Marine ship monitoring method, system, electronic device and readable storage medium

Abstract

The application discloses a kind of offshore warship monitoring method, system, electronic equipment and readable storage medium, it is related to offshore target monitoring technical field.Candidate aircraft's ADS-B track data is acquired, and shipboard aircraft associated with target warship and multiple off-ship events are determined;According to the off-ship position and off-ship heading of each event, a spatial candidate constraint region is constructed, and the motion state of the warship is estimated according to the off-ship position and off-ship time;The constraint regions of different off-ship times are compensated to the same reference time, the candidate position region is determined according to the overlapping degree after superposition, and the position of the warship is updated.The present application can reduce the misplacement of asynchronous position constraints caused by the movement of the warship, and provide a basis for the determination and update of the position of the target warship.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of offshore target monitoring, and particularly relates to an offshore ship monitoring method and system, electronic equipment and readable storage medium. BACKGROUND

[0002] Offshore ship monitoring usually uses AIS track data, satellite remote sensing images and public ship activity information to obtain the position of the target. Due to the interruption of the track, the observation interval of the satellite and the spatiotemporal accuracy of the information, the above information is difficult to continuously provide the position of the target ship, and the position extrapolation based on the historical track may also deviate due to the change of the ship movement.

[0003] For the ship equipped with an aircraft, the ADS-B track data of the aircraft can provide auxiliary monitoring clues. However, the data reflects the activities of the aircraft itself and cannot directly determine the position of the associated ship, and the activities of the shore-based aircraft and the shipboard aircraft may intersect each other, and it is necessary to distinguish the association between the aircraft and the target ship.

[0004] In addition, the movement of the ship will make the activity clues at different times correspond to different ship positions. The position constraint provided by a single clue is limited, and directly integrating the clues at different times may introduce position deviation. Therefore, how to use the related aircraft track information to reduce the influence of the time sequence position difference caused by the ship movement on the monitoring result still needs further research. SUMMARY

[0005] The present application aims to solve the technical problem that the position clues at different times are difficult to be directly integrated when using aircraft track information to monitor offshore ships, which affects the accuracy of the determination and update of the position of the target ship, and provides an offshore ship monitoring method, system, electronic equipment and readable storage medium.

[0006] The first aspect of the present application provides a method for monitoring a marine vessel, comprising: obtaining broadcast automatic dependent surveillance-B (ADS-B) track data of a candidate aircraft, and determining a correlation relationship between the candidate aircraft and a target vessel according to the ADS-B track data to obtain a correlated carrier aircraft; determining a plurality of off-ship events from the ADS-B track data of the correlated carrier aircraft, wherein each off-ship event is characterized by a corresponding off-ship position, off-ship time and off-ship heading; constructing a spatial candidate constraint region of the target vessel at a corresponding off-ship time according to the off-ship position and off-ship heading of each off-ship event; estimating a motion state of the target vessel according to the off-ship positions and off-ship times of the plurality of off-ship events; compensating the spatial candidate constraint regions corresponding to different off-ship times in position according to the motion state, and converting each spatial candidate constraint region to a same reference time; superimposing the converted plurality of spatial candidate constraint regions, and determining a candidate position region of the target vessel according to the overlapping degree between the plurality of spatial candidate constraint regions; and updating the position of the target vessel at the reference time based on the candidate position region.

[0007] The second aspect of the present application provides a system for monitoring a marine vessel, comprising: a vessel correlation module configured to obtain broadcast automatic dependent surveillance-B (ADS-B) track data of a candidate aircraft, and determine a correlation relationship between the candidate aircraft and a target vessel according to the ADS-B track data to obtain a correlated carrier aircraft; an off-ship event determination module configured to determine a plurality of off-ship events from the ADS-B track data of the correlated carrier aircraft, wherein each off-ship event is characterized by a corresponding off-ship position, off-ship time and off-ship heading; a spatial constraint construction module configured to construct a spatial candidate constraint region of the target vessel at a corresponding off-ship time according to the off-ship position and off-ship heading of each off-ship event; a motion state estimation module configured to estimate a motion state of the target vessel according to the off-ship positions and off-ship times of the plurality of off-ship events; a position compensation module configured to compensate the spatial candidate constraint regions corresponding to different off-ship times in position according to the motion state, and convert each spatial candidate constraint region to a same reference time; a candidate region determination module configured to superimpose the converted plurality of spatial candidate constraint regions, and determine a candidate position region of the target vessel according to the overlapping degree between the plurality of spatial candidate constraint regions; and a position updating module configured to update the position of the target vessel at the reference time based on the candidate position region.

[0008] The third aspect of the present application provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor to implement the method of the first aspect.

[0009] The fourth aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method according to the first aspect.

[0010] Compared with the prior art, the present application has the following technical effects: By determining the association relationship between the carrier-based aircraft and the target ship, the off-ship position and off-ship heading of the carrier-based aircraft are converted into the spatial candidate constraints of the target ship, thereby providing auxiliary basis for ship position determination. By estimating the ship motion state by using multiple off-ship events and compensating the spatial candidate constraint regions at different times to the same reference time, the constraint misplacement caused by ship movement can be reduced, and the position clues at different times can be comprehensively utilized under the consistent time reference. On this basis, the candidate position region is determined according to the overlapping degree of each spatial candidate constraint region, and the position of the target ship is updated by comprehensively utilizing the constraints of multiple off-ship events, which helps to reduce the deviation introduced by directly superimposing the position clues at different times, and improves the accuracy of ship position determination and updating. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 is a flowchart of a sea ship monitoring method according to an embodiment of the present application; Figure 2 is a structural schematic diagram of a sea ship monitoring system according to an embodiment of the present application; Figure 3 is a structural schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0012] Unless otherwise explicitly stated, throughout the specification and claims, the term "comprise" or variations such as "comprises" or "comprising" will be understood to imply the inclusion of a stated element or group of elements but not the exclusion of any other element or group of elements.

[0013] The technical solutions of the present application are described below by means of specific examples. It should be understood that the one or more steps mentioned in the present application do not exclude other methods and steps before and after the combination steps, or other methods and steps can be inserted between these explicitly mentioned steps. It should also be understood that these examples are only used to illustrate the present application and do not limit the scope of the present application. Unless otherwise stated, the numbering of each method step is only for the purpose of identifying each method step, and is not intended to limit the arrangement order of each method or to limit the scope of the implementation of the present application. Changes or adjustments of the relative relationship, without substantial technical content changes, can also be considered as the implementation scope of the present application.

[0014] One embodiment of the present application relates to a maritime vessel monitoring method, which can be performed by a maritime vessel monitoring system. The system receives flight path data of aircrafts, and maintains a monitoring record of a target vessel, which contains the target vessel identification, position estimation result and corresponding time.

[0015] The embodiment determines position constraints provided by different off-ship events by using the flight path of the aircraft associated with the target vessel, and converts each position constraint to the same reference time by combining the vessel motion state, for updating the position of the target vessel.

[0016] As shown in the method, the method can specifically include the following steps: Figure 1 S101, obtaining ADS-B flight path data of a candidate aircraft, and determining the association relationship between the candidate aircraft and the target vessel according to the ADS-B flight path data to obtain an associated aircraft.

[0017] The candidate aircraft refers to the aircraft included in the current flight path analysis and to be determined for the association relationship with the target vessel; and the associated aircraft refers to the aircraft having the corresponding relationship with the target vessel after the association judgment.

[0018] The ADS-B flight path data of the candidate aircraft is obtained, the flight path records are collected according to the aircraft identification, and the flight paths of each aircraft are organized according to the time sequence of the flight path points. The aircraft identification, position, time and heading information of the flight path points are retained in the flight path data, so as to distinguish the activities of different aircrafts and the states of the same aircraft at different times.

[0019] The association relationship between the candidate aircraft and the target vessel is judged according to the corresponding relationship between the aircraft activities reflected by the flight path data and the related activities of the target vessel. The association result is recorded as the corresponding relationship between the aircraft identification and the target vessel identification, and the aircraft carrier flight path associated with the target vessel is extracted as the input for determining the subsequent off-ship event.

[0020] For example, when two different aircrafts are both associated with the same target vessel, the flight path records of the two aircrafts are retained respectively, and the subsequent off-ship events determined from the flight paths of the two aircrafts are classified into the same event set of the target vessel.

[0021] S102, determining a plurality of off-ship events from the ADS-B flight path data of the associated aircraft, wherein each off-ship event is characterized by the corresponding off-ship position, off-ship time and off-ship heading.

[0022] ​A departure event describes an activity in which a related carrier-based aircraft leaves a target ship. The departure location is the geographical location of the related carrier-based aircraft at the time of the event, the departure time is the exact moment the event occurs, and the departure heading is the heading information of the related carrier-based aircraft at that moment. These three parameters together describe the spatial location, time, and direction information of the same departure event.

[0023] The target ship position reference used for departure determination adopts historical AIS fusion positioning results or position estimation results obtained from the previous round of multi-source data updates, and is converted to the current departure determination time period based on the corresponding time of the result and existing motion information. This position reference is a priori estimation value and does not use the position update results to be obtained in the current round. Centered on this position reference, a shipborne reference determination area with a fixed threshold is set according to the ship size and shipborne signal coverage. When three consecutive ADS-B sampling points are outside this area, the track continues to move away from the center of the area, and the effective movement speed and the state verification of no reciprocating offset drift of the track pass the test, a departure event is initially detected.

