An AI and big data-based ship transportation safety supervision method and system

Abstract

The present invention discloses a method and system for ship transportation safety supervision based on AI and big data, specifically related to the field of ship traffic supervision, and is used to solve the problems of identifying the position of hidden ships, difficult to dynamically identify blind spots in supervision, and insufficient traffic risk warning ability in the existing ship traffic supervision process; collecting AIS ship dynamic data through base stations to generate a real-time traffic state set, marking suspected hidden ships after detecting the loss of ship AIS signals, analyzing the steering and avoidance behaviors of surrounding ships, inferring the hidden probability position domain through reverse cross-positioning, performing spatio-temporal correlation verification on hidden targets and inferring the range of their motion parameters, and at the same time identifying the migration path and range of blind spots in supervision based on continuous monitoring periods, generating a dynamic supervision risk area in the electronic nautical chart and implementing traffic supervision warnings for ships entering the area, so as to realize the dynamic identification of hidden ship activities and traffic safety supervision.

Description

Technical Field

[0001] This invention relates to the field of ship traffic supervision technology, and more specifically, to a method and system for ship traffic safety supervision based on AI and big data. Background Technology

[0002] With the continuous growth of global maritime transport, the traffic density of ships in port channels, bridge areas and traffic separation zones is constantly increasing. Maritime regulatory authorities usually rely on Automatic Identification System (AIS) to monitor the navigation status of ships in their jurisdiction in real time. They receive dynamic navigation information reported by ships through AIS base stations to achieve traffic organization, navigation safety early warning and accident risk control.

[0003] However, in actual shipping environments, some vessels may maliciously shut down or abnormally interrupt their AIS signals during navigation for reasons such as evading supervision, engaging in illegal operations, or equipment malfunctions. This prevents the monitoring system from continuously acquiring the vessel's real-time position and movement status. Even when a vessel shuts down its AIS in densely trafficked waters, it may still affect surrounding vessels navigating normally, such as causing them to suddenly turn, slow down, or change course. However, due to the lack of direct AIS data support, traditional monitoring systems struggle to identify the location and movement trends of such concealed targets in a timely manner. Furthermore, AIS base station signal coverage often exhibits spatial unevenness in complex waterway environments, potentially creating monitoring blind spots in areas such as bridge areas, shoreline obstructions, or areas far from base stations. When concealed vessels operate in these areas, the monitoring system finds it even more difficult to effectively grasp their movement trajectories and their impact on the surrounding traffic structure, thereby increasing the risks encountered by vessels and the difficulty of traffic organization.

[0004] Therefore, in actual maritime traffic supervision scenarios, how to identify the location of hidden vessels by observing changes in the behavior of surrounding vessels when AIS signals are missing, and how to dynamically identify regulatory blind spots and their migration trends by combining traffic flow changes, has become an important technical issue for improving the ability to supervise maritime traffic safety. Summary of the Invention

[0005] In order to overcome the above-mentioned deficiencies of the prior art, embodiments of the present invention provide a method and system for ship transportation safety supervision based on AI and big data to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for supervising ship transportation safety based on AI and big data includes the following steps:

[0008] S1. The base station collects real-time ship dynamic data from AIS and generates a set of current real-time traffic status data.

[0009] S2. When a vessel loses its AIS signal after entering the traffic-controlled waters and its speed was greater than zero before the signal was lost, the corresponding vessel will be marked as a concealed suspected vessel and its last position and motion parameters will be recorded.

[0010] S3. Extract the real-time traffic status set of all ships within a preset range around the last location of the concealed suspected ship, identify the ship clusters that synchronously turn to avoid it, perform reverse cross-location based on the spatial orientation of the ship clusters avoiding it, and mark the concealment probability location domain.

[0011] S4. Perform spatiotemporal correlation verification between the concealment probability location domain and the last position of the concealed suspected vessel to lock the concealed target, and infer the current speed range and heading range of the concealed target based on the distribution of the avoidance direction of the vessel cluster within the concealed target location domain and the last motion parameters.

[0012] S5. Based on the location sequence of the same concealed target within multiple consecutive monitoring periods, identify the location changes of the regulatory blind spot caused by the continuous movement of the concealed target, and generate the migration path and range of the regulatory blind spot.

[0013] S6. Based on the migration path and scope of the regulatory blind spot, generate traffic dynamic monitoring risk areas on the electronic nautical chart and issue traffic monitoring early warnings.

