Ship intelligent identification management method and system

Abstract

The invention relates to the technical field of maritime affair safety monitoring, in particular to an intelligent ship identification management method and system, which ensures the comprehensiveness of data sources and provides a rich basis for subsequent analysis by collecting original data from a plurality of heterogeneous sensors. Original data are fused to generate a fusion feature vector, effective integration of multi-source data is achieved, ship feature representation is more comprehensive and accurate, and a reliable basis is provided for judgment of unfamiliar ships; whether the target ship is a strange ship or not is judged based on the fusion feature vector, manual experience judgment is replaced with data-driven judgment logic, and subjectivity and negligence are avoided; risk assessment is carried out on the unfamiliar ship, and corresponding early warning is triggered, so that seamless connection of identification and early warning is realized, and rapid risk response to the unfamiliar ship is ensured; and according to the risk level and the real-time state information, a disposal suggestion is generated and linkage response is started, so that a complete process from identification to disposal is formed, and automation and high efficiency of ship management are improved.

Description

Technical Field

[0001] This invention relates to the field of maritime safety monitoring technology, and in particular to a method and system for intelligent identification and management of ships. Background Technology

[0002] In the field of maritime safety monitoring, vessel identification and management are core aspects of ensuring waterway safety. Current vessel monitoring solutions mostly combine data from a limited number of devices such as radar, AIS (Automatic Identification System), and CCTV (Closed-Circuit Television), performing only basic data integration without forming a systematic intelligent analysis process. Specifically, existing technologies typically only capture and save target screenshots via CCTV when radar echoes without AIS signals are detected, relying entirely on manual vessel identification, judgment, and handling afterwards.

[0003] However, the most significant drawback of the aforementioned existing technologies is their extremely low level of intelligence. They lack a deep fusion mechanism for multi-source data and deep learning capabilities, making it impossible for them to automatically learn ship behavior patterns or autonomously integrate multi-dimensional data for intelligent analysis. They must rely on human experience to determine ship identity and behavioral attributes, making it difficult to efficiently and accurately distinguish between normal vessels and unfamiliar / suspicious vessels. This core deficiency of over-reliance on human intervention not only leads to low vessel identification efficiency and susceptibility to missed or incorrect judgments due to human error, but also results in delayed response and inability to meet early warning needs in real time. This seriously affects the reliability and timeliness of maritime safety monitoring, posing potential risks to water safety. Therefore, a technical solution capable of deep fusion of multi-source data and intelligent identification of unfamiliar vessels is urgently needed to address these issues. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a ship intelligent identification and management method and system that can automatically identify and distinguish unfamiliar ships without relying on human experience judgment, thereby improving the level of intelligence in ship identification.

[0005] To solve the above-mentioned technical problems, the present invention adopts the first technical solution as follows:

[0006] A method for intelligent identification and management of ships includes the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

[0007] The second technical solution adopted in this invention is as follows: A ship intelligent identification and management system includes one or more processors and a memory, wherein the memory stores a program that, when executed by the processor, performs the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

[0008] The beneficial effects of this invention are as follows: In step S1 of this method, raw data from multiple heterogeneous sensors is collected, ensuring the comprehensiveness of the data sources and providing a rich foundation for subsequent analysis, breaking the limitations of a single data source. Step S2 fuses the raw data to generate a fused feature vector, achieving effective integration of multi-source data and making the ship feature representation more comprehensive and accurate, providing a reliable basis for the identification of unfamiliar ships. Step S3 determines whether the target ship is an unfamiliar ship based on the fused feature vector, replacing manual experience judgment with data-driven judgment logic, avoiding subjectivity and negligence, and solving the core problem of excessive reliance on manual labor in existing technologies. Step S4 conducts risk assessment for unfamiliar ships and triggers corresponding early warnings, achieving seamless connection between identification and early warning, ensuring rapid response to risks from unfamiliar ships. Step S5 generates disposal suggestions based on risk level and real-time status information and initiates a linkage response, forming a complete process from identification to disposal, improving the automation and efficiency of ship management. Each step is interconnected, jointly achieving intelligent identification and efficient management of unfamiliar ships, and significantly reducing manual intervention. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the steps of the intelligent identification and management method for ships according to the present invention. Figure 2 This is a connection block diagram of the ship intelligent identification management system of the present invention; Label Explanation: 1. Processor; 2. Memory. Detailed Implementation

