Route Extraction Method and System Based on Ship Behavior Patterns

Abstract

The present application discloses a method and system for extracting a route based on ship behavior patterns. By using ship behavior patterns as the classification criterion, the AIS data is classified and filtered to obtain the AIS data corresponding to the berthing mode and the turning mode respectively. Then, the DBSCAN algorithm with self-optimizing parameters is used to process the AIS data corresponding to the berthing mode to determine the berthing area, and the sliding window technology is used to process the AIS data corresponding to the turning mode to determine the waypoints, so as to obtain the berthing area and waypoints related to the determined route, realizing the screening of AIS data and reducing the subsequent data processing volume. Further, by constructing a rasterized route extraction model based on semantic analysis to process the waypoints and berthing area, the target route is determined, which can improve the efficiency of data processing and ensure the reliability of the target route.

Description

Technical Field

[0001] The present invention relates to the technical field of ships, and particularly to a route extraction method and system based on ship behavior patterns. Background Art

[0002] As an important part of China's comprehensive transportation system, waterway transportation plays an important role in supporting the development of the national economy, optimizing the development of the national territory and industrial layout, promoting foreign trade and the improvement of international competitiveness, and safeguarding national rights and interests and economic security. As the main body of waterway transportation, ships are the components of the national comprehensive three-dimensional transportation network, and their intelligent technologies have become the research hotspots in the fields of ship engineering and intelligent navigation.

[0003] In the existing process of obtaining ship routes through AIS data, there are mainly three types of methods: grid-based methods, vector-based methods, and statistical-based methods. However, due to the large amount of AIS data, it is difficult to determine the reliability of the data, resulting in low route extraction efficiency.

[0004] Therefore, in the prior art, in the process of obtaining ship routes, there is a problem of low route extraction efficiency due to the large amount of data processing. Summary of the Invention

[0005] In view of this, it is necessary to provide a route extraction method and system based on ship behavior patterns to solve the problem of large measurement errors in the process of obtaining the fuel loading amount of ships.

[0006] To solve the above problems, the present invention provides a route extraction method based on ship behavior patterns, including:

[0007] Determining ship behavior patterns based on AIS data of a preset water area, where the ship behavior patterns at least include a berthing pattern and a turning pattern;

[0008] Processing the AIS data corresponding to the berthing pattern through the DBSCAN algorithm with self-optimizing parameters to determine the berthing area;

[0009] Processing the AIS data corresponding to the turning pattern through a sliding window technique to determine waypoints;

[0010] Constructing a rasterized route extraction model based on semantic analysis, and determining the target route according to the waypoints and the berthing area.

[0011] Further, determining ship behavior patterns based on AIS data of a preset water area includes:

[0012] Constructing an adaptive ship behavior pattern identification model;

[0013] Obtain the sailing speed, sailing time, and sailing distance corresponding to the AIS data;

[0014] Based on the adaptive ship behavior pattern identification model, compare the sailing speed, sailing time, and sailing distance with the speed threshold, time threshold, and distance threshold respectively, and determine the ship behavior pattern corresponding to the AIS data.

[0015] Furthermore, process the AIS data corresponding to the berthing mode through the DBSCAN algorithm with self-optimizing parameters to determine the berthing area, including:

[0016] Construct a parameter optimization evaluation model;

[0017] Determine the optimal parameters of the DBSCAN algorithm according to the parameter optimization evaluation model, and determine the DBSCAN algorithm with self-optimizing parameters based on the optimal parameters;

[0018] Process the AIS data corresponding to the berthing mode according to the DBSCAN algorithm with self-optimizing parameters to determine the berthing area.

[0019] Furthermore, process the AIS data corresponding to the turning mode through the sliding pane technology to determine the waypoints, including:

[0020] Determine the turning section according to the AIS data corresponding to the turning mode;

[0021] Perform secondary identification on the turning section, reduce the window length to the preset number of track points, and determine the waypoints.

[0022] Furthermore, construct a rasterized route extraction model based on semantic analysis, and determine the target route according to the waypoints and the berthing area, including:

[0023] Construct a rasterized route extraction model based on semantic analysis;

[0024] Perform rasterization processing on the preset water area to determine multiple raster areas of the preset water area;

[0025] Perform semantic extraction on the waypoints and the berthing area according to the rasterized route extraction model based on semantic analysis to determine the ship itinerary semantic objects corresponding to the raster areas;

[0026] Match the ship itinerary semantic objects to determine the target route.

[0027] Furthermore, match the ship itinerary semantic objects to determine the target route, including:

[0028] Obtain the number of objects of the ship itinerary semantic objects in each raster area;

[0029] Determine two destinations according to the ship itinerary semantic object, where the first destination is used as the starting point of the route and the second destination is used as the ending point of the route;

[0030] Obtain the first target grid area and the second target grid area corresponding to the two destinations, and determine the number of objects of the ship itinerary semantic object corresponding to the grid area of any row / column between the first target grid area and the second target grid area;

[0031] Connect the grid areas corresponding to the maximum number of objects in any row / column to determine the target route.

