Inland ship trajectory association method based on radar and AIS data fusion

Abstract

The application discloses a radar and AIS data fusion-based inland ship track correlation method, in the track correlation method in the radar and AIS fusion of the application, the distance between the radar track and the AIS track of the target ship is used as a coefficient matrix, through matrix transformation, trial assignment and the operation of the Nig cover theory, the optimal solution of track matching is finally found out, and the application has the advantages of high accuracy and strong reliability, and in the complex scene of multiple targets, the radar track and the AIS track of the ship are also accurately and reliably correlated. Through the exploration of the similarity between different tracks in the complex scene of multiple targets, the optimal matching scheme of the radar track and the AIS track of the ship is obtained, the AIS and the radar interconnection corresponding relationship of the ship in the navigation process is found out, the dynamic information fusion of the ship is completed, and the dynamic perception and analysis ability of the target ship in the inland navigation is improved.

Description

Technical Field

[0001] This invention relates to a method for fusing radar and AIS data and associating flight tracks, belonging to the field of digital technology for inland waterway shipping. Background Technology

[0002] In inland waterway transportation, radar and AIS are crucial means of target identification for vessels, and their information is complementary. AIS data provides accurate vessel position and attribute information, offering high navigation accuracy and being less affected by factors such as position and weather; however, AIS can only operate at very high frequencies and cannot identify vessels without AIS equipment. Radar can simultaneously track and survey both moving and stationary vessels within its detection area, but it has blind spots, and its echoes are easily affected by weather, sea conditions, and terrain obstruction, resulting in reduced resolution. Therefore, fully utilizing the fused information from radar and AIS is essential.

[0003] The key to radar and AIS data fusion lies in the correlation of target ship tracks, primarily based on the correlation between the target ship's position and motion state. Currently, commonly used methods for track correlation include weighted methods, modified methods, and nearest neighbor methods. In addition, there are detection methods such as likelihood ratio and multivariate hypothesis. However, in situations with dense, overlapping, or highly maneuverable ships, these methods often lead to many erroneous track correlations. Summary of the Invention

[0004] This invention proposes a method for associating inland waterway vessel trajectories based on radar and AIS data fusion. Its purpose is to overcome the technical challenges outlined in the background and solve the trajectory association problem in radar and AIS data fusion. The trajectory association method described in this invention primarily utilizes the distance between the target vessel's radar track and AIS track as a coefficient matrix. Through matrix transformation, trial assignment, and Nigel coverage theory, the optimal trajectory matching solution is ultimately found. This method boasts advantages such as high accuracy and strong reliability, achieving accurate and reliable association between vessel radar tracks and AIS tracks even in complex multi-target scenarios.

[0005] To achieve the above objectives, this invention provides the following technical solution: a method for associating inland waterway vessel trajectories based on radar and AIS data fusion, the implementation steps of which are as follows:

[0006] S1 Assume there are n radar data points {(x r1 ,y r1 ),(x r2 ,y r2 ),…(x rn ,y rn )} and n AIS data {(x a1 ,y a1 ),(xa2 ,y a2 ),…(x an ,y an The distance Q between each radar data point and the AIS data is calculated using the distance formula. ij :

[0007] Q ij =6371·arcos[cos(y ri )·cos(y aj )·cos(x ri -x aj )+sin(y ri )·sin(y aj )]

[0008] Where i = 1 to n; j = 1:n; x rn ,y rn These are the longitude and latitude of the nth radar data point, respectively; x an ,y an These are the longitude and latitude of the nth AIS data point, respectively.

[0009] S2 will form a matrix Q = (Qij)n consisting of distances Qij. ×n As the coefficient matrix for the global trajectory matching problem.

[0010] S3 defines a candidate pair as the ship to be matched in the radar data and each candidate ship in the AIS data, and defines a binary function Xij for each candidate pair:

[0011]

[0012] The mathematical model for trajectory matching is described as follows:

[0013]

[0014] It satisfies:

[0015] S4 transforms the coefficient matrix Q by first subtracting the minimum value of each row from each element, and then subtracting the minimum value of each column from each element. If a row or column in the matrix already contains 0, no further subtraction is needed.

