Ship companion pattern detection method, system and medium based on trajectory feature mining

Abstract

The present application provides a ship accompanying pattern detection method, system and medium based on trajectory feature mining, including the steps of: obtaining ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strong correlation candidate set of ship trajectory accompanying patterns; detecting ship accompanying patterns with multiple correlation strengths from the strong correlation candidate set based on multi-feature grid similarity. It is able to quickly and accurately extract the movement patterns of group targets from massive spatiotemporal trajectory data, mine the ship spatiotemporal accompanying patterns, reduce the mining time complexity, and improve the mining accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of ship accompanying pattern detection, and in particular to a ship accompanying pattern detection method, system and medium based on trajectory feature mining. Background Art

[0002] With the comprehensive development of information technology and the improvement of the accuracy of positioning equipment, massive spatiotemporal data have been preserved. Spatiotemporal trajectory data, as one type of data, contains the activity patterns of mobile targets. Interesting information can be obtained through data mining. However, the characteristics of massive spatiotemporal trajectory data such as high dimension, large granularity variation, and high spatiotemporal correlation have brought challenges to data analysis and mining technology.

[0003] The companion pattern is a spatiotemporal motion pattern of a group of targets, and each moving target moves a certain distance together under the constraints of adjacent time and adjacent space. As one of the most common spatiotemporal patterns in massive spatiotemporal trajectory data, the companion pattern is characterized by the high similarity of moving targets in time, space and other dimensions (moving speed, moving direction, etc.), such as traveling together, fleet escort, animal migration, etc. Mining potential accompanying patterns of moving targets from massive spatiotemporal trajectory data has important research value and significance in abnormal behavior detection, traffic congestion prediction, etc.

[0004] Traditional adjoint pattern mining methods mostly focus on clustering intersection, similarity analysis, motion attribute matrix, etc. When faced with massive spatiotemporal trajectory data, the efficiency and accuracy of these methods are a great challenge. Summary of the invention

[0005] The purpose of the present invention is to overcome the above-mentioned technical deficiencies and provide a ship companion pattern detection method, system and medium based on trajectory feature mining, which can quickly and accurately extract the motion pattern of group targets from massive spatiotemporal trajectory data, mine the spatiotemporal companion pattern of ships, reduce the mining time complexity, and improve the mining accuracy.

[0006] In order to achieve the above technical objectives, in a first aspect, the technical solution of the present invention provides a ship accompanying pattern detection method based on trajectory feature mining, comprising the following steps:

[0007] Obtain the space-time trajectory data of ship AIS (Automatic Identification System);

[0008] Performing data preprocessing and space-time grid sequence conversion on the ship AIS space-time trajectory data to obtain the ship's navigation trajectory;

[0009] Based on the spatiotemporal index statistical characteristics, the ship's navigation trajectory is filtered and accompanied by screening to obtain a strongly associated candidate set of ship trajectory accompanying patterns;

[0010] Based on the multi-feature grid similarity, ship accompanying patterns with multiple association strengths are detected from the strong association candidate set.

[0011] Compared with the prior art, the ship accompanying pattern detection method based on trajectory feature mining provided by the present invention has the following beneficial effects:

[0012] Filtering rules are set by using spatiotemporal trajectory features and accompanying pattern spatiotemporal features, constraints are set on spatiotemporal and voyage features, the scale of computational data is reduced from the source of the data, the overall computational effort of the algorithm is reduced, and the efficiency of the algorithm is improved; a pattern mining method based on the improved longest common subsequence that measures the similarity of grid trajectory sequences is more robust to discrete trajectory points and noise points than other distance measurement methods; a method for mining ship accompanying patterns that expresses the degree of trajectory similarity based on the similarity of multi-feature grid sequences has a better time complexity than traditional mining methods such as clustering-intersection and motion attribute matrix, and is more comprehensive and accurate in detecting accompanying patterns than traditional methods that only consider spatiotemporal attributes.

[0013] According to some embodiments of the present invention, the data preprocessing and space-time grid sequence conversion of the ship AIS space-time trajectory data comprises the steps of:

[0014] De-noising and cleaning the ship AIS spatiotemporal trajectory data to delete format error data, duplicate data, redundant data and outlier points, interpolating the sparse trajectory segments, and sorting the trajectory points of each mobile target in the ship AIS spatiotemporal trajectory data in chronological order to obtain cleaned spatiotemporal trajectory points;

[0015] Based on the size and motion characteristics of the ship, the grid accuracy is selected, and the whole world is gridded to construct a spatiotemporal grid index, and the cleaned spatiotemporal trajectory points are converted into a spatiotemporal grid trajectory sequence;

[0016] The ship navigation trajectory is filtered out from the space-time grid trajectory sequence based on the ship navigation speed characteristics and the space-time grid index.