[0010] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0011] Please refer to Figure 1 The first technical solution adopted in this invention is as follows: A method for intelligent identification and management of ships includes the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

[0012] As can be seen from the above description, the beneficial effects of the present invention are as follows: In step S1 of this method, raw data from multiple heterogeneous sensors is collected, ensuring the comprehensiveness of the data sources and providing a rich foundation for subsequent analysis, breaking the limitations of a single data source. Step S2 fuses the raw data to generate a fused feature vector, achieving effective integration of multi-source data and making the ship feature representation more comprehensive and accurate, providing a reliable basis for the identification of unfamiliar ships. Step S3 determines whether the target ship is an unfamiliar ship based on the fused feature vector, replacing manual experience judgment with data-driven judgment logic, avoiding subjectivity and negligence, and solving the core problem of excessive reliance on manual labor in existing technologies. Step S4 conducts risk assessment for unfamiliar ships and triggers corresponding early warnings, achieving seamless connection between identification and early warning, ensuring rapid response to risks from unfamiliar ships. Step S5 generates disposal suggestions based on risk level and real-time status information and initiates a linkage response, forming a complete process from identification to disposal, improving the automation and efficiency of ship management. Each step is interconnected, jointly achieving intelligent identification and efficient management of unfamiliar ships, and significantly reducing manual intervention.

[0013] Furthermore, step S2 specifically involves: S21. Perform spatiotemporal registration and cleaning on the original data to obtain registered spatiotemporally aligned data; S22. Extract sub-features of the target ship under different data modes from the spatiotemporal aligned data; S23. An attention-based fusion algorithm is used to fuse the sub-features to generate a fused feature vector.

[0014] As described above, the spatiotemporal registration and cleaning process in step S21 ensures the consistency of the original data in time and space, removes invalid and redundant data, and provides a high-quality data foundation for subsequent processing. Step S22 extracts sub-features under different data modalities from the spatiotemporally aligned data, providing targeted feature support for data fusion. Step S23 uses an attention-based fusion algorithm to fuse the features, which can focus on key information and improve the accuracy of the fused feature vector representation. The entire refinement process makes data fusion more logical and operable, providing a more reliable feature basis for subsequent unfamiliar vessel identification, and effectively improving the quality and efficiency of data fusion.

[0015] Furthermore, in step S22, the sub-features include at least two of the following: location features, motion features, identity features, and visual appearance features.

[0016] As described above, by combining multiple types of sub-features, the attributes of the target vessel can be comprehensively characterized from different dimensions, avoiding the problem of one-sided representation by a single type of sub-feature. The clear definition of the sub-feature types makes feature extraction more targeted, and can be flexibly selected according to actual needs, taking into account both recognition accuracy and data processing efficiency. The complementary effect of multiple types of sub-features can improve the integrity and robustness of feature representation. Even if some feature data is missing, it can still be supported by other sub-features for subsequent fusion and judgment, providing more comprehensive feature support for the identification of unfamiliar vessels.

[0017] Furthermore, step S3 specifically includes: S31. The fused feature vector is matched and compared with the records in the preset ship database to obtain the matching and comparison results; S32. The matching comparison results are evaluated according to the preset priority rules. If the evaluation results do not meet the preset confidence conditions, the target vessel is determined to be an unfamiliar vessel.

[0018] As described above, matching and comparing based on a pre-set ship database ensures the objectivity of the judgment and avoids subjective differences caused by human judgment. The introduction of priority rules makes the judgment logic more in line with actual needs. By assigning priorities to different attribute information, the rationality of the judgment results is improved. Pre-set confidence conditions as judgment thresholds realize the quantitative standard of judgment and avoid the problems of missed judgments and misjudgments caused by absolute judgment. The entire judgment process is standardized and objective, which greatly improves the accuracy and consistency of the judgment of unfamiliar ships and gets rid of the dependence on human experience.

[0019] Furthermore, in step S4, a fuzzy logic evaluation model is used to conduct a risk assessment of the unfamiliar vessel based on the fused feature vector.