[0016] S5 performs a trial assignment. For ease of description, each row of the coefficient matrix Q represents a different ship trajectory from radar data, and each column represents a different ship trajectory from AIS data. Starting with a row or column containing only one 0 element, the 0 element in that row or column is marked with @, indicating that the radar ship trajectory represented by that row is only associated with the AIS trajectory represented by that column or row. Then, all other 0s in the column or row containing @ are deleted, denoted as &, indicating that the AIS trajectory represented by this column or row has already been associated with radar, and other radar ship trajectories do not need to be considered. Similarly, the same operation is performed on other rows in the coefficient matrix Q containing only one 0 element and their corresponding columns, until all 0 elements in the matrix are deleted or marked.

[0017] After S4, if the coefficient matrix still contains 0 elements, then following the principle of "least selective elements first," starting from the row with the fewest remaining 0 elements, compare the number of 0 elements in the column or row containing each element in that row or column. Mark the 0 element in the column or row with the fewest 0 elements as @, and then delete all other 0 elements in the same row and column. Repeat this step until all 0 elements in the coefficient matrix Q have been marked or deleted.

[0018] S7 If the order of the coefficient matrix Q is equal to the number of elements marked as @, then the optimal solution is obtained; otherwise, proceed to step S8.

[0019] S8 utilizes the Nigel Coverage Theory and proceeds as follows:

[0020] a) Mark rows without the @ element with an X;

[0021] b) In the row marked with an "X", mark all columns containing 0 elements or including "&" with an "X".

[0022] c) In the column marked with an ×, mark the rows containing the @ element with an ×;

[0023] d) Repeat steps b) and c) until no × signs can be printed in the rows and columns of the matrix;

[0024] e) Finally, draw vertical lines on columns with an × sign in the coefficient matrix and horizontal lines on rows without an × sign.

[0025] f) Suppose we obtain k lines. If k is the same as the order of the matrix, then return to step 5; if k is less than the order of the matrix,

[0026] Then continue with the following steps;

[0027] S9 finds the minimum value among the elements in the matrix not covered by straight lines, subtracts this value from each element in the row marked with an "X", and adds this value to each element in the column marked with an "X", resulting in a new matrix. S5-S7 are then applied to the new matrix in the same manner. If n independent zero elements are obtained, the optimal solution is found; otherwise, return to S8 and repeat the process until the optimal solution is obtained, indicating that the AIS and radar data representing these zero elements match.

[0028] Compared with the prior art, the present invention has the following advantages: The track association algorithm for fusion of radar data and AIS data described in the present invention obtains the optimal matching scheme between ship radar track and AIS track by exploring the similarity between different tracks in complex multi-target scenarios, finds the correspondence between AIS and radar interconnection during the ship's navigation, completes the dynamic information fusion of the ship, and improves the dynamic perception and identification capability of target ships in inland waterway navigation. Detailed Implementation

[0029] The present invention will be described in detail below with reference to the embodiments.

[0030] The implementation process of the technical solution in Embodiment 1 is as follows:

[0031] 1. Assume there are 4 radar data points.

[0032] {(106.6635,29.5975),(106.6651,29.5975),(106.6584,29.5976),(106.6584,29.5963)}

[0033] and 4 AIS data

[0034] {(106.6609,29.5997),(106.6646,29.5977),(106.6603,29.5971),(106.6654,29.5996)}, calculate the distance between each radar data point and the AIS data using the distance formula:

[0035]

[0036] 2. Set Q = (Q ij ) 4*4 As the coefficient matrix for the global trajectory matching problem.

[0037] 3. The ship to be matched in the radar data and each candidate ship in the AIS data are referred to as a candidate pair. A binary function X is defined for each candidate pair. ij :

[0038]

[0039] The mathematical model for trajectory matching is described as follows:

[0040]

[0041] It satisfies:

[0042] 4. Transform the coefficient matrix Q by first subtracting the minimum value of each row from each element, and then subtracting the minimum value of each column from each element. If a row or column in the matrix already contains 0, no further subtraction is needed.