[0017] According to some embodiments of the present invention, the step of filtering out the ship's navigation trajectory from the spatiotemporal grid trajectory sequence based on the ship's navigation speed characteristics and the spatiotemporal grid index comprises the following steps:

[0018] The minimum navigation speed threshold is set, and the continuous grid index of the spatiotemporal grid trajectory sequence of the retention trajectory is combined to maintain consistent spatiotemporal characteristics. The consistency of the spatiotemporal grid index of each ship is checked, and the trajectory points of the stay trajectory points and the anchored trajectory points are filtered out to screen out the navigation trajectory of the ship.

[0019] According to some embodiments of the present invention, the filtering and accompanying screening of the ship's navigation trajectory based on spatiotemporal index statistical features comprises the steps of:

[0020] Eliminate the short-range space-time trajectory of ships in the navigation trajectory of the ship based on the adjoint pattern characteristics;

[0021] Screening the ship's navigation trajectory to obtain an initial candidate set of ship trajectory accompanying patterns screened based on accompanying time features;

[0022] The strong correlation candidate set of the ship trajectory accompanying pattern is obtained based on the spatial grid statistics of the initial candidate set.

[0023] According to some embodiments of the present invention, the method of detecting ship accompanying patterns of multiple association strengths from the strong association candidate set based on multi-feature grid similarity comprises the steps of:

[0024] Construct multi-feature spatiotemporal grids of ship motion based on grid index;

[0025] Extracting associated ships from the strongly associated candidate set according to the multi-feature spatiotemporal grid of ship motion;

[0026] The longest common subsequence of the spatiotemporal grid sequence of the associated ships is calculated as the support of the accompanying pattern, and the pattern support is filtered. When the support meets the accompanying voyage requirement, the trajectory similarity of the associated ships is calculated as the accompanying pattern confidence, and ship accompanying patterns with different association strengths are obtained according to the accompanying pattern confidence.

[0027] According to some embodiments of the present invention, the ship navigation trajectory is screened to obtain an initial candidate set of ship trajectory companion patterns screened based on the accompanying time features, including the steps of:

[0028] The time characteristics of the spatiotemporal trajectory of each ship are processed, and a ship is arbitrarily extracted from the ship navigation trajectory data set as a main target, and other ships in the ship navigation trajectory data set are used as secondary targets;

[0029] Match the common time interval of the main target and the secondary target, select the main target and secondary target whose common time interval is greater than or equal to the accompanying time threshold as associated ships, traverse the entire ship navigation trajectory data set, and generate the initial candidate set of ship trajectory accompanying mode based on the accompanying time feature screening.

[0030] According to some embodiments of the present invention, the strong correlation candidate set of the ship trajectory accompanying pattern is obtained based on the spatial grid statistics of the initial candidate set, including the steps of:

[0031] A sliding window with a step size of 2 is used to deduplicate the spatiotemporal grid trajectory sequence of each ship in the initial candidate set. When the grid indexes in the window are the same, only one is retained to reduce the data size and retain the temporal nature of the grid sequence.

[0032] Counting the grid sequences of each ship after deduplication, calculating the number of common elements between the grid sequences of each associated ship, removing the associated ships whose number of common elements is less than a threshold value from the initial candidate set, taking the associated ships whose number of common elements is greater than or equal to the threshold value as pattern mining candidates, and updating the initial candidate set;

[0033] The initial candidate set is traversed to filter out the ship pairs with low correlation strength in the spatial dimension, and a strongly correlated candidate set of associated ships that meets the requirements of the ship accompanying mode in terms of spatiotemporal characteristics is obtained.

[0034] In a second aspect, the technical solution of the present invention provides a ship accompanying pattern detection system based on trajectory feature mining, comprising:

[0035] Data acquisition module, used to obtain the ship AIS spatiotemporal trajectory data;

[0036] A ship navigation track extraction module is communicatively connected with the data acquisition module, and is used for performing data preprocessing and space-time grid sequence conversion on the ship AIS space-time track data to obtain the ship navigation track;

[0037] A strongly associated candidate set extraction module is communicatively connected to the ship navigation track extraction module, and the strongly associated candidate set extraction module is used to filter and concomitantly screen the ship navigation track based on spatiotemporal index statistical features to obtain a strongly associated candidate set of ship track concomitant patterns;

[0038] The ship accompanying pattern detection module is communicatively connected with the strong correlation candidate set extraction module, and is used for detecting ship accompanying patterns of various correlation strengths from the strong correlation candidate set based on multi-feature grid similarity.

[0039] In the third aspect, the technical solution of the present invention provides a ship accompanying pattern detection system based on trajectory feature mining, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, the ship accompanying pattern detection method based on trajectory feature mining as described in any one of the first aspects is implemented.

[0040] In a fourth aspect, the technical solution of the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute a ship accompanying pattern detection method based on trajectory feature mining as described in any one of the first aspects.