[0024] For tracks that meet the above detection conditions, their ADS-B sampling points are traced back. Combining the position reference, deck layout, and positioning error, the last valid sampling point within the preset error range of the departure end of the corresponding deck is selected. If such a point exists, its coordinates are used as an approximation of the departure position, its timestamp is used as the departure time, and the average heading of two track points before and after the departure threshold is fitted as the departure heading. For tracks where the first receiving point is after departure and there is no historical sampling data from inside the ship, this embodiment does not use them for departure event statistics and parameter calculation.

[0025] When the shipborne signal coverage extends beyond the outer edge of the deck, this coverage area is used for preliminary screening of departure events, and the signal coverage boundary is not directly regarded as the departure end of the deck. Only when the sampled point corresponds to the departure end of the deck and the positional deviation of its relative end can be limited within a preset error range, will the event be included in subsequent deck geometric constraint calculations; if the above conditions are not met or there is a lack of available position references, the event will not be used for position back-calculation in this round.

[0026] Departure events are identified from the flight paths of each associated carrier-based aircraft, and corresponding parameters are extracted. Each departure event is associated with the aircraft identifier and the target ship identifier, and then a departure event sequence for the target ship is formed according to the chronological order of departure times. The number of events corresponds to the number of different departure activities, and multiple flight path records for the same departure activity are used to describe the same event.

[0027] Multiple departure events used for monitoring the same target ship can originate from different departure activities of the same associated carrier-based aircraft, or from different associated carrier-based aircraft. The departure positions of each event are expressed using consistent coordinates, the departure times use a consistent time reference, and the departure headings use a consistent angle reference, so that parameters from different events can be used for subsequent comparisons and calculations.

[0028] S103, based on the departure position and departure course corresponding to each departure event, construct the spatial candidate constraint region of the target ship at the corresponding departure time.

[0029] The spatial candidate constraint region refers to the range of possible positions of a target ship at the time of a certain departure event, determined based on information from that event. Candidate positions within this region represent possible locations of the target ship and correspond to the departure time of that event.

[0030] For each departure event, the departure position is used as the spatial reference for constructing the candidate region. Utilizing the directional information provided by the departure course and combining it with the relative geometric relationship between the departure position and the target ship's position, the bearing range of the candidate ship's position relative to the departure position is determined. Combining these positional and directional constraints, a spatial candidate constraint region corresponding to the departure moment is formed.

[0031] When constructing positional constraints using the departure course, it is necessary to determine the bearing range of the candidate ship position relative to the departure position. The correspondence between this bearing range and the aircraft's departure course depends on the departure direction and the relative geometric relationship between the departure position and the target ship. For events involving departure along the prescribed catapult or ski-jump departure direction, the opposite direction to the departure direction (corrected for ship motion) is used as the reference direction. This is combined with the relative position of the departure end and the center of mass in the fixed deck layout, the ship's orientation range, ADS-B heading error, and ship motion estimation error to determine the candidate bearing range of the center of mass. This is then combined with distance constraints to limit the spatial candidate constraint area. The aircraft's departure course cannot be directly used as this bearing.

[0032] The measured departure course is the direction of the ground track. Under the condition that the ship's course is approximately stable during the critical departure period, the aircraft's ground velocity vector is determined based on the track point during that period. This is then subtracted from the ship's ground velocity vector determined by the existing motion information used to construct the position reference. The direction of the resulting relative velocity vector in the geographic coordinate system is taken as the relative departure direction. If this direction cannot be reliably determined, the event is not used for position back-calculation in this round. Specifically, in the ship's coordinate system, the angle of deviation of the direction pointing from the departure end of each available deck towards the center of mass relative to the specified departure direction of that end is determined. Using the same clockwise angle reference, this angle is superimposed on the opposite direction of the aforementioned relative departure direction and refracted in a 360° cycle to obtain the corresponding candidate bearing of the center of mass. Based on the deck length and width, the allowable departure end range, and the heading and ship motion estimation error, the bearing sector is expanded, and the portion compatible with the ship's external orientation range is retained. This conversion uses the relative geometric relationship of the deck and does not require knowledge of the absolute coordinates of the current center of mass; the extraction of the departure event should still meet the aforementioned position reference and sampling conditions.

[0033] For the purpose of illustrating the definition of azimuth, let the known coordinates of the ship's center of mass be (120.0000° E, 30.0000° N), and the coordinates of the point of departure be (120.0010° E, 30.0002° N). In a local plane coordinate system with the transverse axis pointing east and the longitudinal axis pointing north, the displacement of the center of mass relative to the point of departure is approximately ( 96.4 meters (22.3 meters), located in the southwest direction; with true north as 0° and measured clockwise, the bearing of the point of departure from the ship towards the center of mass is approximately 257°. This example of known coordinates is only used to illustrate the bearing conversion and does not use the coordinates of the center of mass to be determined in this round as known input for reverse position calculation.

[0034] For each departure event, a corresponding region is constructed, and the correspondence between the region and the target ship identifier, departure event, and departure time is preserved. Since the ship may move between different departure times, each region represents the possible location range of the ship at different times.

[0035] Step S104: Estimate the motion state of the target ship based on the departure position and departure time corresponding to multiple departure events.

[0036] In this embodiment, the motion state is used to describe the target ship's speed and direction of motion within a selected time range, and to estimate the ship's position change between different moments. Specifically, the direction of motion describes the direction of the ship's position change, and the departure heading describes the direction of the carrier-based aircraft's departure; these are used for ship motion compensation and spatial candidate constraint region construction, respectively.

[0037] Using the departure time as a time index, the spatial changes between departure positions corresponding to multiple departure events are compared. Combined with the time intervals between departure times, the target ship's speed and direction of motion are estimated. A consistent position definition is used for the departure positions involved in the comparison, ensuring comparability of position changes across different events.

[0038] Through the above processing, the estimated ship motion state corresponding to the selected departure event and its time range is obtained. Based on this result, the position change of the target ship between each departure time and the reference time can be further estimated, serving as the basis for position compensation in the spatial candidate constraint region.

[0039] Step S105: Based on the motion state, perform position compensation on the spatial candidate constraint regions corresponding to different departure times, and transform each spatial candidate constraint region to the same reference time.

[0040] The reference time refers to the time corresponding to the current ship position update result. In this embodiment, the departure time of the first departure event in the departure event sequence is selected as the reference time, and the spatial candidate constraint regions corresponding to the remaining departure events are converted to this time.

[0041] For departure events occurring at the reference time, the spatial candidate constraint region is directly used as the constraint region at the reference time. For departure events occurring after the reference time, the positional change of the target ship from the reference time to the corresponding departure time is determined based on the motion state estimation results, and the candidate positions in that region are then extrapolated back to the reference time.

[0042] For example, to illustrate the compensation direction, assume two ship departure events occur at 08:00 and 08:02 respectively. Motion state estimation results indicate that the target ship moved 100 meters eastward between these two times. When 08:00 is selected as the reference time, the spatial candidate constraint region corresponding to 08:02 is compensated 100 meters westward, while the region corresponding to 08:00 retains its original position, so that both regions represent the possible position range of the target ship at 08:00. The above times and distances are only used to illustrate the position compensation process.

[0043] After compensation, each candidate spatial constraint region has the same temporal meaning and can be compared and superimposed under the same spatial coordinates. Simultaneously, the correspondence between the compensated regions and the original departure events is preserved to clarify the track information upon which each region is based.

[0044] Step S106: Overlay the multiple converted spatial candidate constraint regions and determine the candidate location region of the target ship based on the degree of overlap between the multiple spatial candidate constraint regions.

[0045] Under a unified spatial coordinate system, candidate spatial constraint regions at various reference times are superimposed to determine the coverage of different spatial locations by each region. The degree of overlap describes the extent to which each spatial location is supported by multiple off-ship event constraints.

[0046] For example, in an illustrative scenario where all regions contribute the same constraint, assume that the three compensated spatial candidate constraint regions share a common coverage area. A location within the common coverage area of ​​all three regions is supported by constraints from all three departure events; a location within only one of the regions is supported by constraints from only one departure event. The former has a higher degree of overlap than the latter. In this example, the area shared by the three regions can be extracted as a candidate location region for the target ship.

[0047] By comparing the common support at different positions in the superimposed results, the positional constraints provided by each departure event are combined into a range of candidate positions for the target ship at the reference time.

[0048] Position compensation is used to mitigate regional misalignment caused by ship movement, while region overlay is used to integrate the spatial support provided by different departure events. When the compensated regions completely overlap, the resulting candidate position regions can maintain their original range, and their area does not need to shrink as the number of departure events increases.

[0049] Step S107: Update the position of the target ship at the reference time based on the candidate position area.

[0050] The position estimate of the target ship at a reference time is generated based on the candidate location area, and the result is associated with the target ship's identification and the reference time.

[0051] This embodiment uses a region format to represent the ship's position estimation results, that is, candidate position regions represent the possible position range of the target ship at the reference time. Based on the target ship's identification location and corresponding monitoring records, the candidate position regions and their corresponding reference times are written into these records, thus completing the update of the target ship's position information.