[0043]

[0044] 5. Perform trial assignment. For ease of description, each row of the matrix represents different ship trajectories in radar data, and each column represents different ship trajectories in AIS data. First, start with the row (or column) containing only one 0 element, marking the 0 element in that row as @, indicating that the radar trajectory represented by that row is only associated with the AIS trajectory represented by that column. Then, remove the other 0s from the column containing @, marking it as &, indicating that the AIS trajectory represented by this column has already been associated with radar, and other radar ship trajectories do not need to be considered. Repeat this operation; the result is:

[0045]

[0046] 6. After step 5, if the matrix still contains 0 elements, then, following the principle of "least selective elements first," start with the row with the fewest remaining 0 elements. Compare the number of 0 elements in the column containing each element in that row. Mark the 0 element in the column with the fewest 0 elements as @, and then delete all other 0 elements in the same row and column. Repeat this step until all 0 elements in the matrix have been marked or deleted. The result is:

[0047]

[0048] 7. The number of marked @ elements in the matrix is ​​4, and the matrix order is also 4, thus yielding the optimal solution: the first radar track matches the first AIS track, the second radar track matches the second AIS track, and the third...

[0049] The radar track matches the fourth AIS track, and the fourth radar track matches the third AIS track. The implementation process of this second embodiment is as follows:

[0050] 1. Assume there are 5 radar data points {(106.6584,29.5987),(106.6651,29.5975),(106.6584,29.5976),(106.6573,29.5971),(106.6684,29.5978)} and 5 AIS data points {(106.6584,29.5978),(106.6684,29.5971),(106.6603,29.5971)(106.6573,29.5975),(106.6651,29.5987)}. Calculate the distance between each radar data point and the AIS data point using the distance formula:

[0051]

[0052] 2. Set Q = (Q ij ) 5*5 As the coefficient matrix for the global trajectory matching problem.

[0053] 3. The ship to be matched in the radar data and each candidate ship in the AIS data are referred to as a candidate pair. A binary function X is defined for each candidate pair. ij :

[0054]

[0055] The mathematical model for trajectory matching is described as follows:

[0056]

[0057] It satisfies:

[0058] 4. Transform the coefficient matrix Q by first subtracting the minimum value of each row from each element, and then subtracting the minimum value of each column from each element. If a row or column in the matrix already contains 0, no further subtraction is needed.

[0059]

[0060] 5. Perform trial assignment. For ease of description, each row of the matrix represents different ship trajectories in radar data, and each column represents different ship trajectories in AIS data. First, start with the row (or column) containing only one 0 element, marking the 0 element in that row as @, indicating that the radar trajectory represented by that row is only associated with the AIS trajectory represented by that column. Then, remove the other 0s from the column containing @, marking it as &, indicating that the AIS trajectory represented by this column has already been associated with radar, and other radar ship trajectories do not need to be considered. Repeat this operation; the result is:

[0061]

[0062] 6. After step 5, if the matrix still contains 0 elements, then, following the principle of "least selective elements first," start with the row with the fewest remaining 0 elements. Compare the number of 0 elements in the column containing each element in that row. Mark the 0 element in the column with the fewest 0 elements as @, and then delete all other 0 elements in the same row and column. Repeat this step until all 0 elements in the matrix have been marked or deleted. The result is:

[0063]

[0064] 7. The number of @ elements marked in the matrix above is 4, while the matrix order is 5, which are not equal. Perform the following steps.

[0065] 8. Using Nigel's Coverage Theory, proceed as follows:

[0066] g) Mark lines without the @ element with an X;

[0067] h) In the row marked with an ×, mark all columns containing 0 elements (including &) with an ×;

[0068] i) In the column marked with an ×, mark the rows containing the @ element with an ×;

[0069] j) Repeat steps b) and c) until no × signs can be printed in the rows and columns of the matrix;

[0070] k) Finally, draw vertical lines on the columns marked with an × in the coefficient matrix and horizontal lines on the rows without an ×. The result is:

[0071]

[0072] l) If four lines are obtained, which are less than the matrix order, continue with the following steps;

[0073] 9. Find the minimum value among the elements in the matrix that are not covered by a line. Subtract this value from each element in the row marked with an "×".