[0020] As described above, the fuzzy logic model can effectively handle the fuzziness and uncertainty in multi-dimensional indicators related to ships, and is more comprehensive and scientific than single-indicator assessments. This model can accurately reflect the actual risk level of unfamiliar ships, making risk level classification more in line with reality and providing a reliable basis for graded early warning. The assessment results can directly correspond to the preset risk level, ensuring the pertinence of early warning responses, improving the efficiency and accuracy of risk assessment, and providing solid support for the generation of subsequent handling recommendations.

[0021] Please refer to Figure 2 The second technical solution adopted in this invention is as follows: A ship intelligent identification and management system includes one or more processors 1 and a memory 2. The memory 2 stores a program that, when executed by the processor 1, performs the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

[0022] As can be seen from the above description, the beneficial effects of the present invention are as follows: This system solidifies the intelligent ship identification and management method into an executable program through a "processor 1 and memory 2" architecture, ensuring the stable implementation and large-scale application of the method and solving the problem of the difficulty in engineering implementation of technical solutions. The system supports the operation of one or more processors 1, which can efficiently process multi-source heterogeneous data and adapt to the real-time requirements of maritime monitoring scenarios. The memory 2 can stably store information such as programs, ship databases and handling rules, while supporting data traceability for easy subsequent optimization. The system can flexibly adapt to the hardware deployment of different monitoring areas, with strong compatibility and wide applicability. The program execution ensures the consistency and stability of each step, avoids errors from manual operation, and realizes the automated and standardized operation of ship identification and management.

[0023] Furthermore, when the program is executed by processor 1, it performs the following steps: Step S2 is as follows: S21. Perform spatiotemporal registration and cleaning on the original data to obtain registered spatiotemporally aligned data; S22. Extract sub-features of the target ship under different data modes from the spatiotemporal aligned data; S23. An attention-based fusion algorithm is used to fuse the sub-features to generate a fused feature vector.

[0024] As described above, the spatiotemporal registration and cleaning process in step S21 ensures the consistency of the original data in time and space, removes invalid and redundant data, and provides a high-quality data foundation for subsequent processing. Step S22 extracts sub-features under different data modalities from the spatiotemporally aligned data, providing targeted feature support for data fusion. Step S23 uses an attention-based fusion algorithm to fuse the features, which can focus on key information and improve the accuracy of the fused feature vector representation. The entire refinement process makes data fusion more logical and operable, providing a more reliable feature basis for subsequent unfamiliar vessel identification, and effectively improving the quality and efficiency of data fusion.

[0025] Furthermore, when the program is executed by processor 1, it performs the following steps: In step S22, the sub-features include at least two of the following: location features, motion features, identity features, and visual appearance features.

[0026] As described above, by combining multiple types of sub-features, the attributes of the target vessel can be comprehensively characterized from different dimensions, avoiding the problem of one-sided representation by a single type of sub-feature. The clear definition of the sub-feature types makes feature extraction more targeted, and can be flexibly selected according to actual needs, taking into account both recognition accuracy and data processing efficiency. The complementary effect of multiple types of sub-features can improve the integrity and robustness of feature representation. Even if some feature data is missing, it can still be supported by other sub-features for subsequent fusion and judgment, providing more comprehensive feature support for the identification of unfamiliar vessels.

[0027] Furthermore, when the program is executed by processor 1, it performs the following steps: Step S3 is as follows: S31. The fused feature vector is matched and compared with the records in the preset ship database to obtain the matching and comparison results; S32. The matching comparison results are evaluated according to the preset priority rules. If the evaluation results do not meet the preset confidence conditions, the target vessel is determined to be an unfamiliar vessel.

[0028] As described above, matching and comparing based on a pre-set ship database ensures the objectivity of the judgment and avoids subjective differences caused by human judgment. The introduction of priority rules makes the judgment logic more in line with actual needs. By assigning priorities to different attribute information, the rationality of the judgment results is improved. Pre-set confidence conditions as judgment thresholds realize the quantitative standard of judgment and avoid the problems of missed judgments and misjudgments caused by absolute judgment. The entire judgment process is standardized and objective, which greatly improves the accuracy and consistency of the judgment of unfamiliar ships and gets rid of the dependence on human experience.