[0014] As a further aspect of the present invention, in step S1, generating the current real-time traffic status set specifically includes:

[0015] The real-time ship dynamic navigation data collected by the base station is cleaned to remove abnormal data points with speeds exceeding the threshold corresponding to the ship type and cross-boundary data points with coordinates exceeding the jurisdiction boundary. Data with the same ship identification code is sorted by timestamp to form a sequence of ship motion trajectories.

[0016] The ship motion trajectory sequence is associated with static data including ship length and ship type to generate a dynamic attribute vector for each ship, which includes real-time position, motion parameters and ship type. The motion parameters include instantaneous speed and instantaneous heading.

[0017] Spatially synchronize the dynamic attribute vectors of all ships at the current moment to generate a real-time ship traffic status set.

[0018] As a further aspect of the present invention, in S2, when a vessel loses its AIS signal after entering the traffic-controlled waters and its speed was greater than zero before the loss, marking the corresponding vessel as a suspected concealed vessel and recording its last position and motion parameters specifically includes:

[0019] Real-time monitoring of the AIS message reception interval for each vessel; when the signal loss duration exceeds a preset threshold after a vessel enters the traffic-controlled waters, record the last position coordinates, last speed, last course, and last timestamp of the last frame of the message before the loss.

[0020] Extract the historical speed sequence of the vessel before it was lost, and combine it with the last speed to determine whether the vessel was in a maneuvering state at the time of loss.

[0021] If the vessel is in a maneuvering state, the type of the regulated waterway where the vessel is currently located is queried based on the last position coordinates, and the suspected weight of the vessel's covert behavior is generated based on the type of regulated waterway.

[0022] By associating the last position coordinates, last speed, last course, last timestamp, maneuvering status determination results, and the weight of suspected concealment behavior, a concealment suspicion record of the vessel is generated and added to the list of suspected concealment vessels.

[0023] As a further aspect of the present invention, in step S3, reverse cross-positioning is performed based on the spatial orientation of the ship cluster avoidance, and the concealment probability location domain is marked specifically including:

[0024] The search area is defined with the last location of the concealed suspected vessel as the center. The course and speed time sequences of all vessels in the search area within the preset time before and after the signal loss are extracted from the real-time vessel traffic status set.

[0025] The heading and speed time sequences of each ship are input into a pre-trained collision avoidance behavior recognition model. The collision avoidance behavior recognition model identifies the current motion state of the ship by turning to avoid collision based on the ship type, heading change characteristics, speed change characteristics and time sequence characteristics.

[0026] Starting from the position coordinates of each ship identified as turning to avoid an obstacle at the beginning of the avoidance maneuver, a ray is drawn in the opposite direction of its instantaneous course. The pairwise intersections of all rays are calculated and the coordinates of the intersections are spatially clustered. The cluster center is retained as the estimated location of the hidden interference source.

[0027] Centered on the estimated locations of all concealed interference sources, and combined with the distribution range of all avoidant vessels, an elliptical region containing confidence boundaries is generated and marked as the concealment probability location domain.

[0028] As a further aspect of the present invention, in step S4, the process of inferring the current speed range and heading range of the concealed target based on the distribution of ship cluster avoidance directions within the concealed target's location domain and the last motion parameters specifically includes:

[0029] Extract the last timestamp of the concealed suspected vessel, and at the same time obtain the average start time of the avoidance action of all avoidance vessels within the concealment probability location domain, and calculate the time difference between the two as the duration of the movement.

[0030] The final course of the concealed suspected vessel is compared with the main peak direction of the statistical distribution of the heading direction of the avoidance vessel within the concealment probability location domain. When the angle difference between the two is less than a preset threshold, the course consistency is determined.

[0031] Extract the final speed and combine it with the duration of motion to estimate the theoretical sailing distance. At the same time, calculate the distance between the centroid of the concealment probability location domain and the line connecting the final position. When the ratio of the theoretical sailing distance to the line connecting the final position is within a preset correlation range, the spatiotemporal correlation is determined to be established, the concealment suspected vessel is locked as the concealment target, and the corresponding concealment probability location domain is marked as the concealment target location domain.

[0032] Extract the distance between the centroid of the hidden target's location domain and its last location, estimate the estimated speed by combining the average duration of the avoidance maneuvers of the ship cluster within the range, and take the main peak direction of the avoidance ship's course as the estimated course to generate the current motion parameter range of the hidden target.