[0074] Adding this value to each element in the column marked with an × results in a new matrix:

[0075]

[0076] Repeat steps 5-7 to process the new matrix in the same way, and obtain the result:

[0077]

[0078] 10. The processed new matrix contains 5 @ symbols, which means the optimal solution is obtained: the first radar trajectory matches the first AIS trajectory, the second radar trajectory matches the fifth AIS trajectory, the third radar trajectory matches the fourth AIS trajectory, the fourth radar trajectory matches the third AIS trajectory, and the fifth radar trajectory matches the second AIS trajectory.

Claims

1. A method for associating inland waterway vessel trajectories based on radar and AIS data fusion, characterized in that, The implementation steps of this method are as follows: S1 assumes there are n radar data points. and n AIS data The distance between each radar data point and the AIS data is calculated using a distance formula. : ； in ; These are the longitude and latitude of the nth radar data point, respectively. These are the longitude and latitude of the nth AIS data entry; S2 will be far The matrix formed As the coefficient matrix for the global trajectory matching problem; S3 refers to each candidate ship in the radar data and each candidate ship in the AIS data as a candidate pair, and defines a binary function for each candidate pair. : ； The mathematical model for trajectory matching is described as follows: ； It satisfies: ; ; S4 on the coefficient matrix To perform the transformation, first subtract the minimum value of each row from each element, and then subtract the minimum value of each column from each element. If a row or column in the matrix already contains 0, no further subtraction is needed. S5 performs trial assignment: The coefficient matrix is... Each row represents different ship trajectories in radar data, and each column represents different ship trajectories in AIS data. Starting with a row or column containing only a single 0 element, mark the 0 element in that row or column as @, indicating that the radar ship trajectory represented by that row is only associated with the AIS trajectory represented by that column or row. Then, remove the other 0s from the column or row containing @, denoted as &, indicating that the AIS trajectory represented by this column or row has been associated with radar. Then, convert the coefficient matrix... Perform the same operation on the other rows and their corresponding columns that contain only one 0 element, until all 0 elements in the matrix have been deleted or marked. After step S4, if the coefficient matrix still contains 0 elements, then following the principle of "least selective elements take precedence," starting from the row with the fewest remaining 0 elements, compare the number of 0 elements in the column or row containing each element in that row or column. Mark the 0 element in the column or row with the fewest 0 elements as @, and then delete all other 0 elements in the same row and column. Repeat step S6 until the coefficient matrix... All zero elements are marked or deleted; S7 If the coefficient matrix If the order of the denoted element is equal to the number of elements marked with @, then the optimal solution is obtained; otherwise, proceed to step S8. S8 utilizes the Nigel Coverage Theory and proceeds as follows: a) Mark lines without the @ element with an X; b) In the row marked with an "X", mark all columns containing 0 elements or including "&" with an "X". c) In the column marked with an ×, mark the rows containing the @ element with an ×; d) Repeat steps b) and c) until no × signs can be printed in the rows and columns of the matrix; e) Finally, draw vertical lines on columns with an × sign in the coefficient matrix and horizontal lines on rows without an × sign. f) Suppose we get k lines. If k is the same as the order of the matrix, then return to S5; if k is less than the order of the matrix, then continue with the following steps. S9 finds the minimum value among the elements in the matrix that are not covered by a straight line, subtracts the value from each element in the row marked with an ×, and adds the value to each element in the column marked with an ×, to obtain a new matrix. Repeat the same process for the new matrix according to S5-S7. If n independent 0 elements are obtained, the optimal solution is obtained; otherwise, return to S8 and repeat the process until the optimal solution is obtained. This means that the AIS and radar data of these 0 elements match the ship represented by them.

Images