A Knowledge Graph Construction Method for Maritime Moving Targets

Abstract

The present invention belongs to the field of knowledge graph construction, and discloses a method for constructing a knowledge graph for maritime mobile targets. Based on the dynamic positioning data of maritime mobile targets, the basic motion behavior semantics are defined, and further, the high-order behavior semantics applicable to multiple types of maritime mobile targets in complex environments are defined. A dynamic knowledge graph is constructed by using the dynamic positioning data of maritime mobile targets and the dynamically changing information in the sea area geographical environment information and meteorological information, and a static knowledge graph is constructed by using the static information of the data. The dynamic and static graphs are fused, and knowledge graph reasoning technology is applied to improve the quality of the knowledge graph. A unified knowledge system is used to organize and represent a large amount of heterogeneous AIS data, maritime mobile target motion patterns and behavior knowledge, accurately express the complex behavior semantics of maritime mobile targets, understand different behavior intentions, realize intelligent analysis and decision-making of maritime mobile target behaviors based on dynamic positioning data, and promote the intelligent monitoring and management of maritime mobile targets based on big data.

Description

Technical Field

[0001] The present invention belongs to the field of knowledge graph construction, and more specifically, relates to a method for constructing a knowledge graph for maritime mobile targets. Background Art

[0002] With the rapid development of China's shipping industry, a large amount of Automatic Identification System (AIS) data has been accumulated. AIS periodically sends dynamic and static information of maritime mobile targets, such as ship names, positions, speeds, etc. Analyzing the dynamic positioning data (AIS data) of maritime mobile targets is of great significance for maritime traffic management, route optimization, and collision prevention. However, when using dynamic positioning data to carry out intelligent monitoring and management of maritime mobile targets, there are still many technical bottlenecks: First, existing research focuses more on trajectory prediction and anomaly detection, and less on the semantic connotations behind behavior patterns, making it difficult to accurately understand the intentions of different behaviors; Second, there is a lack of a unified knowledge system to organize and represent massive heterogeneous dynamic positioning data, maritime mobile target knowledge, and behavior knowledge, and the knowledge reasoning ability is insufficient; Third, the differentiated utilization of dynamic and static knowledge is insufficient, and the semantic connection between maritime mobile target entities and their behaviors has not been established.

[0003] To achieve AIS-driven intelligent management of maritime mobile targets, it is urgent to develop a method for constructing a semantic dynamic and static knowledge graph of maritime mobile target behaviors. This method should automatically learn the dynamic information in AIS data, construct a knowledge graph of behavior events, represent the motion associations between maritime mobile targets, establish an ontology knowledge base including maritime mobile target entities and behavior patterns, and form a behavior dynamic knowledge graph; and use domain knowledge and AIS static information to construct a static knowledge graph of maritime mobile target entities. On this basis, knowledge graph fusion is carried out, and the completeness of the knowledge graph is enhanced through knowledge reasoning to accurately predict the behavior semantics of maritime mobile targets and understand complex maritime traffic states.

[0004] In summary, to solve the problem of insufficient expression of maritime mobile target behavior knowledge in existing maritime mobile target management technologies, a method for constructing a knowledge graph for maritime mobile targets is proposed, aiming to achieve intelligent monitoring and management of maritime mobile targets based on AIS data, which has important technical significance and application prospects. Summary of the Invention

[0005] To solve the problem of insufficient expression of maritime mobile target behavior knowledge in existing maritime mobile target management technologies, a method for constructing a knowledge graph for maritime mobile targets is proposed to achieve intelligent monitoring and management of maritime mobile targets based on AIS data, which has important technical significance and application prospects.

