacts flight tracking, operational status identification, and anomaly warnings.

[0003] Current aircraft registration verification typically involves matching the flight number with the ADS-B call sign and comparing the retrieved aircraft registration number with the registration number in the flight status data. However, call signs are temporary identifiers in aviation operations, and multiple aircraft may use the same call sign within the same time frame. If only the latest record or a single aircraft is selected for verification, false alarms are likely due to incorrect candidate selection, as flight status data and ADS-B data originate from different data production links. The message generation time and ground reception time in ADS-B data have different temporal semantics than the actual takeoff time in flight status data and may be affected by reporting delays, transmission delays, and operational time deviations. Directly using these times for trajectory reachability determination can easily lead to spatiotemporal verification errors due to inconsistent time bases. ADS-B data may be repeatedly collected by multiple receiving stations, and scheduled tasks may read the same cached observation data multiple times. Simply accumulating abnormal evidence based on the number of records can easily treat repeated observations as multiple independent pieces of evidence. Meanwhile, existing methods mostly use fixed distance thresholds or single takeoff times to calculate reachability, which is difficult to adapt to deviations in takeoff time and changes in the reasonable reachability of aircraft at different flight stages. When the ADS-B aircraft number is missing or the ICAO address and aircraft number mapping data is incomplete, existing methods are unable to effectively distinguish between abnormal real aircraft numbers and insufficient evidence. Summary of the Invention

[0004] The purpose of this invention is to provide an aircraft registration anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification, thereby solving the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: 1. An aircraft registration anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification, characterized in that it includes: S1. Obtain the flight number, departure airport, actual departure time and aircraft number to be verified of the target flight. Generate a standard call sign based on the flight number, search for aircraft using the standard call sign, and form a first candidate aircraft set. S2. Obtain the ADS-B observation data corresponding to the first candidate aircraft set, using the message generation time as the observation time, and remove data that exceeds the delay threshold based on the time difference between the message generation time and the reception time to obtain the effective observation dataset. S3. Group the data with the same aircraft address and that meet the time and spatial proximity thresholds in the effective observation dataset into independent evidence points, and sort them by observation time to form an observation sequence; S4. Determine the first and second takeoff time intervals based on the actual takeoff time and observation sequence, and perform time constraint fusion based on the intersection and interval interval of the two time intervals to obtain the third takeoff time interval; when the interval between the two time intervals exceeds the preset time tolerance threshold, it is determined that the time evidence is insufficient. S5. Calculate the upper limit of the reachable distance based on the third takeoff time interval, the observation time of the independent evidence point and the aircraft motion parameters, and compare it with the observation distance from the independent evidence point to the takeoff airport to form a spatiotemporal state sequence. S6. Select the first candidate aircraft set based on the independent evidence points that continuously satisfy the spatiotemporal compatibility condition in the spatiotemporal state sequence, and obtain the second candidate aircraft set. S7. Analyze the candidate aircraft numbers of the second candidate aircraft set; if there is a candidate aircraft number that is the same as the aircraft number to be verified, the aircraft number is determined to be consistent; if all candidate aircraft numbers can be determined and none of them are different, the aircraft number is determined to be abnormal and an early warning is generated; if the evidence is insufficient, no early warning is generated.

[0006] As can be seen from the above technical solution, the present invention has the following beneficial effects: This invention distinguishes the actual takeoff time in flight dynamic data from the message generation and reception times in ADS-B data, and constructs a takeoff time interval based on time deviations. This reduces the impact of inconsistent time semantics between different data sources on trajectory verification. By acquiring all candidate aircraft based on standard call signs and filtering the candidate set using trajectory spatiotemporal compatibility, false alarms due to incorrect selection of a single candidate can be avoided when the same call sign corresponds to multiple aircraft. By merging ADS-B observation data that are temporally and spatially adjacent, forming an independent evidence sequence arranged by observation time, the accumulation of evidence duplication caused by repeated reporting from multiple receiving stations and repeated reading of timed tasks can be reduced. By combining takeoff time intervals, observation times, and aircraft motion parameters to determine the upper limit of reachable distances, and filtering trajectory-compatible aircraft based on continuous spatiotemporal states, anomaly judgment can adapt to scenarios with different flight durations and takeoff time deviations. Furthermore, by performing set matching on candidate aircraft numbers and stopping the generation of anomaly warnings when aircraft number mapping is missing or time and trajectory evidence is insufficient, the invention can distinguish between real aircraft number conflicts and insufficient evidence, thereby improving the accuracy, stability, and reliability of aircraft number anomaly identification and warnings. Attached Figure Description

[0007] Figure 1 This is a flowchart of the aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to the present invention; Figure 2 This is a timing diagram of the aircraft registration anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to the present invention; Figure 3 A bar chart comparing the accuracy, false negative rate, and false positive rate of the three methods of this invention in a call sign multiplexing adversarial scenario; Figure 4 Line graph showing the accuracy and decision coverage of the present invention and the variant without time fusion under five levels of business time deviation; Figure 5 This is a graph showing the original observation count and the number of independent evidence points after merging for 1 / 2 / 4 receiving stations in this invention; Figure 6 This is a graph showing the upper limit of the reachable distance of the present invention. Detailed Implementation

[0008] 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.

[0009] Example 1: like Figure 1 and Figure 2 As shown, this invention provides a technical solution: an aircraft registration anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification, comprising: S1. Obtain the flight number, departure airport, actual departure time and aircraft number to be verified of the target flight. Generate a standard call sign based on the flight number, search for aircraft using the standard call sign, and form a first candidate aircraft set. S2. Obtain the ADS-B observation data corresponding to the first candidate aircraft set, using the message generation time as the observation time, and remove data that exceeds the delay threshold based on the time difference between the message generation time and the reception time to obtain the effective observation dataset. S3. Group the data with the same aircraft address and that meet the time and spatial proximity thresholds in the effective observation dataset into independent evidence points, and sort them by observation time to form an observation sequence; S4. Determine the first and second takeoff time intervals based on the actual takeoff time and observation sequence, and perform time constraint fusion based on the intersection and interval interval of the two time intervals to obtain the third takeoff time interval; when the interval between the two time intervals exceeds the preset time tolerance threshold, it is determined that the time evidence is insufficient. S5. Calculate the upper limit of the reachable distance based on the third takeoff time interval, the observation time of the independent evidence point and the aircraft motion parameters, and compare it with the observation distance from the independent evidence point to the takeoff airport to form a spatiotemporal state sequence. S6. Select the first candidate aircraft set based on the independent evidence points that continuously satisfy the spatiotemporal compatibility condition in the spatiotemporal state sequence, and obtain the second candidate aircraft set. S7. Analyze the candidate aircraft numbers of the second candidate aircraft set; if there is a candidate aircraft number that is the same as the aircraft number to be verified, the aircraft number is determined to be consistent; if all candidate aircraft numbers can be determined and none of them are different, the aircraft number is determined to be abnormal and an early warning is generated; if the evidence is insufficient, no early warning is generated.

[0010] In the above scheme, the correspondence between the target flight's operational data and ADS-B surveillance data is first established. The system reads the target flight's flight number, departure airport, actual departure time, and aircraft number to be verified, and standardizes the flight number according to pre-set call sign conversion rules. These call sign conversion rules include airline code correspondence rules, flight number processing rules, character case uniformity rules, and padding character deletion rules. After processing, a standard call sign for ADS-B data retrieval is obtained. The system uses the standard call sign as the search condition to find data records in the ADS-B observation data whose flight identification field matches the standard call sign, and extracts the aircraft address from the search results. Data records with the same aircraft address are grouped into the same candidate aircraft; data records with different aircraft addresses are grouped into different candidate aircraft. All retrieved candidate aircraft are combined into a first candidate aircraft set. Through the above processing, the standard call sign is used to identify aircraft with a call sign association with the target flight, and the aircraft address is used to distinguish specific individual aircraft.

[0011] In the above scheme, ADS-B observation data is extracted for each candidate aircraft in the first candidate aircraft set. Each ADS-B observation data includes message generation time, reception time, aircraft address, and aircraft position data, and existing motion parameters such as ground speed, heading, altitude, and vertical speed recorded in the message are extracted. The system determines the message generation time as the observation time of this ADS-B data, and subsequent time sorting, status judgment, and trajectory spatiotemporal verification are all based on this observation time. For each ADS-B observation data, the message generation time and reception time are read respectively, and the message generation time is subtracted from the reception time to obtain the time length from the generation to the reception of this data. This time length is compared with a pre-set delay threshold. When the time length exceeds the delay threshold, the ADS-B observation data is discarded; when the time length does not exceed the delay threshold, the ADS-B observation data is retained. After performing the above processing on all ADS-B observation data corresponding to the first candidate aircraft set, a valid observation dataset is formed. This processing ensures that subsequent trajectory analysis does not use data exceeding the specified transmission delay range, thereby avoiding misalignment between observation time and actual aircraft movement time caused by reception delay.

[0012] After obtaining the valid observation dataset, observation data with the same aircraft address are grouped into independent evidence points. The system pre-sets temporal proximity thresholds and spatial proximity thresholds. For two observation data sets with the same aircraft address, the absolute time difference between the generation times of the two data packets is first calculated, and then the corresponding spatial distance is calculated based on the location of the two data records. When the time difference does not exceed the temporal proximity threshold and the spatial distance does not exceed the spatial proximity threshold, the two data sets are assigned to the same evidence group; if either the time difference or the spatial distance exceeds the corresponding threshold, the two data sets are assigned to different evidence groups.

[0013] For the same set of evidence, a single point of evidence is formed according to uniform rules.