[0032] Furthermore, before semantic extraction of waypoints and berthing areas based on the rasterized route extraction model based on semantic analysis, it further includes:

[0033] Classify the initial waypoints and initial berthing areas according to the berthing areas, and determine the waypoints and berthing areas corresponding to each trajectory cluster;

[0034] Among them, the berthing areas of the same trajectory cluster are the same.

[0035] To solve the above problems, the present invention also provides a route extraction system based on ship behavior patterns, including:

[0036] A ship behavior pattern determination module for determining the ship behavior pattern based on the AIS data of the preset water area, where the ship behavior pattern at least includes a berthing pattern and a turning pattern;

[0037] A berthing area determination module for processing the AIS data corresponding to the berthing pattern through the DBSCAN algorithm with self-optimized parameters to determine the berthing area;

[0038] A waypoint determination module for processing the AIS data corresponding to the turning pattern through the sliding window technology to determine the waypoints;

[0039] A target route determination module for constructing a rasterized route extraction model based on semantic analysis and determining the target route according to the waypoints and berthing areas.

[0040] To solve the above problems, the present invention also provides an electronic device, including a processor and a memory, where a computer program is stored on the memory, and when the computer program is executed by the processor, it implements the route extraction method based on ship behavior patterns as described in any of the above technical solutions.

[0041] To solve the above problems, the present invention also provides a storage medium storing computer program instructions, and when the computer program instructions are executed by a computer, the computer is made to execute the route extraction method based on ship behavior patterns as described in any of the above technical solutions.

[0042] The beneficial effects of adopting the above embodiments are as follows: The present invention provides a route extraction method and system based on ship behavior patterns. By using ship behavior patterns as the classification criterion, the AIS data is classified and filtered to obtain the AIS data corresponding to the berthing mode and the turning mode respectively. Then, the DBSCAN algorithm with self-optimizing parameters is used to process the AIS data corresponding to the berthing mode to determine the berthing area, and the sliding window technology is used to process the AIS data corresponding to the turning mode to determine the waypoints, thereby obtaining the berthing area and waypoints related to the determined route, realizing the screening of AIS data and reducing the subsequent data processing volume. Further, by constructing a rasterized route extraction model based on semantic analysis to process the waypoints and berthing area, the target route can be determined, which can improve the efficiency of data processing and ensure the reliability of the target route. Description of the Drawings

[0043] Figure 1 It is a schematic flowchart of an embodiment of the route extraction method based on ship behavior patterns provided by the present invention;

[0044] Figure 2 It is a schematic flowchart of an embodiment of determining ship behavior patterns provided by the present invention;

[0045] Figure 3 It is a schematic result diagram of an embodiment of the DBSCAN principle provided by the present invention;

[0046] Figure 4 It is a schematic flowchart of an embodiment of determining the berthing area provided by the present invention;

[0047] Figure 5 It is a schematic flowchart of an embodiment of identifying waypoints provided by the present invention;

[0048] Figure 6 It is a schematic flowchart of an embodiment of determining the target route provided by the present invention;

[0049] Figure 7 It is a schematic result diagram of an embodiment of trajectory cluster classification provided by the present invention;

[0050] Figure 8 It is a schematic flowchart of an embodiment of matching ship travel semantic objects provided by the present invention;

[0051] Figure 9 It is a schematic structural diagram of an embodiment of the route extraction system based on ship behavior patterns provided by the present invention;

[0052] Figure 10 It is a schematic block diagram of an embodiment of an electronic device provided by the present invention. Detailed Embodiments

[0053] The preferred embodiments of the present invention will be specifically described below with reference to the accompanying drawings, in which the drawings form a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, rather than to limit the scope of the present invention.

[0054] Before presenting the embodiments, the AIS data and the STSO model will be elaborated:

[0055] AIS (Automatic Identification System) data, including the real-time dynamic and static information of ships, can well describe the operating state of ships.

[0056] The STSO (Ship Trip Semantic Object) model refers to the ship trip semantic object model, which can model and analyze trajectory data by enriching the semantic information of moving objects.

[0057] On the basis of the continuous iteration and update of the information technology revolution, the information wave continuously affects the global technological development. Multiple information and communication technologies interact with each other, jointly creating the influence of big data and making a large amount of data become new possibilities. As early as 2008, Nature first mentioned the concept of big data in "Big Data", and then Science pointed out in "Dealing with Data" that the era of big data has arrived and elaborated on the importance of big data in future scientific research. With the rise of big data, a large number of data science technologies based on big data have emerged and have been widely applied to multiple fields such as information, energy, and economics, resulting in a large number of scientific research achievements. Driven by the information wave and data science, information cities and digital cities are continuously constructed and developed. However, a large amount of data has been accumulated in this process, and new models need to be proposed to integrate multi-source data in urban development and coordinate various parts.