[0041] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, wherein the abstract drawings are identical to one of the drawings in the specification:

[0043] Figure 1 A flow chart of a ship accompanying pattern detection method based on trajectory feature mining provided by one embodiment of the present invention;

[0044] Figure 2 A flow chart of a method for processing and organizing ship trajectory data based on a spatiotemporal grid provided by another embodiment of the present invention;

[0045] Figure 3 A flow chart of a ship trajectory filtering and accompanying screening method based on spatiotemporal grid statistical characteristics provided by another embodiment of the present invention;

[0046] Figure 4 A flow chart of ship accompanying pattern detection based on multi-feature grid similarity provided by another embodiment of the present invention;

[0047] Figure 5 A schematic diagram of the electronic device structure of a ship accompanying pattern detection system based on trajectory feature mining provided in another embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0049] It should be noted that, although the functional modules are divided in the system schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0050] The present invention provides a ship companion pattern detection method based on trajectory feature mining, which can quickly and accurately extract the motion pattern of a group target from massive spatiotemporal trajectory data, mine the spatiotemporal companion pattern of ships, reduce the mining time complexity, and improve the mining accuracy.

[0051] The embodiments of the present invention are further described below in conjunction with the accompanying drawings.

[0052] Reference Figure 1 , Figure 2 and Figure 3 , Figure 1 A flow chart of a ship accompanying pattern detection method based on trajectory feature mining provided by one embodiment of the present invention; Figure 2 A flow chart of a method for processing and organizing ship trajectory data based on a spatiotemporal grid provided by another embodiment of the present invention; Figure 3 A flow chart of a ship trajectory filtering and accompanying screening method based on spatiotemporal grid statistical characteristics provided by another embodiment of the present invention; Figure 4 A flow chart of ship accompanying pattern detection based on multi-feature grid similarity provided by another embodiment of the present invention; Figure 5 A schematic diagram of the electronic device structure of a ship accompanying pattern detection system based on trajectory feature mining provided in another embodiment of the present invention.

[0053] The ship accompanying pattern detection method based on trajectory feature mining includes but is not limited to steps S110 to S140.

[0054] Step S110, obtaining the ship AIS spatiotemporal trajectory data;

[0055] Step S120, performing data preprocessing and space-time grid sequence conversion on the ship AIS space-time trajectory data to obtain the ship's navigation trajectory;

[0056] Step S130, filtering and accompanying screening the ship's navigation trajectory based on the spatiotemporal index statistical characteristics to obtain a strongly associated candidate set of the ship's trajectory accompanying pattern;

[0057] Step S140 , detecting ship accompanying patterns of various correlation strengths from the strong correlation candidate set based on the multi-feature grid similarity.

[0058] In one embodiment, a ship accompanying pattern detection method based on trajectory feature mining includes the steps of: acquiring ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strongly associated candidate set of ship trajectory accompanying patterns; and detecting ship accompanying patterns of various association strengths from the strongly associated candidate set based on multi-feature grid similarity.

[0059] 1. Ship trajectory data processing and organization based on space-time grid

[0060] The original ship trajectory data hides many noise points due to the accuracy of the positioning equipment and the transmission signal problems. The existence of noise points reduces the efficiency and effect of pattern mining. Therefore, it is of great significance to clean and convert the data. The specific steps are as follows:

[0061] (1) Preprocess the spatiotemporal trajectory data of the ship AIS (Automatic Identification System). De-noise the data, delete the data with obvious format errors, remove redundant data with repeated or few position points, and outliers, interpolate the sparse trajectory segments, sort the trajectory points of each mobile target in chronological order, and sort the overall data in ascending order by the mmsi (Maritime Mobile Service Identity) value.

[0062] (2) Constructing a spatiotemporal grid index. Based on the ship size and motion characteristics, select the appropriate grid accuracy, construct a spatiotemporal grid index through global gridding processing, and convert the cleaned spatiotemporal trajectory points into a spatiotemporal grid trajectory sequence, that is, convert the original trajectory tr = {mmsi, time, latitude, longitude, direction, heading, speed} into TR = {(x, y), mmsi, time, latitude, longitude, direction, heading, speed}, where the grid index grid id = (x, y), x and y are the indices of the grid where the longitude and latitude of the trajectory point are located. Each field represents a whole column of data of each feature of the ship (the same below), for example mmsi = {mmsi1, mmsi2,,, mmsi n}, n represents the overall data size, and mmsi deduplication is the number of ships in the trajectory data. The spatiotemporal index grid sequence is used to replace redundant ship position information, so that the trajectory data can be statistically analyzed through the grid index in the future, reducing the data size and the amount of data calculation.

[0063] (3) Filtering the ship's navigation trajectory based on the ship's navigation speed characteristics and grid index. By setting the minimum navigation speed threshold and combining the continuous grid index of the spatiotemporal grid trajectory sequence of the detention trajectory to maintain consistent spatiotemporal characteristics, the spatiotemporal grid index consistency of each ship is checked, and the trajectory points of the stop and anchor are filtered out to filter out the ship's navigation trajectory to avoid the situation where moving targets are accompanied by detention.

