Satellite internet-based ship-based radar and AIS data fusion method and device

Abstract

The application discloses a ship-based radar and AIS data fusion method and device based on a satellite internet, and relates to the technical field of ship information. The low-orbit satellite is used to collect the self motion, AIS and radar data of multiple ships in real time, a first-order linear ship motion model and an adaptive extended Kalman filter are used for target track tracking and data filtering, an error model containing a radial error coefficient, an azimuth error and an effective detection distance is established for the dynamic system error of the shipborne radar, and the radar-AIS target matching and error correction are realized based on an adaptive differential evolution algorithm and a Hungarian algorithm, and finally, the multi-hypothesis tracking method is used for fusion processing of unknown targets perceived by multiple nodes to form a globally unified target observation result. The method can effectively associate the shipborne radar data and the AIS data, realize real-time monitoring of ship targets in a wide sea area, and provide reliable technical support for maritime supervision, safety early warning and track backtracking.

Description

A method and apparatus for fusing shipborne radar and AIS data based on satellite internet Technical Field

[0001] This invention relates to the field of ship information technology, and in particular to a method and apparatus for fusing ship-based radar and AIS data based on satellite internet. Background Technology

[0002] With the rapid development of low-Earth orbit (LEO) satellite internet, it has become increasingly possible to collect shipborne Automatic Identification System (AIS) data and ship-based navigation radar data over large sea areas with low latency by deploying LEO satellite communication equipment on ships and then centrally processing and analyzing the data in a data center. However, how to fuse massive amounts of multi-source, wide-area shipborne AIS target data and radar target data to achieve high-precision and robust target perception and unified observation presents new difficulties and challenges for the design of data fusion methods.

[0003] AIS is an automatic positioning system widely used by ships. Ships equipped with AIS broadcast key information such as their call sign, position, speed, and course in real time at sea via VHF radio. Shipborne AIS data is mainly used for ship navigation. By collecting information about surrounding ships through the ship's AIS equipment, it is mainly used for the ship's own navigation and collision avoidance. At the same time, the data can also be received by shore-based AIS base stations.

[0004] Ship-based navigation radar is an indispensable device in ship navigation. It detects the surrounding environment, including other ships and obstacles at sea, by emitting electromagnetic waves and receiving their reflected signals. It primarily compensates for visual limitations, ensuring navigational safety and providing collision warnings in adverse weather or poor visibility conditions. With the continuous maturation of Automatic Radar Plotting (APRA) technology, this technology has been widely applied in marine navigation radar systems. ARA can analyze and process radar data in real time, identifying and marking maritime targets. However, in real-world scenarios, due to factors such as equipment performance, parameter settings, usage habits, and the working environment, radar detection suffers from dynamic systematic errors that change with operational status, leading to deviations in the detected target position, speed, and heading from the actual values.

[0005] Current radar and AIS data fusion methods primarily rely on shore-based radar and near-shore AIS data, while ship-based radar and AIS data fusion presents different characteristics and challenges. Compared to shipborne radar, shore-based radar offers higher stability and stronger detection capabilities, producing higher-quality detection data. Shipborne radar, limited by space and power supply on vessels, has relatively lower detection range and accuracy, resulting in data with lower stability and accuracy. Furthermore, different vessels exhibit varying deviations in target detection. However, shipborne radar offers advantages in maritime mobility and real-time performance. Additionally, due to the random distribution of numerous vessels, its sensing range can cover farther sea areas, and it can detect suspicious targets that have actively disabled their AIS equipment.

[0006] Therefore, for the massive amounts of multi-source, wide-area shipborne radar and AIS data collected based on satellite internet, the research focus of data fusion technology should be on correcting shipborne radar errors, matching shipborne radar detection targets with AIS targets, and unifying observation results from different ships for unknown targets that cannot be matched. Summary of the Invention

[0007] The technical problem to be solved by this invention is to provide a method and device for fusing shipborne radar and AIS data based on satellite internet, which addresses the characteristics of multi-source, wide-area, and massive shipborne radar and AIS data collected based on satellite internet. This method and device can effectively fuse multi-source shipborne radar data and AIS data to obtain globally unified target observation and matching results, and provide reliable data support for real-time tracking, monitoring, and track tracing of ship targets in wide-area maritime areas.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0009] A method and apparatus for fusing shipborne radar and AIS data based on satellite internet, comprising the following steps:

[0010] S1. For the continuous target traces recorded in radar data and AIS data, the ship motion is simplified into linear motion and a motion model is constructed. Based on the motion model, an adaptive extended Kalman filter (AEFK) single target tracking and data filtering method is constructed to obtain the stable real-time trajectory of each target.