[0033] As a further aspect of the present invention, in step S5, generating the migration path and range of the regulatory blind spot specifically includes:

[0034] Extract the location domain of the same hidden target locked within multiple consecutive monitoring periods, and obtain the deployment location and signal coverage spatial distribution of all base stations in the traffic monitoring water area;

[0035] For each monitoring period, the ratio of the area of ​​intersection between the target location and the base station signal coverage to the total area of ​​the target location is calculated as the signal coverage. The target location with signal coverage below a preset threshold is marked as a regulatory blind spot.

[0036] Based on the current speed range and heading range of the concealed target obtained in the previous monitoring cycle, calculate the theoretical displacement range of the centroid coordinates of the monitoring blind zone between adjacent monitoring cycles, eliminate abnormal jump points that do not conform to the theoretical displacement range, and retain the monitoring blind zone sequence with continuous motion.

[0037] Connect the centroid coordinates of the regulatory blind spots in chronological order to generate the migration path curve of the regulatory blind spots, calculate the cumulative distance between adjacent centroid coordinates as the cumulative migration distance, extract the spatial radius of the regulatory blind spots in each cycle, and take the maximum value of all cycle radii as the range of the regulatory blind spots.

[0038] As a further aspect of the present invention, in step S6, generating a traffic dynamic monitoring risk area on the electronic nautical chart and issuing a traffic monitoring early warning based on the migration path and range of the monitoring blind spot specifically includes:

[0039] Based on the migration path curves of the regulatory blind spots corresponding to all locked hidden targets in the list of hidden suspected vessels, the regulatory blind spot range is expanded equidistantly to both sides to generate a strip-shaped dynamic risk area that includes the entire path. The coordinates of vessels with AIS normally enabled are extracted from the real-time vessel traffic status set, and traffic supervision warnings are issued for vessels entering the dynamic risk area.

[0040] On the other hand, the present invention provides a ship transportation safety supervision system based on AI and big data, comprising:

[0041] The traffic status construction module is used to receive AIS ship dynamic data collected in real time by the base station, clean, sort and associate the ship dynamic navigation data, and generate a set of current real-time traffic status.

[0042] The concealed suspect identification module is used to monitor the continuity of AIS signals after a vessel enters a traffic-controlled waterway and to identify concealed suspect vessels that have lost signals and are in a maneuvering state.

[0043] The shadow position inversion module is used to extract the course time series and speed time series of ships around the concealed suspected ship, identify ships that turn to avoid them, and reverse cross-locate the concealment probability position domain.

[0044] The concealed target locking module is used to perform spatiotemporal correlation verification between the concealment probability location domain and the last position of the concealed suspected vessel, lock the concealed target, and infer its current speed range and heading range.

[0045] The blind spot monitoring and early warning module is used to identify the migration path and range of the blind spot based on the location domain of the hidden target within a continuous monitoring period, and to generate dynamic monitoring risk areas to implement traffic warnings.

[0046] The technical effects and advantages of the present invention, a method and system for monitoring ship transportation safety based on AI and big data, are as follows:

[0047] This invention combines AIS vessel dynamic data, vessel avoidance behavior characteristics, and base station signal coverage information to achieve indirect identification and dynamic monitoring of concealed vessel activities when AIS signals are abnormal or off. By analyzing the course changes and avoidance behaviors of vessels surrounding suspected concealed vessels, and using the spatial orientation of avoidance behavior for reverse cross-location, the probabilistic location domain of concealed vessels can be inferred even in the absence of direct AIS data, thus compensating for the inability of traditional monitoring systems to locate vessels with missing signals in real time. Simultaneously, by performing spatiotemporal correlation verification between the probabilistic location domain and the vessel's last AIS position, and by inferring the range of motion parameters, concealed targets can be further located and their possible movement trends inferred. Based on this, by analyzing the relationship between the concealed target's position changes and base station signal coverage within a continuous monitoring period, monitoring blind spots are identified and migration paths for these blind spots are generated, enabling the monitoring system to dynamically grasp the impact of concealed vessel activities on the traffic monitoring area. Furthermore, dynamic risk areas are generated based on the migration path and scope of the regulatory blind spots, and warnings are issued to vessels navigating normally entering these areas. This improves the real-time and forward-looking nature of maritime traffic supervision, reduces the risk of regulatory blind spots caused by the lack of AIS signals, and enhances the ability to ensure the safety of vessel traffic in complex water environments. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of a ship transportation safety supervision method based on AI and big data according to the present invention;

[0049] Figure 2 This is a schematic diagram of the structure of a ship transportation safety supervision system based on AI and big data according to the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0051] Example 1

[0052] Figure 1 This invention presents a method for supervising ship transportation safety based on AI and big data, which includes the following steps:

[0053] S1. The base station collects real-time ship dynamic data from AIS and generates a set of current real-time traffic status data.