[0064] By preprocessing and converting trajectory data, worthless points can be basically eliminated, reducing the amount of subsequent pattern mining calculations and improving mining efficiency.

[0065] 2. Ship trajectory filtering and accompanying screening based on spatiotemporal grid statistical characteristics

[0066] The adjoint pattern can be understood as the phenomenon of a moving target moving continuously for a sufficiently long distance under the constraints of the spatiotemporal conditions of adjacent time and adjacent space. Its characteristics are highly correlated in spatiotemporal characteristics and motion characteristics. Therefore, filtering rules can be designed to generate a candidate set of pattern mining objects. The specific steps are as follows:

[0067] (1) Elimination of short-range ship spatiotemporal trajectories based on accompanying pattern characteristics. A sliding window with a step size of 2 is used to deduplicate the spatiotemporal grid trajectory sequence of each ship in the initial candidate set. When the grid index in the window is the same, only one is retained, that is, when each ship continuously sails in the same grid, only one grid index record is retained to reduce the data size and retain the temporal sequence of the grid sequence. By filtering the grid sequence of each ship Grid id = {grid id1, grid id2,…, grid id n} are deduplicated to obtain a new grid sequence, recorded as Grid id_, and the ships whose grid trajectory sequence length len(Grid id) is greater than or equal to the length threshold ΔL are screened out by setting a threshold to eliminate the ship trajectories that cannot meet the requirements of accompanying movement for a certain voyage.

[0068] (2) Generation of the initial candidate set of ship trajectory accompanying patterns based on time dimension features. First, the time features of each ship's spatiotemporal trajectory are processed, and one ship is randomly extracted from the ship navigation trajectory dataset as the main target, and the others are selected as secondary targets. The common time interval of the main target and the secondary target is matched. When the common time interval is less than the time threshold ΔT requirement, the subsequent processing process is skipped, and the main and secondary targets that meet the common time interval greater than or equal to the accompanying time threshold are selected as associated ships. The entire ship navigation trajectory dataset is traversed to generate the initial candidate set of ship trajectory accompanying patterns based on the selection of accompanying time features.

[0069] (3) Generation of strong correlation candidate sets of ship trajectory accompanying patterns based on spatial grid statistics. From the perspective of designing filtering rules for the two fields {grid id, mmsi} to implement proximity constraints, the grid sequence Grid id_ of each ship after deduplication is counted, and the number of common elements between sequences is calculated by traversing the grid sequences of each associated ship. Then, the associated ships that meet the common element number threshold Δσ are recorded as the strong correlation candidate sets of mining objects, and the initial candidate set is updated. For example, the mobile main target mmsi composed of mmsi is generated AThe adjoint pattern strongly associated candidate set is denoted as

[0070] |mmsi A | indicates the value of the maritime mobile communication service identification digital code, and requires | mmsi a |>|mmsi A |, to avoid subsequent duplicate records. This filtering rule reduces the amount of calculation by narrowing the subsequent mining objects.

[0071] Finally, after traversing and updating the initial candidate set of each ship, the ship pairs with low correlation strength in the spatial dimension can be filtered out, and a strong correlation candidate set of associated ships that meets the requirements of the ship accompanying pattern in terms of spatiotemporal characteristics can be obtained, thereby improving the accuracy of pattern mining and reducing the redundancy of mining data.

[0072] 3. Ship Accompanying Pattern Detection Based on Multi-Feature Grid Similarity

[0073] In order to mine the ship companion patterns from the candidate set, based on the fact that the companion patterns have a high correlation with the ship's spatiotemporal dimension and motion attributes, combined with the characteristics of the ship's trajectory spatiotemporal grid, a method of measuring the similarity of multi-feature grid sequences using the longest common subsequence (LCSs) distance is proposed to mine the ship companion patterns. The specific steps are as follows:

[0074] (1) Construct a multi-feature ship motion grid based on grid index. By statistically analyzing the grid index sequence Grid id_ of each moving target after deduplication, the motion information of each moving target in each grid it passes through is calculated, including the average speed and average heading, and a multi-feature ship motion grid is constructed, including There are four features, as shown in Table 1.