[0011] S2. Considering the differences in radar equipment models, parameter settings, crew usage habits, working environments, and conditions on different ships, three dynamic system errors are set for the radar equipment on each ship. , respectively, radial error coefficient Azimuth error and effective detection range ;

[0012] S3. Data processing is performed on each sensing vessel as a unit. A real-time trajectory time span is set, and the real-time trajectory segments of the radar-detected targets on the sensing vessel and the real-time trajectory segments of AIS targets within a radius of the radar's furthest detection distance centered on the sensing vessel are extracted. The real-time trajectory segments are extracted as real-time feature points of the targets. The correlation problem between the two sets of target trajectories is transformed into a matching problem between two sets of point sets. Through the radar-AIS target matching and error correction method based on adaptive differential evolution JADE and Hungarian algorithm, the two sets of point sets are matched one-to-one. The optimal matching relationship and the corresponding system error are obtained. The dynamic system error of the sensing vessel's radar equipment is updated with this system error, and radar-detected targets that have not been matched with AIS targets in this process are recorded as unknown targets.

[0013] S4. Set the data fusion interval. During each data fusion interval, each sensing vessel completes one data processing step in parallel according to step S3 to obtain the matching relationship of targets around each sensing vessel and the unknown targets that have not been matched. Since the unknown targets lack identification information, there is a problem of repeated observation of unknown targets recorded by different sensing vessels. By using the multi-node distributed target sensing result fusion method based on multi-hypothesis tracking MHT, the unknown targets recorded by each sensing vessel are matched and merged to complete the fusion of multi-source wide-area massive ship-based radar and AIS real-time data.

[0014] The AEFK-based single-target tracking and data filtering method in step S1 includes the following steps:

[0015] S1-1. Process all AIS data collected by sensing vessels in a unified manner, using MMSI number as the basis for target differentiation, set an AEFK based on the ship motion model for each target, adaptively adjust the system noise covariance and observation error covariance, optimize the target state estimation and uncertainty quantification, filter out abnormal points and duplicate points that do not conform to the motion model and time series characteristics, and obtain effective target trajectory data collected by AIS.

[0016] S1-2. The radar detection data of each sensing vessel are processed separately. The target batch number is used as the basis for target differentiation. An AEFK based on the ship motion model is set for each target. At the same time, a dynamic smoothing window is set to record the continuous target measurement values, calculate the latitude and longitude changes and time intervals of the latest data, smooth the speed and heading of the real-time target track, and use the smoothed data as input into the AEFK. The system noise covariance and observation error covariance are adaptively adjusted to optimize the target state estimation and uncertainty quantification. Clutter points, non-target points and false target trajectories in radar detection are filtered out to obtain the effective target trajectory data detected by radar.

[0017] The radial error coefficient mentioned in step S2 This refers to the ratio of the actual measured radial distance to the true radial distance of the target, used to correct deviations in the target trajectory in the distance direction; the azimuth error... This refers to the difference in the horizontal angle measured by radar relative to the ship, used to correct for horizontal deviations in the target's trajectory; the effective detection range... This refers to the maximum range within which radar equipment can acquire effective target trajectory data;

[0018] In step S3, the detection range of the shipborne radar is usually greater than the range of data received by the shipborne AIS equipment. Therefore, the real-time trajectory segment of the AIS target observed uniformly within the radar coverage area is obtained from the AIS trajectory aggregated by the data center to ensure that it matches the radar's farthest detection range.

