A ship lingering AIS trajectory quick simplification method

Abstract

The present application belongs to the cross field of maritime affairs, ship, shipping, data science, machine learning and artificial intelligence, and particularly relates to a ship lingering AIS trajectory rapid simplification method, which comprises: arranging and processing AIS data of a ship in time sequence to form an AIS trajectory of the ship; cutting the AIS trajectory to obtain a plurality of trajectory segments; performing self-intersection detection on each trajectory segment; constructing a trajectory directed graph for the trajectory segment with intersection points; and using a graph shortest distance solving algorithm to find out a simplified trajectory and output the same. The method provided by the present application can quickly remove and correct trajectory redundancy and disorder caused by lingering, can not change the trajectory form of other normal navigation segments, and can maximize the retention of useful information of the original trajectory; the constructed directed graph is a very sparse graph, and the operation speed is very efficient and fast, without the need of manually fine-tuning parameters, and is suitable for automatically and rapidly processing a large amount of trajectory data.

Description

TECHNICAL FIELD

[0001] The application belongs to the cross field of maritime affairs, ships, shipping, data science, machine learning and artificial intelligence, and particularly relates to a ship lingering AIS trajectory rapid simplification method. BACKGROUND

[0002] Ship AIS trajectory refers to a trajectory line reflecting the ship's movement path formed by connecting the ship's AIS information in chronological order in the geographic coordinate system. Ship lingering trajectory simplification is to correct the chaotic trajectory formed by the ship's lingering movement in a small range, remove the complex and redundant trajectory segments, and form a simple and intuitive trajectory, which involves the identification and extraction of ship lingering movement trajectory, the removal of redundant trajectory and the repair of missing trajectory segments.

[0003] Ship lingering refers to the ship's repeated movement in a small range within a period of time, such as back and forth, turning, and passing through, etc., and its AIS trajectory often presents characteristics such as cluster shape and dense overlap. Ship lingering movement may be due to the execution of a certain task requirement (such as maritime law enforcement), or it may be due to idling without specific tasks (such as waiting for port entry or operation). The chaotic and redundant characteristics of ship lingering trajectory can seriously interfere with subsequent tasks based on ship trajectory analysis, such as calculating effective range.

[0004] The current mainstream ship trajectory simplification methods mainly include the following:

[0005] (1) Douglas-Peucker algorithm: This is a recursive algorithm that can efficiently simplify curves by removing unimportant points while preserving the overall shape of the curve. It is based on a distance threshold, which removes points with a distance less than the set threshold from a straight line.

[0006] (2) Angle simplification method: This method is based on angle difference and can remove points with small angle changes, thus better preserving the turning points of the trajectory.

[0007] (3) Top-Down time slicing method: This method divides the trajectory into multiple segments at equal intervals according to time, and uses the Douglas-Peucker algorithm to simplify each segment. This method preserves the time characteristics and performs spatial simplification.

[0008] (4) Best fitting algorithm: This method uses statistical techniques to perform optimal fitting of the original trajectory using polynomials, cubic spline curves, etc., to obtain a simplified curve.

[0009] (5) Range filtering method: This is a time series filtering method that can effectively smooth the trajectory and eliminate noise interference. It is often used in conjunction with the Douglas-Peucker algorithm.

[0010] There are two main problems in the above trajectory simplification method when used for wandering trajectory simplification: (1) these methods change the trajectory shape of the normal navigation section while simplifying the wandering section trajectory, which causes certain interference to the subsequent trajectory analysis task; (2) the ship wandering trajectory is usually distributed in a chaotic cluster in a small range, and the existing trajectory simplification method is difficult to process such chaotic trajectory, or needs to select appropriate parameters very carefully to simplify the chaotic wandering trajectory, which is difficult to operate, and is not suitable for batch automatic processing of a large amount of trajectory data. SUMMARY

[0011] The present application solves the problems of the existing trajectory simplification method that changes the trajectory shape of the normal navigation section while simplifying the wandering section trajectory, which causes interference to the subsequent trajectory analysis, and is not suitable for batch automatic processing of a large amount of trajectory data, and provides a ship wandering AIS trajectory rapid simplification method.