[0054] S2. When a vessel loses its AIS signal after entering the traffic-controlled waters and its speed was greater than zero before the signal was lost, the corresponding vessel will be marked as a concealed suspected vessel and its last position and motion parameters will be recorded.

[0055] S3. Extract the real-time traffic status set of all ships within a preset range around the last location of the concealed suspected ship, identify the ship clusters that synchronously turn to avoid it, perform reverse cross-location based on the spatial orientation of the ship clusters avoiding it, and mark the concealment probability location domain.

[0056] S4. Perform spatiotemporal correlation verification between the concealment probability location domain and the last position of the concealed suspected vessel to lock the concealed target, and infer the current speed range and heading range of the concealed target based on the distribution of the avoidance direction of the vessel cluster within the concealed target location domain and the last motion parameters.

[0057] S5. Based on the location sequence of the same concealed target within multiple consecutive monitoring periods, identify the location changes of the regulatory blind spot caused by the continuous movement of the concealed target, and generate the migration path and range of the regulatory blind spot.

[0058] S6. Based on the migration path and scope of the regulatory blind spot, generate traffic dynamic monitoring risk areas on the electronic nautical chart and issue traffic monitoring early warnings.

[0059] In step S1, a set of current real-time traffic conditions is generated.

[0060] The system receives AIS message data continuously collected by the base station. Each AIS message contains at least a ship identification code, latitude and longitude coordinates, ground speed, ground heading, and timestamp information. The data is then categorized based on the ship identification code, and ship type information is read to determine the corresponding reasonable speed range. Speed ​​thresholds are preset according to ship type; for example, the speed threshold is set to 25 knots for large cargo ships, 22 knots for oil tankers, and 30 knots for passenger ships. When the speed in a message exceeds the speed threshold for the corresponding ship type, the data point is determined to be abnormal and removed. Simultaneously, a regulatory area boundary is established based on the spatial scope of the traffic-regulated waterway. The regulatory boundary is marked using latitude and longitude coordinates on an electronic nautical chart. For example, a regulatory waterway can be defined as a rectangular area between 118°30′ and 118°55′ east longitude and 24°20′ and 24°45′ north latitude. When a ship's coordinates exceed this area, the data point is determined to be out-of-bounds data and removed. After cleaning up abnormal and out-of-bounds data, the remaining data is grouped according to the vessel identification code and sorted from earliest to latest based on the timestamp, thus forming a continuous sequence of vessel movement trajectories. This trajectory sequence records the changes in the spatial position and navigation status of vessels within the regulated waters in chronological order, accurately reflecting the navigation trajectory of vessels in the regulated waters.

[0061] The system reads static ship data from the ship registration information database. This static data includes fixed attribute information such as ship identification code, ship length, and ship type. The static data is then associated with the corresponding ship motion trajectory sequence according to the ship identification code. Subsequently, the latest AIS record for the current moment is read from the trajectory sequence, and its latitude and longitude coordinates are used as the ship's real-time position. The corresponding ground speed and ground heading are used as the ship's current motion parameters. Simultaneously, the ship length and ship type information are integrated to form a dynamic attribute vector describing the current navigation state of a single ship. This dynamic attribute vector includes at least the ship identification code, real-time position coordinates, instantaneous speed, instantaneous heading, ship length, and ship type. Based on this, the dynamic attribute vectors of all ships at the current moment are uniformly time-aligned. All ship dynamic attributes collected at the same moment are aggregated and organized according to spatial location to form a real-time ship traffic status set for the current moment. This set records the real-time navigation status of all ships within the regulated waters in a spatially distributed form, including the number of ships, their position distribution, speed status, and heading changes, providing basic data for subsequent identification of abnormal ship behavior and analysis of traffic structure.

[0062] In S2, when a vessel loses its AIS signal after entering the traffic-controlled waters and its speed was greater than zero before the signal was lost, the corresponding vessel is marked as a suspected concealed vessel and its last position and motion parameters are recorded.

