CN111179638B

CN111179638B - Ship AIS target navigation monitoring method based on time sequence

Info

Publication number: CN111179638B
Application number: CN202010018366.7A
Authority: CN
Inventors: 崔威威; 张秉致; 刘硕; 鲍鹏飞
Original assignee: 724th Research Institute of CSIC
Current assignee: 724th Research Institute of CSIC
Priority date: 2020-01-08
Filing date: 2020-01-08
Publication date: 2022-03-11
Anticipated expiration: 2040-01-08
Also published as: CN111179638A

Abstract

The invention provides a time sequence-based ship AIS target navigation monitoring method, which comprises the steps of firstly completing AIS data deduplication based on geographic range, MMSI, time and space information, secondly arranging AIS sequences according to a time ascending sequence, calculating the speed and the course of a target at two time points in front and at the back, completing AIS target MMSI deception judgment and verifying and correcting AIS navigation speed and course key information; based on the target AIS updating time, dynamically updating and generating a statistical grid, realizing detection and identification of a concomitant target group, and providing statistical information such as appearance/disappearance of regional ship navigation, speed and course distribution and the like to the outside; and finally, evaluating the AIS target navigation state based on a time series method. The invention can provide the risk early warning information of important areas and targets for the maritime supervision department.

Description

Ship AIS target navigation monitoring method based on time sequence

Technical Field

The invention belongs to the field of AIS data processing.

Background

With the rapid and continuous development of the economy in China and the implementation of the strategy of the economy zone of the Yangtze river, the water traffic of the gold water channel in the Yangtze river and the coastal areas is rapidly developed, the economic loss and the environmental damage risk caused by potential water traffic accidents are increasingly non-negligible, and the maritime supervision departments in China continuously perfect the construction of supervision facilities such as ship automatic identification systems, water traffic radars and the like and try to improve the supervision capability of the ship activities. The AIS equipment is necessary to be equipped for large ships, has the characteristics of low cost, identifiability, high precision, quick updating and quasi real-time performance, is an important information source for ship identity identification, behavior monitoring and rule statistics, and is also an important dependence for improving ship supervision and risk monitoring of a marine management department.

Aiming at AIS information processing, a large number of data processing means combined with regional information of ports and the like are formed at present, and regional statistical analysis such as density, flow, hot spots and the like is emphasized, such as speed analysis under the conditions of port entry and exit and berthing of ships. Due to the fact that the AIS information acquisition equipment is limited in working principle and equipment use and the forwarding, recording and managing mechanisms of the data processing system, the AIS information has serious data quality problems which are represented by the problems of information repetition, tampering, error, missing and the like, and how to manage the data is achieved, so that ships in the jurisdiction can be finely monitored, dangerous sailing states of the ships can be recognized, and corresponding research reports are not seen yet.

Disclosure of Invention

The invention aims to provide a method for distinguishing MMSI deception of an AIS target of a ship, splicing tracks, counting regional target navigation and evaluating the navigation state of the ship in a low-quality AIS data scene.

The invention firstly provides the MMSI, time and longitude and latitude data deduplication based on the AIS. To reduce computing storage resource consumption. Therefore, when the MMSI call sign, the time and the longitude and latitude are consistent, the data are considered to be repeated, and the repeated data record is deleted; and deleting the field data beyond the geographic range according to the geographic information.

Secondly, based on the MMSI call signs of the ship AIS target, grouping and sequencing according to time are carried out preliminarily, the course speed between AIS data is dynamically calculated by utilizing the updated longitude and latitude and time information of the AIS, MMSI deception discrimination and track fragment splicing are completed, and the speed and the course dynamic information of the ship AIS target are verified and corrected.

Generating a center coordinate again, dynamically generating/updating a statistical grid through AIS information of a ship target, and acquiring statistical information such as the appearance/disappearance time of a regional target, the target navigational speed, the course and the like; if two targets appear in a plurality of grids at the same time and the appearance time difference is not more than 20 minutes, the two targets are considered to have the accompanying relation, and the accompanying target group is detected based on the accompanying transfer hypothesis.

And finally, splitting the ship target AIS sequence into a plurality of subsequences according to the updating time interval, respectively extracting noise items of the speed and course difference of the subsequences, and obtaining a ship navigation state metric value through non-dimensionalization processing to realize target navigation state evaluation.

Drawings

Fig. 1 is a ship navigation information safety evaluation process.

Fig. 2 illustrates the MMSI spoofing decision principle.

Fig. 3 is a process of MMSI spoofing decision and piecemeal track splicing of a ship.

FIG. 4 illustrates grid generation and target association recognition.

FIG. 5 is a schematic diagram of the speed and heading differential sequence splitting principle determined by time intervals.

Detailed Description

The key steps of the invention comprise data cleaning, MMSI camouflage judgment of the ship, grid generation and information statistics and navigation state evaluation.