[0012] The technical solution claimed in the present application is as follows:

[0013] A ship wandering AIS trajectory rapid simplification method, comprising the following steps:

[0014] S1: arranging the AIS data of the ship in time sequence to form an AIS trajectory point sequence of the ship, and connecting the trajectory point sequence into a line according to the latitude and longitude position to form an AIS trajectory of the ship;

[0015] S2: cutting the AIS trajectory formed in S1: deleting the berthing points in the AIS trajectory, and cutting the AIS trajectory into different trajectory sections; the trajectory section is composed of a series of AIS data arranged in time sequence;

[0016] S3: self-intersection detection: judging the number of AIS data in each trajectory section obtained in S2, if there are more than 4, detecting the self-intersection point in each trajectory section, if there is a self-intersection point, jumping to S3; otherwise, the trajectory section is the simplified trajectory, and jumping to S6;

[0017] S4: constructing a trajectory directed graph, the trajectory directed graph comprising a node set and an edge set, the node set comprising the positions of all AIS data in the trajectory section and all self-intersection points, each AIS data position or self-intersection point being a graph node; the edge set being an edge added in time sequence between adjacent two AIS data corresponding graph nodes, or between a self-intersection point corresponding graph node and a related AIS data corresponding graph node, or between two self-intersection point corresponding graph nodes;

[0018] S5: using the graph shortest distance solving algorithm, finding out the graph shortest path between the graph node corresponding to the first AIS data of the track segment and the graph node corresponding to the last AIS data of the track segment from the track directed graph constructed in S4; the shortest path is the simplified track;

[0019] S6: outputting the simplified track in S3 or S5.

[0020] The AIS data in S1 comprises dynamic information and static information of the ship; the dynamic information comprises position, time stamp, speed, heading and navigation state; the static information comprises ship name, MMSI number and destination port; the position comprises longitude and latitude.

[0021] In some embodiments of the present application, the berthing point in S2 refers to the AIS data with the navigation state being berthing or anchoring, and in another embodiments of the present application, when the AIS data navigation state information is missing, the SOG field is used for judgment, and when the SOG is equal to 0 or less than 0.1 knots in the continuous AIS data arranged in time sequence, the positions corresponding to the AIS data are the berthing points.

[0022] In the above method, the self-intersection in S3 refers to the overlapping or coincidence of the different time parts of a track segment.

[0023] The detection method of the self-intersection point in S3 is that the position coordinates of the adjacent two AIS data in the track segment are marked as Pi=(xi, yi) and Pj=(xj, yj), and Pi is earlier than Pj in time, a first directed line segment is formed by connecting them, and is marked as Pij, a second directed line segment is formed by processing other adjacent AIS data in the track segment, and is marked as Pmn, the Pmn represents the directed line segment formed by the position coordinates Pm and Pn, and whether the two directed line segments Pij and Pmn intersect is judged, wherein the sufficient and necessary condition for the intersection at point P is that there exist 0≤a≤1 and 0≤b≤1 satisfying the following equations:

[0024] P=Pj+a*(Pi-Pj)

[0025] P=Pn+b*(Pm-Pn)

[0026] The position of the intersection point P is solved by the above formula.

[0027] The edge set in S5 is calculated in the following several cases:

[0028] (1) the positions of the two AIS data adjacent in time sequence in the track segment are marked as Pi and Pj, and it is assumed that Pi is earlier than Pj, then an edge is added in time sequence from Pi to Pj, and the weight of the edge is the distance between Pi and Pj;

[0029] (2) For each self-intersection point P, assuming it is the intersection of directed line segments Pij and Pmn, then add the following 4 directed edges, respectively from Pi to P, from P to Pj, from Pm to P, from P to Pn, and the weight of each edge is the distance between the two end nodes;

[0030] (3) Determine whether multiple line segments and the same line segment intersect: assuming that three directed line segments all intersect Pij, the intersection points are Px, Py, and Pz, respectively, and according to the time sequence of the three directed line segments, the intersection points of Pij are sorted as Qx, Qy, and Qz, then add directed edges: from Qx to Qy, Qy to Qz, and the weight of each edge is the distance between the two end nodes; the time of the three directed line segments is the earlier time of the two end nodes.

[0031] Preferably, the distance in the above (1), (2) and (3) is the ground distance or the Euclidean distance.

[0032] Preferably, the calculation of the intersection edge set of multiple line segments and the same line segment in the above (3) is consistent with the calculation method of three directed line segments intersecting with one directed line segment.