[0029] Furthermore, when the program is executed by processor 1, it performs the following steps: In step S4, a fuzzy logic evaluation model is used to conduct a risk assessment of the unfamiliar vessel based on the fused feature vector.

[0030] As described above, the fuzzy logic model can effectively handle the fuzziness and uncertainty in multi-dimensional indicators related to ships, and is more comprehensive and scientific than single-indicator assessments. This model can accurately reflect the actual risk level of unfamiliar ships, making risk level classification more in line with reality and providing a reliable basis for graded early warning. The assessment results can directly correspond to the preset risk level, ensuring the pertinence of early warning responses, improving the efficiency and accuracy of risk assessment, and providing solid support for the generation of subsequent handling recommendations.

[0031] Please refer to Figure 1 Embodiment 1 of the present invention is as follows: A method for intelligent identification and management of ships includes the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

[0032] Step S2 is as follows: S21. Perform spatiotemporal registration and cleaning on the original data to obtain registered spatiotemporally aligned data; S22. Extract sub-features of the target ship under different data modes from the spatiotemporal aligned data; S23. An attention-based fusion algorithm is used to fuse the sub-features to generate a fused feature vector.

[0033] In step S22, the sub-features include at least two of the following: location features, motion features, identity features, and visual appearance features.

[0034] Step S3 is as follows: S31. The fused feature vector is matched and compared with the records in the preset ship database to obtain the matching and comparison results; S32. The matching comparison results are evaluated according to the preset priority rules. If the evaluation results do not meet the preset confidence conditions, the target vessel is determined to be an unfamiliar vessel.

[0035] In step S4, a fuzzy logic evaluation model is used to conduct a risk assessment of the unfamiliar vessel based on the fused feature vector.

[0036] The specific implementation steps of the intelligent ship identification and management method designed in this scheme are as follows: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; The core of this step is to achieve synchronous acquisition and timestamp marking of multi-source data to ensure data integrity and time consistency.

[0037] Sensor selection and deployment: Deploy Beidou shipborne positioning modules, radar detection modules, AIS receiving modules, and high-definition video monitoring modules (i.e., CCTV) within the monitoring area. Each sensor covers the same monitoring area to ensure that the data of the target ship can be captured synchronously by multiple source sensors.

[0038] Data collection content: Beidou shipborne equipment positioning module: receives basic positioning and motion data of the target ship, such as latitude and longitude coordinates, speed, and heading angle; Radar detection module: Employs millimeter-wave radar to detect data such as the distance, azimuth, and relative speed between the target vessel and the radar. AIS receiving module: Collects ship identification information (such as MMSI, ship name, call sign), ship type, size, draft, navigation status and other data; High-definition video surveillance module: captures real-time images and video stream data of the target vessel, including visual information such as the vessel's appearance, hull color, and deck equipment.

[0039] Data time synchronization: All raw data collected by sensors are marked with a unified timestamp, with timestamp accuracy controlled at the millisecond level, ensuring the time synchronization of data from different sources and laying the foundation for subsequent spatiotemporal registration.

[0040] S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; This step achieves deep integration of multi-source data through a three-level process of "data preprocessing - feature extraction - fusion computation," specifically including: S21. Perform spatiotemporal registration and cleaning on the original data to obtain registered spatiotemporally aligned data: Data cleaning: Data filtering and deduplication algorithms are used to remove obvious erroneous data (such as abnormal coordinates outside the monitoring area, invalid data with negative speed), duplicate data (such as the same data collected continuously by the same sensor), and redundant data (such as environmental noise data that is not related to ship identification) from the original data. Time registration: The Kalman filter algorithm is used to correct the time deviation of data from different sensors to ensure that the multi-source data of the same target ship are fully aligned in the time dimension. For example, the timestamps of radar detection data and Beidou positioning data are uniformly calibrated. Spatial registration: The position data under different coordinate systems are uniformly converted into the WGS-84 coordinate system, including the latitude and longitude coordinates of Beidou shipborne equipment and the relative position data detected by radar. All of these are converted into absolute position information under a unified coordinate system to achieve spatial alignment. Data format standardization: Convert the cleaned and registered data into a unified data format (such as JSON format) to facilitate subsequent feature extraction and fusion processing.