[0058] Waterway transportation, as an important part of China's comprehensive transportation system, plays an important role in supporting the development of the national economy, optimizing the development of the national territory and industrial layout, promoting foreign trade and enhancing international competitiveness, and safeguarding national rights and interests and economic security.

[0059] At the same time, how to perceive the traffic situation from a large amount of ship AIS data to escort the decision-making and planning of managers is the key. In the existing process of obtaining ship routes through AIS data, there are mainly three types of methods: grid-based methods, vector-based methods, and statistical-based methods. However, due to the large quantity of AIS data, it is difficult to determine the reliability of the data, resulting in low efficiency in route extraction.

[0060] Therefore, in the prior art, during the process of obtaining the ship route, there is a problem that the route extraction efficiency is low due to the large amount of data processing.

[0061] To solve the above problems, the present invention provides a method and system for extracting a route based on the ship behavior pattern. By systematically focusing on the intrinsic correlation mechanism between ship attributes, ship behaviors, and ship behaviors, a ship spatio-temporal behavior feature correlation model system is constructed to support ship route extraction and intention inference.

[0062] Specifically, based on ship AIS data, the present application uses spatio-temporal correlation analysis method to study the ship spatio-temporal behavior feature correlation model, and reveals the ship spatio-temporal behavior feature correlation mechanism; furthermore, uses clustering and related route extraction methods to study the autonomous design method of the ship planned route, and establishes a ship route database driven by the destination. Based on the background of the development of ship intelligence, the excavation of sea area channel patterns and behavior patterns can not only extract interesting features from a large amount of ship track data for relevant analysis, but also provide a theoretical basis and technical support for subsequent ship position inference and anomaly detection and other fields.

[0063] As Figure 1 shown, Figure 1 is a schematic flow chart of an embodiment of the method for extracting a route based on the ship behavior pattern provided by the present invention, including:

[0064] Step S101: Based on the AIS data of a preset water area, determine the ship behavior pattern, where the ship behavior pattern at least includes a berthing pattern and a turning pattern;

[0065] Step S102: Process the AIS data corresponding to the berthing pattern by the DBSCAN algorithm with self-optimized parameters to determine the berthing area;

[0066] Step S103: Process the AIS data corresponding to the turning pattern by the sliding window technology to determine the waypoints;

[0067] Step S104: Construct a rasterized route extraction model based on semantic analysis, and determine the target route according to the waypoints and the berthing area.

[0068] In this embodiment, first, based on the AIS data of a preset water area, determine the ship behavior pattern, where the ship behavior pattern at least includes a berthing pattern and a turning pattern; then, process the AIS data corresponding to the berthing pattern by the DBSCAN algorithm with self-optimized parameters to determine the berthing area, and process the AIS data corresponding to the turning pattern by the sliding window technology to determine the waypoints; finally, construct a rasterized route extraction model based on semantic analysis, and determine the target route according to the waypoints and the berthing area.

[0069] In this embodiment, by using the ship behavior mode as the classification criterion, the AIS data is classified and filtered to obtain the AIS data corresponding to the berthing mode and the turning mode respectively. Then, the DBSCAN algorithm with self-optimizing parameters is used to process the AIS data corresponding to the berthing mode to determine the berthing area, and the sliding window technology is used to process the AIS data corresponding to the turning mode to determine the waypoints, so as to obtain the berthing area and waypoints related to the determined route, realizing the screening of the AIS data and reducing the subsequent data processing volume. Further, by constructing a rasterized route extraction model based on semantic analysis to process the waypoints and berthing area, the target route is determined, which can improve the efficiency of data processing and ensure the reliability of the target route.

[0070] As a preferred embodiment, in step S101, the ship behavior mode further includes a straight-line mode. To determine the ship behavior mode, as Figure 2 shown, Figure 2 is a schematic flowchart of an embodiment for determining the ship behavior mode provided by the present invention, including:

[0071] Step S111: Construct an adaptive ship behavior mode identification model;

[0072] Step S112: Obtain the navigation speed, navigation time, and navigation distance corresponding to the AIS data;

[0073] Step S113: Based on the adaptive ship behavior mode identification model, compare the relationships between the navigation speed, navigation time, and navigation distance with the speed threshold, time threshold, and distance threshold respectively, and determine the ship behavior mode corresponding to the AIS data.

[0074] In this embodiment, by constructing an adaptive ship behavior mode identification model to process and determine the navigation speed, navigation time, and navigation distance corresponding to the AIS data, the ship behavior mode is correspondingly determined, which can improve the efficiency of data processing.

[0075] It should be noted that since the setting of the speed threshold, time threshold, and distance threshold depends relatively on manual experience, in order to reduce the influence of human factors and avoid the situation that the recognition result is inaccurate due to subjective factors, during the model training process of the adaptive ship behavior mode identification model, its own parameters will be adjusted backward according to the recognition result to improve the reliability and accuracy of the recognition result.

[0076] As a preferred embodiment, in step S102, the DBSCAN algorithm is an unsupervised machine learning method based on density clustering. It does not require setting the number of clustering clusters in advance. It reflects the tightness of data distribution through two important parameters, the radius (ε) and the neighborhood density threshold (Z), and finds clusters with irregular shapes. This algorithm labels densely packed scattered points as one class and can identify scattered points with low density as noise.