[0033] Preferably, the graph shortest distance solving algorithm in S5 includes Dijkstra, Floyd, Bellman-Ford, and SPFA.

[0034] Beneficial effects:

[0035] The application provides a ship lingering AIS trajectory rapid simplification method, AIS data of a ship is arranged in time sequence to form an AIS trajectory point sequence of the ship, and the trajectory point sequence is connected into a line according to longitude and latitude positions to form an AIS trajectory of the ship; stop points in the AIS trajectory are deleted, the AIS trajectory is divided into different trajectory segments, and the different trajectory segments are processed subsequently, so that the time characteristics between AIS points are retained, the space is simplified, all AIS data of non-lingering parts in navigation are retained, and interference on subsequent processing is avoided; the number of AIS data in each trajectory segment is judged, if the number is more than 4, self-intersection points in each trajectory segment are detected, ship lingering trajectories usually show a disordered cluster, trajectories of front and rear time overlap each other, the self-intersection points of the trajectories are detected to effectively locate the trajectories of the lingering segments, if the self-intersection points exist, a directed graph of the trajectories is constructed, the directed graph of the trajectories removes redundant and disordered trajectories while not affecting the trajectory form of the normal navigation segments, and the problem that the existing trajectory simplification method changes the trajectory form of the normal navigation segments while simplifying the trajectories of the lingering segments, thereby causing interference on subsequent trajectory analysis is solved; through actual measurement, the constructed directed graph of the trajectories is a very sparse graph, therefore, the operation speed is very efficient and fast, the method does not need to manually and finely adjust parameters, is suitable for automatically and rapidly processing a large amount of trajectory data, and solves the problem that the existing trajectory simplification method is not suitable for automatically processing a large amount of trajectory data. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The figure is a flow chart of the ship lingering AIS trajectory rapid simplification method in the embodiment of the application.

[0037] Figure 2 The figure is an AIS trajectory schematic diagram of a merchant ship with an MMSI number of 241579000 from November 16, 2022 to December 12, 2022 in the embodiment of the application.

[0038] Figure 3 The figure is an enlarged display diagram of lingering navigation segment trajectories in the AIS trajectory of the merchant ship with the MMSI number of 241579000 from November 16, 2022 to December 12, 2022 in the embodiment of the application, and the lingering navigation segments marked by numbers 1, 2 and 3 in the figure correspond to the lingering navigation segments from left to right. Figure 2

[0039] Figure 4 The figure is a trajectory segment schematic diagram obtained after the AIS trajectory is divided in the embodiment of the application.

[0040] Figure 5 The figure is a self-intersection point obtained by detecting the self-intersection points of the trajectories in the embodiment of the application, and the self-intersection points of the lingering navigation segments marked by numbers 1, 2 and 3 in the figure correspond to the lingering navigation segments from left to right, wherein the red points are the self-intersection points. Figure 2 ​​

[0041] Figure 6 For the overall effect of trajectory simplification in the embodiment of the present application, from left to right in turn corresponds to Figure 4 segment 0 and segment 3.

[0042] Figure 7 For the local enlarged result of the wandering trajectory simplification in the embodiment of the present application, from left to right in turn corresponds to Figure 2 the wandering segment marked by the numbers 1, 2 and 3. DETAILED DESCRIPTION

[0043] The present application will be further described below in conjunction with the accompanying drawings:

[0044] A ship wandering AIS trajectory rapid simplification method, as shown in Figure 1 , comprising the following steps:

[0045] S1: arranging the AIS data of the ship in time sequence to form an AIS trajectory point sequence of the ship, and connecting these trajectory point sequences into a line according to the latitude and longitude position to form an AIS trajectory of the ship; the AIS data includes dynamic information and static information of the ship; the dynamic information includes position lat-lon, time stamp postime, speed sog, heading cog and navigation status status; the static information includes ship name, MMSI number and destination port; the position includes longitude and latitude.