[0075]

[0076] Table 1 Movement information of each moving target in the grid it passes through

[0077] (2) Trajectory multi-feature grid sequence distance metric based on the improved longest common subsequence algorithm. The traditional longest common subsequence algorithm calculates the number of identical characters between two strings to express the similarity between the two strings. For example, given two sets X and Y with lengths i and j respectively, the longest common subsequence calculation formula is:

[0078]

[0079] Where LCSs[i][j] represents the length of the LCS of Xi and Yj. However, the elements in the above two sets are one-dimensional. In order to adapt to the similarity measurement of multi-feature grid sequences, the algorithm needs to be improved. When matching the associated ship trajectories, it is necessary to require not only the consistency of the grid index, but also the correlation of the motion features of the associated ships in each public grid. The motion features in the same grid are required to be similar, that is, the average heading difference and the average speed difference of the two moving targets in the same grid meet the constraints of the thresholds Δθ and Δv respectively. At this time, the improved longest common subsequence algorithm can be obtained. For example, given a pair of associated ship grid trajectory sequences

[0080] TR A ={(x, y) A , mmsi A , time A , latitude A , longitude A direction A , heading A ,speed A},

[0081] TR a ={(x, y) a , mmsi a , time a , latitude a , longitude a direction a , heading a ,speed a}, the grid indexes are grid id A =(x, y) A , grid id a =(x,y) a , the calculation formula of the longest common subsequence of two ship grid trajectory sequences is:

[0082]

[0083] Where LCSs[i][j] represents TR A and TR a The length of the longest common subsequence is calculated. If and only if all the judgment conditions meet the requirements, the length of the longest common subsequence is increased by 1. Since the order of the elements of the longest common subsequence in the two sets is consistent, that is, the longest common subsequence is a vector, the situation where two moving targets are sailing towards each other can be ruled out. The multi-feature spatiotemporal grid trajectory sequence distance between each associated ship is obtained by calculation.

[0084] (3) Ship trajectory accompanying pattern detection based on multi-feature grid sequence distance. The longest common subsequence of the improved longest common subsequence of the ship pair with a certain correlation strength is As the support of the accompanying mode, the mode support is filtered. When the support meets the requirements of the accompanying voyage, the similarity of the two trajectories is calculated as the accompanying mode confidence. The calculation formula for the similarity of the associated ship trajectories is:

[0085]

[0086] The generated companion pattern associated ship information is sorted in descending order of confidence, and the ship companion patterns of various association strengths can be obtained. The identified companion pattern moving target spatiotemporal trajectory information is sorted and stored.

[0087] Filtering rules are set by spatiotemporal trajectory features and accompanying pattern spatiotemporal features. By setting constraints on spatiotemporal and voyage features, the scale of calculated data is reduced from the source of the data, the overall amount of algorithm calculation is reduced, and the efficiency of the algorithm is improved. The pattern mining method based on the improved longest common subsequence and measuring the similarity of grid trajectory sequences is robust to discrete trajectory points and noise points compared to other distance measurement methods. The method for mining ship accompanying patterns based on the expression of trajectory similarity by multi-feature grid sequence similarity has better time complexity than traditional mining methods such as clustering-intersection and motion attribute matrix, and is more comprehensive and accurate in accompanying pattern detection than traditional methods that only consider spatiotemporal attributes.

[0088] In one embodiment, a ship accompanying pattern detection method based on trajectory feature mining includes the steps of: acquiring ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strongly associated candidate set of ship trajectory accompanying patterns; and detecting ship accompanying patterns of various association strengths from the strongly associated candidate set based on multi-feature grid similarity. The ship AIS spatiotemporal trajectory data is subjected to data preprocessing and spatiotemporal grid sequence conversion, comprising the steps of: denoising and cleaning the ship AIS spatiotemporal trajectory data to delete format error data, duplicate data, redundant data and outlier points, interpolating sparse trajectory segments, and sorting the trajectory points of each moving target in the ship AIS spatiotemporal trajectory data in chronological order to obtain cleaned spatiotemporal trajectory points; based on the ship size and motion characteristics, the grid accuracy is selected, the world is gridded to construct a spatiotemporal grid index, and the cleaned spatiotemporal trajectory points are converted into a spatiotemporal grid trajectory sequence; and the ship navigation trajectory is screened out from the spatiotemporal grid trajectory sequence based on the ship navigation speed characteristics and the spatiotemporal grid index.

[0089] In one embodiment, a ship accompanying pattern detection method based on trajectory feature mining includes the steps of: acquiring ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strongly associated candidate set of ship trajectory accompanying patterns; and detecting ship accompanying patterns of various association strengths from the strongly associated candidate set based on multi-feature grid similarity. The ship AIS spatiotemporal trajectory data is preprocessed and converted into a spatiotemporal grid sequence, including the following steps: denoising and cleaning the ship AIS spatiotemporal trajectory data to delete format error data, duplicate data, redundant data and outlier points, interpolating the sparse trajectory segments, and sorting the trajectory points of each moving target in the ship AIS spatiotemporal trajectory data in chronological order to obtain the cleaned spatiotemporal trajectory points; based on the ship size and motion characteristics, the grid accuracy is selected, the global grid is processed to construct a spatiotemporal grid index, and the cleaned spatiotemporal trajectory points are converted into a spatiotemporal grid trajectory sequence; the ship navigation trajectory is screened out from the spatiotemporal grid trajectory sequence based on the ship navigation speed characteristics and the spatiotemporal grid index. The ship navigation trajectory is screened out from the spatiotemporal grid trajectory sequence based on the ship navigation speed characteristics and the spatiotemporal grid index, including the following steps: setting a minimum navigation speed threshold, combining the continuous grid index of the spatiotemporal grid trajectory sequence of the retention trajectory to maintain consistent spatiotemporal characteristics, checking the consistency of the spatiotemporal grid index of each ship, filtering out the trajectory points of the stay trajectory points and the anchored trajectory points, and screening out the ship navigation trajectory.