[0077] The definition of the DBSCAN algorithm is as follows:

[0078] ① ε-neighborhood: The neighborhood of a given object p within radius ε is called the ε-neighborhood of this object, that is

[0079] N ε (P) = {q ∈ D | ρ(p, q) ≤ ε}

[0080] where D is the data set; ρ(p, q) is the distance between object p and object q; N ε (P) contains all objects in the data set D whose distance from object p is not greater than ε.

[0081] ② Core object: If the ε-neighborhood of object p contains at least the minimum number Z of objects, then this object is called a core object, that is

[0082] |N ε (p)| ≥ Z

[0083] ③ Density-direct reach: In the data set D, if object q is within the ε-neighborhood of object p and object p satisfies condition ②, then it is said that object p is density-directly reachable from object q.

[0084] ④ Density-reachable: If there exists an object chain p1, p2,..., p i ,..., p n , satisfying p1 = p and p n = p, p i is density-directly reachable from p i+1 with respect to ε and Z, then object p is density-reachable from object q with respect to ε and Z.

[0085] ⑤ Density-connected: If there exists an object O ∈ D such that object O is density-reachable from both object p and object q, then object p and object q are density-connected.

[0086] ⑥ Cluster and noise: Arbitrarily select an object p from the data set D, and search in the data set D starting from object p for all points that satisfy the ε and Z conditions and are density-reachable to form a cluster. Objects that do not belong to any cluster are marked as noise points.

[0087] As Figure 3 shown, Figure 3Schematic diagram of the results of an embodiment of the DBSCAN principle provided by the present invention. Among them, the circular white dots represent core points because the number of points in their neighborhood within the radius ε is 6; the triangular points are within the neighborhood of the core points and the number of points within their neighborhood radius ε does not exceed Z, so they are border points.

[0088] When the DBSCAN algorithm determines clusters, it searches the neighborhood of each test object in the dataset. If the number of objects in the neighborhood exceeds the minimum value (Z), a new cluster with this object as the core will be created. Then the algorithm starts from the core object, finds all density-reachable objects, and merges them into one cluster. The process will terminate until no points in the dataset can be added to any cluster. Points that do not fall into any cluster are considered noise or outliers.

[0089] However, although the DBSCAN algorithm has the advantages of being able to automatically determine the number of clustering clusters, being able to identify clusters of any shape, and separating noise, etc., the recognition results of the DBSCAN algorithm are highly dependent on the parameters ε and Z. Therefore, in order to improve the reliability of the DBSCAN algorithm, it is necessary to control the optimization parameters of the DBSCAN algorithm to improve the reliability of the berthing area.

[0090] The parameters to be optimized in the DBSCAN algorithm include the radius and the neighborhood density threshold, as Figure 4 shown, Figure 4 Schematic diagram of the process of an embodiment for determining the berthing area provided by the present invention, including:

[0091] Step S121: Construct a parameter optimization evaluation model;

[0092] Step S122: Determine the optimal parameters of the DBSCAN algorithm according to the parameter optimization evaluation model, and determine the DBSCAN algorithm with self-optimizing parameters based on the optimal parameters;

[0093] Step S123: Process the AIS data corresponding to the berthing mode according to the DBSCAN algorithm with self-optimizing parameters to determine the berthing area.

[0094] In this embodiment, first, evaluate the parameters to be optimized in the DBSCAN algorithm based on the parameter optimization evaluation model, and through iterative calculation, determine the optimal parameters of the DBSCAN algorithm to obtain the DBSCAN algorithm with self-optimizing parameters based on the optimal parameters; then, determine the berthing area according to the DBSCAN algorithm with self-optimizing parameters.

[0095] The DBSCAN algorithm with self-optimizing parameters can effectively remove the noise data of berthing points, then divide the ship position data into multiple berthing point clusters of any shape, then introduce the edge calculation method of the concave area, and finally generate the berthing area based on the AIS data.

[0096] In a specific embodiment, since in an actual water area, the area corresponding to the berthing area is the anchorage area, a parameter optimization evaluation model composed of two evaluation indexes, namely the proportion of the area coverage and the reciprocal of the offset distance of the center point, is established. By comparing the results of the model under different parameter selections, the optimal parameters are selected and substituted into the DBSCAN algorithm to obtain the clustering result of the berthing area obtained by the self-optimizing parameter model.

[0097] As a preferred embodiment, in step S103, in order to determine the waypoints, the modeling idea of a sliding window is introduced. First, according to the AIS data corresponding to the turning mode, the turning section is determined; then, the turning section is identified again, and the window length is reduced to the preset number of track points to determine the waypoints.