[0046] S2: cutting the AIS trajectory formed by S1: deleting the berthing point in the AIS trajectory, cutting the AIS trajectory into different trajectory segments; assuming that any trajectory segment is recorded as S, the trajectory segment S is composed of a series of AIS data arranged in time sequence; the berthing point refers to the AIS data with navigation status status being 1 (berthing) or 5 (anchoring), and the AIS data with navigation status status being 0 in the AIS trajectory is retained; in the specific embodiment of the present application, the status field of the AIS data is missing, when the status field is missing, the SOG field is used for judgment, when the SOG of the continuous AIS data arranged in time sequence is equal to 0 or less than 0.1 knots, the position corresponding to these AIS data is the berthing point.

[0047] S3: Self-intersection detection: determine the number of AIS data in each trajectory segment obtained in S2, if more than 4, detect the self-intersection point in each trajectory segment, if there is a self-intersection point, jump to S3; otherwise (i.e. the number of AIS data in the trajectory segment is less than or equal to 4, or the number of AIS data in the trajectory segment is greater than 4 but there is no self-intersection point), no need to process, the trajectory segment is the simplified trajectory, jump to S6, directly output the simplified trajectory; the self-intersection refers to the overlap or coincidence of different time parts of a trajectory line; the detection method of the self-intersection point is: mark the position coordinates of the adjacent two AIS data in the trajectory segment as Pi=(xi, yi) and Pj=(xj, yj), and Pi is earlier than Pj in time, connect them to form a first directed line segment, denoted as Pij, and other adjacent AIS data in the trajectory segment are processed in the same way to form a second directed line segment, denoted as Pmn, which represents the directed line segment formed by position coordinates Pm and Pn, and determine whether any two directed line segments Pij and Pmn intersect, wherein the sufficient and necessary condition for the intersection of the two at point P is that there exist 0≤a≤1 and 0≤b≤1 satisfying the following equations:

[0048] P=Pj+a*(Pi-Pj)

[0049] P=Pn+b*(Pm-Pn)

[0050] The position of the intersection point P can be solved by the above two formulas.

[0051] S4: Construct a trajectory directed graph G(V, E), which includes a node set V and an edge set E, the node set V includes all AIS data positions and all self-intersection points in the trajectory segment, and each AIS data position or self-intersection point is a graph node; the edge set E is added between the graph nodes corresponding to adjacent two AIS data, or between the graph nodes corresponding to a self-intersection point and the AIS data related to it, or between the graph nodes corresponding to two self-intersection points in time sequence; the edge set is calculated as follows:

[0052] (1) For the positions of the adjacent two AIS data in the trajectory segment, denoted as Pi and Pj, assuming that Pi is earlier than Pj, a directed edge is added from Pi to Pj in time sequence, and the weight of the edge is the distance between Pi and Pj; the distance is the ground distance or the Euclidean distance;

[0053] (2) For each self-intersection point P, assuming that it is the intersection of directed line segments Pij and Pmn, the following four directed edges are added, respectively from Pi to P, from P to Pj, from Pm to P, and from P to Pn, and the weight of each edge is the distance between the two end nodes; the distance is the ground distance or the Euclidean distance.

[0054] (3) Determine whether multiple line segments and the same line segment intersect: assume that three directed line segments all intersect with Pij, and the intersection points are Px, Py, and Pz, respectively. According to the time sequence of the three directed line segments, the intersection points of the three directed line segments with Pij are sorted as Qx, Qy, and Qz, respectively. Then, directed edges are added from Qx to Qy and from Qy to Qz, and the weight of each edge is the distance between the two end nodes. The time of the three directed line segments is the earlier time of the two end nodes. The distance is the ground distance or the Euclidean distance. In the specific embodiment of the present application, it is determined whether multiple line segments and the same line segment intersect. If so, the following operations are performed: assume that P12 (P1 is earlier than P2, and P1 points to P2), P34 (P3 is earlier than P4, and P3 points to P4), and P56 (P5 is earlier than P6, and P5 points to P6) are three directed line segments that all intersect with Pij (the processing method for more line segments intersecting with the same Pij is the same). The intersection points are Px, Py, and Pz, respectively. According to the time sequence of P1, P3, and P5, the intersection points of the three directed line segments with Pij are sorted as Qx, Qy, and Qz, respectively. Then, directed edges are added from Qx to Qy and from Qy to Qz, and the weight of each edge is the distance between the two end nodes. The processing method for multiple line segments intersecting with the same line segment is consistent with the processing method for three directed line segments all intersecting with the same directed line segment. The condition for determining whether two directed line segments intersect and the determination of the position of the intersection point are shown in the two formulas in S3. Through the position of the intersection point, it can be determined whether multiple directed line segments intersect.