[0014] Specifically, the earliest message generation time within the evidence group is used as the observation time of the independent evidence point, and the corresponding aircraft position is used as the position of the independent evidence point. Each evidence group forms only one independent evidence point. The system arranges the independent evidence points according to their observation time sequence to form the observation sequence of the corresponding candidate aircraft. Therefore, repeated observation data of the same aircraft within a specified time and space range will not be repeatedly counted as multiple independent pieces of evidence.

[0015] In the above scheme, a first takeoff time interval is established based on the actual takeoff time of the target flight. The system pre-sets forward and backward deviations for the takeoff time. Subtracting the forward deviation from the actual takeoff time yields the start time of the first takeoff time interval; adding the backward deviation forward yields the end time. The first takeoff time interval is formed from the start time to the end time. This time interval represents the allowable time range corresponding to the actual takeoff time recorded by the business system. Simultaneously, a second takeoff time interval is determined based on the observation sequence of candidate aircraft. The system pre-stores the airport association area corresponding to the takeoff airport and individually determines the positional relationship between independent evidence points in the observation sequence and the airport association area. When an independent evidence point is located within the airport association area, it is marked as airport-associated; when it is located outside the airport association area, it is marked as departing.

[0016] The system checks the observation sequence from beginning to end according to the observation time. When a change from airport-associated state to departure state is detected, the system extracts the observation time of the last independent evidence point of airport-associated state before the state transition, and simultaneously extracts the observation time of the first independent evidence point of departure state after the state transition. The previous observation time is used as the start time of the second takeoff time interval, and the next observation time is used as the end time of the second takeoff time interval, thus forming the second takeoff time interval. This time interval represents the time range within which the aircraft transitions from airport-associated state to departure state. After obtaining the first and second takeoff time intervals, time constraint fusion is performed on the two time intervals.

[0017] First, determine if there is an intersection between the first and second takeoff time intervals. Specifically, compare the start times of the first and second takeoff time intervals, and take the later start time as the intersection start time; compare the end times of the first and second takeoff time intervals, and take the earlier end time as the intersection end time. If the intersection start time is not later than the intersection end time, it is determined that the first and second takeoff time intervals have an intersection, and the time range between the intersection start time and the intersection end time is determined as the third takeoff time interval.

[0018] When the start time of the intersection is later than the end time of the intersection, it is determined that the first takeoff time interval and the second takeoff time interval do not intersect, and the interval interval between the two time intervals is further calculated. When the first takeoff time interval is earlier than the second takeoff time interval, the interval interval is calculated by subtracting the end time of the first takeoff time interval from the start time of the second takeoff time interval; when the second takeoff time interval is earlier than the first takeoff time interval, the interval interval is calculated by subtracting the end time of the second takeoff time interval from the start time of the first takeoff time interval.

[0019] The interval is compared with a preset time tolerance threshold. When the interval does not exceed the time tolerance threshold, it is considered that although there is no direct intersection between the time constraint corresponding to the actual takeoff time and the time constraint corresponding to the ADS-B trajectory departure time, their time deviation is still within the allowable range. In this case, the earlier of the two start times of the first and second takeoff time intervals is taken as the start time of the third takeoff time interval, and the later of the two end times is taken as the end time of the third takeoff time interval, thereby forming the smallest continuous time interval that can continuously cover the first takeoff time interval, the second takeoff time interval, and the interval between them, and this smallest continuous time interval is determined as the third takeoff time interval. In this embodiment, the time tolerance threshold is 90s.

[0020] When the first takeoff time interval and the second takeoff time interval do not intersect, and the interval between them exceeds the time tolerance threshold, it is determined that there is a time conflict exceeding the allowable range between the actual takeoff time evidence and the ADS-B trajectory departure time evidence. In this case, a valid third takeoff time interval is not formed, and the takeoff time evidence of the corresponding candidate aircraft is determined to be insufficient. The candidate aircraft is not used to generate an abnormal aircraft number warning.

[0021] In this embodiment, when the third takeoff time interval is valid, the system calculates the aircraft's permissible movement time based on the third takeoff time interval and the observation time of the independent evidence point. For each independent evidence point requiring post-departure spatiotemporal verification, the system reads the observation time of that independent evidence point and the start time of the third takeoff time interval. The system subtracts the start time of the third takeoff time interval from the observation time of the independent evidence point to obtain the longest movement time of the aircraft from the permissible takeoff time to the observation time of that independent evidence point. If the observation time of an independent evidence point is earlier than the start time of the third takeoff time interval, that independent evidence point is not included in the post-departure reachability calculation.

[0022] For independent evidence points that enter the reachable distance calculation, the system calculates the upper bound of the reachable distance by combining the aircraft motion parameters.

[0023] Specifically, the system pre-sets an upper limit for aircraft speed and a distance tolerance. The upper limit for aircraft speed is multiplied by the longest movement time corresponding to the independent evidence point to obtain the basic reachable distance for the aircraft to continuously move at the upper speed limit within that time range. Then, the pre-set distance tolerance is added to the basic reachable distance to obtain the upper bound of the reachable distance corresponding to the independent evidence point. The distance tolerance is used to cover the allowable error between the takeoff airport reference position, the ADS-B position, and the actual aircraft movement path and the distance calculation path.

[0024] When the system uses motion parameters that include speed variation constraints, it calculates the upper limit of the reachable distance according to pre-set speed variation rules. The system first reads the upper limit of the aircraft's initial motion speed and the upper limit of speed increase per unit time. Based on the longest motion time corresponding to the independent evidence point, it calculates the allowed speed increase within that time range. Then, based on the initial motion speed, the speed increase process, and the longest motion time, it determines the achievable distance of the aircraft within that time range. Finally, a preset distance tolerance is added to obtain the upper limit of the reachable distance. During a single aircraft registration verification process, all candidate aircraft are calculated using the same set of preset motion parameter rules to ensure consistent screening criteria. The system then calculates the observation distance between the independent evidence point and the takeoff airport. First, it reads the pre-stored reference position of the takeoff airport, then reads the aircraft position recorded by the independent evidence point, and calculates the distance between the two using a unified geographical distance calculation rule. This distance is then determined as the observation distance from the independent evidence point to the takeoff airport.

[0025] For each independent evidence point, the observed distance is compared with the corresponding upper bound of the reachable distance. When the observed distance does not exceed the upper bound, it indicates that the aircraft can reach the location corresponding to the independent evidence point under the takeoff conditions and preset motion parameters defined in the third takeoff time interval, and the independent evidence point is marked as spatiotemporally compatible. When the observed distance exceeds the upper bound of the reachable distance, it indicates that the aircraft cannot reach the location corresponding to the independent evidence point from the takeoff airport under the aforementioned time and motion parameter constraints, and the independent evidence point is marked as spatiotemporally incompatible. The system records the state results of all independent evidence points in the order of observation time, forming a spatiotemporal state sequence of the corresponding candidate aircraft.

[0026] In the above scheme, a first set of candidate aircraft is selected based on continuous spatiotemporal compatibility evidence in the spatiotemporal state sequence. The system pre-sets a threshold for the number of continuous spatiotemporal compatibility independent evidence points. For each candidate aircraft, checks are performed sequentially starting from the first independent evidence point in its spatiotemporal state sequence. When the number of consecutively occurring independent evidence points for spatiotemporal compatibility reaches the preset threshold, the candidate aircraft is retained; when the number of consecutive spatiotemporal compatibility independent evidence points has not reached the preset threshold after traversing the entire spatiotemporal state sequence, the candidate aircraft is removed from the first set of candidate aircraft. After all candidate aircraft are selected, the remaining candidate aircraft form a second set of candidate aircraft.

[0027] After obtaining the second set of candidate aircraft, the system resolves the corresponding candidate aircraft number based on the aircraft address of each candidate aircraft and queries the pre-stored correspondence between aircraft addresses and aircraft numbers. If a unique aircraft number can be found for an aircraft address at the corresponding observation time, that aircraft number is identified as a candidate aircraft number and marked as determinable; if an aircraft address cannot find an aircraft number, or if a unique aircraft number cannot be determined at the corresponding observation time, that candidate aircraft number is marked as undeterminable.

[0028] Subsequently, each candidate aircraft number in the second candidate aircraft set is compared character-wise with the aircraft number to be verified. If one candidate aircraft number in the second candidate aircraft set is identical to the aircraft number to be verified, the aircraft number to be verified is determined to match the aircraft number corresponding to the ADS-B trajectory evidence, and no aircraft number anomaly warning is generated. If each candidate aircraft number in the second candidate aircraft set is uniquely identifiable, and after comparison, no candidate aircraft number matches the aircraft number to be verified, the aircraft number to be verified is determined to be abnormal, and an aircraft number anomaly warning is generated. The warning information is associated with the target flight, and the aircraft number to be verified and the candidate aircraft numbers parsed from the second candidate aircraft set are recorded as the basis for anomaly verification.

[0029] When the third takeoff time interval cannot be formed, the effective observation data cannot form an observation sequence that meets the requirements, there are no spatiotemporally compatible independent evidence points in the spatiotemporal state sequence that reach the preset continuous number threshold, the second candidate aircraft set is empty, or there are aircraft in the second candidate aircraft set whose candidate aircraft numbers cannot be uniquely determined, the current verification result will be determined as insufficient evidence, and no abnormal aircraft number warning will be generated.

[0030] In this embodiment, the system completes aircraft number verification in the following order: standard call sign reverse lookup, ADS-B data delay verification, independent evidence point merging, service takeoff time and trajectory departure time fusion, trajectory spatial reachability verification, continuous spatiotemporal evidence screening, and candidate aircraft number comparison. The system only generates an anomaly warning when call sign association, time constraints, spatial constraints, and candidate aircraft number parsing results all meet the anomaly judgment conditions. This avoids directly judging aircraft numbers as anomalies based on missing ADS-B data, missing candidate aircraft numbers, or insufficient time evidence.