[0006] The technical solution adopted by the present invention is as follows:

[0007] A method for constructing a knowledge graph for maritime moving targets, comprising the following steps:

[0008] S1. For AIS data with an original quality lower than the threshold, perform data cleaning, remove abnormal points, thin out dense points, and then perform upsampling according to the time interval;

[0009] S2. Based on the longitude and latitude position information of the maritime moving targets, as well as the fields of the status, speed, and course of the maritime moving targets in the processed AIS data, define the basic behavioral semantics of the maritime moving targets, and perform semantic detection and recognition on the trajectories of the maritime moving targets to endow the trajectory semantic information;

[0010] S3. Based on the trajectory semantic information of the maritime moving targets, the coastline information, and the island information, define the high-order behavioral semantics of the movement of the maritime moving targets;

[0011] S4. Extract the static fixed attributes of the maritime moving target information fields, as well as the island and coastline geographic information in the AIS data as static nodes in the knowledge graph to construct a static knowledge graph; extract the attributes that change dynamically over time during the navigation of the maritime moving targets in the AIS data, as well as meteorological and environmental factors as dynamic nodes in the knowledge graph, and combine the high-order behavioral semantics to construct a dynamic knowledge graph, and dynamically update the dynamic knowledge graph according to the movement information of the maritime moving targets and the changes in the surrounding sea area environment;

[0012] S5. Combine the dynamic knowledge graph and the static knowledge graph, and connect and fuse the nodes in each graph in the form of sub-nodes of the maritime moving targets with the maritime moving target nodes respectively;

[0013] S6. According to the knowledge graph constructed in step S5, use ontology-based and rule-based methods to reason about the knowledge graph, complete the knowledge graph, and update the dynamic node information in real time according to the dynamic positioning data of the maritime moving target voyages, mine and discover the hidden connections between the maritime moving targets, and perform link prediction on the nodes in the dynamic knowledge graph and the static knowledge graph.

[0014] Further, the high-order behavioral semantics of the maritime moving targets in step S3 include:

[0015] Waiting to enter the port means being at a seaport or other dock location, and due to certain reasons, temporarily staying or slowly moving before entering the port. The movement characteristics are manifested as the maritime moving target sailing at a speed lower than the set speed, turning at a speed lower than the set speed, and cruising around. The trajectory points are spatially aggregated;

[0016] Escort means that different maritime mobile targets move forward together at a set distance in an escort formation within a set time range. The movement characteristics are that the trajectories of the maritime mobile targets show geometric formations at different time snapshots, including single-file and triangular formations. The distance and speed between targets are relatively stable at different moments. Among them, the escort formations include single-file and triangular formations.

[0017] Round-trip transportation means that a maritime mobile target goes to a certain place due to a certain task and then returns to the departure place. The movement characteristics are that the end point and the starting point on the trajectory coincide within a certain range.

[0018] Approaching reconnaissance means the process in which a maritime mobile target approaches a target from a distance for real-time observation, data collection, and information gathering. The movement characteristics are approaching the target at a speed lower than the set speed, turning at a speed lower than the set speed, and repeatedly circling.

[0019] Concealment means that a maritime mobile target does not send signals from its radio transmitting equipment within a certain area. The movement characteristics are that the trajectory disappears, is interrupted, or suddenly appears on the trajectory.

[0020] Trajectory deviation means that the trajectory of a maritime mobile target deviates from the normal historical route. The movement characteristics are that the trajectory deviates from the normal route or the trajectories of other maritime mobile targets on the trajectory.

[0021] Island circumvention patrol means that a maritime mobile target conducts patrol activities around an island. The movement characteristics are that the trajectory of the maritime mobile target surrounds the island, and the geometric shape of the trajectory during the movement shows a Z-shaped pattern.

[0022] Departure from port means the action of a maritime mobile target leaving a dock or berth and heading for the ocean or another port. The movement characteristics are that the starting point of the trajectory of the maritime mobile target is the dock or port.

[0023] Furthermore, in step S4, the static fixed attributes extracted from the AIS data include the ship name, call sign, MMSI, ship length, ship width, and draft depth of the maritime mobile target. Taking the maritime mobile target as the node center, a static knowledge graph is constructed; the attributes that change dynamically with time during the navigation of the maritime mobile target are extracted, including the heading, course, speed, navigation status, reception time, and destination of the maritime mobile target. Taking the maritime mobile target as the node center, a dynamic knowledge graph is constructed.