[0063] The system continuously receives vessel dynamic messages uploaded by AIS base stations within its jurisdiction, classifies and records these messages based on vessel identification codes, and continuously monitors the reception time interval between consecutive messages from the same vessel. The normal transmission cycle of AIS messages is identified according to international AIS standards. For example, vessels typically send dynamic messages every 2 to 10 seconds at speeds greater than 14 knots, and at lower speeds, the transmission interval is typically 10 to 30 seconds. When a vessel enters the traffic-monitored waters, if no AIS messages are received from that vessel for several consecutive cycles, the signal loss time is accumulated. When the accumulated loss time exceeds a preset threshold, it is considered an abnormal AIS signal interruption. This threshold is set through historical AIS data statistics. For example, selecting the average interval of messages from AIS communication data in the monitored waters over the past three months; if the average interval is 12 seconds, the signal loss threshold can be set to 120 seconds, meaning that a signal loss is considered to have occurred if no messages are received for 120 consecutive seconds. After confirming the signal loss, the last frame of the message before the signal interruption is extracted from the ship's historical AIS records. The latitude and longitude coordinates corresponding to the message are read as the ship's last position coordinates. At the same time, the corresponding ground speed, ground heading and message timestamp are read and recorded in a unified manner.

[0064] Speed ​​records within a certain time range are extracted from the historical data of the vessel before the AIS signal was lost, forming a continuous speed sequence. The length of the historical time is set based on regulatory experience; for example, speed data from 5 to 10 minutes before the signal loss can be selected as the basis for judgment. After obtaining the speed sequence, the speed at each moment in the sequence is read and combined with the last speed before the signal loss for comprehensive judgment. If the speed sequence remains above 1 knot and the last speed is greater than 2 knots, the vessel is determined to be in a maneuvering state; if the speed remains close to 0 knots, it is determined to be in an anchored or stationary state and is not included in the scope of suspected concealment. Once the vessel is confirmed to be in a maneuvering state, the type of traffic-regulated waterway where the vessel is located is queried on the electronic nautical chart based on the last position coordinates. The type of regulated waterway is divided by pre-marking, such as channel curvature areas, lane-separating intersection areas, bridge areas, and ordinary navigation areas. Different waterways have different levels of traffic risk, and therefore different sensitivities to concealment behavior. Based on the type of water area, a weight is assigned to the suspected concealment behavior of the vessel. For example, the weight is set to 0.9 in military-controlled waters, 0.8 in economically controlled waters, and 0.5 in ordinary navigation areas.

[0065] The acquired information is integrated to create a complete concealment suspicion record for each vessel exhibiting AIS signal anomalies. This record includes the vessel's identification code, last coordinates, last speed, last course, signal interruption time, maneuvering status, and the weight of suspected concealment behavior, all organized according to a unified data structure. A concealment suspicion vessel list is then created in the monitoring records, centrally storing information on all vessels meeting the criteria for concealment behavior. Once a vessel's information is integrated, its record is added to the concealment suspicion vessel list, arranged chronologically, allowing monitoring personnel to visually view all vessels exhibiting AIS signal anomalies in the current waters. In actual operation, if the same vessel remains in a signal loss state for multiple monitoring periods, its record is continuously updated to maintain the real-time information in the list. This method generates a complete concealment suspicion vessel list, providing data support for subsequent analysis of vessel avoidance behavior to determine the location of concealed vessels and ensuring the continuity and consistency of target identification during the monitoring process.

[0066] In step S3, reverse cross-positioning is performed based on the spatial orientation of the ship cluster avoidance, and the concealment probability location domain is marked.

[0067] After identifying a vessel as a suspected concealment vessel, a spatial search area is delineated on the electronic nautical chart, centered on the vessel's last position coordinates before the loss of its AIS signal, to analyze surrounding traffic behavior. This search area is constructed as a circular area, with the suspected vessel's last position coordinates at its center, and the radius set based on the typical avoidance distance of vessels. This radius is determined through historical traffic data statistics; for example, by selecting vessel encounter records within the regulated waters over the past year. When most avoidance behaviors occur within 1.5 to 3 nautical miles, the search radius can be set to 3 nautical miles to ensure coverage of potentially affected vessels.