[0031] The experiment was constructed based on simulated ADS-B observation data. The simulation parameters were set according to the statistical characteristics of real ADS-B monitoring data. All data was reproducible (with a fixed random seed). Furthermore, the experimental process was directly reproduced in the laboratory environment by accessing real historical ADS-B messages and flight dynamic data, as shown in Table 1. Table 1 Simulation Elements and Basis

[0032] Evaluation indicators: Judgment accuracy: The proportion of judgment results in a sample that match the true value; Abnormal false negative rate: The proportion of real serial number abnormal samples that are judged to be consistent; False alarm rate: The proportion of normal samples with genuine serial numbers that are judged as abnormal; Judgment coverage: The proportion of samples that complete the serial number matching / abnormal judgment out of all samples; Insufficient evidence rate: The percentage of samples for which a valid judgment cannot be formed (the invention explicitly marks such results as insufficient evidence without generating a warning); Evidence compression ratio: the ratio of the original number of observations before merging to the number of independent evidence points after merging, as shown in Table 2.

[0033] Table 2 Comparison Method Settings

[0034] The values ​​of the core threshold parameters are shown in Table 3.

[0035] Table 3 Core Threshold Parameters

[0036] Experiment 1: Performance Comparison of Serial Number Anomaly Detection Experimental Design: Construct a sample of 100 departing flights (each experiment uses an independent random seed and can be repeated): Of these, 20% were abnormal aircraft registration numbers: the aircraft registration number declared in the business system was set to be the same as the registration number of a certain interfering candidate aircraft (a taxiing aircraft within the airport or a remote, unrelated aircraft), that is, the actual aircraft registration number was inconsistent with the declared aircraft registration number and the declared aircraft registration number happened to appear in the call sign associated candidate set; 8% of the samples had missing ADS-B data from the actual aircraft: the actual aircraft of the target flight had no ADS-B trajectory due to transponder malfunction / not being switched on, etc., and only interfering candidate data existed; Each flight is accompanied by two potential interference aircraft (a ground taxiing aircraft with a call sign reused + an unrelated aircraft 40km away), and the operational takeoff time differs from the actual aircraft takeoff time by ±60s.

[0037] This scenario specifically addresses the core pain point of false alarms / missed alarms caused by a single candidate selection error when the same call sign corresponds to multiple aircraft, and is the scenario where the comparison method is most likely to fail.

[0038] Table 4 Experimental Results

[0039] As shown in Table 4, Method A only relies on a single observation for call sign matching. In call sign reuse scenarios, it is very easy to mistake unrelated aircraft taxiing in the airport for target flights: 98.8% of normal flights are falsely reported as abnormal flight numbers, and 55.0% of abnormal flights are missed, almost losing the ability to distinguish between them. Although Method B obtains all call sign association candidates, it does not perform any spatiotemporal filtering and directly performs static aircraft number comparison: because the declared aircraft number happens to be the interference candidate aircraft number, 100% of abnormal flights are missed, and it is impossible to distinguish between real aircraft and call sign reused aircraft; Method C (the method of this invention) eliminates taxiing candidates within the airport (without departure state transition) and distant irrelevant candidates (observation distance exceeding the upper limit of reachable distance) one by one through S4 time interval fusion and S5 spatiotemporal reachability verification, retaining only real aircraft candidates with spatiotemporally compatible trajectories. It achieves 100.0% accuracy, 0.0% false negatives, and 0.0% false positives on all identifiable samples; 8% of the samples (mainly real aircraft ADS-B missing samples) are explicitly judged as insufficient evidence, and no warning is generated, avoiding the misjudgment of missing data as abnormal aircraft registration. Experiment 1 example: Figure 3 As shown.

[0040] Experiment 2: Robustness of Business Time Deviation Experimental Design: The invention clearly points out that the actual takeoff time in flight dynamic data has different temporal semantics from the ADS-B message generation and reception time. It is necessary to construct takeoff time intervals through time deviations to reduce their impact on trajectory verification. This experiment constructs 100 flights (50% of flight numbers are abnormal) under five different service time deviations (±30s, ±60s, ±90s, ±120s, ±180s) for each deviation, and compares the results. Method C (the method of this invention): Completely execute S4 time interval fusion (take the intersection of the first takeoff time interval ±120s and the second takeoff time interval; if there is no intersection but the interval is ≤90s, take the smallest continuous interval covering both intervals); Method C' (variant without time fusion): S4 is cancelled, and only the business take-off time is used as the take-off time. The rest of the process is the same, as shown in Table 5.

[0041] Table 5. Experimental results corresponding to the time deviation of the two methods.

[0042] Method C' has a 0% coverage rate across all deviation levels: When there is a deviation between the service takeoff time and the actual aircraft trajectory time, using a single point in the service time as the takeoff time will result in no common time range between the departure state transition interval and the takeoff time. All samples will enter the insufficient evidence branch, and the system will completely lose its early warning capability. Method C maintains a 100.0% accuracy rate across all five deviation levels, with a coverage rate of no less than 97%. S4 achieves time constraint fusion by connecting the first takeoff time interval with the second takeoff time interval (allowing a ±90s tolerance). This unifies the two different semantics of time evidence—business time and trajectory time—into a third takeoff time interval, thus enabling stable trajectory spatiotemporal verification even when there is a deviation within ±180s in business time. This result quantitatively verifies the impact of temporal semantic inconsistencies between different data sources on trajectory verification and enables anomaly detection to adapt to scenarios with varying flight durations and takeoff times. Figure 4 As shown.

[0043] Experiment 3: Merging of Independent Evidence Points under Multiple Receiving Stations' Repeated Reporting Experimental Design: In actual ADS-B networks, messages from the same aircraft at the same location are often repeatedly received and reported by multiple ground receiving stations (with inter-station time differences at the sub-second level). If the business system's timed tasks repeatedly read these messages, the cumulative duplication will be further amplified. This experiment constructs 100 departing flights (50% of which have abnormal flight numbers) under three overlapping coverage conditions: 1, 2, and 4 receiving stations. Statistics are as follows: Original ADS-B observations before consolidation (average per flight); The number of independent evidence points after S2 delay verification + S3 temporal / spatial proximity merging (average per flight); Evidence compression ratio = Number of original observations ÷ Number of independent evidence points; The accuracy and false negative rate before and after the merger are shown in Table 6.

[0044] Table 6. Experimental Results of Judgment Accuracy and Abnormal Missing Rate Before and After Merging

[0045] As the number of receiving stations increased from 1 to 4, the number of original observations increased from 468 per flight to 1872 per flight (a 3.0-fold increase), but the number of independent evidence points after S3 consolidation only increased from 413 to 475 (a 15% increase), and the evidence compression ratio improved from 1.13× to 3.94×. Merging was performed based on adjacent observation time differences ≤ 5s and great circle distances ≤ 0.6km, grouping them into the same observation cluster, with a cluster time span ≤ 10s and a spatial span ≤ 1.0km. Multi-station forwarding of the same message (sub-second time difference, near-overlapping positions) was correctly merged into an independent evidence point, while departure location messages at different times (time differences of tens of seconds, distances of several kilometers) remained independent evidence points without loss of trajectory information. Merging, while compressing 3.94× observation data, maintained a 100.0% accuracy rate and a 0.0% false negative rate, indicating that independent evidence point merging did not sacrifice trajectory verification accuracy. This result quantitatively verifies the reduction of evidence duplication caused by repeated reporting from multiple receiving stations and repeated readings by timed tasks. The experimental results are as follows. Figure 5 As shown.

[0046] Experiment 4: Sensitivity of the threshold for the number of continuous spatiotemporal compatible evidence points Experimental Design: The above method uses a preset threshold for the number of consecutive spatiotemporally compatible independent evidence points as the candidate selection criterion. To test the sensitivity of the method to the threshold value, the accuracy, false negative rate, false positive rate, and insufficient evidence rate were examined on 200 flights (50% abnormal) when the threshold for the number of consecutive compatible evidence points n∈{1,2,3,4,5} were used, as shown in Table 7.

[0047] Table 7. Experimental Results of Judgment Accuracy, False Alarm Rate, False Alarm Rate, and Insufficient Evidence Rate

[0048] Within the range of n∈[1,5], the accuracy, false negative rate, and false positive rate of this method all remain at the optimal level (100.0% / 0.0% / 0.0%), and the insufficient evidence rate never exceeds 1.5%. This indicates that the threshold for the number of continuous spatiotemporally compatible evidence points does not affect the method performance when the value is within a reasonable range, and the method parameter selection is robust and engineering-friendly. At the same time, this criterion also ensures that candidate screening does not depend on a single evidence point, avoiding misjudgments caused by single-point false detections.

[0049] Figure 6The determination principle of the spatiotemporal reachability model is as follows: taking the starting point of the third takeoff time interval as the benchmark, the longest motion time T is calculated for each independent evidence point = evidence point observation time − takeoff interval starting point, and the upper limit curve of reachable distance is plotted according to the formula: upper limit of reachable distance = upper speed limit × T + distance tolerance (250m / s and 5km are taken in this experiment); the observed distance from the evidence point to the takeoff airport is compared with this upper limit. If the observed distance falls below the upper limit, it is spatiotemporally compatible, and if it falls above the upper limit, it is spatiotemporally incompatible.

[0050] Example of compatibility evidence: After leaving the field, the motion time T = 240s, the observation distance is 32km, and the upper limit is 65km, which is considered spatiotemporally compatible; Example of incompatible evidence point: T=300s, observation distance 82km, corresponding to an upper limit of 80km, judged as spatiotemporally incompatible - this evidence point cannot be reached from the takeoff airport under all possible takeoff times limited by the third takeoff time interval.