[0024] Furthermore, the combination of the dynamic knowledge graph and the static knowledge graph in step S5 is specifically as follows:

[0025] Based on the dynamic knowledge graph and the static knowledge graph established in step S4, they are combined. Taking the ship name as the node center, the equivalent classes and equivalent attributes in the dynamic knowledge graph and the static knowledge graph are aligned and fused to construct a static-dynamic knowledge graph containing static and dynamic information.

[0026] The present invention has the following advantages compared with the prior art:

[0027] Based on the dynamic positioning data of marine mobile targets, the present invention defines the semantic information of their basic motion behaviors, and further defines the high-order behavior semantics applicable to multiple types of marine mobile targets in complex environments. A knowledge graph integrating marine mobile targets, environment, meteorology, and behavior semantics is established, and the knowledge graph is inferred and complemented to realize the intelligent analysis and decision-making of the behaviors of marine mobile targets based on dynamic positioning data, promoting the intelligent monitoring and management of marine mobile targets based on big data. Brief Description of the Drawings

[0028] Figure 1 It is the overall process framework design diagram of the present invention.

[0029] Figure 2 It is the schematic diagram of the knowledge graph constructed by the present invention. Detailed Embodiment

[0030] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0031] The present invention provides a method for constructing a knowledge graph for marine mobile targets. Among them, it automatically learns the dynamic semantic information in AIS data, constructs a knowledge graph of behavior events representing the motion associations between marine mobile targets, establishes an ontology knowledge base including marine mobile target entities and behavior patterns, and forms a dynamic knowledge graph of marine mobile target behaviors; at the same time, it uses domain expert knowledge and static information in AIS data to construct a knowledge graph representing the static attributes of marine mobile targets; on this basis, it fuses the constructed dynamic behavior knowledge graph and entity static knowledge graph, and enhances the completeness of the knowledge graph through knowledge reasoning, aiming to solve the problem of insufficient expression of marine mobile target behavior knowledge in the existing marine mobile target management technology, and realize the intelligent monitoring and management of marine mobile targets based on AIS data. As Figure 1 shown, the specific steps are as follows:

[0032] S1. For the original AIS data with low quality, data cleaning is performed, abnormal points and outliers in space are removed, dense points in time are thinned, and finally upsampling is performed at equal time intervals according to the time interval, and the time interval is set to 15 minutes to form high-quality AIS data;

[0033] S2. Based on the longitude and latitude position information of the maritime mobile targets in the processed AIS data, as well as fields such as the status, speed, and heading of the maritime mobile targets, define the basic behavioral semantics of the maritime mobile targets, and conduct semantic detection and recognition on the trajectories of the maritime mobile targets, and assign trajectory semantic information, including seven basic behavioral semantics: start, normal navigation, stay, detour / avoidance, turn back, surround, and end;

[0034] S3. Based on the trajectory semantic information of the maritime mobile targets, the coastline information, and the island information, define the high-order behavioral semantics of the movement of the maritime mobile targets. The types of vessels applicable to the high-order semantics include commercial vessels, law enforcement vessels, warships, fishing vessels, and special operation vessels, etc. The high-order behavioral semantics include:

[0035] Waiting to enter the port is defined as a state of temporarily staying or moving slowly before entering the port due to various reasons at locations such as seaports or other docks. The movement characteristics are manifested as the maritime mobile target sailing slowly, turning slowly, cruising around, etc., and the trajectory points are aggregated in space;

[0036] Leaving the port is defined as the action of the maritime mobile target leaving the dock or berth and heading for the ocean or other ports. The movement characteristics are that the starting point of the trajectory of the maritime mobile target is generally the dock or port.