[0055] S5: Shortest distance search: for the trajectory directed graph constructed in S4, a graph shortest distance solving algorithm is used to find the graph shortest path between the graph node corresponding to the first AIS data of the trajectory segment and the graph node corresponding to the last AIS data of the trajectory segment. The shortest path is the simplified trajectory. In the specific embodiment of the present application, for the constructed directed graph G(V, E), a graph shortest distance solving algorithm (such as Dijkstra, Floyd, Bellman-Ford, and SPFA) is used to find the graph shortest path between the first AIS data and the last AIS data in S. The path is taken as the simplified trajectory of the original trajectory S.

[0056] S6: Output the simplified trajectory: output the simplified trajectory described in S3 or S5.

[0057] The present application is further described below by taking the AIS trajectory of a merchant ship with MMSI number 241579000 from November 16, 2022 to December 12, 2022 as an example.

[0058] As shown in Figure 2 , wherein the circular dots represent the positions of the AIS data points, and the figure includes three lingering segments, the positions of which are as shown in Figure 2The three wandering segments are shown in the middle black boxes (1, 2, 3). The zoomed-in display of the three wandering segments is shown in FIG. 4B. Figure 3 As can be seen from FIG. 4B, the trajectories of the wandering segments are distributed in a group in a disordered manner. The method of the present application automatically simplifies these disordered trajectories while keeping the shapes of other parts of the trajectories unchanged.

[0059] Trajectory segmentation: AIS data with status field being 1 or 5 indicates that the ship is at anchor or at berth when the AIS data is sent. The ship trajectory is disconnected from the anchor or the berth (if there is no status field information, the SOG field in the AIS data can also be used for judgment, such as when the SOG in a plurality of AIS data arranged in time sequence is equal to 0 or less than 0.1 knots, the positions corresponding to these AIS data are the docking points, which need to be deleted from the trajectory), so that the whole trajectory is segmented into a plurality of one-way sailing trajectory segments. The AIS trajectory of the merchant ship with MMSI number 241579000 from 2022-11-16 to 2022-12-12 is disconnected by status being 1 or 5, and 4 trajectory segments segment 0, segment 1, segment 2 and segment 3 are obtained as shown in FIG. 4C. Figure 4 Figure 4 As can be seen from FIG. 4C, segment 2 is relatively short and is located in the lower right corner and is hidden by segment 3, and segment 1 and segment 2 are simple one-way motion trajectories of the ship and do not have wandering motion and self-intersection points, so they do not need to be simplified subsequently.

[0060] Trajectory self-intersection detection: self-intersection detection is performed on each trajectory segment obtained by trajectory segmentation to obtain self-intersection points as shown in FIG. 4D. Figure 5 Figure 5 As can be seen from FIG. 4D, the self-intersection detection results in segment 0 and segment 3 are shown, and the red points in the figure represent the detected self-intersection points. Figure 4

[0061] Trajectory simplification: for the trajectory segment having the self-intersection points, a directed graph of the trajectory segment is constructed, and then Dijkstra’s algorithm is used to obtain the shortest path from the first AIS of the trajectory segment to the last AIS of the trajectory segment, which is the simplified AIS trajectory. The wandering trajectory simplification result is shown in FIG. 4E, and the red trajectory in the figure represents the simplified trajectory. As can be seen from FIG. 4E, the redundant trajectories of all the wandering segments are simplified, the new trajectory is more concise, and the original trajectory shape outside the wandering segments is retained. The zoomed-in result of the simplified wandering segments is shown in FIG. 4F, and the red trajectory in the figure represents the simplified trajectory. Figure 6 Figure 7

[0062] ​​​​​It should be noted that the above detailed description of the specific embodiments of the present application is not intended to limit the present application in any way. Thus, while the present application has been described in detail with reference to specific embodiments thereof, it will be apparent to those skilled in the art that various modifications and changes can be made thereto without departing from the spirit and scope of the present application.