[0051] This method employs a closed-loop process—standard call sign reverse lookup, dynamic delay verification, independent evidence point merging, takeoff time interval fusion, spatiotemporal reachability verification, continuous spatiotemporal compatibility screening, and candidate aircraft number set matching—to achieve 100.0% accuracy and 0% false alarm / missed alarm rate under real-world disturbances such as call sign reuse adversarial, operational time deviation, and multiple station duplicate reporting. Furthermore, it explicitly marks samples that cannot be reliably determined as insufficient evidence rather than false alarms. This systematic approach outperforms baseline methods such as single-point call sign matching and static aircraft number comparison.

[0052] Example 2: In this embodiment, S1 includes: Normalize the format of the target flight number and extract the airline code and flight suffix; Based on the airline code mapping relationship, the airline code is converted into an ICAO call sign prefix and combined with the flight suffix to form a standard call sign; Based on the actual takeoff time, a call sign retrieval time window is constructed according to preset forward and backward time thresholds, and ADS-B records that match the standard call sign within the call sign retrieval time window are retrieved. The aircraft addresses in the ADS-B records are extracted and deduplicated, and the aircraft corresponding to different aircraft addresses form a first candidate aircraft set.

[0053] In the above implementation, S1 is used to determine the standard call sign corresponding to the target flight number, and to retrieve the aircraft that actually uses the standard call sign from the ADS-B record within a limited time range.

[0054] After obtaining the target flight number, the flight number is first normalized. Specifically, full-width English letters and numbers are converted to their corresponding half-width characters, lowercase English letters are converted to uppercase English letters, leading and trailing spaces and spaces used for padding between characters are removed, while maintaining the original order of the numeric characters and leading zeros. The normalized flight number is used as the parsing object for the airline code and flight suffix. The extraction of the airline code is based on a pre-established airline code mapping table. The airline code mapping table records at least the airline code used for the operational flight number and the corresponding ICAO call sign prefix. During parsing, starting from the first character of the normalized flight number, prefix matching is performed with the airline codes in the airline code mapping table; if two or more prefixes match simultaneously, the longest matching result is taken as the airline code corresponding to that flight number. All characters in the normalized flight number following the airline code are determined as the flight suffix.

[0055] When the airline code in the normalized flight number fails to find a match in the airline code mapping table, the current target flight lacks the conversion basis required to determine the standard call sign. Therefore, the current verification result of the target flight is deemed insufficient evidence, and the unconverted flight number is not directly used to replace the standard call sign for ADS-B data retrieval. After obtaining the airline code, the corresponding ICAO call sign prefix in the airline code mapping table is read using this airline code as the query key, and characters are concatenated in the order of ICAO call sign prefix first, followed by flight suffix. No spaces or other connecting characters are inserted during the concatenation process, thus forming the standard call sign. Leading zeros are retained when present in the flight suffix; English letters are converted to uppercase when present in the flight suffix. After the standard call sign is formed, a call sign retrieval time window is constructed based on the actual departure time. The forward and backward time thresholds can be predetermined based on verified historical flight data. The historical flight data used to determine the time thresholds must have at least a valid actual departure time, a confirmed actual operating aircraft, and an ADS-B record for the corresponding standard call sign.

[0056] For each historical flight, the forward and backward time deviations of the message generation time of the ADS-B record corresponding to the standard call sign relative to the actual takeoff time can be calculated separately. Based on the historical time deviation distribution of the corresponding data source, forward and backward time thresholds are determined. The determined forward and backward time thresholds are then stored in association with the corresponding ADS-B data source and parameter version, remaining unchanged under the same parameter version.

[0057] When performing a target flight search, a forward time threshold is subtracted from the actual departure time, and the resulting time is used as the start time of the callsign search time window; a backward time threshold is added to the actual departure time, and the resulting time is used as the end time of the callsign search time window. Whether an ADS-B record is within the callsign search time window is determined by the message generation time of the corresponding ADS-B record. For ADS-B records within the callsign search time window, their flight identification field is first processed to standardize characters, including removing leading and trailing spaces, filling spaces, and converting all English letters to uppercase. Then, the standardized flight identification field is compared character by character with the standard callsign. When the number of characters is the same and all corresponding characters are identical, the ADS-B record is determined to be consistent with the standard callsign; if any character is different or the number of characters is different, the ADS-B record is not used as the basis for extracting candidate aircraft.

[0058] For ADS-B records with identical call signs, the corresponding aircraft addresses are read. Aircraft addresses are extracted according to the complete address field specified by the data source, and the case of English letters is standardized before comparison, with leading zeros retained. The normalized complete aircraft address is used as the deduplication key. A candidate aircraft is created for the first occurrence of an aircraft address; if the aircraft address of a subsequent ADS-B record is exactly the same as an already created aircraft address, it is only included in the corresponding existing candidate aircraft, without being added again; if it is different from an existing aircraft address, a new candidate aircraft is created. After processing all ADS-B records within the call sign retrieval time window, different aircraft addresses correspond to different candidate aircraft, which together constitute the first candidate aircraft set. When the same standard call sign corresponds to only one aircraft address within the call sign retrieval time window, the first candidate aircraft set includes one candidate aircraft; when the same standard call sign corresponds to two or more different aircraft addresses, the aircraft corresponding to each aircraft address are all included in the first candidate aircraft set.

[0059] This step does not directly determine the actual aircraft corresponding to the target flight based on the number of ADS-B records, the order in which the records appear, or any single aircraft address. Instead, it retains all candidate aircraft that meet the standard call sign and retrieval time conditions, and continues to perform delay data filtering, independent evidence point merging, takeoff time constraints, spatiotemporal reachability verification, and continuous spatiotemporal compatibility evidence filtering through S2 to S6.

[0060] Example 3: In this embodiment, S2 includes: Obtain the corresponding ADS-B observation data based on the aircraft addresses in the first candidate aircraft set; Extract the message generation time and reception time respectively, calculate the reception delay value between the reception time and the message generation time, and form a reception delay sequence according to the message generation time; Calculate the median and median absolute deviation of the received delay sequence within a preset sliding time window; The dynamic delay threshold is determined based on the median, median absolute deviation, and preset dynamic delay threshold coefficient. ADS-B observation data with reception delay values ​​exceeding the dynamic delay threshold are discarded, and the remaining ADS-B observation data are determined as valid observation datasets.

[0061] In the above implementation, S2 is executed for each candidate aircraft in the first candidate aircraft set obtained by S1.

[0062] Using the aircraft address corresponding to the candidate aircraft as the search criteria, observation records with completely identical aircraft addresses are extracted from ADS-B data. Each ADS-B observation record includes at least the message generation time, reception time, aircraft position, and motion parameters already present in the original message.

[0063] ADS-B observation data corresponding to different aircraft addresses are formed into independent data groups, and the corresponding reception delays are calculated separately to avoid mixing of delay data corresponding to different aircraft or different data reception links.

[0064] For each ADS-B observation data, the message generation time and reception time are uniformly converted to the same time base and the same timing unit, and the reception delay value of the corresponding ADS-B observation data is obtained by subtracting the message generation time from the reception time.

[0065] After calculating the reception delay of all ADS-B observation data corresponding to the same aircraft address, the data are arranged from earliest to latest according to the message generation time; if there are messages with the same generation time, they are arranged from earliest to latest according to the reception time, thus forming a reception delay sequence.

[0066] In this embodiment, the length of the sliding time window is set to 5 minutes. For any ADS-B observation data to be determined in the received delay sequence, the message generation time of the ADS-B observation data is taken as the end time of the current sliding time window, and the start time of the current sliding time window is determined by going back 5 minutes.

[0067] The ADS-B observation data whose message generation time is within the current sliding time window and whose aircraft address is the same as the ADS-B observation data to be judged constitute the delay calculation sample corresponding to the current sliding time window.

[0068] Within the current sliding time window, the received delay values ​​are arranged in ascending order. When the number of received delay values ​​is odd, the middle value after arrangement is taken as the median M; when the number of received delay values ​​is even, the average of the two middle values ​​after arrangement is taken as the median M. After obtaining the median M, the median M is subtracted from each received delay value within the current sliding time window, and the absolute value of the difference is taken to form an absolute deviation sequence. The absolute deviation sequence is arranged in ascending order, and the median absolute deviation MAD is determined according to the same calculation rules as the median M.

[0069] The dynamic delay threshold Dth is determined based on the median M, median absolute deviation MAD, and dynamic delay threshold coefficient α within the current sliding time window, and satisfies: Dth = M + α × 1.4826 × MAD.

[0070] Where 1.4826 is the coefficient used for scaling the median absolute deviation, and α is the delay dynamic threshold coefficient. In this embodiment, α is taken as 2.5.

[0071] Therefore, the dynamic delay threshold in this embodiment satisfies: Dth = M + 2.5 × 1.4826 × MAD.

[0072] For example, if the received delay values ​​within a 5-minute sliding time window are 1s, 1s, 2s, 2s, and 3s respectively, then the median M is 2s. The absolute deviations of each received delay value relative to the median M are 1s, 1s, 0s, 0s, and 1s respectively, and their corresponding median absolute deviation (MAD) is 1s. At this point: Dth=2+2.5×1.4826×1=5.7065s.

[0073] This yields the dynamic delay threshold corresponding to the current ADS-B observation data to be determined.

[0074] When the MAD in the current sliding time window is 0, the minimum deviation can be set according to the minimum timing accuracy of the corresponding data source timestamp. The minimum deviation is used to replace 0 in the dynamic delay threshold calculation to avoid the dynamic delay threshold lacking the allowable deviation corresponding to the timestamp accuracy when all received delay values ​​are the same.