[0037] Escort is defined as different maritime mobile targets moving forward together at a certain distance in a formation (such as in a line, triangle, etc.) by warships or law enforcement vessels within a certain time range. The movement characteristics are manifested as the trajectories of the maritime mobile targets presenting special geometric queues (in a line, triangle) in different time snapshots, and the distances and speeds between the targets are relatively stable at different times;

[0038] Round-trip transportation is defined as the maritime mobile target going to a certain place due to a certain task and then returning to the departure place. The movement characteristics are manifested as the end point and the starting point on the trajectory coinciding within a certain range;

[0039] Approaching for reconnaissance is defined as the process of the maritime mobile target approaching the target (such as an island, national border) from a distance for real-time observation, data collection, and information gathering. The movement characteristics are manifested as slowly approaching the target, slightly turning, and repeatedly circling;

[0040] Concealment is defined as the behavior of the maritime mobile target not sending signals from its radio transmitting equipment in a certain area to avoid being discovered of its behavioral intention. The movement characteristics are manifested as the trajectory disappearing, interrupting, or suddenly appearing on the trajectory;

[0041] Trajectory deviation means that the trajectory of the maritime mobile target deviates from the normal historical route. The movement characteristics are manifested as deviating from the normal route or the trajectories of other maritime mobile targets on the trajectory;

[0042] The island-circling patrol is defined as the patrol activity of a moving target at sea around an island. In terms of motion characteristics, it is manifested as the trajectory of the moving target at sea surrounding the island, and the geometric shape of the trajectory during the movement may be in a zigzag shape;

[0043] S4. Use Portege to construct a knowledge graph. Extract the fixed attributes of the moving target nodes at sea from AIS data, including static information such as the ship name, call sign, MMSI, ship length, ship width, and draft depth of the moving target at sea, as well as static fixed attributes of geographical information such as islands and coastlines. Taking the moving target at sea as the node center, construct a static knowledge graph of the moving target at sea; Extract the attributes that change dynamically with time during the navigation of the moving target at sea from AIS data, including the heading, course, speed, navigation status, reception time of the moving target at sea, and the destination, meteorological, and environmental factors as dynamic nodes in the knowledge graph. Taking the moving target at sea as the node center, combine high-order behavior semantics to construct a dynamic knowledge graph of the moving target at sea, and dynamically update the knowledge graph for the motion information of the moving target at sea and the changes in the surrounding sea area environment in the dynamic knowledge graph; As Figure 2 shown;

[0044] S5. Use graph matching, ontology alignment, and semantic fusion algorithms to fuse the dynamic knowledge graph and the static knowledge graph. For example, map the nodes in the two types of graphs to the same ontology, fuse the static nodes such as the moving target at sea name and MMSI to eliminate redundant nodes, and construct a complete static and dynamic knowledge graph; As Figure 2 shown;

[0045] S6. According to the knowledge graph constructed in step S5, use ontology-based and rule-based methods to reason about the knowledge graph and complete the relationship between nodes. For example, define the IsA relationship between the ship name node and the MMSI, destination, draft depth, ship width, and moving target at sea type node. By analyzing the similarity of each node between ship A and ship B, the link relationship between the ship name node of ship A and the moving target at sea type node can be completed based on the relationship between ship B and the moving target at sea type node.

[0046] By analyzing the PartOf relationship between the ship name node and nodes such as speed, basic behavior semantics, high-order behavior semantics, coastline, and meteorology, it is inferred that there is a link relationship between the ship name node and the high-order behavior semantics in the next stage. And according to the AIS data of the moving target at sea navigation, the dynamic node information is updated in real time, the high-order semantic behavior of the moving target at sea in the future is accurately predicted, the knowledge graph is completed, and the quality of graph construction is improved;

[0047] Although the above-described illustrative embodiments of the present invention have been described to facilitate understanding of the present invention by those skilled in the art, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions made using the concept of the present invention are within the scope of protection.