[0052] By simultaneously recording the location range and corresponding time, the spatial meaning and temporal attribution of this update result are clarified, which can be read, compared and updated in subsequent monitoring processes.

[0053] This invention determines the relationship between carrier-based aircraft and target ships, transforming the departure positions and headings of the aircraft into spatial candidate constraints for the target ship, thus providing auxiliary data for ship position determination. By estimating the ship's motion state using multiple departure events and compensating the spatial candidate constraint regions at different times to the same reference time, constraint misalignment caused by ship movement can be mitigated, allowing position cues from different times to be comprehensively utilized under a consistent time reference. Furthermore, candidate position regions are determined based on the degree of overlap between the spatial candidate constraint regions, and the target ship's position is updated by integrating the constraints of multiple departure events. This helps reduce the bias introduced by directly superimposing position cues from different times, improving the accuracy of ship position determination and updating.

[0054] In one embodiment, the association between the candidate aircraft and the target ship is further explained. A candidate ship refers to a ship that may have an affiliation with a candidate aircraft, including the target ship whose position information needs to be updated. For the same candidate aircraft, it is matched with each candidate ship to obtain a corresponding affiliation matching score, and then the associated carrier-based aircraft is determined based on this score.

[0055] The features used for matching include at least two of the following: aircraft type, takeoff and landing area, flight path round-trip characteristics, and historical formation relationships. Aircraft type and historical formation relationships can be obtained from existing aircraft files or historical activity records based on aircraft identification; takeoff and landing area and flight path round-trip characteristics are determined by combining ADS-B flight path data. Candidate ship information used for comparison is derived from existing ship files, observation records, or historical activity information.

[0056] Aircraft type features are used to determine the compatibility between the type of candidate aircraft and the type of aircraft carried by candidate ships. For example, the aircraft type of a candidate aircraft is compared with the aircraft types recorded in the candidate ship's file to obtain the matching results for this type of feature.

[0057] The takeoff and landing activity area feature is used to determine whether there is a spatial correspondence between the geographical range of the aircraft's takeoff and landing activities and the existing activity range of the candidate ship during the corresponding time period. This feature is used for regional-level correlation determination; the specific departure location and departure time in subsequent departure events will be determined separately.

[0058] Flight path round-trip features are used to describe the flight path of an aircraft after it leaves an activity area and returns to the relevant area. By combining the time and area corresponding to the round-trip flight path, the correspondence between the round-trip activity and the existing activity information of the candidate ships can be determined.

[0059] Historical formation relationship features are used to describe the formation or affiliation relationships of candidate aircraft and candidate ships in historical activities. By querying the historical records associated with aircraft and ship identifiers, the support of this type of feature for the current affiliation relationship can be obtained.

[0060] Based on the available feature information, at least two categories of the above features are selected for matching. For comparisons between the same candidate aircraft and different candidate ships, the same feature combinations and scoring rules are used to ensure comparability of the scores. The attribution matching score indicates the degree to which the selected features support the attribution relationship between the aircraft and the candidate ship; a higher score indicates a higher degree of support.

[0061] As a simplified scoring example, we can determine whether each selected feature category conforms to a corresponding relationship, and use the proportion of feature categories that conform to the corresponding relationship to the total number of selected feature categories as the attribution matching score. In this example, all selected features have information that can be used for judgment, and the score ranges from 0 to 1; when all selected features conform to the corresponding relationship, the attribution matching score is 1.

[0062] After obtaining the attribution matching score, a matching threshold corresponding to the scoring rule is set as the score boundary for determining whether association is supported. When the attribution matching score between a candidate aircraft and a candidate ship reaches the matching threshold, the candidate ship is designated as the attribution candidate for that aircraft.

[0063] For aircraft tracks corresponding to takeoff and landing activities and known shore-based sites, if their affiliation matching scores with each candidate ship are all below the matching threshold, they are identified as shore-based aircraft tracks unrelated to the candidate ships and are removed from the tracks used for ship-aircraft association. The determination of shore-based activities is combined with the determination of ship affiliation to distinguish between shore-based activity clues and track clues that can support ship association.

[0064] For the remaining candidate aircraft, if only one candidate ship's score reaches the matching threshold, that candidate ship is identified as an associated ship. If multiple candidate ships' scores reach the matching threshold, an assignment conflict exists. When the highest score can be uniquely determined, the candidate ships are compared according to their assignment matching scores, and the candidate ship with the highest score is identified as the associated ship for that aircraft. If multiple candidate ships have the same highest score, the aircraft's track is marked as having an undetermined assignment, and it is not assigned an associated ship or participates in subsequent position back-calculation. Once new observation data updates the matching score and eliminates the conflict of the same highest score, the valid track with a determined assignment will be included in the position back-calculation.

[0065] For example, to illustrate the above judgment process, two types of features are selected for matching: takeoff and landing activity area and flight track round-trip features, and the matching threshold is assumed to be 0.5. Aircraft A matches both types of features with candidate ship A, and its matching score is 1; it matches only the takeoff and landing activity area feature with candidate ship B, and its matching score is 0.5. At this point, both candidate ships have reached the matching threshold. Based on the score comparison, candidate ship A is determined to be an associated ship of aircraft A.

[0066] For example, if the takeoff and landing activities of aircraft B correspond to a known shore-based site, and neither of the two selected features can support its affiliation with candidate ships A and B, resulting in a matching score of 0, then the corresponding track of aircraft B will be filtered out as a shore-based irrelevant track. The scores and thresholds mentioned above are only used to illustrate the scoring and judgment process.

[0067] After completing the association determination, candidate aircraft whose target ship is the associated ship are identified as associated carrier-based aircraft. The aircraft identifier, target ship identifier and corresponding attribution matching score are saved, and the ADS-B track of the associated carrier-based aircraft is provided to the subsequent departure event determination steps.

[0068] Aircraft tracks whose attribution is pending, those suspected of being fraudulent, or whose track quality does not meet the minimum valid threshold will not participate in this round of position back-dive. Valid tracks selected must be able to provide multiple departure events required for subsequent processing; if this is not met, the process will proceed to the evidence invalidation backoff branch described later.

[0069] In one embodiment, the distance limitation of the spatial candidate constraint region in step S103 is further explained. For each departure event, a distance constraint region is established based on the corresponding departure position, so that the constructed spatial candidate constraint region is located within the distance constraint region.

[0070] In this embodiment, the center of mass of the target ship is used as the reference point to characterize the ship's position. The candidate position of the target ship's center of mass refers to the location where the target ship's center of mass may be located, inferred from the departure event. The departure position and the candidate position of the center of mass are represented by a local planar coordinate system of the monitored sea area, and all relevant distances are calculated according to the planar distances in this coordinate system.

[0071] The preset distance threshold is determined based on the distance from the target ship's center of mass to the end of the deck. The end of the deck refers to the end of the deck corresponding to the respective departure activity, and the distance from the center of mass to the end of the deck can be determined based on the target ship's structural dimensions and deck layout parameters. This distance reflects the ship's own geometry and its determination is not predicated on the known current geographical location of the target ship.

[0072] In the geometric model used in this embodiment, the associated carrier-based aircraft departs from the corresponding deck end, and the departure position determined in step S102 is used as an approximation of the position of that end at the moment of departure. This geometric model is only applicable to events where the sampled departure points can represent the corresponding deck end position. For such events, when the departure position estimation error is negligible, the planar distance from the target ship's center of mass to the deck end is used as a preset distance threshold; when there is a non-negligible error, an upper bound of the position error relative to the deck end is superimposed on this geometric distance, and the resulting distance is used as the preset distance threshold. A circular region containing the circumferential boundary is constructed with the departure position as the center and the preset distance threshold as the radius, and this region is used as the distance constraint region for the corresponding departure event.

[0073] Specifically, let the first The departure location of each departure event is The preset distance threshold is The corresponding distance constraint region is:

[0074] in, The coordinates of the candidate position of the target ship's center of mass. , , , and Use a consistent unit of length. The distance between the candidate position and the departure position should not exceed [a certain value]. At that time, the distance constraint of the departure event is satisfied; the distance is greater than At that time, it does not belong to the distance constraint area.

[0075] The distance constraint region defines the maximum distance between the candidate position of the target ship's center of mass and its departure position. Step S103, which constructs the spatial candidate constraint region based on the departure position and departure heading, should also satisfy the aforementioned distance constraint. Let this spatial candidate constraint region be... Then we have:

[0076] Therefore, being located within the distance constraint area only indicates that the corresponding candidate position meets the distance condition. Whether it is included in the spatial candidate constraint area also depends on whether it meets other spatial constraints determined based on the departure course and the relative positional relationship between the ship and the aircraft.

[0077] For example, without considering the estimation error of the departure position, assuming the planar distance from the target ship's center of mass to the corresponding deck end is 150 meters, and setting the preset distance threshold to 150 meters, and using the departure position of a certain departure event as the local coordinate origin, then the distance constraint region for that event is a circular area with the origin as the center and a radius of 150 meters. Candidate positions with coordinates (100 meters, 0 meters) satisfy the distance constraint and can be further evaluated in conjunction with other spatial constraints; candidate positions with coordinates (160 meters, 0 meters) exceed the preset distance threshold and should be excluded. The above values ​​are used to illustrate the construction and evaluation process of the distance constraint region.