[0098] As Figure 5 shown, Figure 5 is a schematic flow chart of an embodiment of the waypoint identification provided by the present invention. For the trajectory lines tra i ={tp1, tp2, …, tp n} of different ships in different voyages, first, the idea of a sliding window is introduced to identify the turning section. Starting from the first trajectory point of the trajectory line tra i , a window of a certain length is slid in sequence, and the vector included angle value is calculated. Assuming that the size of the sliding window is p ship trajectory points, the vector included angle value discrimination formula is shown in Formulas 1 and 2:

[0099]

[0100] |cosθ j | ≤ υ T (2)

[0101] Wherein, is the vector from point tp k to point tp k-a ; is the vector from point tp k+b to point tp k ; υ T is the cosine threshold of the included angle.

[0102] If the points within the sliding window simultaneously satisfy Formulas 1 and 2, it is considered that they form a turning section, that is, the trajectory segment i formed by all the trajectory points between point tp k-a and point tp k+b on the trajectory line tra (m represents the mth turning section identified) has a bending feature and is regarded as a turning section.

[0103] In a specific embodiment, the size p of the sliding window and the cosine threshold υ can be adjusted according to the actual situation T to achieve an adaptive adjustment of the judgment conditions for the turning section.

[0104] Then, since the key turning points in the turning section are the waypoints defined in the model, it is necessary to perform a secondary recognition on this turning section. Similarly, the idea of a sliding window is introduced, and the window length is reduced to 3 trajectory points. For this turning section the vector included angle values of the secondary trajectory segments are calculated in turn. The calculation formula for the vector included angle value is shown in Equation 3:

[0105]

[0106] where is the cosine value of the turning included angle of the nth secondary trajectory segment within the recognized turning section ; is the vector from point tp k-4+r to point tp k-5+r ; is the vector from point tp k-3+r to point tp k-4+r ; r is an integer with a value ranging from 1 to p - 2, indicating the sliding state of the secondary trajectory segment on the turning section.

[0107] Since a turning section consists of p trajectory points and the secondary trajectory segment is defined as 3 trajectory points in length, a turning section can be divided into p - 2 secondary trajectory segments in total. That is, the maximum value of r is p - 2, expressed as Then, search for the minimum value among them to identify the trajectory interval with the most obvious turning characteristics, and consider the secondary trajectory segment represented by k-4+r as the part with the largest turning amplitude within this turning section. The middle trajectory point tp of this secondary trajectory segment is the waypoint obtained through recognition in the turning section i and is represented as the tth waypoint on tra

[0108] After traversing all the trajectory points of this trajectory line, connect all the recognized waypoints in sequence to obtain the simplified trajectory line, denoted as

[0109] It should be noted that the subsequent trajectory points that have been recognized as the turning section will no longer be included in the turning feature judgment. That is, the next trajectory segment to be judged is composed of p trajectory points from tp k+5 to tp k+14 . Additionally, if this trajectory segment does not satisfy the vector included angle value discriminant formula 2, continue to check tra iA trajectory segment consisting of consecutive p trajectory points starting from the next trajectory point, and shifting the starting trajectory point in this pattern until a new trajectory segment meets the conditions, then judging the turning characteristics of the sub-trajectory segments inside it and performing corresponding operations. Complete the identification of all trajectory points in the research water area according to the above steps, and generate the waypoint set corresponding to each trajectory line

[0110] In this embodiment, by using a sliding window of a certain length to capture the turning segments in the trajectory, and then identifying the trajectory points that can best represent the turning characteristics of the ship from the turning segments, the identification of waypoints is realized.

[0111] As a preferred embodiment, in step S104, on the basis of determining the berthing area and waypoints, in order to determine the target route, as Figure 6 shown Figure 6 is a schematic flowchart of an embodiment for determining the target route provided by the present invention, including:

[0112] Step S141: Construct a rasterized route extraction model based on semantic analysis;

[0113] Step S142: Perform rasterization processing on the preset water area to determine multiple grid areas of the preset water area;

[0114] Step S143: Perform semantic extraction on the waypoints and berthing areas according to the rasterized route extraction model based on semantic analysis to determine the ship travel semantic objects corresponding to the grid areas;

[0115] Step S144: Match the ship travel semantic objects to determine the target route.

[0116] In this embodiment, by performing rasterization processing on the preset water area to determine multiple grid areas of the preset water area, and performing semantic extraction on the waypoints and berthing areas corresponding to each grid area according to the rasterized route extraction model based on semantic analysis, the ship travel semantic objects corresponding to the grid areas are determined; finally, by matching the ship travel semantic objects, the target route is determined.

[0117] As a preferred embodiment, in step S141, the rasterized route extraction model based on semantic analysis models and analyzes the trajectory data by enriching the semantic information of moving objects.

[0118] Among them, a moving object refers to an object whose position changes over time. For example, pedestrians, cars on the road, and ships sailing at sea. The motion object pattern is an abstraction of the motion of a certain object, which endows key features with rich semantics to represent specific motion behaviors, simplifies the trajectory expression, and extracts motion laws.

[0119] Based on AIS data containing a large amount of ship movement information, identify the key features of ship trajectories, and construct an STSO to enhance the understanding of the semantic information of ship trajectories.