[0075] For each ADS-B observation data to be judged, its reception delay value is compared with the corresponding dynamic delay threshold Dth. When the reception delay value is greater than the dynamic delay threshold Dth, the ADS-B observation data is determined to be data that exceeds the current allowable range of reception delay and is discarded; when the reception delay value is less than or equal to the dynamic delay threshold Dth, the ADS-B observation data is retained.

[0076] After performing delayed screening on all candidate aircraft in the first candidate aircraft set in the manner described above, the ADS-B observation data that were not eliminated were determined as the valid observation dataset.

[0077] The valid observation data continues to retain the corresponding aircraft address, message generation time, reception time, aircraft position, and existing motion parameters, and in S3, the message generation time continues to be used as the observation time.

[0078] Example 4: In this embodiment, S3 includes: The valid observation dataset is grouped according to aircraft address and sorted by observation time to form an initial observation sequence; Calculate the observation time difference and great circle distance between adjacent ADS-B observation data in sequence, and classify ADS-B observation data that simultaneously meet the time proximity threshold and spatial proximity threshold into the same observation cluster. Before adding the current ADS-B observation data to the observation cluster, it is further determined whether the maximum observation time span and the maximum spatial span after addition do not exceed the preset cluster time threshold and cluster space threshold, respectively. When either span exceeds the corresponding threshold, a new observation cluster is established. The ADS-B observation data with the earliest message generation time was selected from each observation cluster as an independent evidence point; An observation sequence is formed by arranging the observation times of each independent evidence point.

[0079] In the above implementation, S3 performs a merging process on the effective observation dataset obtained in S2 to reduce duplicate ADS-B observation data generated by the same aircraft in adjacent time and adjacent spatial ranges.

[0080] First, the aircraft addresses are read from each valid observation data. Data with identical aircraft addresses are grouped into the same data group, while data with different aircraft addresses are processed separately.

[0081] For data groups corresponding to the same aircraft address, the message generation time is used as the observation time, and the data is arranged from earliest to latest. If there are identical message generation times, they are then arranged from earliest to latest according to the reception time, thus forming the initial observation sequence for the corresponding candidate aircraft. Observation clusters are established sequentially according to the arrangement of the initial observation sequence. The first ADS-B observation data in the initial observation sequence is used as the first data in the first observation cluster, and subsequent ADS-B observation data are read sequentially. For two adjacent ADS-B observation data, the observation time of the subsequent ADS-B observation data is subtracted from the observation time of the preceding ADS-B observation data to obtain the adjacent observation time difference. The spatial distance between adjacent ADS-B observation data is calculated using great circle distance. Specifically, the longitude and latitude corresponding to the two ADS-B observation data are read respectively, and the longitude and latitude are converted from the degree system to the radian system. The geocentric angle between the two locations is calculated based on the difference in latitude and longitude and the corresponding latitude value. Then, the geocentric angle is multiplied by the average radius of the Earth, 6,371,000m, to obtain the great circle distance between the corresponding locations of the two ADS-B observation data.

[0082] When the time difference between adjacent observations does not exceed the temporal proximity threshold, and the great circle distance between adjacent locations does not exceed the spatial proximity threshold, the current ADS-B observation data is included in the overall span judgment of the observation cluster.

[0083] In this embodiment, the temporal proximity threshold is 5s and the spatial proximity threshold is 0.6km.

[0084] Before adding current ADS-B observation data to an observation cluster, calculate the maximum observation time span after addition. Read the earliest and latest observation times in the proposed observation cluster, and subtract the earliest observation time from the latest observation time. The resulting time length is the maximum observation time span.

[0085] Simultaneously, the great circle distance between any two ADS-B observation data in the proposed observation cluster is calculated, and the maximum value among all location distances is taken as the maximum spatial span of the observation cluster.

[0086] In this embodiment, the cluster time threshold is 10s and the cluster space threshold is 1.0km.

[0087] When the maximum observation time span formed after the current ADS-B observation data is added to the observation cluster does not exceed 10 seconds and the maximum spatial span does not exceed 1.0 km, the current ADS-B observation data will be added to the existing observation cluster.

[0088] When the maximum observation time span exceeds 10 seconds or the maximum spatial span exceeds 1.0 km after being added, the current ADS-B observation data will not be added to the existing observation cluster. Instead, a new observation cluster will be established with the current ADS-B observation data as the first data.

[0089] When the observation time difference between adjacent ADS-B observation data exceeds 5 seconds or the great circle distance exceeds 0.6 km, a new observation cluster is established using the current ADS-B observation data.

[0090] The above-described method is used to process all valid observation data corresponding to the current aircraft address sequentially until all corresponding observation clusters are formed. By setting constraints on the overall time span and overall spatial span of the observation clusters, it is possible to avoid multiple adjacent ADS-B observation data points, although satisfying the time and spatial proximity conditions point by point, being merged into an observation cluster with an excessively large span due to continuous transmission during the continuous movement of the aircraft. After each observation cluster is formed, according to the independent evidence point formation method determined in Example 1, the ADS-B observation data with the earliest message generation time is selected from each observation cluster as an independent evidence point. The message generation time of the selected ADS-B observation data is used as the observation time of the corresponding independent evidence point, the aircraft position corresponding to the ADS-B observation data is used as the position of the independent evidence point, and its original motion parameters continue to be used as the motion parameters corresponding to the independent evidence point.

[0091] When two or more ADS-B observation data sets with the same earliest message generation time exist within the same observation cluster, one of the ADS-B observation data sets can be further determined as an independent evidence point according to its reception time from earliest to latest. Each observation cluster forms only one independent evidence point. After determining all independent evidence points corresponding to the same aircraft address, they are arranged according to their observation time from earliest to latest to form the observation sequence of the corresponding candidate aircraft.

[0092] Example 5: In this embodiment, S4 includes: The first takeoff time interval is formed based on the actual takeoff time and the preset forward deviation boundary and backward deviation boundary; Identify the state transition segment from airport-associated state to departure state based on the positional relationship between independent evidence points in the observation sequence and the associated area of ​​the takeoff airport; The second takeoff time interval is formed based on the observation time of two independent evidence points before and after the state transition section. When the first takeoff time interval and the second takeoff time interval intersect, the intersection of the two is determined as the third takeoff time interval; When the first takeoff time interval and the second takeoff time interval do not overlap but the interval interval does not exceed the preset time tolerance threshold, the smallest continuous time interval covering the first takeoff time interval, the second takeoff time interval, and the interval interval between them will be determined as the third takeoff time interval. When the interval exceeds a preset time tolerance threshold, it is determined that the time evidence is insufficient.

[0093] In the above implementation, S4 is used to fuse the actual takeoff time recorded by the business system with the aircraft departure time reflected by the ADS-B observation sequence, and obtain the third takeoff time interval for the spatiotemporal reachability verification of S5.

[0094] First, a first takeoff time interval is formed based on the actual takeoff time. Forward and backward deviation boundaries for the takeoff time are preset. In this embodiment, both the forward and backward deviation boundaries are set to 120 seconds.

[0095] The start time of the first takeoff time interval is obtained by subtracting 120 seconds from the actual takeoff time; the end time of the first takeoff time interval is obtained by adding 120 seconds from the actual takeoff time. The first takeoff time interval is defined by the start and end times.

[0096] The forward deviation boundary and the backward deviation boundary can be determined based on the verified historical flights of the corresponding data source, and after determination, they are associated with and saved with the corresponding data source and parameter version.

[0097] The second takeoff time interval was then determined using the observation sequence formed by S3.

[0098] The system predetermines the airport association area corresponding to the departure airport. In this embodiment, the airport association area can be determined with the departure airport reference location as the center and a radius of 2.5km.

[0099] For each independent evidence point in the observation sequence, its position is read, and the distance from the independent evidence point to the departure airport reference position is calculated using the same great circle distance calculation method as S3. When the distance from the independent evidence point to the departure airport reference position does not exceed the radius of the airport association area, the independent evidence point is determined to be in an airport association state; when the distance exceeds the radius of the airport association area, the independent evidence point is determined to be in a departure state. The system checks the observation sequence sequentially from earliest to latest according to the observation time. When two adjacent independent evidence points are in an airport association state and the latter is in a departure state, the adjacent segment is determined to be a state transition segment from airport association state to departure state.

[0100] Based on the order of observation time, the first occurrence of the airport-related state transitioning to the departure state can be taken as the takeoff state transition segment for the current candidate aircraft. After determining the state transition segment, the observation time of the independent evidence point preceding the state transition segment is taken as the start time of the second takeoff time interval, and the observation time of the independent evidence point following the state transition segment is taken as the end time of the second takeoff time interval, thus forming the second takeoff time interval.

[0101] When there is no state transition segment in the observation sequence that transitions from the airport-associated state to the departure state, a second takeoff time interval cannot be formed, and the time evidence for the corresponding candidate aircraft is determined to be insufficient.

[0102] Optionally, without changing the basic determination relationship of the state transition segment based on the positional relationship between the airport's associated area and the surrounding area, the positional and temporal changes between adjacent independent evidence points can be calculated, and the movement speed of adjacent segments can be obtained accordingly for auxiliary verification of the determined state transition segments. After the first and second takeoff time intervals are formed, it is first determined whether there is an intersection between them. The start times of the two time intervals are compared, and the later start time is taken as the intersection start time; the end times of the two time intervals are compared, and the earlier end time is taken as the intersection end time. When the intersection start time is not later than the intersection end time, it is determined that the two time intervals have an intersection, and the time range between the intersection start time and the intersection end time is determined as the third takeoff time interval.