[0041] S22. Extract sub-features of the target ship under different data modalities from the spatiotemporal aligned data: From the standardized spatiotemporally aligned data, extract the following sub-features (containing at least two) according to data modality classification: Location characteristics: Based on BeiDou positioning data and radar spatial registration data, extract characteristic parameters such as the target ship's absolute latitude and longitude, distance relative to the monitoring area boundary, and whether it is close to sensitive areas (such as the core area of ​​the port or the border guard line); Motion characteristics: Based on the speed and heading data from BeiDou positioning and the relative speed data detected by radar, dynamic characteristics such as the target ship's speed, acceleration, rate of change of heading, and turning frequency are extracted. Identification features: Based on AIS received data, extract unique identifiers or core identity information such as the ship's MMSI, ship name, call sign, ship type, and port of registration; Visual appearance features: Based on high-definition video surveillance data, visual features such as ship size, appearance outline, deck equipment layout, and ship color distribution are extracted through convolutional neural networks (CNN) to generate visual feature vectors.

[0042] S23. Using an attention-based fusion algorithm, the sub-features are fused to generate a fused feature vector: Feature encoding: Normalize each sub-feature (e.g., normalize continuous features such as speed and distance to the [0,1] interval) and convert them into feature vectors of uniform dimension through their respective encoding networks (e.g., convert them all to 256-dimensional vectors). Attention weight calculation: Construct a multimodal fusion network based on the attention mechanism and calculate the attention weight of each sub-feature vector. The network automatically learns the importance of different sub-features in ship identification. For example, in the identity recognition scenario, the weight of identity identification features is higher than that of visual appearance features. In the scenario without AIS signal, the weights of visual appearance features and motion features are automatically increased. Feature fusion: Based on the calculated attention weights, the sub-feature vectors are weighted and summed to generate a unified fused feature vector (e.g., 256-dimensional), which comprehensively represents the static attributes and dynamic behavior characteristics of the target ship.

[0043] S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S31. Match and compare the fused feature vector with records in the preset ship database; Ship database construction: The preset ship database contains information on all legally registered ships within the monitoring area. The database records include the ship's identification features (MMSI, ship name, etc.), static features (size, type, appearance feature vector), dynamic features (historical navigation trajectory, normal speed range), etc., which correspond one-to-one with the feature dimensions of the fused feature vector. Matching calculation: Using the cosine similarity algorithm or the Euclidean distance algorithm, the similarity between the fused feature vector of the target ship and each record in the database is calculated to obtain a matching score (0-100 points, the higher the score, the higher the matching degree).

[0044] S32. The matching comparison results are evaluated according to the preset priority rules. If the evaluation results do not meet the preset confidence conditions, the target vessel is determined to be an unfamiliar vessel. Priority rule settings: First priority (weight 60%): Identification features (MMSI, ship name). If the matching score of this type of feature is ≥95, it is directly judged as a high matching degree. Second priority (weight 30%): Static features (size, type, appearance features). If the matching score of this type of feature is ≥85, it is judged as a medium match. Third priority (weight 10%): dynamic features (normal speed, navigation trajectory). If the matching score of this type of feature is ≥70, it is judged as a low matching degree. Overall assessment score calculation: Overall assessment score = First priority score × 60% + Second priority score × 30% + Third priority score × 10%; Confidence condition determination: The preset confidence threshold is 80 points. If the comprehensive evaluation score is <80 points, it is considered that "the evaluation result does not meet the preset confidence condition" and the target vessel is determined to be an unfamiliar vessel. If the score is ≥80 points, it is determined to be a registered and legal vessel.