[0078] For different departure events, distance constraint regions are established for each departure position at the corresponding departure time, and spatial candidate constraint regions are defined accordingly. During subsequent position compensation, the spatial candidate constraint regions that have met the distance constraints are collectively transformed to the same reference time and then superimposed. In this way, the ship's own geometric dimensions are transformed into distance boundaries for the candidate position range, providing spatial constraints for the subsequent determination of candidate position regions.

[0079] In one embodiment, the estimation process of the target ship's motion state in step S104 is further explained. Multiple departure events are sorted according to their departure times. Two departure events are selected where the departure points are approximately at the same position relative to the target ship but at different departure times. Using a uniform linear motion model, the estimated velocity vector of the target ship is calculated based on the displacement between the two departure positions and the corresponding time interval.

[0080] The uniform linear motion model refers to approximating the motion of the target ship as a motion in which the magnitude and direction of velocity remain constant within the time range under consideration. In this embodiment, the aircraft departure points corresponding to the two selected departure events are approximately the same relative to the ship, and the ship maintains an approximately stable heading between the two departures. Therefore, the displacement between the two departure positions can be used as an approximation of the target ship's displacement within that time interval.

[0081] During the calculation, the two departure positions are transformed to the same local plane coordinate system, and the time reference for departure is unified. The east direction is taken as the positive horizontal axis and the north direction as the positive vertical axis. The earlier departure time is assumed to be... The corresponding departure position is The later departure time was The corresponding departure position is ,in, .

[0082] The displacement vector from the earlier departure position to the later departure position, and the time interval between the two departure times, are as follows:

[0083]

[0084] in, It simultaneously indicates the magnitude and direction of the positional change. The value must be positive. Two events with the same departure time cannot be used as an event pair in this calculation to avoid the time interval being zero.

[0085] Dividing the displacement vector by the time interval yields an estimate of the target ship's velocity vector.

[0086] in, The two components represent the estimated velocities of the target ship along the east-west and north-south directions, respectively. When the coordinates are in meters and the time interval is in seconds, the unit of the velocity vector is meters per second. The magnitude of the velocity vector represents the estimated speed, and its direction represents the estimated direction of motion; when the velocity vector is zero, it means that the estimated displacement given by the model is zero.

[0087] For example, to illustrate the above calculation process, assume that two departure events occur at 08:00 and 08:02 respectively. The earlier departure position in the local coordinate system is (0 meters, 0 meters), and the later departure position is (100 meters, 0 meters). At this time, the displacement vector is (100 meters, 0 meters), the time interval is 120 seconds, and the calculated velocity vector estimate is (…). (0 m / s, 0 m / s), the estimated speed is 3 km / h, and the estimated direction of motion is due east. The above data is only used to illustrate the calculation method of the velocity vector.

[0088] After obtaining the velocity vector estimate, it is used as the motion state of the target ship within the examined time range and provided to step S105 for position compensation. Thus, the position changes corresponding to different departure times are converted into motion parameters that can be used to calculate the ship's displacement, providing a basis for transforming each spatial candidate constraint region to the same reference time.

[0089] In one embodiment, the position compensation in step S105 and the region overlay and candidate position region determination process in step S106 are further explained in conjunction with the velocity vector estimate obtained in step S104.

[0090] Let the estimated velocity vector of the target ship be... , No. The departure time of each departure event is The corresponding spatial candidate constraint region is The selected reference time is Based on the uniform linear motion model, calculate the position compensation amount for the candidate constraint region in this space:

[0091] in, This is a translation vector with magnitude and direction, representing the displacement required to transform the position at the moment of departure to the reference moment. Velocity and time difference use matched units, ensuring that the position compensation amount has the same length unit as the coordinates of the spatial candidate constraint region.

[0092] By translating each point within the spatial candidate constraint region according to the same position compensation amount, the spatial candidate constraint region at the reference time is obtained:

[0093] in, To compensate for the location points in the previous region, This is the compensated region. The global translation preserves the shape and size of the region, ensuring that the positional constraints represented by this region correspond to the reference time.

[0094] When the reference time is later than the departure time, the time difference is positive, and the ship moves in the estimated direction of motion; when the reference time is earlier than the departure time, the time difference is negative, and the ship moves in the opposite direction of the estimated direction of motion; when the two times are the same, the position compensation is zero.

[0095] Following the previous calculation example, the estimated velocity vector of the target ship is ( (0 m / s, 0 m / s), taking 08:00 as the reference time. For the spatial candidate constraint region corresponding to 08:02, the time difference is... 120 seconds, the position compensation is ( (100 meters, 0 meters), meaning the entire area is shifted 100 meters westward. The position compensation of the spatial candidate constraint area corresponding to 08:00 is zero. Therefore, the spatial candidate constraint areas corresponding to the two ship departure events are both converted to 08:00.

[0096] After location compensation is completed, a unified geographic raster is established within the spatial range covering each compensated area. Each geographic raster uses a consistent coordinate reference and raster size. This embodiment uses a square raster, and whether an area covers a corresponding raster is determined by whether the raster's center point is located within the compensated spatial candidate constraint area; when dividing the raster, each area participating in the overlay is ensured to contain at least one raster center point.

[0097] Assign corresponding constraint weights to each spatial candidate constraint region. Constraint weights are positive values, representing the constraint contribution of that region to its coverage location. For the same geographic raster, sum the constraint weights corresponding to each compensated region covering that raster to obtain the cumulative constraint weight:

[0098] in, For the first Cumulative constraint weights for each geographic raster The number of spatial candidate constraint regions to participate in the overlay. For the first The constraint weights corresponding to the nth region; when the nth region... The center point of each geographic raster is located at At that time, ,otherwise .

[0099] In one equal-weighting implementation, the constraint weights corresponding to each region are all set to 1. In this case, the cumulative constraint weight is equal to the number of regions covering the geographic raster. For example, a raster covered by only one region has a cumulative constraint weight of 1; a raster covered by two regions has a cumulative constraint weight of 2; and a raster covered by three regions has a cumulative constraint weight of 3. Through the above accumulation method, the common constraint support obtained by the raster represents the degree of overlap at corresponding locations.

[0100] Determine the maximum value of the cumulative constraint weight across all geographic rasters:

[0101] Select cumulative constraint weight equal to All geographic rasters are divided according to spatial connectivity. In this embodiment, two adjacent rasters with a common edge are considered connected; the highest-weighted raster that can be continuously connected through the above-mentioned adjacency relationship is assigned to the same connected region. Two rasters that only touch at their corners are not considered connected in this respect.

[0102] When all the maximum weighted grids are interconnected, a connected region is obtained; when the maximum weighted grids are distributed in positions that are not interconnected, multiple connected regions are obtained. The coverage area of ​​the grids contained in each connected region is determined as the candidate position region of the target ship at the reference time and provided to subsequent position update steps.

[0103] The above processing determines the candidate location region based on the maximum value of the cumulative constraint weights. When multiple compensated spatial candidate constraint regions completely overlap and use the same constraint weights, the cumulative constraint weights at the overlapping locations increase, and the resulting candidate location region can maintain the raster representation range of the original overlapping region. Therefore, whether the candidate location region shrinks depends on the distribution of the compensated regions and their common coverage.

[0104] In one embodiment, before updating the position of the target ship in step S107, the position back-inference confidence is determined based on the ship-aircraft association results, track quality, spatial constraint effectiveness, and abnormal situations, and the ADS-B position probability distribution of the target ship at the reference time is adjusted accordingly.

[0105] The position back-dive confidence score is used to characterize the reliability of the position back-dive based on the associated carrier-based aircraft tracks. Taking multiple departure events of the same associated carrier-based aircraft as an example, the attribution matching score between the associated carrier-based aircraft and the target ship saved in step S101 is read. Combined with the ADS-B track data associated with the aircraft and spatial constraints, the single-aircraft position back-dive confidence score is calculated according to the following formula:

[0106] in, To infer the confidence level from the location of a single machine, For the belonging matching score, For track quality factor, As a spatial constraint effectiveness factor, The non-negative magnitude of the abnormal penalty item is represented by its weighted value, which is entered with a negative sign. , , and These are the corresponding non-negative weight coefficients. The attribution matching score, track quality factor, and spatial constraint effectiveness factor are all normalized to the interval 0 to 1. A weighted result greater than 1 is set to 1, less than 0 is set to 0, and the original value is retained in other cases to ensure that the output confidence level is within the 0 to 1 interval.

[0107] The track quality factor is normalized and assigned a value based on the sampling continuity of ADS-B track points, the estimated positioning error, and the data update frequency within the time range involved in this location back-reasoning. Sampling continuity reflects track interruptions or lost points, the estimated positioning error reflects the uncertainty of track point positions, and the data update frequency reflects the update status of available track points per unit time.

[0108] Under the same statistical range and normalization rules, the more continuous the sampling, the smaller the estimated positioning error, and the higher the data update frequency, the closer the track quality factor is to 1; conversely, the value of this factor should be reduced.

[0109] The spatial constraint effectiveness factor is determined by normalizing the residuals of the ship-aircraft geometric constraints and motion consistency constraints corresponding to the associated carrier-based aircraft. It is used to characterize the degree of consistency between the trajectory and the adopted spatial constraints and motion model. The smaller the residual, the closer the spatial constraint effectiveness factor is to 1.