[0103] When the start time of the intersection is later than the end time of the intersection, it is determined that the first takeoff time interval and the second takeoff time interval do not intersect, and the interval interval between the two time intervals is further calculated. When the first takeoff time interval is earlier than the second takeoff time interval, the interval interval is obtained by subtracting the end time of the first takeoff time interval from the start time of the second takeoff time interval; when the second takeoff time interval is earlier than the first takeoff time interval, the interval interval is obtained by subtracting the end time of the second takeoff time interval from the start time of the first takeoff time interval.

[0104] In this embodiment, the time tolerance threshold is 90s.

[0105] When the interval does not exceed 90 seconds, the deviation between the operational takeoff time constraint and the ADS-B trajectory departure time constraint is considered to be within the allowable range. In this case, the earlier start time of the two time intervals is taken as the start time of the third takeoff time interval, and the later end time of the two time intervals is taken as the end time of the third takeoff time interval, thus forming the minimum continuous time interval covering the first takeoff time interval, the second takeoff time interval, and the interval between them, and the minimum continuous time interval is determined as the third takeoff time interval.

[0106] When the interval exceeds 90 seconds, it is determined that there is a time conflict exceeding the allowable range between the actual takeoff time evidence and the ADS-B trajectory departure time evidence. The time evidence of the corresponding candidate aircraft is determined to be insufficient, and no effective third takeoff time interval is formed for S5 spatiotemporal reachability judgment.

[0107] Therefore, S4 completes the time constraint fusion by the intersection of the first takeoff time interval and the interval interval, and unifies the takeoff time information corresponding to different data sources into the third takeoff time interval.

[0108] Example 6: In this embodiment, S5 includes: Obtain the start time of the third takeoff time interval and subtract it from the observation time of each independent evidence point to determine the maximum possible flight duration corresponding to each independent evidence point; Calculate the upper bound of the reachable distance for the corresponding independent evidence point based on the maximum possible flight time, the upper limit of the aircraft's speed, and the distance tolerance. The observation distance from the independent evidence point to the reference position of the takeoff airport is calculated using the great circle distance algorithm; The observation distance is compared with the corresponding upper bound of the reachable distance to obtain the spatiotemporal compatible state or spatiotemporal incompatible state of the corresponding independent evidence point; Arrange the spatiotemporal states according to the observation time of independent evidence points to form a spatiotemporal state sequence.

[0109] In the above implementation, S5 uses the third takeoff time interval obtained in S4 as the time constraint for the possible takeoff time of the aircraft, and performs spatiotemporal reachability judgment on each independent evidence point of the candidate aircraft one by one.

[0110] For any independent evidence point, the start time of the third takeoff time interval and the observation time of that independent evidence point are read. The start time of the third takeoff time interval is subtracted from the observation time of the independent evidence point to obtain the longest movement time between the earliest allowed takeoff time of the aircraft within the third takeoff time interval and the observation time of that independent evidence point. This longest movement time is determined as the maximum possible flight duration Tmax corresponding to that independent evidence point.

[0111] When the observation time of an independent evidence point is earlier than the start time of the third takeoff time interval, the maximum possible flight duration is less than 0, and the independent evidence point will not be used as evidence of spatiotemporal accessibility after departure to participate in the subsequent continuous spatiotemporal state screening of S6.

[0112] In this embodiment, the system pre-sets a uniform upper limit for aircraft speed. To ensure that different candidate aircraft in the same registration verification process use the same trajectory spatiotemporal screening criteria, the upper limit for aircraft speed Vmax in this embodiment is set to 250 m / s.

[0113] Meanwhile, the system pre-sets a distance tolerance Dc. In this embodiment, the distance tolerance Dc is set to 5 km.

[0114] The distance tolerance is used to cover the spatial deviation between the reference position of the takeoff airport and the actual takeoff position of the aircraft, the ADS-B position observation error, and the allowable spatial deviation between the actual movement path of the aircraft and the simplified distance calculation path.

[0115] The distance tolerance and the airport association area radius used in S4 to determine the airport association status are set separately, and they have different technical uses. The airport association area radius is used to determine whether an independent evidence point is within the airport association range, while the distance tolerance is used to calculate the upper limit of the reachable distance of the aircraft.

[0116] For any independent point of evidence, multiply the upper limit of the aircraft's speed Vmax by the maximum possible flight time Tmax to obtain the basic reachable distance, and add a distance tolerance Dc to the basic reachable distance to obtain the upper limit of the reachable distance Dmax: Dmax = Vmax × Tmax + Dc.

[0117] In this embodiment: Vmax = 250 m / s; Dc=5km.

[0118] When all distances are uniformly calculated using meters as the unit of measurement, the calculation relationship can be expressed as: Dmax = 250 × Tmax + 5000.

[0119] As the maximum possible flight time Tmax increases, the upper bound of the reachable distance Dmax increases monotonically. Therefore, the reachable distance calculation satisfies the condition that the reachable distance does not decrease monotonically with the allowable flight time.

[0120] The observation distance (Dobs) from the independent evidence point to the takeoff airport was then calculated.

[0121] Specifically, the longitude and latitude corresponding to the independent evidence point and the longitude and latitude corresponding to the reference position of the takeoff airport are read, the longitude and latitude are converted into radians, the corresponding geocentric angle is determined according to the latitude difference, longitude difference and latitude cosine value between the two positions, and the geocentric angle is multiplied by the average radius of the earth, 6371000m, to obtain the observation distance Dobs from the independent evidence point to the takeoff airport.

[0122] Compare the observation distance Dobs with the corresponding upper bound of the reachable distance Dmax.

[0123] When Dobs≤Dmax, it indicates that the aircraft can reach the position corresponding to the independent evidence point under the conditions of the earliest allowed takeoff time within the third takeoff time interval and the preset upper limit of motion speed, and the independent evidence point is determined to be in a spatiotemporally compatible state.

[0124] When Dobs > Dmax, it indicates that even if the aircraft takes off at the earliest allowed takeoff time within the third takeoff time interval and continues to move at the preset upper limit of motion speed, it still cannot reach the corresponding position at the observation time of the independent evidence point. Therefore, the independent evidence point is determined to be in a spatiotemporally incompatible state.

[0125] Optionally, the distance ratio R between the observed distance Dobs and the upper bound of the reachable distance Dmax can also be calculated: R = Dobs / Dmax. When the reachability threshold is set to 1, R ≤ 1 corresponds to a spatiotemporal compatible state, and R > 1 corresponds to a spatiotemporal incompatible state. Therefore, using the distance ratio to make a judgment has the same result as directly comparing Dobs and Dmax.

[0126] Since the start time of the third takeoff time interval corresponds to the maximum possible flight time within the range defined by the third takeoff time interval, and the reachable distance monotonically does not decrease with the flight time, when the observation distance of the independent evidence point still exceeds the upper limit of the reachable distance calculated according to the maximum possible flight time, it can be determined that the independent evidence point also cannot meet the spatiotemporal reachability conditions under other possible takeoff times defined by the third takeoff time interval.

[0127] After completing the spatiotemporal reachability assessment of each independent evidence point for the same candidate aircraft, the corresponding spatiotemporal compatible and spatiotemporal incompatible states are arranged from earliest to latest according to the observation time of the independent evidence points, forming the spatiotemporal state sequence of the candidate aircraft.

[0128] Alternatively, in other embodiments, the upper limit of reachable distance can be calculated using motion rules that include the starting speed and the upper limit of speed change, provided that reliable aircraft motion parameters are available. However, in the same aircraft registration verification process, all candidate aircraft should use the same set of preset motion parameter rules to ensure that the screening criteria among candidate aircraft are consistent.

[0129] Example 7: In this embodiment, S6 includes: Read the spatiotemporal state sequence corresponding to each first candidate aircraft in the order of observation time; Count the independent evidence points of consecutively occurring spatiotemporally compatible states; When the number of independent evidence points of continuous spatiotemporal compatibility reaches the preset continuous number threshold, the corresponding first candidate aircraft will be retained. When the number of independent evidence points of continuous spatiotemporal compatible states does not reach the preset continuous number threshold after traversing the entire spatiotemporal state sequence, the corresponding first candidate aircraft is eliminated; the retained first candidate aircraft are collected to form a second candidate aircraft set.

[0130] In the above implementation, S6 uses the spatiotemporal state sequence formed in S5 as input to perform continuous spatiotemporal evidence screening for each candidate aircraft in the first candidate aircraft set. For any candidate aircraft, starting from the independent evidence point with the earliest observation time in the corresponding spatiotemporal state sequence, the spatiotemporal state corresponding to each independent evidence point is read sequentially from earliest to latest according to the observation time.

[0131] Set the continuous compatibility count value N, and initialize N to 0.

[0132] When the current independent evidence point corresponds to a spatiotemporal compatible state, the continuous compatibility count value N is increased by 1.

[0133] When the current independent evidence point corresponds to a spatiotemporally incompatible state, the continuous compatibility count value N is reset to 0.

[0134] Then continue reading the next independent evidence point and update the consecutive compatibility count value N in the same way.

[0135] The system pre-sets a threshold Nth for the number of continuous spatiotemporally compatible independent evidence points. In this embodiment: Nth=3.

[0136] When the continuous compatibility count N reaches 3, it indicates that the current candidate aircraft has formed at least 3 independent evidence points that are consecutively arranged according to observation time and all meet the spatiotemporal reachability conditions, thus forming continuous spatiotemporal compatible trajectory evidence that meets the requirements. The candidate aircraft is then retained and included in the second candidate aircraft set. For candidate aircraft that have reached the aforementioned continuous count threshold, the continuous compatibility determination of the current spatiotemporal state sequence of the corresponding candidate aircraft can be terminated, or the remaining spatiotemporal states can be traversed to save the complete trajectory state record. Neither result changes the determination that the candidate aircraft has met the continuous spatiotemporal compatibility conditions.