[0090] In one embodiment, a ship accompanying pattern detection method based on trajectory feature mining includes the steps of: obtaining ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strong correlation candidate set of ship trajectory accompanying patterns; detecting ship accompanying patterns of multiple correlation strengths from the strong correlation candidate set based on multi-feature grid similarity. The ship navigation trajectory is filtered and accompanied by screening based on spatiotemporal index statistical features, including the steps of: eliminating short-range ship spatiotemporal trajectories in the ship navigation trajectory based on the accompanying pattern features; screening the ship navigation trajectory to obtain an initial candidate set of ship trajectory accompanying patterns screened based on the accompanying time features; and obtaining a strong correlation candidate set of ship trajectory accompanying patterns based on spatial grid statistics of the initial candidate set.

[0091] In one embodiment, a ship accompanying pattern detection method based on trajectory feature mining includes the steps of: obtaining ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strong correlation candidate set of ship trajectory accompanying patterns; detecting ship accompanying patterns of multiple correlation strengths from the strong correlation candidate set based on multi-feature grid similarity. Detecting ship accompanying patterns of multiple correlation strengths from the strong correlation candidate set based on multi-feature grid similarity includes the steps of: constructing a multi-feature spatiotemporal grid of ship motion based on grid index; extracting associated ships from the strong correlation candidate set according to the multi-feature spatiotemporal grid of ship motion; calculating the longest common subsequence of the spatiotemporal grid sequence of associated ships as the support of the accompanying pattern, filtering the pattern support, and when the support meets the accompanying voyage requirement, calculating the trajectory similarity of the associated ship as the accompanying pattern confidence, and obtaining ship accompanying patterns of different correlation strengths according to the accompanying pattern confidence.

[0092] In one embodiment, a ship accompanying pattern detection method based on trajectory feature mining includes the steps of: obtaining ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strong correlation candidate set of ship trajectory accompanying patterns; detecting ship accompanying patterns of multiple correlation strengths from the strong correlation candidate set based on multi-feature grid similarity. The ship navigation trajectory is filtered and accompanied by screening based on spatiotemporal index statistical features, including the steps of: eliminating short-range ship spatiotemporal trajectories in the ship navigation trajectory based on the accompanying pattern features; screening the ship navigation trajectory to obtain an initial candidate set of ship trajectory accompanying patterns screened based on the accompanying time features; and obtaining a strong correlation candidate set of ship trajectory accompanying patterns based on spatial grid statistics of the initial candidate set.

[0093] The accompanying screening process of the ship navigation trajectory is performed to obtain an initial candidate set of the ship trajectory accompanying mode based on the accompanying time screening, including the steps of: processing the time characteristics of each ship's spatiotemporal trajectory, arbitrarily extracting one ship from the ship navigation trajectory data set as the main target, and the others as secondary targets, matching the common time interval of the main target and the secondary target, screening out the main and secondary targets that meet the common time interval greater than or equal to the accompanying time threshold as associated ships, traversing the entire ship navigation trajectory data set, and generating an initial candidate set of the ship trajectory accompanying mode based on the accompanying time feature screening.

[0094] In one embodiment, a ship accompanying pattern detection method based on trajectory feature mining includes the steps of: obtaining ship AIS spatiotemporal trajectory data; performing data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; filtering and accompanying screening the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strong correlation candidate set of ship trajectory accompanying patterns; detecting ship accompanying patterns of multiple correlation strengths from the strong correlation candidate set based on multi-feature grid similarity. The ship navigation trajectory is filtered and accompanied by screening based on spatiotemporal index statistical features, including the steps of: eliminating short-range ship spatiotemporal trajectories in the ship navigation trajectory based on accompanying pattern features; accompanying screening the ship navigation trajectory to obtain an initial candidate set of ship trajectory accompanying patterns based on accompanying time screening; processing the initial candidate set based on spatial grid statistics to obtain a strong correlation candidate set of ship trajectory accompanying patterns. The initial candidate set is processed based on spatial grid statistics to obtain a strongly associated candidate set of ship trajectory accompanying patterns, including the following steps: counting the grid sequences of each ship after deduplication, calculating the number of common elements between the grid sequences of each associated ship, removing the associated ships whose number of common elements is less than a threshold value from the initial candidate set, and then taking the associated ships whose number of common elements is greater than the threshold value as pattern mining candidates to update the initial candidate set; traversing the initial candidate set to filter out ship pairs with low correlation strength in the spatial dimension, and obtaining a strongly associated candidate set of associated ships that meet the requirements of the ship accompanying pattern in terms of spatiotemporal characteristics.