Claims

1. A method for fast and simple simplification of a ship's meandering AIS trajectory, characterized in that, The method comprises the following steps: S1: arranging the AIS data of the ship in time sequence to form an AIS track point sequence of the ship, and connecting the track point sequences in line to form an AIS track of the ship according to the latitude and longitude positions; S2: cutting the AIS track formed in S1: deleting the berthing points in the AIS track, and cutting the AIS track into different track segments; the track segment is composed of a series of AIS data arranged in time sequence; S3: self-intersection detection: judging the number of AIS data in each track segment obtained in S2, if the number is more than 4, detecting the self-intersection points in each track segment, if there is a self-intersection point, jumping to S4; otherwise, the track segment is the simplified track, and jumping to S6; S4: constructing a track directed graph, the track directed graph comprises a node set and an edge set, the node set comprises all the positions of the AIS data in the track segment and all the self-intersection points, each position of the AIS data or self-intersection point is a graph node; the edge set is an edge added in time sequence between the graph nodes corresponding to two adjacent AIS data, or between the graph nodes corresponding to a self-intersection point and the graph nodes corresponding to the AIS data related to the self-intersection point, or between the graph nodes corresponding to two self-intersection points; S5: finding the graph shortest path between the graph node corresponding to the first AIS data of the track segment and the graph node corresponding to the last AIS data of the track segment by using a graph shortest distance solving algorithm on the track directed graph constructed in S4; the shortest path is the simplified track; S6: outputting the simplified track in S3 or S5.

2. The method of claim 1, wherein, The AIS data in S1 comprises dynamic information and static information of the ship; the dynamic information comprises position, time stamp, speed, heading and navigation state; the static information comprises ship name, MMSI number and destination port; the position comprises longitude and latitude.

3. The method of claim 2, wherein, The berthing point in S2 refers to the AIS data with the navigation state being berthing or anchoring.

4. The method of claim 3, wherein, When the navigation state information of the AIS data is missing in S2, the SOG field is used for judgment, when the SOG of a plurality of AIS data arranged in time sequence is equal to 0 or less than 0.1 knots, the positions of the AIS data are berthing points.

5. The method of claim 1, wherein, The self-intersection in S3 refers to the overlapping or coincidence of the front and rear time parts of a track segment.

6. The method of claim 5, wherein, The detection method of the self-intersection point in S3 is: marking the position coordinates of the two adjacent AIS data in the track segment as Pi = (xi, yi) and Pj = (xj, yj), and Pi is earlier than Pj in time, connecting them to form a first directed line segment, denoted as Pij, marking the other adjacent AIS data in the track segment to form a second directed line segment, denoted as Pmn, the Pmn represents the directed line segment formed by the position coordinates Pm and Pn, and judging whether any two directed line segments Pij and Pmn intersect, wherein the sufficient and necessary condition for the two directed line segments intersecting at point P is that there exist 0 ≤ a ≤ 1 and 0 ≤ b ≤ 1 satisfying the following equations: ​ P = Pn + b*(Pm-Pn) The position of the intersection point P is solved by the above equation.

7. The method of claim 6, wherein, The edge set in S5 is calculated as follows: (1) For the positions of two adjacent AIS data in the trajectory segment, denoted as Pi and Pj, and assuming that Pi is earlier than Pj, an edge is added from Pi to Pj in time sequence, and the weight of the edge is the distance between Pi and Pj; (2) For each self-intersection point P, assuming that it is the intersection point of directed line segments Pij and Pmn, the following four directed edges are added, respectively from Pi to P, from P to Pj, from Pm to P, and from P to Pn, and the weight of each edge is the distance between the two end nodes; (3) Determine whether multiple line segments and the same line segment intersect: assuming that three directed line segments all intersect with Pij, the intersection points are Px, Py, and Pz, and according to the time sequence of the three directed line segments, the intersection points of Pij are sorted as Qx, Qy, and Qz, then directed edges are added from Qx to Qy and from Qy to Qz, and the weight of each edge is the distance between the two end nodes; the time of the three directed line segments is the time of the earlier one.

8. The method of claim 7, wherein, The distance in (1), (2), and (3) is the ground distance or the Euclidean distance.

9. The method of claim 7, wherein, The calculation of the edge set of multiple line segments and the same line segment in (3) is consistent with the calculation method of three directed line segments and one directed line segment.

10. The method of claim 1, wherein, The shortest distance solving algorithm of the graph in S5 includes Dijkstra, Floyd, Bellman-Ford, and SPFA.

Images