[0120] In a specific embodiment, the STSO is defined as the movement itinerary of a ship from one berthing point to another. Therefore, the ship itinerary can be simplified by the berthing point (S) and waypoints (M) as follows:

[0121] TR p =(S i ,M1,…,M n ,S i+1 )

[0122] It should be noted that the route extraction method of this application is studied based on the STSO endowed with semantic information.

[0123] As a preferred embodiment, in step S143, since the same waypoint and berthing area may involve multiple different ship trajectories, problems such as disordered route connection may occur during subsequent data processing due to large spans and excessive intersections of ship trajectories. Therefore, before performing semantic extraction on waypoints and berthing areas according to the rasterized route extraction model based on semantic analysis, it is also necessary to classify the waypoints and berthing areas.

[0124] Specifically, classify the initial waypoints and initial berthing areas according to the berthing area, and determine the waypoints and berthing areas corresponding to each trajectory cluster;

[0125] Among them, the berthing areas of the same trajectory cluster are the same.

[0126] In a specific embodiment, each ship trajectory is identified separately to obtain berthing endpoints, waypoints, and entrance / exit endpoints. Among them, the berthing endpoints and entrance / exit endpoints are the start and end points of a single ship trajectory, and the waypoints are the intermediate key nodes within the trajectory. According to the clustering characteristics of ship behavior, the obtained endpoints are clustered and identified to obtain the distribution of special areas within the preset water area, such as berthing areas and entrance / exit positions. These two types of special areas can distinguish ship trajectories with different starting and ending positions, that is, voyage plans with significant differences, and thus the trajectory data can be classified. It is defined that the berthing area and the entrance / exit position both belong to the departure and arrival area, and it is considered that the trajectory lines passing through the same group of departure and arrival areas should have the same ship behavior pattern and follow similar navigation routes.

[0127] By analyzing the navigation behavior patterns of ships in a preset water area in combination with the departure and arrival areas, it can be summarized as follows: Ships heading for ports within the water area enter the research water area from the entrance and exit positions, then sail at a low speed in the berthing area, and then head for the port; Ships leaving ports within the water area start from the port, also enter the berthing area for berthing operations, and finally leave the research water area from the entrance and exit positions. Therefore, ships passing through the same set of entrance and exit positions and berthing areas should have the same motion pattern and similar navigation routes.

[0128] Based on the above analysis, the identified anchorages and entrances and exits are combined, and the ship trajectories passing through the same set of anchorages and entrances and exits are extracted to form each trajectory cluster. For example, ships passing through anchorage area i and entrance and exit position loc i belong to the same trajectory cluster indicating that the ship trajectories under this combination have the same ship behavior pattern and follow similar navigation routes. According to the above method and idea, the classification of all trajectory data is realized, and the trajectory cluster set corresponding to each group of departure and arrival areas is generated as:

[0129]

[0130] In addition, since the identified berthing area is a polygon with a certain area and often shows an overlapping distribution form within the research scope, it is possible that a small number of clusters in the above trajectory clusters will overlap. In this regard, it is necessary to retain and eliminate the overlapping trajectory clusters.

[0131] Among them, the update principle is: retain the trajectory cluster with a relatively large water area coverage rate and delete the trajectory cluster with a relatively small coverage rate. In addition, for the trajectory cluster whose number of included trajectories is much lower than the average number of trajectories in each cluster, it is also reasonably deleted. As Figure 7 shown, Figure 7 is a schematic diagram of the result of an embodiment of the trajectory cluster classification provided by the present invention.

[0132] Through the above method, the trajectory data is classified according to different starting and ending positions. On the one hand, applying the maritime domain knowledge obtained by the ship behavior pattern recognition method to the trajectory data set can effectively separate the trajectory clusters with large differences in behavior patterns. On the other hand, trajectory classification can reduce the intersection of traffic flows within each trajectory cluster, which helps the introduction of subsequent route extraction algorithms and improves the accuracy of the extraction results.

[0133] As a preferred embodiment, in step S144, in order to match the ship itinerary semantic objects, as Figure 8 shown, Figure 8 is a schematic flowchart of an embodiment of matching the ship itinerary semantic objects provided by the present invention, including:

[0134] Step S1441: Obtain the number of objects of the ship travel semantic objects in each grid area;

[0135] Step S1442: Determine two destinations according to the ship travel semantic objects, where the first destination is used as the starting point of the route and the second destination is used as the ending point of the route;

[0136] Step S1443: Obtain the first target grid area and the second target grid area corresponding to the two destinations, and determine the number of objects of the ship travel semantic objects corresponding to the grid areas in any row / column between the first target grid area and the second target grid area;

[0137] Step S1444: Connect the grid areas corresponding to the maximum value of the number of objects in any row / column to determine the target route.

[0138] In this embodiment, based on the number of objects in the grid area, the positions where ships appear more frequently are determined, and the final target route is determined by connecting the corresponding positions.