[0095] The present invention also provides a ship accompanying pattern detection system based on trajectory feature mining, comprising: a data acquisition module, used to acquire ship AIS spatiotemporal trajectory data; a ship navigation trajectory extraction module, which is communicatively connected to the data acquisition module, and the ship navigation trajectory extraction module is used to perform data preprocessing and spatiotemporal grid sequence conversion on the ship AIS spatiotemporal trajectory data to obtain the ship navigation trajectory; a strongly associated candidate set extraction module, which is communicatively connected to the ship navigation trajectory extraction module, and the strongly associated candidate set extraction module is used to perform filtering processing and accompanying screening on the ship navigation trajectory based on spatiotemporal index statistical features to obtain a strongly associated candidate set of ship trajectory accompanying patterns; a ship accompanying pattern detection module, which is communicatively connected to the strongly associated candidate set extraction module, and the ship accompanying pattern detection module is used to detect ship accompanying patterns of multiple association strengths from the strongly associated candidate set based on multi-feature grid similarity.

[0096] The present invention also provides a ship accompanying pattern detection system based on trajectory feature mining, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the ship accompanying pattern detection method based on trajectory feature mining as described above is implemented.

[0097] The processor and the memory may be connected via a bus or other means.

[0098] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0099] It should be noted that the ship accompanying pattern detection system based on trajectory feature mining in this embodiment may include a business processing module, an edge database, a server version information register, and a data synchronization module. When the processor executes the computer program, it implements the ship accompanying pattern detection method based on trajectory feature mining as applied to the ship accompanying pattern detection system based on trajectory feature mining as mentioned above.

[0100] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, an embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are executed by a processor or a controller, for example, by a processor in the above-mentioned terminal embodiment, so that the above-mentioned processor can execute the ship accompanying pattern detection method based on trajectory feature mining in the above-mentioned embodiment.

[0102] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0103] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions without violating the spirit of the present invention. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present invention.

[0104] The specific implementation of the present invention described above does not constitute a limitation on the protection scope of the present invention. Any other corresponding changes and modifications made based on the technical concept of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A ship accompanying pattern detection method based on trajectory feature mining, characterized in that: The following steps are involved: Obtain ship AIS spatiotemporal trajectory data; Performing data preprocessing and space-time grid sequence conversion on the ship AIS space-time trajectory data to obtain the ship's navigation trajectory; Based on the spatiotemporal index statistical characteristics, the ship's navigation trajectory is filtered and accompanied by screening to obtain a strongly associated candidate set of ship trajectory accompanying patterns; Detecting ship accompanying patterns of various correlation strengths from the strong correlation candidate set based on multi-feature grid similarity; The filtering and accompanying screening of the ship's navigation trajectory based on the spatiotemporal index statistical characteristics comprises the following steps: Eliminate the short-range space-time trajectory of ships in the navigation trajectory of the ship based on the adjoint pattern characteristics; Screening the ship's navigation trajectory to obtain an initial candidate set of ship trajectory accompanying patterns screened based on accompanying time features; Based on spatial grid statistics, the strong correlation candidate set of the ship trajectory accompanying pattern is obtained for the initial candidate set; The method of detecting ship accompanying patterns of multiple association strengths from the strong association candidate set based on multi-feature grid similarity comprises the following steps: Construct multi-feature spatiotemporal grids of ship motion based on grid index; Extracting associated ships from the strongly associated candidate set according to the multi-feature spatiotemporal grid of ship motion; The longest common subsequence of the spatiotemporal grid sequence of the associated ships is calculated as the support of the accompanying pattern, and the pattern support is filtered. When the support meets the accompanying voyage requirement, the trajectory similarity of the associated ships is calculated as the accompanying pattern confidence, and ship accompanying patterns with different association strengths are obtained according to the accompanying pattern confidence.

2. A ship accompanying pattern detection method based on trajectory feature mining according to claim 1, characterized in that: The ship AIS spatiotemporal trajectory data is subjected to data preprocessing and spatiotemporal grid sequence conversion, comprising the steps of: De-noising and cleaning the ship AIS spatiotemporal trajectory data, deleting format error data, duplicate data, redundant data and outlier points, interpolating sparse trajectory segments to generate the ship spatiotemporal trajectory, and sorting the trajectory points of each mobile target in the ship AIS spatiotemporal trajectory data in chronological order to obtain the cleaned spatiotemporal trajectory points; Based on the size and motion characteristics of the ship, the grid accuracy is selected, and the whole world is gridded to construct a spatiotemporal grid index, and the cleaned spatiotemporal trajectory points are converted into a spatiotemporal grid trajectory sequence; The ship navigation trajectory is filtered out from the space-time grid trajectory sequence based on the ship navigation speed characteristics and the space-time grid index.