[0019] The features of the target real-time feature point mentioned in step S3 include velocity, heading, and the length and width of the trajectory. First, the latitude and longitude range of the real-time trajectory segment is calculated, and the midpoint of the latitude and longitude range is taken as the position of the feature point. The longitude range is the trajectory length, and the latitude range is the trajectory width. The average velocity of all points recorded within the real-time trajectory segment is the velocity feature v, and the average heading is the heading feature. ;

[0020] The radar-AIS target matching error correction method based on adaptive differential evolution JADE and the Hungarian algorithm described in step S3 includes the following steps:

[0021] S3-1. Define the objective function as based on the current system error parameters. After correction, the average weighted Euclidean distance between radar target feature points and AIS feature point matching pairs is used, with Euclidean distance weights set. for ,in For positional weights, For speed weights, For course weighting, For long trajectory weights, For trajectory width weighting;

[0022] Let the corrected radar target feature set be... ,in It is the first The feature points of each radar target, and the AIS feature point set are: ,in yes Feature points of each AIS target;

[0023] Corresponding radar system error parameters The matching relationship of point sets is represented by a matching matrix. express:

[0024]

[0025] in express and match, 0 represents and Mismatch;

[0026] Then the objective function The calculation formula is:

[0027]

[0028] S3-2. Based on the perceived position of the ship, the feature points of surrounding radar-detected targets are transformed into polar coordinates. Based on the system error parameters, the radial distance of the feature points is divided by the radial error coefficient. The azimuth angle of the feature point plus the azimuth error coefficient The corrected range will be within the radar's effective detection range. After removing the non-standard feature points, the corrected feature points are transformed into Cartesian coordinates to obtain the corrected radar target feature point set. ;

[0029] S3-3 Calculate the corrected radar target feature point set With AIS feature point set Weighted distance matrix ,by The cost matrix is ​​used to calculate the matching relationship based on the Hungarian algorithm. At the same time, the set of unknown targets that do not match the AIS feature points is obtained;

[0030] S3-4, Based on matching relationships Calculate the objective function obtained based on the current system error parameters. ;

[0031] S3-5, Based on the JADE algorithm, using radar system error To optimize the parameters, loops S3-2, S3-3, and S3-4 are performed to optimize the objective function. Adjusting the descent direction to correct radar system error Set the number of loops The stopping condition is to minimize the objective function. ,in To find the optimal radar error parameters that minimize the objective function, retain... Corresponding matching relationship .

[0032] Step S4, the multi-node target perception result fusion method based on multi-hypothesis tracking MHT, includes the following steps:

[0033] S4-1. Cluster the unknown targets from all sensing vessels according to their location to divide them into different MHT regions. Each MHT region must meet the constraint that the Euclidean distance between targets does not exceed a preset threshold. For each independent MHT region, establish a local multi-hypothesis tracker to fuse the sensing results within the region.

[0034] S4-2. Within an MHT region, using the previous time step's unknown target perception results as the target set to be matched and the current time step's unknown target perception results as the observation set, initialize all global matching hypotheses between the target set to be matched and the current time step's observation set, denoted as independent targets. False targets and matching existing targets ,in This indicates that the unknown target and a target in the measurement set are the same target;

[0035] S4-3. Calculate the sum of the confidence scores of all global matching hypotheses in the region. Set fixed values ​​for independent targets and false targets. For matching targets, calculate the weighted Euclidean distance as the confidence score and retain the global matching hypothesis with the highest confidence score.

[0036] S4-4. Based on the global matching hypothesis with the highest confidence at the current time in this region, the hypothesis of independent target is not processed, the hypothesis of false target is removed, and the hypothesis of matching an existing target is merged into a single target. The processed result is then output.

[0037] S4-5. Based on the parallel calculations of S4-2, S4-3, and S4-4 above, the fused observation results of unknown targets in each MHT region are obtained. Combined with the observation results of known targets, the wide-area unified target observation results at the current moment are obtained.

[0038] A computer-readable storage medium having a computer program stored thereon, characterized in that, when executed by a processor, the program implements the steps of the satellite internet-based shipborne radar and AIS data fusion method as described above.

[0039] The present invention has the following beneficial effects:

[0040] This invention leverages the high-speed communication capabilities of satellite internet to utilize randomly distributed cooperative vessels across a vast sea area as sensing nodes. These vessels collect shipborne AIS and radar data, which are then uploaded in real-time to a land-based data center. The proposed method fuses this sensing data from multiple nodes, overcoming detection errors in shipborne radar data, correcting the original radar detection data, and achieving real-time correlation and matching between radar-detected target trajectories and AIS target trajectories. This results in a unified target observation result covering a vast sea area, enabling all-weather, all-round marine target monitoring. Attached Figure Description

[0041] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. In the drawings:

[0042] Figure 1 is a simplified flowchart of the steps of a method for fusing ship-based radar and AIS data based on satellite internet according to the present invention.