[0110] The geometric constraint residuals of the ship and aircraft are obtained by comparing the relative bearing and distance between the departure point and the candidate centroid position with the theoretical bearing and theoretical distance defined by the deck layout. During calculation, the center of each grid within the candidate position area is used as the centroid position to be tested at the reference time, and converted to the corresponding departure time based on the aforementioned ship motion state; the bearing is uniformly adopted as the direction from the departure point to the centroid. The bearing residual is taken as the minimum circumferential angle difference between two bearings, and the distance residual is taken as the absolute value of the difference between the two distances; when the theoretical bearing or distance is within an allowable range, the minimum deviation of the observed value from that range is taken, and the residual is zero when it is within the range. Here, the candidate centroid position to be tested is used, and it is not required to obtain the actual centroid position in advance.

[0111] The motion consistency constraint residuals are obtained by comparing the actual departure speed and rate of change of heading determined by the ADS-B track with the theoretical departure motion envelope. The actual departure speed is calculated based on the displacement and time interval of track points near the departure threshold, while the rate of change of heading is calculated based on the directional heading difference and time interval between adjacent track points after circular reversal processing. Using the typical departure motion envelope of the corresponding aircraft type as the relative motion constraint, and combining it with the aforementioned ship velocity vector estimate, it is transformed to a geographic coordinate system consistent with the ADS-B track, yielding the allowable ranges for the theoretical departure speed and rate of change of heading used for comparison. When the observed value lies within the corresponding allowable range, this residual is zero; otherwise, the absolute difference to the nearest boundary of the corresponding range is taken.

[0112] As a normalization and fusion implementation, the residuals of bearing, range, speed, and heading rate of change are divided by their corresponding positive allowable deviation scales to eliminate the influence of dimensions. Each scale is pre-set based on deck geometry error, ADS-B positioning and heading error, and the motion envelope of the corresponding aircraft type. At each candidate centroid position, the arithmetic mean of the four standardized residuals is first taken, and then the average of each effective departure event of the same associated carrier-based aircraft is taken to obtain the comprehensive residual corresponding to that aircraft. The reciprocal of the sum of the comprehensive residuals is taken as the validity score at that candidate position, and normalized weighted according to the position weights obtained by the constraint superposition of each grid in the candidate area to obtain the spatial constraint validity factor of the aircraft. This factor is located in the range of 0 to 1, and increases when the residual decreases; the position weights used are determined before confidence calculation and probability smoothing.

[0113] Anomaly penalties are determined based on the occurrence of track interruptions, ADS-B silent trigger backoff circuit breakers, and suspected anomalous spoofing tracks. Track interruptions are determined based on corresponding interruption records; ADS-B silent trigger backoff circuit breakers refer to a process where, after a continuous failure to obtain usable ADS-B data and the corresponding triggering conditions are met, the system switches to a state where the use of this data as a reliable positional basis is reduced or suspended; suspected anomalous spoofing tracks are determined based on corresponding markers in the track anomaly identification results. Anomaly penalties are obtained based on the occurrence of the above anomalies and corresponding penalty rules. As the anomaly penalty increases, the calculated position back-inference confidence decreases. Suspected spoofing track markers are generated based on the verification results of sudden velocity changes, position jumps, and motion characteristics exceeding the typical motion envelope of a carrier-based aircraft; when multiple verification indicators simultaneously exceed preset thresholds, corresponding markers are generated, and the track is removed from the valid input of this round. The magnitude of the anomaly penalty is zero or a finite positive value, and its negative contribution is reflected by the aforementioned minus sign.

[0114] For example, to illustrate the confidence score calculation, assume the attribution matching score, track quality factor, spatial constraint effectiveness factor, and anomaly penalty are 0.9, 0.8, 0.7, and 0.5 respectively, with corresponding weighting coefficients of 0.4, 0.3, 0.3, and 0.2. The weighted result will be in the range of 0 to 1, and no changes to the values ​​are needed. The confidence score derived from the single-aircraft position is:

[0115] When multiple associated carrier-based aircraft participate in reverse thrust, the confidence level of each aircraft's individual position reverse thrust is calculated according to the above rules, and then the confidence levels of the effective carrier-based aircraft's individual position reverse thrust are weighted and fused. Let the number of effective carrier-based aircraft be... , and The index for a valid carrier-based aircraft takes a value from 1 to... . No. The confidence level of the single-aircraft position back-difference of a carrier-based aircraft is When the sum of the individual confidence scores is greater than zero, the normalized value of each individual confidence score is used as the fusion weight.

[0116] in, For the first The fusion weight of an effective carrier-based aircraft The overall position confidence score is calculated after multi-aircraft fusion, for subsequent probability smoothing. All weights are non-negative and sum to 1. The higher the confidence score of a single aircraft, the greater its contribution to the overall confidence score. When there is only one valid carrier-based aircraft, the overall confidence score equals the confidence score of that aircraft. If the individual confidence scores of all valid carrier-based aircraft are zero, the overall confidence score is set to zero, and weight normalization is not performed. If there are no valid tracks or departure events that meet the requirements of this round of back-calculation, the process enters the evidence invalidation backoff branch.

[0117] For example, when the individual confidence levels of two effective carrier-based aircraft are 0.8 and 0.6 respectively, the normalization weights are 4 / 7 and 3 / 7 respectively, and the overall confidence level is approximately 0.714. This value is only used to illustrate the aggregation rules.

[0118] Location back-inference confidence is used to determine the processing intensity of location distribution, and the location probability of each geographic grid is obtained separately through location weight normalization.

[0119] Specifically, candidate location regions are mapped to geographic rasters, and the location weight of the corresponding raster is determined based on the superposition of multiple spatial candidate constraint regions transformed to the same reference time. As one implementation, the constraint weights provided by each spatial candidate constraint region covering the corresponding raster can be added together to obtain the location weight of that raster.

[0120] Establish a finite grid range to represent the location probability distribution, which includes the candidate location region and its surrounding region. Let it contain... The first geographic raster, the The position weight of each grid cell is: In the initial distribution, the position weights of the grid cells corresponding to the candidate location regions are obtained based on the overlay results, while the position weights of the remaining grid cells within the range are set to zero. When the sum of the position weights is greater than zero, normalization yields:

[0121] in, This indicates that the target ship was located at the reference time. The ADS-B location probabilities of each geographic raster are given, where the location probabilities of each raster are non-negative and their sum is 1.

[0122] Inversely infer confidence level from location With preset information threshold Comparison. When When the above normalized result is used as the ADS-B location probability distribution for this location update, then... At that time, the smoothing strength is determined based on the confidence level, the distribution is probabilistically smoothed, and the smoothing result is normalized again.

[0123] As an alternative probabilistic smoothing implementation, the initial position probability distribution can be mixed with a uniform distribution within the aforementioned finite grid range. For cases below a preset confidence threshold, the smoothing strength can be determined by the following formula:

[0124] in, To smooth the intensity adjustment parameters, The value of is greater than 0 and less than 1. All other things being equal, the lower the confidence level of the position deduction, the greater the smoothing strength. This function directly calculates the smoothing strength based on the difference between the confidence level and the threshold.

[0125] The determined smoothing intensity is used to process the position probability of each grid cell:

[0126]

[0127] in, The probability is uniformly distributed across all grid cells. To smooth out the values, This is the ADS-B location probability after re-normalization. By assigning probabilities within a grid area that includes the candidate region and its surroundings, low-confidence results can be represented with a more dispersed location distribution.

[0128] Under the same initial distribution and the same grid range, for a non-uniform initial distribution, as the smoothing intensity increases, the probabilities of each grid tend to become more uniform, and the peak probability decreases. In this embodiment, the uniformity of probability allocation characterizes the dispersion of the distribution, and the maximum value among the probabilities of each grid characterizes the peak probability. When the initial distribution is already uniform, it remains uniform after smoothing.

[0129] For example, suppose the grid range used for demonstration contains four grid cells, with initial position probabilities of 0.5, 0.5, 0, and 0, respectively. When the smoothing intensity is 0.2, the processed probabilities are 0.45, 0.45, 0.05, and 0.05, respectively; when the smoothing intensity increases to 0.6, the processed probabilities are 0.35, 0.35, 0.15, and 0.15, respectively. As the smoothing intensity increases, grid cells outside the original candidate positions receive more probabilities, and the peak probability decreases from 0.45 to 0.35, indicating an increase in uncertainty in the position deduction result. The above values ​​are only used to illustrate the smoothing process.

[0130] After completing the above processing, the target ship identifier, reference time, and processed ADS-B position probability distribution are associated and saved, and provided to subsequent position update steps, so that track association, data quality, and anomalies have a real impact on the reliability of the position back-inference results.

[0131] In addition to the normal ADS-B position backpropagation process, an evidence failure rollback branch can be set up. Each associated carrier-based aircraft is checked individually: if no valid ADS-B track points are received within a preset continuous observation period, or if the received track fails verification for multiple consecutive frames and both the track quality factor and spatial constraint validity factor are below the minimum valid threshold, the corresponding track is determined to be silent or failed. If the remaining valid tracks can still provide the multiple departure events required for this round of backpropagation, the remaining valid tracks continue to be used; if the valid tracks and departure events required for this position backpropagation cannot be obtained, the failed evidence path is closed, the previously formed local position distribution is stopped from constraining this position update, and this triggering situation is recorded as an abnormal event. In this case, evidence failure rollback is used, and position weights are no longer calculated based on the missing candidate position regions.