[0137] When the continuous compatibility count N does not reach 3 after traversing the entire spatiotemporal state sequence, it indicates that there is no spatiotemporally compatible independent evidence point sequence in the current observation trajectory of the candidate aircraft that meets the preset continuous number requirement, and the candidate aircraft is removed from the first candidate aircraft set.

[0138] For example, the spatiotemporal state sequence corresponding to a candidate aircraft is as follows: The sequence is: Spatiotemporal compatible, Spatiotemporal compatible, Spatiotemporal incompatible, Spatiotemporal compatible, Spatiotemporal compatible, Spatiotemporal compatible. The first two independent evidence points of spatiotemporal compatibility cause the continuous compatibility count to become 1 and 2 respectively; the third independent evidence point is a spatiotemporal incompatible state, so the continuous compatibility count is reset to 0; the subsequent three independent evidence points are all spatiotemporally compatible states, causing the continuous compatibility count to reach 1, 2, and 3 respectively. Therefore, this candidate aircraft meets the threshold requirement for the number of independent evidence points of continuous spatiotemporal compatibility and is retained. After processing each candidate aircraft in the first candidate aircraft set in the above manner, all candidate aircraft that meet the continuous spatiotemporal compatibility condition are gathered to form the second candidate aircraft set.

[0139] The threshold for the number of continuous spatiotemporally compatible independent evidence points can be preset according to the ADS-B observation update frequency, data integrity, and application scenario requirements. In this embodiment, 3 is used as the baseline threshold, and the experimental part further evaluates the parameter sensitivity of the continuous number threshold range of 1 to 5.

[0140] By using continuous spatiotemporally compatible independent evidence points to screen candidate aircraft, we can avoid directly retaining candidate aircraft based on a single spatiotemporally compatible observation point. This requires candidate aircraft to meet takeoff time constraints and spatial reachability conditions within a continuous trajectory range.

[0141] Example 8: In this embodiment, S7 includes: Based on the aircraft addresses of each candidate aircraft in the second candidate aircraft set, query the mapping relationship between aircraft addresses and aircraft numbers, and resolve the corresponding candidate aircraft numbers; The candidate machine numbers that can be uniquely identified are formed into a set of candidate machine numbers; Match each candidate number in the candidate number set with the number to be verified. If any candidate number is the same as the number to be verified, determine that the number to be verified is the same and terminate the current number abnormality judgment. If there is no candidate aircraft number that is the same as the aircraft number to be verified, determine whether the candidate aircraft numbers corresponding to each candidate aircraft in the second candidate aircraft set can be uniquely determined. When there are candidate aircraft whose candidate aircraft numbers cannot be uniquely determined, the current verification result will be determined as insufficient evidence, and no abnormal warning will be generated; When all candidate aircraft numbers of the second candidate aircraft can be uniquely determined and are all different from the aircraft number to be verified, the aircraft number to be verified is determined to be abnormal and an abnormal warning is generated.

[0142] Furthermore, the generation of the anomaly warning includes: Obtain snapshots of target flight business fields, sequences of independent evidence points, and version information of aircraft address and registration number mapping data used for anomaly detection to form an anomaly evidence snapshot; form a unique evidence identifier based on the target flight identifier, warning type, observation time window, and mapping data version; The existing anomaly warning records are queried using the unique identifier of the evidence. When a corresponding anomaly warning record exists, the newly added independent evidence point is merged with the independent evidence point in the existing anomaly warning record to remove duplicates. When no corresponding anomaly warning record exists, a new anomaly warning record is created and associated with the anomaly evidence snapshot. The flight business fields, independent evidence points, and mapping data versions actually used when judging anomalies are saved so that the anomaly warning can be reproduced according to the data state at the time of generation.

[0143] In the above implementation, S7 is used to parse the candidate aircraft number corresponding to the candidate aircraft based on the second candidate aircraft set having already undergone call sign association, time constraints, trajectory spatiotemporal constraints and continuous spatiotemporal compatibility evidence screening, and to complete the final determination of whether the aircraft number to be verified is consistent, abnormal, or insufficient.

[0144] For each candidate aircraft in the second candidate aircraft set, its aircraft address is first read, and then the pre-stored aircraft address-to-registration mapping data is queried using that aircraft address. The aircraft address-to-registration mapping data may include at least the aircraft address, registration number, record effective time, record expiration time, and data version.

[0145] When querying, the actual departure time of the target flight can be used as the time to determine the validity of the mapping relationship.

[0146] If the effective time of the corresponding mapping record is no later than the actual departure time of the target flight, and the expiration time is later than the actual departure time of the target flight, the mapping record is determined to be valid at the corresponding time. For mapping records without an expiration time, they can be treated as continuously valid records. If only one valid mapping record is found for the same aircraft address under the same actual departure time, and a unique aircraft number can be obtained from the valid mapping record, then that aircraft number is determined as the candidate aircraft number for the corresponding candidate aircraft.

[0147] If no valid aircraft address and aircraft number mapping record is found, the candidate aircraft number of the candidate aircraft is marked as undetermined.

[0148] When more than two valid mapping records are found under the same actual takeoff time, and the corresponding aircraft numbers are inconsistent, instead of arbitrarily selecting one from the multiple aircraft numbers as a candidate aircraft number, the candidate aircraft number of the candidate aircraft is marked as undetermined.

[0149] Optionally, when the corresponding ADS-B data source can directly provide aircraft number information that has been confirmed and can uniquely correspond to the current aircraft address, the aircraft number information can be used as a supplementary source of aircraft address and aircraft number mapping data, but this does not change the judgment requirement that the candidate aircraft number must be uniquely determined.

[0150] For candidate machine numbers that can be identified, remove leading and trailing spaces, convert all English letters to uppercase, and maintain the original order of numeric characters and valid connectors.

[0151] After resolving the aircraft numbers of all candidate aircraft in the second candidate aircraft set, deduplication is performed on the candidate aircraft numbers that can be determined according to the unified complete character sequence.

[0152] When two candidate aircraft numbers have the same number of characters and the characters at each corresponding position are identical, they are considered to be the same candidate aircraft number, and one of them is retained in the candidate aircraft number set. At the same time, all aircraft addresses corresponding to that candidate aircraft number are saved.

[0153] Different candidate machine numbers are retained separately, thus forming a set of candidate machine numbers.

[0154] The machine number to be verified adopts the same character standard processing method as the candidate machine number, removing leading and trailing spaces, converting English letters to uppercase, and maintaining the order of numeric characters and valid connectors.

[0155] Then, each candidate machine number in the candidate machine number set is compared with the machine number to be verified one by one.

[0156] When the number of characters in the candidate number is the same as the number to be verified, and the characters in each corresponding position are identical, the candidate number is determined to be the same as the number to be verified.

[0157] If at least one candidate aircraft number in the candidate aircraft number set is the same as the aircraft number to be verified, it is determined that the aircraft number to be verified is consistent with the aircraft number corresponding to the trajectory spatiotemporal evidence, and the current aircraft number anomaly judgment is terminated, and no aircraft number anomaly warning is generated.

[0158] If there is no candidate aircraft number in the candidate aircraft number set that is the same as the aircraft number to be verified, the resolution status of the candidate aircraft number corresponding to each candidate aircraft in the second candidate aircraft set is further checked.

[0159] If a candidate aircraft number cannot be uniquely determined, the current verification result will be determined as insufficient evidence. Missing candidate aircraft numbers or conflicting mapping relationships will not be directly identified as abnormal aircraft numbers, and no abnormal aircraft number warning will be generated.

[0160] When all candidate aircraft in the second candidate aircraft set can obtain a unique candidate aircraft number, and all candidate aircraft numbers are different from the aircraft number to be verified, the aircraft number to be verified is determined to be abnormal, and the abnormality warning generation process begins. When generating an abnormality warning, an abnormality evidence snapshot is first created. The target flight business field snapshot uses the business data read at the start of this aircraft number verification, including at least the target flight identifier, flight number, departure airport, actual departure time, and aircraft number to be verified. After the abnormality evidence snapshot is created, the business fields remain unchanged. Subsequent modifications to the fields in the business system will not overwrite the already saved abnormality evidence snapshot.

[0161] The sequence of independent evidence points uses the independent evidence points corresponding to the second candidate aircraft involved in this anomaly assessment. For each independent evidence point, the aircraft address, observation time, longitude, latitude, corresponding reception delay, and spatiotemporal state determined by S5 can be saved, and the arrangement relationship is saved from earliest to latest according to the observation time.

[0162] This ensures that the trajectory evidence saved in the anomaly warning is consistent with the data actually used in the judgment from S3 to S6. Aircraft address and serial number mapping data is saved according to the data version actually used in this candidate serial number resolution. A corresponding fixed version identifier is generated each time mapping data is released, maintaining the correspondence between aircraft address, serial number, effective time, and expiration time within the same version. In addition to saving the mapping data version, the anomaly evidence snapshot can also save the mapping record actually used in this candidate serial number resolution, so that candidate serial numbers can be obtained again according to the same data state, thus forming a unique evidence identifier. The target flight identifier, the warning type corresponding to the serial number anomaly, the observation time window involved in this anomaly judgment, and the mapping data version are obtained and combined according to a preset field order to form a unique evidence identifier.

[0163] Specifically, the earliest observation time among all independent evidence points involved in this anomaly assessment can be used as the start time of the observation time window, and the latest observation time can be used as the end time of the observation time window. The same target flight identifier, the same warning type, the same observation time window, and the same mapping data version correspond to the same unique evidence identifier.

[0164] Use the aforementioned evidence as a unique identifier to query existing abnormal warning records.