[0068] After defining the spatial range, navigation records of all vessels within the search area are extracted from the real-time vessel traffic status dataset within a certain time range before and after signal loss. This time range is set according to the typical duration of vessel avoidance behavior, for example, selecting navigation records from 3 minutes before to 3 minutes after signal loss, and forming a heading time series and a speed time series in chronological order. Subsequently, the heading sequence, speed sequence, and vessel type information corresponding to each vessel are fed as input data into a pre-trained avoidance behavior recognition model. This recognition model is constructed using a multi-layer time-series neural network structure, including a feature extraction layer, a time-series encoding layer, and a behavior classification layer. The feature extraction layer encodes the magnitude of heading changes, speed changes, and the duration of these changes; the time-series encoding layer uses a two-layer long short-term memory network structure to extract temporal features from continuous navigation state changes; and the behavior classification layer outputs the current behavior category of the vessel through a fully connected network. During the model training phase, manually labeled obstacle avoidance behavior samples from historical traffic monitoring data were selected as the training dataset. The training samples included normal navigation samples and turning obstacle avoidance samples. Approximately 20,000 vessel navigation sequences were extracted from traffic monitoring records over the past two years as training data. Through multiple rounds of iterative training, the model was able to identify typical obstacle avoidance behavior features such as sudden changes in course, speed reduction, and continuous course deviation. After training, real-time acquired vessel navigation sequences were input into the model for inference and recognition. When the recognition result indicated a turning obstacle avoidance behavior, the corresponding vessel was marked as the obstacle avoidance vessel.

[0069] After identifying the vessels that steered to avoid a collision, a spatial reverse analysis is performed on the navigation directions of all vessels to infer the location of potential interference sources. Specifically, the latitude and longitude coordinates of each vessel at the start of its avoidance maneuver are first read, and these coordinates are used as the starting point of a ray. The instantaneous heading of the vessel at that moment is then read. Subsequently, a spatial ray is constructed in the opposite direction of this heading. This reverse ray represents the direction of the potential target being avoided by the vessel's avoidance maneuver. After constructing reverse rays for all vessels, the spatial intersection points between any two rays are calculated, excluding parallel rays. Due to differences in the avoidance directions of different vessels, ray intersection points typically form a set of spatially distributed points, and these concentrated areas of intersection points often correspond to the possible locations of concealed vessels. Then, spatial clustering is performed on all intersection point coordinates. A density-based spatial clustering method is used, grouping closely spaced intersection points into the same cluster. For example, intersection points less than 200 meters apart are considered to be in the same cluster. This method can filter out discrete intersection points caused by deviations in the headings of individual vessels. After clustering, the center position of each intersection cluster is calculated by averaging the coordinates of all intersection points within the cluster. This center point serves as the presumed location of the concealed interference source. When multiple intersection clusters exist, the cluster center containing the most intersection points is selected as the primary presumed location, as this region represents the common direction of most ships' avoidance paths, thus more accurately reflecting the potential spatial location of concealed vessels.

[0070] Spatial coordinates of all intersection point cluster centers were collected and statistically analyzed. Spatial distribution characteristics were obtained by calculating the dispersion of each coordinate point along the longitude and latitude directions. A two-dimensional spatial covariance structure was then constructed based on these characteristics to reflect the diffusion of the estimated location in different directions. After obtaining the covariance structure, an elliptical probability region was constructed on the electronic nautical chart. The center of this elliptical region was set as the average position of all cluster center coordinates. The major axis of the ellipse was set according to the direction of the most concentrated intersection point distribution, and the minor axis was perpendicular to the major axis. The lengths of the major and minor axes of the ellipse were determined based on the spatial distribution range of the intersection points. For example, when the intersection points were relatively dispersed in the east-west direction, the major axis length was set to 1.2 times the maximum span of the intersection points, and the minor axis length was set to 1.2 times the north-south span, thus forming a probability region covering the main estimated locations. Subsequently, the boundaries of the elliptical region were corrected based on the spatial distribution of avoidance vessels. When the avoidance vessels were mainly concentrated on one side, the center position of the elliptical region was appropriately offset so that the region could cover most ray intersection points. Finally, an elliptical region with spatial confidence boundaries is generated and marked as the concealment probability location domain. This region represents the spatial range in which the concealed vessel is most likely to exist during the current monitoring period.

[0071] In S4, the current speed range and heading range of the concealed target are inferred from the distribution of the avoidance directions of the ship cluster within the concealed target's location domain and the last motion parameters.