[0045] S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. Construction of Fuzzy Logic Assessment Model: Establish a risk assessment system based on fuzzy logic, and set input variables (from fused feature vectors), fuzzy rules and output variables (risk level). Input variables include multiple dimensions such as the distance between the vessel and the sensitive area, the rate of change of the vessel's speed, whether it deviates from the regular waterway, meteorological and hydrological conditions, and the vessel's historical violation records; Fuzzy rules: For example, "If a ship is less than 1 nautical mile from a sensitive area and its rate of change of speed is greater than 5 m / s², it is considered high risk"; "If a ship deviates from its regular course but is more than 3 nautical miles from a sensitive area, it is considered medium risk". Risk level classification: Three risk levels are calculated using a fuzzy logic model: High risk: Unknown vessels exhibit obvious suspicious behavior (such as approaching sensitive areas at high speeds or frequently changing course to evade surveillance). Medium risk: Unknown vessels behaving abnormally but not posing a direct threat (such as deviating from regular shipping lanes or showing no obvious intention to approach sensitive areas). Low risk: Unknown vessels behave generally normally (e.g., sailing at a constant speed in non-sensitive areas, without abnormal turning); Tiered warning trigger: High risk: Triggering visual alarm (red flashing on the monitoring center's large screen), sound alert (high-pitched alarm), pop-up notification (pushed to the on-duty personnel's mobile APP), and SMS warning (synchronized to the maritime law enforcement officer); Medium risk: Triggers visual alarm (flashing yellow) and pop-up notification; Low risk: Only records the vessel's trajectory, triggering a minor visual cue (blue indicator).

[0046] S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response. Real-time status information supplement: Obtain real-time location, speed, heading, and surrounding environment information (such as the location of nearby law enforcement vessels and the deployment of drones) of unfamiliar vessels; Intelligent response suggestion generation: Based on the built-in rule engine and machine learning algorithm, personalized response plans are generated according to the risk level; If the risk is high, it is recommended to immediately dispatch the nearest law enforcement vessel and plan the shortest route; dispatch drones to the target area for close-range aerial photography and evidence collection; and generate standardized broadcast messages (such as "This is the XX Maritime Monitoring Center. Please report your identity information immediately and stop approaching the sensitive area"). If the risk level is medium, it is recommended to deploy drones for long-distance monitoring and continuously track the vessel's trajectory; and to attempt to contact the vessel via VHF radio to verify its identity information. If the risk is low, it is recommended to continuously monitor the ship's trajectory, record navigation data, and refrain from taking any proactive intervention measures. Linkage response activation means that the system achieves linkage control with external devices through API interfaces, as detailed below: Send navigation route instructions to the law enforcement vessel dispatch system and receive real-time vessel position feedback; Send flight, cruise, and aerial photography commands to the drone control system and receive real-time video streams; Send control commands to VHF radios and audible and visual warning devices to execute announcements and warnings; The handling process is recorded automatically by the system, including the content of the handling suggestions, the response status of the linked equipment, and the subsequent changes in the behavior of the unfamiliar vessel, forming a complete handling file for easy traceability later.

[0047] Please refer to Figure 2 Embodiment two of the present invention is as follows: A ship intelligent identification and management system includes one or more processors 1 and a memory 2. The memory 2 stores a program that, when executed by the processor 1, performs the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

[0048] When this program is executed by processor 1, it performs the following steps: Step S2 is as follows: S21. Perform spatiotemporal registration and cleaning on the original data to obtain registered spatiotemporally aligned data; S22. Extract sub-features of the target ship under different data modes from the spatiotemporal aligned data; S23. An attention-based fusion algorithm is used to fuse the sub-features to generate a fused feature vector.

[0049] When this program is executed by processor 1, it performs the following steps: In step S22, the sub-features include at least two of the following: location features, motion features, identity features, and visual appearance features.

[0050] When this program is executed by processor 1, it performs the following steps: Step S3 is as follows: S31. The fused feature vector is matched and compared with the records in the preset ship database to obtain the matching and comparison results; S32. The matching comparison results are evaluated according to the preset priority rules. If the evaluation results do not meet the preset confidence conditions, the target vessel is determined to be an unfamiliar vessel.

[0051] When this program is executed by processor 1, it performs the following steps: In step S4, a fuzzy logic evaluation model is used to conduct a risk assessment of the unfamiliar vessel based on the fused feature vector.