[0132] If subsequent ADS-B tracks that meet the minimum quality threshold are received consecutively, valid departure events can be stably extracted, and spatial constraint verification returns to normal, the use of the corresponding track is restored. After the set of valid tracks once again meets the event quantity and parameter requirements of this round of back-reasoning, the candidate location region and ADS-B location probability distribution are recalculated, and the evidence path is restored.

[0133] For implementations using a unified grid for multi-source fusion, the failed ADS-B location evidence can be set to a uniform distribution covering the grid area:

[0134] in, To ensure a uniform number of rasters within a given area, each raster has the same ADS-B location probability. When used for subsequent Bayesian fusion, this uniformity factor cancels out during normalization, preventing ADS-B evidence from providing a relative preference for different locations. At this point, location updates are then performed using available AIS location evidence, geographic priors formed from open-source information, and satellite imagery location evidence.

[0135] The evidence failure rollback and low confidence smoothing correspond to different data states: when there are available tracks and can form candidate location areas, the aforementioned confidence calculation and smoothing process is followed; when the tracks are completely invalid, the invalid evidence pathway is closed according to this branch or set to uniform distribution.

[0136] In one embodiment, the process of updating the target ship's position in step S107 is further explained. By combining the target ship's historical AIS track data, the ADS-B position probability distribution derived from the associated carrier-based aircraft tracks, satellite remote sensing imagery, and open-source information, the position probability is updated under a unified temporal and spatial reference. The corresponding position output result is determined based on whether there are conflicts between the position evidence.

[0137] In this embodiment, the imaging time of the satellite remote sensing imagery is selected as the reference time. The position extrapolation of historical AIS tracks and the position compensation of the aforementioned ADS-B spatial candidate constraints are both mapped to this reference time. A unified geographic grid covering various types of location evidence and related sea areas of activity is established, with each grid using the same coordinate reference and spatial scale. It is assumed that the unified grid contains... The first geographic raster, the The center coordinates of each grid are .

[0138] Based on the target ship's identification, historical AIS track data not used in this image candidate selection or ship-aircraft association is obtained. Valid track records prior to the reference time are selected, and the position, speed, heading, and corresponding time are read from these records. In one implementation of position extrapolation, uniform extrapolation is performed according to the velocity vector represented by the record to obtain the AIS extrapolated position at the reference time.

[0139] in, For the location of the selected AIS record, For the time of this record, For reference time, This is the velocity vector determined based on AIS speed and heading. With the positive horizontal axis pointing due east, the positive vertical axis pointing due north, and the heading angle measured clockwise from due north, the speed is... , heading angle is At that time, the two components of the velocity vector are respectively and .

[0140] Position weights are assigned around the AIS extrapolated position to represent historical positioning errors and the uncertainties introduced by position extrapolation. As an optional implementation, the weights of each grid cell can be constructed using a two-dimensional Gaussian form.

[0141] in, The position error scale uses the same unit of length as the coordinates. Its value is determined by considering the AIS's own positioning error, time registration error, and extrapolation uncertainty, and is also set in conjunction with historical track quality and extrapolation time intervals. The weights of each grid are normalized to obtain the AIS position probability distribution at the reference time:

[0142] For satellite remote sensing imagery, candidate locations of ship targets and corresponding detection confidence scores are obtained, and these candidate locations are converted into geographic coordinates. Detection confidence scores characterize the degree to which the corresponding candidate target is supported by ship detection results, while the location probability distribution represents the relative support level of the target ship in different geographical locations.

[0143] Before constructing the probability distribution of satellite image locations, a convolutional neural network trained on ship samples can be used to detect ship targets in satellite remote sensing images, obtaining the image location of the ship target and the detection confidence level. Combining image resolution and the identifiability of target features, identifiable features such as ship size, island location, deck outline, and wake are extracted and compared with corresponding features in existing archives or historical image records of the target ship to screen image candidates that may belong to the target ship.

[0144] When available AIS data is available, the AIS records can be verified against the existing identity information of the target ship based on the Maritime Mobile Service Identifier (MMSI). This verification, combined with the AIS location and ship activity information at the time of imaging, will check the spatiotemporal correspondence between the image candidates and the target ship. When available AIS data is lacking, corresponding candidates are retained based on identifiable ship features in the image and existing target activity records. If the identity cannot be uniquely determined, multiple possible candidates and their detection confidence levels are retained for subsequent fusion. Candidates entering the image location distribution construction process should have identification criteria corresponding to the target ship. AIS observations and open-source activity records used for this image candidate selection will no longer be used as independent inputs for this round of location fusion. The same information already used in the ship-aircraft association phase, as well as historical observations used to construct the departure location benchmark for this round, will not be repeatedly counted as independent location evidence for this round.

[0145] Ship detection confidence and target identity correspondence results are recorded separately: detection confidence reflects the degree of support for the image target belonging to the ship category, and identity correspondence results reflect the association between the candidate ship and the target ship monitored this time. Based on the selected candidate locations and corresponding detection confidence, the following satellite image location probability distribution is constructed.

[0146] As an optional implementation, a spatial weight distribution is constructed centered on each candidate location, and then weighted and superimposed using detection confidence. Let the candidate locations in the image participating in this location update be denoted as . The corresponding detection confidence level is Then the first The image location weights of a geographic raster can be expressed as:

[0147] in, The number of candidate positions. All values ​​are non-negative and at least one is greater than zero. A position error scale is set to integrate image positioning error and temporal registration error. Dividing by the sum of the image location weights of all grid cells yields the satellite image location probability distribution. .

[0148] The ADS-B location probability distribution uses the processing results obtained in the aforementioned embodiments. When the location back-inference confidence level reaches a preset confidence threshold, the distribution obtained by location weight normalization is used; when it is below the preset confidence threshold, the distribution after probability smoothing and re-normalization is used. This distribution is denoted as... And make it correspond to the same geographic raster as the AIS location probability distribution and the satellite image location probability distribution.

[0149] A priori geographic probability distribution was constructed based on the ship activity area, mission time, and formation information from open-source information not used in this image candidate selection or ship-aircraft association. Specifically, mission time was used to determine whether relevant activity records were applicable to the reference time, formation information was used to determine the correspondence between relevant activity records and target ships, and the ship activity area was used to determine the spatial range supported by the corresponding records.

[0150] In one prior weighting implementation, areas with valid activity records are assigned higher prior weights, while other feasible locations within the unified monitoring range are reserved with lower non-zero background weights. The prior weights of each grid are normalized to obtain the initial prior distribution. ,in, Non-zero background weights are retained to ensure that subsequent location evidence can provide updated information for locations not covered by open-source data. When independently usable open-source activity records are lacking, a uniform initial prior is used within the established unified monitoring grid, without reusing open-source information already processed in the pre-processing stage for weighting.

[0151] This example uses AIS location probability distribution, ADS-B location probability distribution, and satellite imagery location probability distribution as three types of location evidence. It differentiates the information used in preliminary candidate screening, ship-aircraft association, departure determination, and location fusion: the same AIS observation or open-source information already used in preliminary processing is no longer used to construct the location likelihood or geographical prior for this round of fusion. In this example, AIS location evidence is constructed when independently usable historical AIS observations are available; a uniform location benchmark is used when constructing each single-source location distribution, and independently usable open-source priors are uniformly introduced during fusion. A conditional independence approximation is used for the three types of evidence given the target location. Under this modeling approach, each single-source location distribution is used as a normalized location likelihood, and Bayesian probability updates are performed sequentially.

[0152] in, To introduce the first After the type of location evidence, the first The posterior probability of each grid cell; This represents the normalized position likelihood of this type of location evidence on the corresponding raster. In this embodiment, AIS, ADS-B, and satellite imagery location evidence are sequentially introduced, and after three updates, the posterior probability distribution of the position at the reference time is obtained. The denominator for each update is the sum of the unnormalized weights of all rasters.

[0153] The target ship's position update at the reference time is determined based on the posterior probability distribution of its position. As an optional implementation, the high-probability regions of each piece of position evidence are judged to be mutually exclusive based on their own position probability distribution. As another optional judgment method, a high-probability region for a certain type of evidence is formed by grouping grids with probabilities not lower than a preset proportion of their maximum grid probability. Let this proportion be... ,in Then the first The high-probability region of similar evidence consists of a grid that satisfies the following formula:

[0154] Compare high-probability regions of different pieces of evidence. When two high-probability regions do not share a common grid and the minimum spatial distance between them is greater than a preset conflict distance threshold, they are determined to be misaligned, and the corresponding two types of location evidence are marked as conflicting evidence. The minimum spatial distance is calculated based on the minimum distance between the centers of the grids contained in the two regions, and the conflict distance threshold is consistent with the coordinate units used. The error scale of each evidence source is determined by combining its positioning error and time registration error. When comparing two types of evidence, the conflict distance threshold is set based on the sum of the error scales of the two types of evidence with a safety margin.