1, a ship AIS data cleaning and checking method:

the existing ship AIS database records have data quality problems in a large range, such as data repetition, MMSI camouflage, abnormal speed and course, ship name vacancy, position jumping, time jumping and the like, and data cleaning treatment is required to be carried out, so that the speed and the quality of data processing are improved. The basic method for data cleaning is as follows: carrying out rough grouping according to the MMSI, arranging according to the ascending order of time, and deleting the next data until no repetition occurs if the updating time and the latitude and longitude information of the adjacent AIS are repeated; when the geographical information such as accurate navigation paths, river channels and the like cannot be obtained, the approximate region range of the ship activity is given, and the outlier information is deleted.

2-vessel MMSI masquerading judgment

Because the state of the ship AIS terminal equipment is editable, AIS track jumping data exists under the condition that MMSI call signs are the same, namely AIS tampering and disguising conditions exist; meanwhile, the critical information of the air speed and the course of the AIS information has a large amount of abnormal updating conditions which are represented by data non-updating and inaccurate speed and course, and the data are required to be processed uniformly. The process is shown in figure 3.

(1) As shown in fig. 2, the L AIS data with the same MMSI are sorted in ascending order according to time, and the time, heading and speed of the adjacent AIS are calculated; if the speed obtained at e (k) (1, 2, … N-1, e (N) (L)) is greater than 30m/s, then the AIS sequence is segmented at e (k) to obtain track fragments, and the traversal is performed until all track fragments are obtained [ s (k), e (k) ], k (1, 2, … N:

the invention adopts the standard sphere to improve the data positionLet us say that the two A, B coordinates are (Lat)₁,Lon₁) And (Lat)₂,Lon₂) And calculating the distance dis, the course court and the navigational speed velo of the B relative to the A according to the following formula, wherein A is a central point:

Lat_i＝90-Lat_i,i＝1,2

c＝sin(Lat₁)*sin(Lat₂)*cos(Lon₁-Lon₂)+cos(Lat₁)*cos(Lat₂)

dis＝R*acos(c)

cos(c)＝cos(90-Lat₂)*cos(90-Lat₁)+sin(90-Lat₂)*sin(90-Lat₁)*cos(Lon₂-Lon₁)

sin(c)＝sqrt(1-(cos(c))²)

cour＝arcsin(sin(90-Lat₂)*sin(Lon₂-lon₁)/sin(c))

velo＝dis/t

wherein, R is (6378.14+5253.755)/2 is approximately equal to 6316KM, and the heading court is normalized and mapped to 0-360 degrees, if in the second quadrant, the heading court is 360+ court, and if in the third and fourth quadrants, the heading court is 180-court.

(2) Traversing and calculating the speed from the AIS point e (k) at the end of each track to the starting point of each segmentation point, namely calculating the speeds from e (k) to s (k +2), s (k +3) and s (N), and performing track merging and state updating when the speed is less than a threshold value; if the speed between e (k) and s (l) is less than the threshold value, combining the flight path fragments [ s (k), e (k)) ] and [ s (l), e (l)) ] and updating the starting and ending positions of the flight path at the same time until a plurality of ship AIS target queues are obtained, and uniformly distributing batch numbers BatchIdx.

3 grid generation, information statistics and target satellite group detection

The grid division is a basic means for carrying out track area statistics, the area statistics and the target group monitoring are realized based on the grid division, and the processing flow is shown in figure 4. The specific implementation method comprises the following steps:

(1) if the length and width of the default grid are 500 meters, it is known that the latitude extension is ExtLat ═ 0.0045 degrees, and the grid numbers IdxLat, IdxLon, and the longitude extension ExtLat are:

IdxLat＝floor(lat_i/ExtLat),i＝1,2,3,…,n

ExtLon＝500/(R*cos(IdxLat*ExtLat)*2*pi/360)

IdxLon＝floor(lon_i/ExtLon)

and generating a Hash value by utilizing the longitude and latitude coordinates of the AIS information, storing the Hash value into the grid information, and storing the Hash value to perform grid retrieval.

(2) And after the grid is obtained by calculation according to AIS information of the track, dynamically counting the entering and exiting time, the average speed, the course, the track flow and the density of the target in the grid, thereby completing the counting of the ship navigation area.

(3) And if the time difference of the two targets in the grid does not exceed 20 minutes, generating the Hash value pairwise according to the BatchIdx number, and simultaneously storing the Hash value and the BatchIdx of the targets. And uniformly counting the occurrence times of the Hash values, and if the same Hash value occurs for not less than 10 times, determining that a target companion pair occurs. The present invention recognizes that the companion relationship is transitive if there is a companion target pair [ idx_a,idx_b]、[idx_a,idx_c]Then [ idx ]_a,idx_b,idx_c]Is a companion group, and finally all companion groups are obtained.