[0165] When an anomaly warning record with the same unique identifier already exists, instead of creating a new record, the newly formed independent evidence point is merged with the existing records to remove duplicates. During deduplication, aircraft address, observation time, longitude, and latitude can be used as comparison criteria for independent evidence points. If all of these are consistent, the record is not written again; if any is inconsistent, the independent evidence point is added as new evidence to the existing anomaly warning record. After merging, the corresponding independent evidence points are rearranged according to observation time, and the anomaly evidence snapshot is updated.

[0166] When no corresponding anomaly warning record with the same unique evidence identifier does exist, a new anomaly warning record is created. The newly created anomaly warning record must at least store the target flight identifier, warning type, aircraft number to be verified, observation time window, mapping data version, unique evidence identifier, and anomaly warning generation time, and establish a correlation between the anomaly warning record and the anomaly evidence snapshot.

[0167] After an anomaly alert is generated, the target flight business fields, independent evidence point sequences, and aircraft address-to-registration mapping data versions actually used in this anomaly assessment are permanently saved. When it is necessary to reproduce the anomaly assessment result later, the corresponding anomaly evidence snapshot is read from the anomaly alert record, and the saved target flight business fields, independent evidence point sequences, and corresponding versions of aircraft address-to-registration mapping data are used to re-execute candidate registration number parsing and verification registration number comparison, thus ensuring that the data state used in the reproduction process is consistent with the data state at the time the anomaly alert was generated.

[0168] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An aircraft registration anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification, characterized in that, include: S1. Obtain the flight number, departure airport, actual departure time and aircraft number to be verified of the target flight. Generate a standard call sign based on the flight number, search for aircraft using the standard call sign, and form a first candidate aircraft set. S2. Obtain the ADS-B observation data corresponding to the first candidate aircraft set, using the message generation time as the observation time, and remove data that exceeds the delay threshold based on the time difference between the message generation time and the reception time to obtain the effective observation dataset. S3. Group the data with the same aircraft address and that meet the time and spatial proximity thresholds in the effective observation dataset into independent evidence points, and sort them by observation time to form an observation sequence; S4. Determine the first and second takeoff time intervals based on the actual takeoff time and observation sequence, and perform time constraint fusion based on the intersection and interval interval of the two time intervals to obtain the third takeoff time interval; when the interval between the two time intervals exceeds the preset time tolerance threshold, it is determined that the time evidence is insufficient. S5. Calculate the upper limit of the reachable distance based on the third takeoff time interval, the observation time of the independent evidence point and the aircraft motion parameters, and compare it with the observation distance from the independent evidence point to the takeoff airport to form a spatiotemporal state sequence. S6. Select the first candidate aircraft set based on the independent evidence points that continuously satisfy the spatiotemporal compatibility condition in the spatiotemporal state sequence, and obtain the second candidate aircraft set. S7. Analyze the candidate aircraft numbers of the second candidate aircraft set; If a candidate machine number with the same number to be verified exists, the machine number is determined to be consistent; if all candidate machine numbers are identifiable and none are identical, the machine number is determined to be abnormal and an alert is generated; if the evidence is insufficient, no alert is generated.

2. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 1, characterized in that, S1 includes: Normalize the format of the target flight number and extract the airline code and flight suffix; Based on the airline code mapping relationship, the airline code is converted into an ICAO call sign prefix and combined with the flight suffix to form a standard call sign; Based on the actual takeoff time, a call sign retrieval time window is constructed according to preset forward and backward time thresholds, and ADS-B records that match the standard call sign within the time window are retrieved. The aircraft addresses in the ADS-B records are extracted and deduplicated, and the aircraft corresponding to different aircraft addresses form a first candidate aircraft set.

3. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 1, characterized in that, S2 includes: Obtain the corresponding ADS-B observation data based on the aircraft addresses in the first candidate aircraft set; Extract the message generation time and reception time respectively, calculate the reception delay value of the two, and form a reception delay sequence according to the message generation time; The median and median absolute deviation of the received delay sequence are calculated within a preset sliding time window, and a dynamic delay threshold is determined based on the two. ADS-B observation data with reception delay values ​​exceeding the dynamic delay threshold are discarded, and the remaining data are determined as the valid observation dataset.

4. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 1, characterized in that, S3 includes: The valid observation datasets are grouped according to aircraft addresses and sorted by observation time to form an initial observation sequence; Calculate the observation time difference and great circle distance between adjacent observation data in sequence, and classify the data that simultaneously meet the time proximity threshold and spatial proximity threshold into the same observation cluster; Before adding data to the observation cluster, it is further determined whether the maximum observation time span and the maximum spatial span after addition are less than the preset cluster time threshold and cluster space threshold, respectively. If either threshold is exceeded, a new observation cluster is established. The observation data with the smallest reception delay is selected from each observation cluster as an independent evidence point; An observation sequence is formed by arranging the observation times of independent evidence points.

5. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 1, characterized in that, S4 includes: The first takeoff time interval is formed by determining the front and back deviation boundaries based on the actual takeoff time and the historical time deviation of the corresponding data source. Calculate the distance from independent evidence points to the departure airport and the speed of movement between adjacent independent evidence points in the observation sequence, and identify the state transition segment from the airport-associated state to the departure state based on the airport association distance threshold and the departure speed threshold; The second takeoff time interval is formed based on the observation time of independent evidence points before and after the state transition segment; When the first takeoff time interval and the second takeoff time interval intersect, the intersection is determined as the third takeoff time interval; when the two do not intersect but the interval interval does not exceed the time tolerance threshold, the smallest continuous time interval covering the two intervals is determined as the third takeoff time interval. When the interval exceeds the time tolerance threshold, it is determined to be a time evidence conflict, and the current verification result is transferred to the insufficient evidence state.

6. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 1, characterized in that, S5 includes: Obtain the starting time boundary of the third takeoff time interval and subtract it from the observation time of each independent evidence point to determine the maximum possible flight duration corresponding to each independent evidence point; Based on the maximum possible flight time, the maximum reasonable speed of the aircraft, and the distance compensation parameters, the corresponding upper bound of the reachable distance is calculated using the spatiotemporal reachability model. The observation distance from the independent evidence point to the takeoff airport is calculated using the great circle distance algorithm, and the ratio of the observation distance to the corresponding upper bound of the reachable distance is calculated. The distance ratio is compared with a preset reachability threshold to obtain the spatiotemporal compatible state or spatiotemporal incompatible state of the corresponding independent evidence point. Arrange the spatiotemporal states according to the observation time of independent evidence points to form a spatiotemporal state sequence.

7. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 1, characterized in that, S6 includes: A sliding evidence window is established on the spatiotemporal state sequence corresponding to each first candidate aircraft according to the preset number of evidence points. Statistically analyze the number of evidence points in spatiotemporally compatible states, the proportion of compatible states, and the duration of continuous compatibility within each sliding evidence window; The number of evidence points, the proportion of compatible states, and the duration of continuous compatibility are compared with the evidence number threshold, the compatibility ratio threshold, and the duration threshold, respectively. The first candidate aircraft with a sliding evidence window that simultaneously satisfies all the thresholds is identified as a trajectory compatible aircraft. The trajectory-compatible aircraft are then grouped together to form a second candidate aircraft set.

8. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 1, characterized in that, S7 includes: Obtain the ADS-B aircraft number of each second candidate aircraft in the second candidate aircraft set; if the ADS-B aircraft number is missing, query the mapping relationship between the aircraft address and the aircraft number based on the aircraft address to form a candidate aircraft number set; The candidate number set is matched with the number to be verified. If any candidate number is the same as the number to be verified, the number to be verified is determined to be the same and the abnormal number judgment is terminated. In the absence of duplicate candidate aircraft numbers, determine whether the candidate aircraft numbers of each second candidate aircraft can be resolved; If there is a second candidate aircraft whose candidate aircraft number cannot be resolved, the current verification result will be determined as insufficient evidence, and no abnormal warning will be generated; When all candidate aircraft numbers of the second candidate aircraft can be resolved and are different from the aircraft number to be verified, the aircraft number to be verified is determined to be abnormal and an abnormal warning is generated.

9. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 6, characterized in that, The spatiotemporal reachability model satisfies that the reachable distance monotonically does not decrease with flight duration, and the determination of the upper bound of the reachable distance includes: The time difference between the observation time of the independent evidence point and the two time boundaries of the third takeoff time interval is calculated to obtain the flight duration interval; Select the maximum possible flight time within the specified flight time range; Input the maximum possible flight time and aircraft motion parameters into the spatiotemporal reachability model to calculate the corresponding upper bound of reachable distance; When the observation distance of an independent evidence point still exceeds the upper limit of the reachable distance, it is determined that the independent evidence point does not meet the spatiotemporal reachability condition under all possible takeoff times defined in the third takeoff time interval.

10. The aircraft registration number anomaly early warning method based on call sign reverse lookup and trajectory spatiotemporal verification according to claim 8, characterized in that, The generation of the anomaly warning includes: Obtain snapshots of the target flight business fields, sequences of independent evidence points, and version information of the aircraft address and aircraft number mapping data used for anomaly detection to form an anomaly evidence snapshot; The target flight identifier, warning type, observation time window, and the version of the mapping data are combined to form a unique evidence identifier; The existing abnormal warning records are queried using the unique identifier of the evidence. When a corresponding record exists, the newly added independent evidence point is merged into the existing abnormal warning record. If no corresponding record exists, a new anomaly warning record is created and associated with the aforementioned anomaly evidence snapshot; Save the flight business fields, independent evidence points, and mapping data versions used in the anomaly detection so that the anomaly alert can be reproduced according to the data state at the time of generation.

Images