[0072] After completing the concealment probability location domain identification, the navigation time characteristics of the suspected concealing vessel and the time characteristics of its surrounding avoidance behaviors are jointly analyzed to determine the duration of the concealing vessel's movement after signal loss. First, the timestamp corresponding to the last frame of the suspected concealing vessel's AIS signal before loss is read and used as the last time the concealing vessel was monitored. Simultaneously, all vessels identified as engaging in turning and avoidance behavior within the concealment probability location domain are statistically analyzed. The course change curve of each avoiding vessel is scanned, and the start time of the avoidance action is determined by identifying the moment when the continuous course change begins. Subsequently, the start times of all avoiding vessels are statistically analyzed, and the average start time of the avoidance behavior is obtained by averaging. For example, if there are 5 avoiding vessels in a certain monitoring period, and their avoidance action start times are 12:01:10, 12:01:12, 12:01:15, 12:01:11, and 12:01:14 respectively, then the statistically obtained average start time is approximately 12:01:12. The time difference between this average time and the last AIS time of the concealed suspected vessel is calculated to obtain the duration of the concealed vessel's potential influence on surrounding vessels. After obtaining the duration, the statistical distribution of the concealed suspected vessel's last course and the bearing lines of the avoidance vessels within the concealment probability location domain is compared. In practice, the reverse course directions of all avoidance vessels are statistically analyzed and grouped according to their angle distribution. The direction with the highest frequency is selected as the main peak direction of avoidance behavior. For example, if the reverse course of most avoidance vessels is concentrated around the 225° direction, this direction is identified as the main peak direction. Subsequently, the angle difference between the concealed suspected vessel's last course and this direction is compared. When the angle difference is less than a preset threshold, the two directions are considered to be consistent. This threshold is set through historical navigation data statistics, for example, 30° based on the typical vessel turning deviation range. When the angle difference is less than 30°, the course consistency is determined to be established.

[0073] After confirming course consistency, the reasonableness of the movement distance of the suspected concealed vessel is verified to determine whether there is a real correlation between the probability position domain and the suspected vessel. During implementation, the last speed of the suspected concealed vessel before losing its AIS signal is first read, and the theoretical distance the vessel could have traveled during that time period is estimated by combining this with the previously obtained movement duration. For example, if the last speed was 10 knots and the duration was approximately 3 minutes, the corresponding theoretical distance is approximately 0.5 nautical miles. Then, the geometric center position of the concealment probability position domain is read, and the spatial distance between this center position and the last AIS position of the suspected concealed vessel is calculated. This distance is converted to the actual navigation distance using latitude and longitude coordinates, for example, using a nautical chart scale or a latitude and longitude distance conversion table. After completing the distance calculation, the theoretical navigation distance and the actual spatial distance are compared proportionally. To avoid errors caused by factors such as ocean currents and ship handling deviations, a correlation interval needs to be set to determine whether the two match. This interval is determined through historical navigation data statistics; for example, after statistically analyzing the navigation trajectories of normal vessels to obtain a typical distance error range, the distance ratio interval can be set between 0.5 and 2.0. When the ratio of the actual distance to the theoretical distance falls within this range, it is determined that there is a reasonable spatiotemporal correlation between the concealment probability location domain and the concealed suspected vessel. After completing the correlation verification, the concealed suspected vessel is officially identified as a concealed target, and the corresponding concealment probability location domain is marked as the concealed target location domain, indicating that this area is the spatial range in which the concealed vessel is most likely to be located during the current monitoring period.

[0074] After confirming the correlation between the concealed target and its location domain, the current motion parameters of the concealed target are further estimated to obtain its navigation status during the period of AIS signal loss. First, the centroid coordinates of the concealed target's location domain are read, and the spatial distance between this centroid and the concealed target's last AIS position is calculated. This distance reflects the actual displacement that the concealed vessel might have experienced after the signal loss. Then, the duration of the avoidance actions of all vessels within the location domain is statistically analyzed, and the durations are averaged to obtain the average duration of the vessel group's avoidance behavior. For example, if the avoidance durations of multiple vessels in a certain monitoring period are 2.8 minutes, 3.1 minutes, 3.0 minutes, and 2.9 minutes respectively, the average duration is approximately 3 minutes. Using the above spatial distance and average duration, the concealed target's speed is estimated. The approximate value of the concealed vessel's current speed is obtained through the ratio of distance to time, and this speed is used as the center value of the concealed target's speed range. Simultaneously, combined with the aforementioned statistical results of the avoidance behavior direction, the main peak direction in the distribution of the avoidance vessel's heading is used as the estimated direction of the concealed vessel's heading. To reflect the uncertainty of the course, a certain angular range is set on both sides of the main peak direction, for example, within ±20° of the main peak direction as the possible course range of the concealed target. This ultimately forms a description of the concealed target's motion parameters, including speed and course ranges, used to represent the possible motion state of the concealed vessel within the current monitoring period. This method enables continuous estimation of the motion trend of concealed vessels even in the absence of AIS signals.