[0043] Figure 2 is a detailed flowchart of the steps of the method for fusing ship-based radar and AIS data based on satellite internet according to the present invention.

[0044] Figure 3 is a flowchart of the radar-AIS target matching and error correction method based on adaptive differential evolution JADE and Hungarian algorithm in this invention.

[0045] Figure 4 is a flowchart of the steps of the multi-node target perception result fusion method based on multi-hypothesis tracking MHT in this invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0047] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0048] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0049] To address the challenge of fusing and processing multi-source, heterogeneous, wide-area, and massive target perception data collected via low-Earth orbit satellite internet, this invention provides a method for fusing ship-based radar and AIS data based on satellite internet, as shown in Figure 1. This method includes the following steps S1-S4:

[0050] S1. For the continuous target traces recorded in radar data and AIS data, the ship motion is simplified into linear motion and a motion model is constructed. Based on this motion model, an adaptive extended Kalman filter single target tracking and data filtering method is constructed to obtain the stable real-time trajectory of each target.

[0051] S2. Set three dynamic system errors for each ship's radar equipment: radial error coefficient, azimuth error, and effective detection range.

[0052] S3. Data is processed on a per-ship basis. Real-time trajectory segments of radar targets and trajectory segments of surrounding AIS targets are extracted and transformed into a point set matching problem. Then, one-to-one matching is performed using adaptive differential evolution and the Hungarian algorithm to obtain the optimal matching relationship and system error. The radar dynamic error is updated, and unmatched radar targets are recorded as unknown targets.

[0053] S4. Set the data fusion interval time, and each sensing vessel processes the data in parallel to obtain the target matching relationship and unknown targets of different sensing vessels at the current moment. Then, use the multi-hypothesis tracking method to perform distributed target fusion, eliminate redundancy, and realize wide-area fusion of multi-source radar and AIS data.

[0054] Figure 2 shows a detailed flowchart of a method for fusing ship-based radar and AIS data based on satellite internet.

[0055] First, data from each sensing vessel is transmitted back to a land-based data center in real time via low-orbit satellite internet. Then, real-time data is obtained from the land-based data center, including the vessel's own motion data, shipborne AIS data, and shipborne radar detection data.

[0056] Build longitude-based structures for all ships ,latitude ,speed ,course A uniform linear target ship motion model in four states, where Indicates the current moment. Indicates the next moment, For time intervals, Given the Earth's average radius, its state update formula is... as follows:

[0057]

[0058] Construct an AEFK based on a ship motion model and calculate the state update formula. The Jacobian matrix as the prediction matrix The AIS and radar measurements have a linear relationship with the state vector, and a measurement model is set. If AEFK is the identity matrix, then the update formula for AEFK is as follows:

[0059]

[0060] In the formula Output the target state value for AEFK. for Predicted state at any given time for The target state measurement value at time t. In order to be in The covariance matrix of the state vector at time step, Let be the covariance matrix of the state recursion noise. 'For in Kalman gain of the state vector at time step. To measure the covariance matrix of the noise, and Adaptive adjustment is achieved through the Sage-Husa method.

[0061] For shipborne AIS data, all data from sensed ships are processed uniformly, and the MMSI number is used as the basis for target differentiation. An AEFK based on the ship motion model is set for each target, and abnormal and repetitive points that do not conform to the motion model and time sequence characteristics are filtered out to obtain the target trajectory data of all AIS.

[0062] For shipborne radar data and ship motion data, parallel processing is performed on a per-ship basis. The target batch number in the radar data is used as the basis for target differentiation. A dynamic smoothing window and an AEFK based on the ship motion model are set for each target. The smoothing window inputs the smoothed real-time target trace as the measurement value into the AEFK to filter out clutter points, non-target points and false target trajectories in the radar detection, and finally obtains the target trajectory data detected by the ship's radar.

[0063] Each sensing vessel enters a matching process in parallel, which includes the following steps:

[0064] For each radar device used to sense a ship, three corresponding dynamic system errors are obtained. , respectively, radial error coefficient Azimuth error and effective detection range .