[0139] It should be noted that for each trajectory cluster, the matching of the ship travel semantic objects needs to be performed separately, that is, a target route can be obtained for each trajectory cluster.

[0140] In a specific embodiment, for the trajectory cluster where the entrance and exit positions are distributed at the boundary of the research water area, its main route should have the characteristic that the traffic density is also relatively the highest at the boundary.

[0141] Based on this, first, the entrance and exit blocks of the main route at the boundary are located by searching for the critical threshold of the boundary traffic density. Specifically, to determine the critical threshold of the boundary traffic density, a set of increasing sequences are input as the threshold of the number of trajectories stored in each grid. If the number of trajectories stored in the grid is less than the threshold, the value of the grid is set to zero, otherwise it remains unchanged. As the threshold value continues to increase, the distribution of the non-zero grid values in the boundary row of the adjusted trajectory data matrix gradually becomes sparse until the grid values of all grids in the grid row where the boundary is located are all 0, then the threshold search is stopped, and it is considered that this threshold is the critical threshold corresponding to the main route of this trajectory cluster.

[0142] Then, generate the trajectory data matrix when the value in the set of sequences is slightly less than the threshold, and calculate the midpoint of the non-zero grid values continuously distributed in the critical row to locate the end of the main route. The same method is applied to calculate the midpoint positions of the non-zero grid values in other rows of this data matrix.

[0143] For example, a certain trajectory cluster The corresponding trajectory data matrix is matrix k, the hottest sailing interval of the data matrix is determined by the method of searching for the critical threshold of boundary traffic density when the threshold value of the data matrix is α, that is, the distribution interval of the valid grid cells retained after setting the threshold to zero. The valid grid cells retained in the boundary row are grid i-1,R and grid i,R and grid i+1,R (assuming that the research water area is divided into an R×C grid area), then the calculation formula for the position points of the obtained main route at the boundary is as follows:

[0144]

[0145]

[0146] where and are the coordinate points of the main route distribution at the boundary row (the R-th row) respectively; a, b, and c are the longitude and latitude coordinates of the corresponding coordinate points of the research area with the longitude coordinate range of [a, b] and the latitude coordinate range of [c, d]. Finally, the calculated position points are connected in sequence to generate the main route corresponding to the trajectory cluster

[0147] The set of main routes extracted from each trajectory cluster is represented as:

[0148]

[0149] In summary, by using the ship behavior pattern as the classification criterion, the AIS data is classified and screened to obtain the AIS data corresponding to the berthing mode and the turning mode respectively; then, the DBSCAN algorithm with self-optimizing parameters is used to process the AIS data corresponding to the berthing mode to determine the berthing area, and the sliding window technology is used to process the AIS data corresponding to the turning mode to determine the waypoints, so as to obtain the berthing area and waypoints related to the determined route, realizing the screening of the AIS data and reducing the subsequent data processing volume; furthermore, by constructing a rasterized route extraction model based on semantic analysis to process the waypoints and the berthing area, the target route is determined, which can improve the efficiency of data processing and ensure the reliability of the target route.

[0150] To solve the above problems, the present invention also provides a route extraction system based on ship behavior patterns, as Figure 9 shown Figure 9 is a schematic structural diagram of an embodiment of the route extraction system based on ship behavior patterns provided by the present invention. The route extraction system 900 based on ship behavior patterns includes:

[0151] The ship behavior pattern determination module 901 is configured to determine the ship behavior pattern based on the AIS data of a preset water area, where the ship behavior pattern at least includes a berthing pattern and a turning pattern;

[0152] The berthing area determination module 902 is configured to process the AIS data corresponding to the berthing pattern through the DBSCAN algorithm with self-optimized parameters to determine the berthing area;

[0153] The waypoint determination module 903 is configured to process the AIS data corresponding to the turning pattern through the sliding window technique to determine the waypoints;

[0154] The target route determination module 904 is configured to construct a rasterized route extraction model based on semantic analysis and determine the target route according to the waypoints and the berthing area.

[0155] The present invention also correspondingly provides an electronic device, as Figure 10 shown, Figure 10 is a structural block diagram of an embodiment of the electronic device provided by the present invention. The electronic device 1000 may be a computing device such as a mobile terminal, a desktop computer, a notebook, a palm computer, and a server. The electronic device 1000 includes a processor 1001 and a memory 1002, where a route extraction program 1003 based on the ship behavior pattern is stored on the memory 1002.

[0156] The memory 1002 may be an internal storage unit of the computer device in some embodiments, such as the hard disk or memory of the computer device. The memory 1002 may also be an external storage device of the computer device in other embodiments, such as a plug-in hard disk equipped on the computer device, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 1002 may also include both the internal storage unit and the external storage device of the computer device. The memory 1002 is used to store the application software installed on the computer device and various types of data, such as the program code installed on the computer device. The memory 1002 may also be used to temporarily store the data that has been output or will be output. In one embodiment, the route extraction program 1003 based on the ship behavior pattern may be executed by the processor 1001, so as to implement the route extraction method based on the ship behavior pattern in various embodiments of the present invention.