[0155] Conflicting evidence was subjected to confidence attenuation. The confidence level of each piece of evidence corresponded to its source: the confidence level of ADS-B evidence was determined by back-calculating the aforementioned location; the confidence level of AIS evidence was determined based on historical track quality and extrapolated time; and the confidence level of satellite imagery evidence was determined based on the corresponding detection results. The confidence level of each piece of evidence was limited to the range of 0 to 1.

[0156] In one alternative attenuation implementation, let the original confidence level of the conflicting evidence be . According to the preset attenuation amount Obtain the confidence level after decay This ensures that the confidence level after attenuation is not lower than zero. The attenuation amount increases with the increase of spatial deviation between pieces of evidence according to a preset rule, and the smoothing intensity is determined based on the actual decrease in confidence level after truncation.

[0157] in, For the smoothing adjustment parameters corresponding to the evidence, For the same evidence, the greater the decrease in confidence, the greater the smoothing strength.

[0158] By employing the aforementioned uniform mixing method, the probability distribution of conflicting evidence within a unified grid is smoothed, meaning its probability is equal to that of each grid cell. The uniform distribution is mixed according to the smoothing intensity and then normalized again to obtain the updated conflict evidence. Thus, confidence decay reduces the degree of concentrated support for the original high-probability locations by changing the position probability of each grid.

[0159] The updated conflict evidence is used together with the remaining location evidence for refusion. The refusion preserves the initial prior distribution. Starting from the above Bayesian update formula, various types of evidence are introduced in sequence. Each type of evidence is used once in this round of calculation to obtain the recalculated posterior probability distribution of the location.

[0160] When conflicting evidence exists, arrange the geographic grids from high to low according to the newly obtained posterior probability, and select the top... There are several different grids, among which... , The preset number of candidate locations is output. The geographic coordinates corresponding to the center of the selected raster are used as multiple candidate locations, and their respective posterior probabilities and reference times are output together.

[0161] For example, to illustrate the location selection process, assuming the posterior probabilities of the three candidate rasters after re-fusion are 0.55, 0.40, and 0.05 respectively, and the preset output quantity is 2, the geographic coordinates corresponding to the centers of the first two rasters are output, and the posterior probabilities of 0.55 and 0.40 are associated with them respectively, to represent the two location candidates that are still retained.

[0162] When conflicting evidence exists, conflict alarms can be output along with multiple candidate locations, recording the source of conflicting location evidence, corresponding time, and confidence decay status, so that the location update results can reflect the situation where multiple sources of evidence have not formed a consistent location judgment.

[0163] When no conflicting evidence exists, the geographic grid with the highest probability is selected from the initially obtained posterior probability distribution, and the geographic coordinates corresponding to its center are determined as the updated position of the target ship at the reference time. If multiple grids have the same highest probability, this embodiment selects one of them according to the preset grid numbering order.

[0164] Through the above processing, the aircraft track back-inference results are mapped to the same reference time and geographic raster as AIS extrapolation, satellite image detection and open source activity information, and the centralized constraint effect is reduced when evidence conflicts occur, and the uncertainty in the location update results is expressed with multiple candidate locations.

[0165] Another embodiment of the present invention relates to a maritime ship monitoring system, such as Figure 2 As shown, the system may specifically include: The ship-aircraft association module 20 is used to acquire the Automatic Dependent Surveillance-Broadcast (ADS-B) track data of candidate aircraft, and determine the association relationship between the candidate aircraft and the target ship based on the ADS-B track data to obtain the associated carrier-based aircraft. The departure event determination module 21 is used to determine multiple departure events from the ADS-B track data of the associated carrier-based aircraft, wherein each departure event is characterized by a corresponding departure position, departure time and departure course. The spatial constraint construction module 22 is used to construct the spatial candidate constraint region of the target ship at the corresponding departure time based on the departure position and departure course corresponding to each departure event. The motion state estimation module 23 is used to estimate the motion state of the target ship based on the departure position and departure time corresponding to the multiple departure events; The position compensation module 24 is used to perform position compensation on the spatial candidate constraint regions corresponding to different departure times according to the motion state, and to convert each spatial candidate constraint region to the same reference time. The candidate region determination module 25 is used to superimpose multiple converted spatial candidate constraint regions and determine the candidate position region of the target ship based on the degree of overlap between the multiple spatial candidate constraint regions. The position update module 26 is used to update the position of the target ship at the reference time based on the candidate position region.

[0166] Optionally, in this embodiment, the ship-aircraft association module 20 includes: The first determining submodule is used to determine the attribution matching score between the candidate aircraft and each candidate ship, including the target ship, based on at least two types of characteristics among the candidate aircraft type, take-off and landing activity area, flight track round-trip characteristics and historical formation relationship. The second determination submodule is used to filter out shore-based aircraft tracks that are not associated with the candidate ships based on the attribution matching score. When the same candidate aircraft has an attribution conflict with multiple candidate ships and the highest attribution matching score can be uniquely determined, the corresponding candidate ship is determined as its associated ship. When multiple candidate ships have the same highest score, the track of the candidate aircraft is marked as an attribution pending track and is not used for subsequent position back-inference. Its associated ship will be determined after the attribution conflict is eliminated by new observation data. The third determination submodule is used to determine the candidate aircraft of the associated ship as the target ship as the associated carrier-based aircraft.

[0167] Optionally, in this embodiment, the spatial candidate constraint region is located within the corresponding distance constraint region; the distance constraint region consists of candidate positions of the target ship's center of mass with a distance from the corresponding departure position not greater than a preset distance threshold; the preset distance threshold is determined based on the distance from the target ship's center of mass to the end of the deck.

[0168] Optionally, in this embodiment, the motion state estimation module 23 includes: The ratio determination submodule is used to select two departure events where the departure point is approximately the same relative to the target ship but at different departure times. According to the uniform linear motion model, the ratio of the displacement vector between the departure positions corresponding to the two departure events to the time interval between the two departure times is used as the estimated value of the velocity vector of the target ship. The displacement vector points from the departure position corresponding to the earlier departure time to the departure position corresponding to the later departure time, and the time interval is the difference between the later departure time and the earlier departure time.

[0169] Optionally, in this embodiment, the position compensation includes translating the corresponding spatial candidate constraint region according to the position compensation amount, wherein the position compensation amount is the product of the velocity vector estimate and the time difference obtained by subtracting the corresponding departure time from the reference time; the degree of overlap is characterized by the cumulative constraint weight of the geographic grid, wherein the cumulative constraint weight is the sum of the constraint weights corresponding to each compensated spatial candidate constraint region covering the same geographic grid; the geographic grid with the cumulative constraint weight reaching the maximum value is divided into one or more connected regions according to spatial connectivity, and the one or more connected regions are determined as the candidate position regions of the target ship.

[0170] Optionally, in this embodiment, it further includes: The confidence determination module is used to, before updating the position of the target ship at the reference time, perform a weighted summation of the attribution matching score between each associated carrier-based aircraft and the target ship, the track quality factor of the corresponding ADS-B track data, and the corresponding spatial constraint validity factor, subtract the weighted value of the anomaly penalty term, and truncate the calculation result to the interval of 0 to 1 to obtain the single-aircraft position back-dive confidence of each associated carrier-based aircraft; and perform a weighted fusion of the single-aircraft position back-dive confidence of each associated carrier-based aircraft participating in this back-dive to obtain the position back-dive confidence. The first processing module is used to map the candidate location region to a geographic grid, determine the location weight of the corresponding geographic grid based on the superposition result of the multiple spatial candidate constraint regions, and normalize the location weight to obtain the ADS-B location probability distribution of the target ship at the reference time. The second processing module is used to perform probability smoothing on the ADS-B location probability distribution based on the location back-inference confidence level when the location back-inference confidence level is lower than a preset confidence threshold, and to normalize the smoothing result. Under the same initial distribution and grid range, the lower the location back-inference confidence level, the greater the smoothing intensity. The track quality factor is determined by normalization based on the sampling continuity of track points, the estimated positioning error, and the data update frequency; the spatial constraint effectiveness factor is determined by normalization based on the residuals of ship-aircraft geometric constraints and motion consistency constraints; the anomaly penalty term is a non-negative penalty magnitude, determined based on the occurrence of track interruption, ADS-B silent trigger backoff circuit breaker, and suspected abnormal deception track.

[0171] Optionally, the confidence determination module is also used to use the normalized value of the individual position back-inference confidence of each effectively associated carrier-based aircraft as the fusion weight; when the confidence of each individual aircraft is zero, the overall position back-inference confidence is set to zero. Tracks with pending attribution, suspected deception markers, or track quality below the minimum effective threshold are not included in this round of position back-inference.