[0065] The time span of the real-time trajectory is set, the position of the sensing vessel is obtained from its own motion data, the AIS target trajectory within the range is extracted from the AIS trajectories of all vessels with the maximum detection range of the radar equipment as the center, and together with the radar detection target trajectory of the sensing vessel, two sets of real-time trajectory segments to be associated are formed.

[0066] The two sets of real-time target trajectory segments to be associated are extracted as real-time target feature points.

[0067] With the current dynamic system error The parameters are input into the radar-AIS target matching and error correction method based on adaptive differential evolution JADE and Hungarian algorithm, and the two sets of points are matched one-to-one.

[0068] During this matching process, the optimal matching relationship and the current optimal system error are retained. The optimal matching relationship is recorded in the unified observation of the matched target in the global unified fusion observation, and the optimal system error is used as the new system error of the shipborne radar. At the same time, unknown targets that have not been matched with the AIS target in the radar's detected targets are recorded.

[0069] The unknown target perception results from all sensing vessels in the matching process are obtained. The matching and merging of all unknown targets are completed by the multi-node target perception result fusion method based on multi-hypothesis tracking MHT, and recorded in the unified observation of unknown targets in the global unified fusion observation.

[0070] Figure 3 shows the flowchart of the radar-AIS target matching and error correction method based on adaptive differential evolution JADE and Hungarian algorithm.

[0071] Based on the current system error of the radar, the process enters the adaptive differential evolution algorithm to minimize the objective function.

[0072] Initialize the JADE parameter vector population, and initialize the mutation factor F and crossover probability CR for each parameter vector individual.

[0073] Each individual parameter vector from the JADE parameter vector population is extracted as a radar system error and fed into the Hungarian algorithm to calculate the matching relationship.

[0074] The corrected radar target feature point set is obtained by using the radar system error correction input as the radar target feature point set.

[0075] The weighted Euclidean distance between the points is calculated between the corrected radar target feature point set and the input AIS target feature point set to obtain the cost matrix D of these two point sets.

[0076] The cost matrix is ​​added to the virtual point balance to form a square matrix, and then input into the Hungarian algorithm to calculate the point set matching relationship with the individual parameter vector as input.

[0077] The point set matching relationship corresponding to each parameter vector individual in the population is calculated according to the above process, and the corresponding objective function value is calculated based on the point set matching relationship.

[0078] For each individual parameter vector, individual mutation is performed based on the mutation factor F to generate a mutated parameter vector, and then individual crossover is performed based on the crossover probability CR to generate an experimental parameter vector.

[0079] Based on the above Hungarian algorithm for calculating the matching relationship, the radar system error with the experimental parameter vector as input is used to calculate the target function value corresponding to the experimental parameter vector.

[0080] Compare the objective function value of the original parameter vector individual with the objective function value of the experimental parameter vector, select the individual with the smaller objective function value to enter the next generation population, and record the F and CR of the individual.

[0081] Finally, the mutation, crossover, and selection steps of all individual parameter vectors are completed, the dynamic system error parameter population is updated, and the new F and CR of each individual are adaptively updated based on the F and CR of all individuals in the population.

[0082] By iterating N times according to the above steps, the process of minimizing the objective function of the adaptive differential evolution algorithm is completed. The individual with the parameter vector that minimizes the objective function in the population is taken as the optimal system error parameter output of this method.

[0083] The point set matching relationship corresponding to the optimal system error is obtained, as well as the unknown targets that failed to complete the matching in the radar target point set.

[0084] Figure 4 shows the flowchart of the multi-node target perception result fusion method based on multi-hypothesis tracking MHT.

[0085] Unknown targets from all sensing vessels are clustered into different MHT regions based on their location. Each MHT region must satisfy the constraint that the Euclidean distance between targets does not exceed a preset threshold. For each independent MHT region, a local multi-hypothesis tracker is established to fuse the sensing results within the region.

[0086] Within an MHT region, the unknown target perception results of the previous time step are used as the target set to be matched, and the unknown target perception results of the current time step are used as the observation set. All global matching hypotheses between the target set to be matched and the observation set of the current time step are initialized as independent target, false target, and matched existing target, respectively, where it means that the unknown target and a certain target in the measurement set are the same target.

[0087] Calculate the sum of the confidence scores of all global matching hypotheses in the region. Set fixed values ​​for independent targets and false targets. For matching targets, calculate the weighted Euclidean distance as the confidence score and retain the global matching hypothesis with the highest confidence score.