[0157] The processor 1001 may be a central processing unit (CPU), a microprocessor, or other data processing chips in some embodiments, and is used to run the program code stored in the memory 1002 or process data, such as executing the route extraction program based on the ship behavior pattern.

[0158] This embodiment also provides a computer-readable storage medium, on which a route extraction program based on ship behavior patterns is stored. When the program is executed by a processor, the route extraction method based on ship behavior patterns described in any of the above technical solutions is implemented.

[0159] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Any reference to a memory, storage, database, or other storage medium used in the various embodiments provided in this application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0160] As described above, the above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

Claims

1. A route extraction method based on ship behavior patterns, characterized in that, Including: Determine the ship behavior pattern based on the AIS data of the preset water area, where the ship behavior pattern at least includes a berthing pattern and a turning pattern; Process the AIS data corresponding to the berthing pattern through the DBSCAN algorithm with self-optimized parameters to determine the berthing area; Process the AIS data corresponding to the turning pattern through the sliding window technique to determine waypoints; Construct a rasterized route extraction model based on semantic analysis, and determine the target route according to the waypoints and the berthing area; The process of processing the AIS data corresponding to the turning pattern through the sliding window technique to determine waypoints includes: Determine the turning section according to the AIS data corresponding to the turning pattern; Perform secondary identification on the turning section, reduce the window length to the preset number of track points, and determine waypoints; The construction of the rasterized route extraction model based on semantic analysis and the determination of the target route according to the waypoints and the berthing area include: Construct the rasterized route extraction model based on semantic analysis; Perform rasterization processing on the preset water area to determine multiple raster areas of the preset water area; Perform semantic extraction on the waypoints and the berthing area according to the rasterized route extraction model based on semantic analysis to determine the ship travel semantic objects corresponding to the raster areas; Match the ship travel semantic objects to determine the target route; The matching of the ship travel semantic objects to determine the target route includes: Obtain the object quantity of the ship travel semantic objects in each raster area; Determine two destinations according to the ship travel semantic objects, where the first destination is used as the route start point and the second destination is used as the route end point; Obtain the first target raster area and the second target raster area corresponding to the two destinations, and determine the object quantity of the ship travel semantic objects corresponding to the raster areas in any row / column between the first target raster area and the second target raster area; Connect the raster areas corresponding to the maximum object quantity in any row / column to determine the target route.

2. The route extraction method based on the ship behavior pattern according to claim 1, characterized in that The determination of the ship behavior pattern based on the AIS data of the preset water area includes: Construct an adaptive ship behavior pattern identification model; Obtain the navigation speed, navigation time and navigation distance corresponding to the AIS data; Based on the adaptive ship behavior pattern identification model, compare the navigation speed, the navigation time and the navigation distance with the speed threshold, the time threshold and the distance threshold respectively, and determine the ship behavior pattern corresponding to the AIS data.

3. The route extraction method based on the ship behavior pattern according to claim 1, wherein The process of processing the AIS data corresponding to the berthing pattern through the DBSCAN algorithm with self-optimized parameters to determine the berthing area includes: Construct a parameter optimization evaluation model; Determine the optimal parameters of the DBSCAN algorithm according to the parameter optimization evaluation model, and determine the DBSCAN algorithm with self-optimized parameters based on the optimal parameters; Process the AIS data corresponding to the berthing pattern according to the DBSCAN algorithm with self-optimized parameters to determine the berthing area.

4. The method for extracting a shipping route based on a ship behavior pattern according to claim 1, wherein Before semantic extraction of the waypoints and the berthing area based on the semantic analysis-based rasterized route extraction model, it further includes: Classify the initial waypoints and the initial berthing area according to the berthing area, and determine the waypoints and the berthing area corresponding to each trajectory cluster; Wherein, the berthing areas of the same trajectory cluster are the same.

5. A route extraction system based on ship behavior patterns, applicable to the route extraction method based on ship behavior patterns according to any one of claims 1 to 4, characterized in that It includes: A ship behavior pattern determination module, configured to determine a ship behavior pattern based on AIS data of a preset water area, wherein the ship behavior pattern at least includes a berthing pattern and a turning pattern; A berthing area determination module, configured to process the AIS data corresponding to the berthing pattern through a DBSCAN algorithm with self-optimized parameters to determine a berthing area; A waypoint determination module, configured to process the AIS data corresponding to the turning pattern through a sliding window technique to determine waypoints; A target route determination module, configured to construct a semantic analysis-based rasterized route extraction model, and determine a target route according to the waypoints and the berthing area.

6. An electronic device, characterized in that, It includes a processor and a memory, and a computer program is stored on the memory. When the computer program is executed by the processor, the route extraction method based on ship behavior patterns according to any one of claims 1-4 is implemented.

7. A storage medium, characterized in that, The storage medium stores computer program instructions. When the computer program instructions are executed by a computer, the computer is caused to execute the route extraction method based on ship behavior patterns according to any one of claims 1 to 4.

Images