4 ship navigation state evaluation based on time series

The invention provides a ship navigation state evaluation method based on a time sequence, which is specifically realized as follows:

in the movement under ideal conditions, the speed and the course of the target do not change frequently or violently; once the ship target has severe and continuous motion state changes, the navigation condition changes, the ship condition is abnormal, and the subjective intention causes the ship steering nonlinear changes, which often mean potential supervision risks. And evaluating the target navigation state by extracting the navigation speed, the course change and the time sequence of the target navigation message:

(1) arranging AIS sequences belonging to the same target according to a time ascending order, and obtaining a speed, course change and time interval sequence for the speed, course and time difference obtained in the step 2; the following processing is carried out on the heading:

(2) as shown in fig. 5, the AIS update time interval is non-uniform, and when the update time is long, the motion state of the target is considered to be discontinuous and unstable, and the corresponding speed and heading difference needs to be split into a plurality of subsequences; in practice, if position e_w(k)+1(k＝1,2,…M-1,e_w(M)＝n_kThe time interval of-1) is greater than 30 minutes, then at e_w(k) The AIS sequence is split until all subsequences [ s ] are obtained_w(k),e_w(k)]Wherein:

if the sequence [ s ]_w(k),e_w(k)]Length n of_kIf the distance is larger than 4, adding a nonlinear characteristic distance measurement calculation process.

(3) Subsequence [ s ]_w(k),e_w(k)]Corresponding course and speed differential feature sequence

Will be provided with

Considering a correlation time sequence with the window length d being 4, calculating a trend term

For i ═ 1,2,3,4, the deviation was calculated:

average value w of_k,iObtaining the season item s_i：

Thus, the noise term is obtained:

(4) carrying out normalization processing on the noise item to carry out dimension removal and normalization processing:

calculating the current subsequence evaluation value determined by noise:

and finally obtaining a dangerous navigation evaluation value of the current track:

in the above formula, D_ck、D_ckAre respectively composed of_w(k),e_w(k)]And evaluating value components of the obtained heading and speed difference subsequence.

Claims

1. A ship AIS target navigation monitoring method based on time series is characterized in that:

step 1: arranging AIS sequences with the same call signs in ascending order according to time, deleting repeated data if the longitude and latitude and the updating time of two AIS before and after the current sequence are the same, and realizing wild value elimination according to the geographical range information distributed by the AIS;

step 2: and (3) arranging the sequences with the same call sign in an ascending order, and calculating the distance between adjacent AIS points to obtain the navigational speed and the course:

wherein lat₁、lon₁、lon₂Respectively are latitude and longitude coordinates of two adjacent AIS points, dis represents the distance between the two adjacent AIS points, when the speed between the adjacent AIS points is more than 30m/s, the front and the rear AIS points are taken as dividing points e (k-1) and s (k), thus obtaining flight path fragments [ s (k), e (k)]A queue; traversing and calculating s (k-1) and e (j), j>And (k) combining the flight path fragments when the flight speed is less than the threshold value until all flight path batch information is obtained, finally realizing MMSI deception judgment, and correcting the flight speed and course key information of the AIS sequence;

and step 3: calculating the extensions of latitude and longitude ExtLat and ExtLon and coordinates [ IdxLat, IdxLon ] according to the length of 500 m serving as a grid:

wherein (lat)_i,lon_i) I is 1,2,3, …, n represents latitude and longitude coordinates of the AIS sequence, R is about 6316km which is an approximate value of the radius of the earth, and after the grid is obtained, the appearance/disappearance time, the target speed/course and the flow distribution of the track target in the grid are counted; when the time difference of two targets appearing in not less than 20 grids at the same time is not more than 20 minutes, the targets are considered to appear to accompany, and an accompanying target group is obtained.

2. The vessel AIS target voyage monitoring method based on the time series according to claim 1, characterized in that: carrying out differential processing on time, navigational speed and course sequence of a ship target AIS in a specified time period, and carrying out the following processing on the navigational speed:

if the time difference between the two AIS is more than 30 minutes, the AIS sequence belonging to the same target is split into a plurality of subsequences_w(k),e_w(k)]And obtaining a noise item of the AIS data based on a time series method, and calculating the estimated values of the difference sequence of the navigational speed and the heading according to the following formula:

wherein N is_kIs a subsequence [ s ]_w(k),e_w(k)]Length of (a) y_k,tIs a pair of subsequence [ s_w(k),e_w(k)]And finally evaluating the AIS target navigation state of the ship by adopting the normalized evaluation value obtained after time series processing of the navigation speed or the navigation course:

D_ck、D_ckrespectively by differential subsequences [ s ]_w(k),e_w(k)]And evaluating value components of the obtained heading and speed difference subsequence.