3. A ship accompanying pattern detection method based on trajectory feature mining according to claim 2, characterized in that: The method of selecting the ship's navigation track from the space-time grid track sequence based on the ship's navigation speed characteristics and the space-time grid index comprises the following steps: The minimum navigation speed threshold is set, and the continuous grid index of the spatiotemporal grid trajectory sequence of the retention trajectory is combined to maintain consistent spatiotemporal characteristics. The consistency of the spatiotemporal grid index of each ship is checked, and the trajectory points of the stay trajectory points and the anchored trajectory points are filtered out to screen out the navigation trajectory of the ship.

4. The ship accompanying pattern detection method based on trajectory feature mining according to claim 1 is characterized in that: The screening process of the ship's navigation track to obtain an initial candidate set of ship track accompanying patterns screened based on accompanying time features comprises the following steps: The time characteristics of the spatiotemporal trajectory of each ship are processed, and a ship is arbitrarily extracted from the ship navigation trajectory data set as a main target, and other ships in the ship navigation trajectory data set are used as secondary targets; Match the common time interval of the main target and the secondary target, select the main target and secondary target whose common time interval is greater than or equal to the accompanying time threshold as associated ships, traverse the entire ship navigation trajectory data set, and generate the initial candidate set of ship trajectory accompanying mode based on the accompanying time feature screening.

5. The ship accompanying pattern detection method based on trajectory feature mining according to claim 1 is characterized in that: The step of obtaining the strongly associated candidate set of the ship trajectory accompanying pattern based on spatial grid statistics of the initial candidate set comprises the following steps: A sliding window with a step size of 2 is used to deduplicate the spatiotemporal grid trajectory sequence of each ship in the initial candidate set. When the grid indexes in the window are the same, only one is retained to reduce the data size and retain the temporal nature of the grid sequence. Counting the grid sequences of each ship after deduplication, calculating the number of common elements between the grid sequences of each associated ship, removing the associated ships whose number of common elements is less than a threshold value from the initial candidate set, taking the associated ships whose number of common elements is greater than or equal to the threshold value as pattern mining candidates, and updating the initial candidate set; The initial candidate set is traversed to filter out the ship pairs with low correlation strength in the spatial dimension, and a strongly correlated candidate set of associated ships that meets the requirements of the ship accompanying mode in terms of spatiotemporal characteristics is obtained.

6. A ship accompanying pattern detection system based on trajectory feature mining, characterized in that: include: Data acquisition module, used to obtain the ship AIS spatiotemporal trajectory data; A ship navigation track extraction module is communicatively connected with the data acquisition module, and is used for performing data preprocessing and space-time grid sequence conversion on the ship AIS space-time track data to obtain the ship navigation track; A strongly associated candidate set extraction module is communicatively connected to the ship navigation track extraction module, and the strongly associated candidate set extraction module is used to filter and concomitantly screen the ship navigation track based on spatiotemporal index statistical features to obtain a strongly associated candidate set of ship track concomitant patterns; A ship accompanying pattern detection module is communicatively connected to the strong correlation candidate set extraction module, and the ship accompanying pattern detection module is used to detect ship accompanying patterns of multiple correlation strengths from the strong correlation candidate set based on multi-feature grid similarity; The filtering and accompanying screening of the ship's navigation trajectory based on the spatiotemporal index statistical characteristics comprises the following steps: Eliminate the short-range space-time trajectory of ships in the navigation trajectory of the ship based on the adjoint pattern characteristics; Screening the ship's navigation trajectory to obtain an initial candidate set of ship trajectory accompanying patterns screened based on accompanying time features; Based on spatial grid statistics, the strong correlation candidate set of the ship trajectory accompanying pattern is obtained for the initial candidate set; The method of detecting ship accompanying patterns of multiple association strengths from the strong association candidate set based on multi-feature grid similarity comprises the following steps: Construct multi-feature spatiotemporal grids of ship motion based on grid index; Extracting associated ships from the strongly associated candidate set according to the multi-feature spatiotemporal grid of ship motion; The longest common subsequence of the spatiotemporal grid sequence of the associated ships is calculated as the support of the accompanying pattern, and the pattern support is filtered. When the support meets the accompanying voyage requirement, the trajectory similarity of the associated ships is calculated as the accompanying pattern confidence, and ship accompanying patterns with different association strengths are obtained according to the accompanying pattern confidence.

7. A ship accompanying pattern detection system based on trajectory feature mining, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the ship accompanying pattern detection method based on trajectory feature mining as described in any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the ship accompanying pattern detection method based on trajectory feature mining as described in any one of claims 1 to 5.

Images