[0088] Based on the global matching hypothesis with the highest confidence at the current time in this region, the hypothesis of independent target is not processed, the hypothesis of false target is removed, and the hypothesis of matching an existing target is merged into a single target. The processed result is then output.

[0089] Based on the above process, the fused observation results of unknown targets in each MHT region are calculated in parallel. Combined with the observation results of known targets, the wide-area unified target observation results at the current moment are obtained.

[0090] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of the invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave.

[0091] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0092] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0093] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for fusing shipborne radar and AIS data based on satellite internet, characterized in that, The method includes the following steps: S1. For the continuous target traces recorded in radar data and AIS data, the ship motion is simplified into linear motion and a motion model is constructed. Based on this motion model, an adaptive extended Kalman filter (AEFK) single-target tracking and data filtering method is constructed to obtain the stable real-time trajectory of each target; S2. Considering the differences in radar equipment models, parameter settings, crew usage habits, working environment, and status of different ships, three dynamic system errors are set for the radar equipment of each ship. , respectively, radial error coefficient Azimuth error and effective detection range S3. Data processing is performed on a per-vessel basis. A real-time trajectory time span is set, and the real-time trajectory segments of targets detected by the radar on the sensing vessel, as well as the real-time trajectory segments of AIS targets within a radius centered on the sensing vessel and defined by the radar's furthest detection distance, are extracted. These real-time trajectory segments are then extracted as real-time target feature points. The correlation problem between the two sets of target trajectories is transformed into a matching problem between two sets of point sets. Using a radar-AIS target matching and error correction method based on Adaptive Differential Evolution (JADE) and the Hungarian algorithm, the two sets of point sets are matched one-to-one. The optimal matching relationship and the corresponding system error are obtained, and this system error is used to update the dynamics of the sensing vessel's radar equipment. System errors are recorded, and radar-detected targets that did not match AIS targets during this processing are recorded as unknown targets; S4, set the data fusion interval time. During each data fusion interval, each sensing vessel completes one data processing in parallel according to step S3, and obtains the matching relationship of targets around each sensing vessel and the unknown targets that did not match. Since the unknown targets lack identification information, there is a problem of repeated observation of unknown targets recorded by different sensing vessels. By using the multi-node distributed target sensing result fusion method based on multi-hypothesis tracking MHT, the unknown targets recorded by each sensing vessel are matched and merged to complete the fusion of multi-source wide-area massive ship-based radar and AIS real-time data.

2. The method for fusing ship-based radar and AIS data based on satellite internet according to claim 1, characterized in that, The single-target tracking and data filtering method based on AEFK in step S1 includes the following steps: S1-1, uniformly process the AIS data collected by all sensing vessels, using the MMSI number as the target distinction criterion, set an AEFK based on the ship motion model for each target, adaptively adjust the system noise covariance and observation error covariance, optimize the target state estimation and uncertainty quantification, filter out abnormal points and duplicate points that do not conform to the motion model and time series characteristics, and obtain effective target trajectory data collected by AIS; S1-2, process the radar detection data of each sensing vessel separately, using the target batch number as the target distinction criterion, set an AEFK based on the ship motion model for each target, set a dynamic smoothing window, record continuous target measurement values, calculate the latitude and longitude changes and time intervals of the latest data, smooth the speed and heading of the target's real-time points, and use the smoothed data as input into the AEFK, adaptively adjust the system noise covariance and observation error covariance, optimize the target state estimation and uncertainty quantification, filter out clutter points, non-target points and false target trajectories in radar detection, and obtain effective target trajectory data detected by radar.

3. The method for fusing ship-based radar and AIS data based on satellite internet according to claim 1, characterized in that, The radial error coefficient mentioned in step S2 This refers to the ratio of the actual measured radial distance to the true radial distance of the target, used to correct deviations in the target trajectory in the distance direction; the azimuth error... This refers to the difference in the horizontal angle measured by radar relative to the ship, used to correct for horizontal deviations in the target's trajectory; the effective detection range... It refers to the maximum range within which radar equipment can acquire effective target trajectory data.