[0052] In summary, this invention provides a ship intelligent identification and management method and system. Step S1 collects raw data from multiple heterogeneous sensors, ensuring the comprehensiveness of data sources and providing a rich foundation for subsequent analysis, breaking the limitations of a single data source. Step S2 fuses the raw data to generate a fused feature vector, effectively integrating multi-source data and making ship feature representation more comprehensive and accurate, providing a reliable basis for identifying unfamiliar ships. Step S3 determines whether the target ship is an unfamiliar ship based on the fused feature vector, replacing manual experience judgment with data-driven judgment logic, avoiding subjectivity and negligence, and solving the core problem of excessive reliance on manual labor in existing technologies. Step S4 conducts risk assessment for unfamiliar ships and triggers corresponding early warnings, achieving seamless connection between identification and early warning, ensuring rapid response to risks from unfamiliar ships. Step S5 generates disposal suggestions and initiates linkage response based on risk level and real-time status information, forming a complete process from identification to disposal, improving the automation and efficiency of ship management. Each step is interconnected, jointly achieving intelligent identification and efficient management of unfamiliar ships, significantly reducing manual intervention.

[0053] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for intelligent identification and management of ships, characterized in that, Includes the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

2. The intelligent identification and management method for ships according to claim 1, characterized in that, Step S2 is as follows: S21. Perform spatiotemporal registration and cleaning on the original data to obtain registered spatiotemporally aligned data; S22. Extract sub-features of the target ship under different data modes from the spatiotemporal aligned data; S23. An attention-based fusion algorithm is used to fuse the sub-features to generate a fused feature vector.

3. The intelligent identification and management method for ships according to claim 2, characterized in that, In step S22, the sub-features include at least two of the following: location features, motion features, identity features, and visual appearance features.

4. The intelligent identification and management method for ships according to claim 1, characterized in that, Step S3 is as follows: S31. The fused feature vector is matched and compared with the records in the preset ship database to obtain the matching and comparison results; S32. The matching comparison results are evaluated according to the preset priority rules. If the evaluation results do not meet the preset confidence conditions, the target vessel is determined to be an unfamiliar vessel.

5. The intelligent identification and management method for ships according to claim 1, characterized in that, In step S4, a fuzzy logic evaluation model is used to conduct a risk assessment of the unfamiliar vessel based on the fused feature vector.

6. A ship intelligent identification and management system, characterized in that, Includes one or more processors and a memory, wherein the memory stores a program that, when executed by the processor, performs the following steps: S1. Collect raw data about target ships within the same monitoring area from multiple heterogeneous sensors; S2. Perform fusion processing on the original data to generate the fusion feature vector of the target ship; S3. Based on the fused feature vector, determine whether the target vessel is an unfamiliar vessel; S4. If the target vessel is an unfamiliar vessel, a risk assessment is performed on the unfamiliar vessel based on the fused feature vector to determine the risk level and trigger a corresponding early warning. S5. Based on the risk level determined in step S4 and the real-time status information of the unfamiliar vessel, generate a handling suggestion and initiate the corresponding linkage response.

7. The ship intelligent identification management system according to claim 6, characterized in that, When the program is executed by the processor, it performs the following steps: Step S2 is as follows: S21. Perform spatiotemporal registration and cleaning on the original data to obtain registered spatiotemporally aligned data; S22. Extract sub-features of the target ship under different data modes from the spatiotemporal aligned data; S23. An attention-based fusion algorithm is used to fuse the sub-features to generate a fused feature vector.

8. The ship intelligent identification management system according to claim 7, characterized in that, When the program is executed by the processor, it performs the following steps: In step S22, the sub-features include at least two of the following: location features, motion features, identity features, and visual appearance features.

9. The ship intelligent identification management system according to claim 6, characterized in that, When the program is executed by the processor, it performs the following steps: Step S3 is as follows: S31. The fused feature vector is matched and compared with the records in the preset ship database to obtain the matching and comparison results; S32. The matching comparison results are evaluated according to the preset priority rules. If the evaluation results do not meet the preset confidence conditions, the target vessel is determined to be an unfamiliar vessel.

10. The ship intelligent identification management system according to claim 6, characterized in that, When the program is executed by the processor, it performs the following steps: In step S4, a fuzzy logic evaluation model is used to conduct a risk assessment of the unfamiliar vessel based on the fused feature vector.

Images