[0172] Optionally, in this embodiment, the location update module 26 includes: The AIS position probability distribution determination submodule is used to extrapolate the position, speed and heading of the target ship based on the position, speed and heading of the historical Automatic Identification System (AIS) track data of the ship that were not used for the current image candidate screening or ship-aircraft association, so as to obtain the AIS position probability distribution at the reference time. The satellite image position probability distribution determination submodule is used to determine the satellite image position probability distribution at the reference time based on the candidate positions of ship targets in the satellite remote sensing image and the corresponding detection confidence. The prior geographic probability distribution construction submodule is used to construct the prior geographic probability distribution of the reference time based on the ship activity sea area, mission time and formation information in the open source information not used for this image candidate screening or ship-aircraft association. The geographic raster allocation submodule is used to use the AIS location probability distribution, the ADS-B location probability distribution and the satellite image location probability distribution as location evidence, and to map each location evidence to the same geographic raster along with the prior geographic probability distribution. The update submodule is used to use the prior geographical probability distribution as the initial prior distribution, perform Bayesian probability iteration update based on each location evidence, obtain the posterior probability distribution of the target ship's position at the reference time, and determine the position update result of the target ship at the reference time based on the posterior probability distribution of the position. Optionally, the update submodule includes: a first update position determination unit, configured to determine the corresponding position evidence as conflicting evidence when the high-probability regions of different position evidences deviate from each other, perform confidence attenuation on the conflicting evidence, and perform probability smoothing and normalization on its position probability distribution based on the attenuated confidence level to update the conflicting evidence; perform the Bayesian probability iteration update again based on the initial prior distribution and the updated position evidence, and select multiple candidate positions from high to low posterior probability according to the newly obtained position posterior probability distribution as the position update result of the target ship at the reference time; The second update position determination unit is used to determine the geographic coordinates corresponding to the geographic grid with the highest probability in the posterior probability distribution of the position as the updated position of the target ship at the reference time when there is no conflict evidence.

[0173] Optionally, the position update module is also used to set the ADS-B position evidence to a uniform distribution within a unified grid range when it is impossible to obtain the valid ADS-B tracks and departure events required for this round of position back-reasoning; after the track quality, departure event extraction and spatial constraint verification are restored to normal and meet the requirements of this round of back-reasoning, the ADS-B position probability distribution is recalculated to restore the evidence path.

[0174] Another embodiment of the present invention relates to an electronic device, such as Figure 3 As shown, it includes: at least one processor 301; and a memory 302 communicatively connected to the at least one processor 301; wherein the memory 302 stores instructions executable by the at least one processor 301, the instructions being executed by the at least one processor 301 to enable the at least one processor 301 to perform the maritime ship monitoring methods in the above-described method embodiments.

[0175] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0176] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0177] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0178] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0179] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0180] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process.Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0181] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for monitoring naval vessels at sea, characterized in that, include: Acquire the Automatic Dependent Surveillance-Broadcast (ADS-B) track data of the candidate aircraft, and determine the correlation between the candidate aircraft and the target ship based on the ADS-B track data to obtain the associated carrier-based aircraft; Multiple departure events are determined from the ADS-B track data of the associated carrier-based aircraft, wherein each departure event is characterized by its corresponding departure position, departure time, and departure course. Based on the departure position and departure course corresponding to each departure event, a spatial candidate constraint region for the target ship at the corresponding departure time is constructed. Based on the departure positions and departure times corresponding to the multiple departure events, the motion state of the target ship is estimated; Based on the motion state, position compensation is performed on the spatial candidate constraint regions corresponding to different departure times, and each spatial candidate constraint region is converted to the same reference time. The multiple converted spatial candidate constraint regions are superimposed, and the candidate position region of the target ship is determined based on the degree of overlap between the multiple spatial candidate constraint regions. The position of the target ship at the reference time is updated based on the candidate position region.

2. The method according to claim 1, characterized in that, Determining the association between the candidate aircraft and the target ship to obtain the associated carrier-based aircraft includes: Based on at least two of the following characteristics of the candidate aircraft: aircraft type, take-off and landing area, flight path round-trip characteristics, and historical formation relationship, the affiliation matching score between the candidate aircraft and each candidate ship, including the target ship, is determined. Based on the attribution matching score, shore-based aircraft tracks that are not associated with the candidate ships are filtered out. When the same candidate aircraft has an attribution conflict with multiple candidate ships and the highest attribution matching score can be uniquely determined, the corresponding candidate ship is identified as its associated ship. When multiple candidate ships have the same highest score, the track of the candidate aircraft is marked as an attribution pending track and is not used for subsequent position back-inference. Its associated ship will be determined after new observation data eliminates the attribution conflict. Candidate aircraft of the associated ship that are the target ship are identified as the associated carrier-based aircraft.

3. The method according to claim 1, characterized in that, The spatial candidate constraint region is located within the corresponding distance constraint region; The distance constraint region consists of candidate locations of the target ship's centroid that are no more than a preset distance threshold from their corresponding off-ship positions. The preset distance threshold is determined based on the distance from the target ship's center of mass to the end of the deck.

4. The method according to claim 1, characterized in that, The step of estimating the motion state of the target ship based on the departure positions and departure times corresponding to the multiple departure events includes: Two departure events are selected, whose departure points are approximately the same relative to the target ship but whose departure times are different. According to the uniform linear motion model, the ratio of the displacement vector between the departure positions corresponding to the two departure events to the time interval between the two departure times is used as the estimated value of the velocity vector of the target ship. The displacement vector points from the departure position corresponding to the earlier departure time to the departure position corresponding to the later departure time, and the time interval is the difference between the later departure time and the earlier departure time.

5. The method according to claim 4, characterized in that, The position compensation includes translating the corresponding spatial candidate constraint region according to the position compensation amount, wherein the position compensation amount is the product of the velocity vector estimate and the time difference obtained by subtracting the corresponding departure time from the reference time; The degree of overlap is characterized by the cumulative constraint weight of the geographic raster, which is the sum of the constraint weights corresponding to each compensated spatial candidate constraint region covering the same geographic raster. Geographic grids whose cumulative constraint weights reach the maximum value are divided into one or more connected regions according to spatial connectivity, and the one or more connected regions are determined as candidate location regions for the target ship.

6. The method according to claim 2, characterized in that, Before updating the position of the target ship at the reference time, the following steps are also included: The affiliation matching score between each associated carrier-based aircraft and the target ship, the track quality factor of the corresponding ADS-B track data, and the corresponding spatial constraint effectiveness factor are weighted and summed, the weighted value of the anomaly penalty term is subtracted, and the calculation result is truncated to the interval of 0 to 1 to obtain the single-aircraft position back-inference confidence of each associated carrier-based aircraft; the single-aircraft position back-inference confidence of each associated carrier-based aircraft participating in this back-inference is weighted and fused to obtain the position back-inference confidence. The candidate location regions are mapped to geographic grids, and the location weights of the corresponding geographic grids are determined based on the superposition results of the multiple spatial candidate constraint regions. The location weights are then normalized to obtain the ADS-B location probability distribution of the target ship at the reference time. When the confidence level of the location back-inference is lower than the preset confidence threshold, the probability distribution of the ADS-B location is smoothed according to the confidence level of the location back-inference, and the smoothing result is normalized. Under the same initial distribution and grid range, the lower the confidence level of the location back-inference, the greater the smoothing intensity. The track quality factor is determined by normalization based on the sampling continuity of track points, the estimated positioning error, and the data update frequency; the spatial constraint effectiveness factor is determined by normalization based on the residuals of ship-aircraft geometric constraints and motion consistency constraints; the anomaly penalty term is a non-negative penalty magnitude, determined based on the occurrence of track interruption, ADS-B silent trigger backoff circuit breaker, and suspected abnormal deception track.

7. The method according to claim 6, characterized in that, Updating the position of the target ship at the reference time based on the candidate position region includes: The position probability distribution at the reference time is obtained by extrapolating the position, speed, and heading of the target ship from its historical Automatic Identification System (AIS) track data that was not used in this image candidate selection or ship-aircraft association. Based on the candidate ship target positions in the satellite remote sensing images and the corresponding detection confidence levels, the probability distribution of the satellite image positions at the reference time is determined. Based on the ship activity area, mission time, and formation information in the open-source information not used for this image candidate selection or ship-aircraft association, a priori geographical probability distribution for the reference time is constructed. The AIS location probability distribution, the ADS-B location probability distribution, and the satellite image location probability distribution are used as location evidence, and each location evidence is mapped to the same geographic raster along with the prior geographic probability distribution. Using the prior geographical probability distribution as the initial prior distribution, Bayesian probability iteration is performed based on the evidence of each location to obtain the posterior probability distribution of the target ship's position at the reference time. The position update result of the target ship at the reference time is determined based on the posterior probability distribution of the position.

8. A maritime ship monitoring system, characterized in that, include: The ship-aircraft association module is used to acquire the Automatic Dependent Surveillance-Broadcast (ADS-B) track data of candidate aircraft, and determine the association relationship between the candidate aircraft and the target ship based on the ADS-B track data to obtain the associated carrier-based aircraft. The departure event determination module is used to determine multiple departure events from the ADS-B track data of the associated carrier-based aircraft, wherein each departure event is characterized by a corresponding departure position, departure time and departure course. The spatial constraint construction module is used to construct the spatial candidate constraint region of the target ship at the corresponding departure time based on the departure position and departure course corresponding to each departure event. The motion state estimation module is used to estimate the motion state of the target ship based on the departure position and departure time corresponding to the multiple departure events; The position compensation module is used to perform position compensation on the spatial candidate constraint regions corresponding to different departure times according to the motion state, and to convert each spatial candidate constraint region to the same reference time. The candidate region determination module is used to superimpose multiple converted spatial candidate constraint regions and determine the candidate position region of the target ship based on the degree of overlap between the multiple spatial candidate constraint regions. The position update module is used to update the position of the target ship at the reference time based on the candidate position region.

9. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, which, when executed by the processor, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images