4. The method for fusing ship-based radar and AIS data based on satellite internet according to claim 1, characterized in that, In step S3, the detection range of the shipborne radar is usually greater than the range of data received by the shipborne AIS equipment. The real-time trajectory segments of the AIS targets observed uniformly within the radar coverage area are obtained from the AIS trajectories aggregated by the data center to ensure that they match the radar's farthest detection range.

5. The method for fusing ship-based radar and AIS data based on satellite internet according to claim 1, characterized in that, The features of the target real-time feature points described in S3 include velocity. ,course Length of the trajectory Hekuan First, calculate the latitude and longitude range of the real-time trajectory segment. Take the midpoint of the latitude and longitude range as the location of the feature point. The longitude range is the trajectory length, and the latitude range is the trajectory width. The average velocity of all points recorded within the real-time trajectory segment is the velocity feature v, and the average heading is the heading feature. 。 6. The method for fusing ship-based radar and AIS data based on satellite internet according to claim 1, characterized in that, The radar-AIS target matching and error correction method based on adaptive differential evolution JADE and Hungarian algorithm described in step S3 includes the following steps: S3-1, Define the objective function as the function based on the current system error parameters. After correction, the average weighted Euclidean distance between radar target feature points and AIS feature point matching pairs is calculated; the weights for the Euclidean distance are set. for: ,in For positional weights, For speed weights, For course weighting, For long trajectory weights, Let the trajectory width be weighted; let the corrected radar target feature point set be... ,in It is the first The feature points of each radar target, and the AIS feature point set are: ,in yes Feature points of each AIS target; corresponding radar system error parameters The matching relationship of point sets is represented by a matching matrix. express: ,in express and match, 0 represents and If there is no match, then the objective function... The calculation formula is: S3-2. Based on the perceived position of the ship, the feature points of surrounding radar-detected targets are transformed into polar coordinates. Based on the system error parameters, the radial distance of the feature points is divided by the radial error coefficient. The azimuth angle of the feature point plus the azimuth error coefficient The corrected range will be within the radar's effective detection range. After removing the non-standard feature points, the corrected feature points are transformed into Cartesian coordinates to obtain the corrected radar target feature point set. S3-3, Calculate the corrected radar target feature point set. With AIS feature point set Weighted distance matrix ,by The cost matrix is ​​used to calculate the matching relationship based on the Hungarian algorithm. Simultaneously, the set of unknown targets that do not match AIS feature points is obtained; S3-4, based on matching relationships Calculate the objective function obtained based on the current system error parameters. S3-5, Based on the JADE algorithm, using radar system error To optimize the parameters, loops S3-2, S3-3, and S3-4 are performed to optimize the objective function. Adjusting the descent direction to correct radar system error Set the number of loops The stopping condition is to minimize the objective function: ,in To find the optimal radar error parameters that minimize the objective function, retain... Corresponding matching relationship 。 7. The method for fusing ship-based radar and AIS data based on satellite internet according to claim 1, characterized in that, Step S4, the multi-node target perception result fusion method based on multiple hypothesis tracking (MHT), includes the following steps: S4-1, clustering unknown targets from all sensing vessels according to their location to divide them into different MHT regions. Each MHT region must satisfy the constraint that the Euclidean distance between targets does not exceed a preset threshold. For each independent MHT region, establish a local multiple hypothesis tracker and perform perception result fusion within the region; S4-2, within an MHT region, using the unknown target perception results from the previous time step as the set of targets to be matched and the unknown target perception results from the current time step as the set of observations, initialize all global matching hypotheses between the set of targets to be matched and the set of observations from the current time step, which are independent targets. False targets and matching existing targets ,in This indicates that the unknown target and a target in the measurement set are the same target; S4-3, calculate the sum of the confidence scores of all global matching hypotheses in the region, set fixed values ​​for independent targets and false targets, and calculate the weighted Euclidean distance as the confidence score between matching targets, and retain the global matching hypothesis with the highest confidence score; S4-4. Based on the global matching hypothesis with the highest confidence in the current time of this region, the hypothesis of independent targets is not processed, the hypothesis of false targets is removed, and the hypothesis of matching an existing target is merged into a single target. The processed result is then output. S4-5. Based on the above S4-2, S4-3, and S4-4, the fused observation results of unknown targets in each MHT region are calculated in parallel. Combined with the observation results of known targets, the wide-area unified target observation results at the current time are obtained.

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

Images