Claims

1. A method for constructing a knowledge graph for maritime moving targets, characterized in that, The following steps are involved: S1. For AIS data whose original quality is lower than the threshold, data cleaning is performed to remove abnormal points and thin out the dense points, and then upsample according to the time interval; S2. Define the basic behavior semantics of the maritime mobile target based on the latitude and longitude position information of the maritime mobile target in the processed AIS data, as well as the maritime mobile target state, speed and heading fields, and perform semantic detection and recognition on the maritime mobile target trajectory to assign trajectory semantic information; S3. Define the high-order behavior semantics of the movement of the mobile target at sea based on the trajectory semantic information, coastline information and island information of the mobile target at sea; S4, extracting AIS dynamic positioning data fields and static fixed attributes of island and coastline geographic information as static nodes in the knowledge graph to construct a static knowledge graph; Extract the attributes of the maritime mobile targets that change dynamically over time during navigation in the AIS data, as well as the meteorological and environmental factors as dynamic nodes in the knowledge graph, build a dynamic knowledge graph in combination with high-order behavioral semantics, and update the dynamic knowledge graph based on the movement information of the maritime mobile targets and the changes in the surrounding sea environment; S5. Combine the dynamic knowledge graph with the static knowledge graph, and connect and fuse the nodes in each graph with the maritime mobile target node in the form of maritime mobile target sub-nodes; S6. Based on the knowledge graph constructed in step S5, the knowledge graph is inferred using ontology-based and rule-based methods to complete the knowledge graph, and the dynamic node information is updated in real time according to the dynamic positioning data, the hidden connections between mobile targets at sea are mined and discovered, and the links of each node in the dynamic knowledge graph and the static knowledge graph are predicted; Among them, the high-level behavior semantics of the marine mobile target in step S3 includes: Waiting to enter the port means that the target temporarily stops or travels slowly before entering the port due to some reasons. The movement characteristics are that the target sails at a speed lower than the set speed, turns at a speed lower than the set speed, and cruises around. The track points are clustered in space. Escort means that different ships advance together in a set distance in an escort formation within a set time range. The movement characteristics are that the ship trajectories present geometric queues at different time snapshots, including a straight line and a triangle. The distance and speed between targets are relatively stable at different times. Among them, the escort formation includes a straight line and a triangle. Round trip transport is a mobile target at sea that goes to a certain place for a certain task and then returns to the starting point. The movement feature is that the end point and the starting point on the trajectory coincide within a certain range. Close reconnaissance is the process of a ship approaching a target from a distance to conduct real-time observation, data collection, and information gathering. The movement characteristics are approaching the target at a speed lower than the set speed, turning at a speed lower than the set speed, and repeatedly maneuvering. Concealment means that the radio transmitting equipment of a mobile target at sea does not send signals in a certain area, and the movement characteristics are that the trajectory disappears, is interrupted, or suddenly appears; Track deviation refers to the deviation of the trajectory of a moving target at sea from the normal historical route. The movement characteristics are manifested as deviations from the normal route or the trajectory of other ships. The island-circling patrol refers to the patrol activities of ships around the islands. In terms of motion characteristics, it is manifested as the trajectory of the moving target at sea surrounding the island, and the geometric shape of the trajectory during the motion is in a Z shape. Departing from the port means the action of a moving target at sea leaving the dock or berth and heading for the ocean or other ports. The motion characteristic is that the starting point of the trajectory of the moving target at sea is the dock or port.

2. The method for constructing a knowledge graph for maritime moving targets according to claim 1, characterized in that In step S4, the static fixed attributes extracted from the AIS data include the ship name, call sign, MMSI, ship length, ship width, and draft depth of the moving target at sea. Taking the moving target at sea as the node center, a static knowledge graph is constructed; the attributes that change dynamically with time during the navigation of the moving target at sea are extracted, including the heading, course, speed, navigation status, reception time, and destination of the moving target at sea. Taking the moving target at sea as the node center, a dynamic knowledge graph is constructed.

3. A method for constructing a knowledge graph for maritime mobile targets according to claim 1, characterized in that In step S5, the combination of the dynamic knowledge graph and the static knowledge graph is specifically as follows: Based on the dynamic knowledge graph and the static knowledge graph established in step S4, they are combined. Taking the ship name as the node center, the equivalent classes and equivalent attributes in the dynamic knowledge graph and the static knowledge graph are aligned and fused to construct a static-dynamic knowledge graph containing static and dynamic information.

Images