[0075] In step S5, the migration path and range of the regulatory blind spot are generated.

[0076] The location domains of the same concealed target within a continuous monitoring period are extracted according to the monitoring time sequence, and the spatial range and center coordinates of each location domain are read. Subsequently, the deployment locations of all AIS base stations within the traffic-monitored waters are obtained, and the signal coverage range of each base station is determined based on the base station equipment parameters. The base station signal coverage range is determined through on-site communication testing and historical communication record statistics. For example, based on the distribution of AIS message reception success rates over the past year, the stable reception range of each base station in different directions is determined, and the signal coverage area is plotted on the electronic nautical chart. When there are many base stations, the coverage areas of all base stations are superimposed to form the overall communication coverage range. Then, the location domains of concealed targets within each monitoring period are spatially superimposed with the communication coverage range, and the area of ​​the overlapping region is calculated and compared proportionally with the total area of ​​the location domains of concealed targets to obtain the signal coverage degree of that location domain. For example, if the area of ​​a location domain is approximately 0.8 square nautical miles, and the overlapping area with the base station coverage area is 0.16 square nautical miles, then the signal coverage degree of that location domain is 0.2. To identify areas with insufficient communication surveillance, a coverage threshold needs to be set. This threshold is determined through historical communication quality statistics; for example, an area with insufficient communication coverage is identified when the coverage is below 0.3. Therefore, when the coverage of a certain monitoring period is less than 0.3, the corresponding concealed target location area is marked as a surveillance blind spot, and its spatial center coordinates and area range are recorded. This is used to describe the area where concealed targets may be active at the current moment, but AIS has difficulty receiving stable reception.

[0077] After obtaining the location of the blind spot in multiple monitoring cycles, its spatial changes are continuously filtered to eliminate abnormal jumps caused by positioning errors or data fluctuations. First, the estimated speed and heading range of the concealed target in the previous monitoring cycle are read, and these motion parameters are used as a reference for the possible displacement of the concealed vessel between two monitoring cycles. Then, the time interval between two adjacent monitoring cycles is calculated, for example, if the monitoring cycle is set to 30 seconds or 60 seconds, and the theoretical displacement range that the concealed vessel may generate within that time period is estimated based on the speed range and time interval. For example, when the speed range is 8 to 12 knots and the time interval is 1 minute, the corresponding theoretical displacement range is approximately 250 to 370 meters. After obtaining the theoretical displacement range, the distance between the centroid coordinates of the blind spot in two adjacent monitoring cycles is calculated to obtain the actual spatial displacement between the two centroids. When the actual displacement falls near the theoretical displacement range, the change in the blind spot is considered to conform to the continuous motion characteristics of the concealed vessel; if the actual displacement is much greater or much less than this range, for example, the displacement reaches more than 1 nautical mile or is close to zero displacement, it is determined to be an abnormal jump point. Monitoring points identified as exhibiting abnormal jumps were eliminated, retaining only the locations of regulatory blind spots whose displacement conformed to the continuity of movement. After this screening process, a sequence of regulatory blind spots that continuously changed in time and space was obtained, which could accurately reflect the movement trajectory of communication blind spots caused by concealed vessel activities.

[0078] After obtaining a continuous sequence of regulatory blind spots, their spatial change trajectories are organized to form a complete migration path for the regulatory blind spots, and the overall spatial impact range of the regulatory blind spots is determined. The centroid coordinates of the regulatory blind spots for each monitoring period are read sequentially over time, and these coordinate points are connected on the electronic nautical chart to form a continuous spatial trajectory line, which is the migration path curve of the regulatory blind spots. Subsequently, the spatial distances between adjacent centroid coordinates are statistically analyzed, and all adjacent distances are summed sequentially to obtain the cumulative migration distance, representing the overall movement distance of the regulatory blind spots caused by the concealed vessel over multiple monitoring periods. For example, if five blind spot locations are continuously monitored, with adjacent distances of 180 meters, 220 meters, 210 meters, and 200 meters respectively, the cumulative migration distance is approximately 810 meters. After completing the path calculation, the spatial extent of the regulatory blind spots within each monitoring period is analyzed. Specifically, the spatial scale of the elliptical or probability region corresponding to each regulatory blind spot is read, and the equivalent radius of the region is obtained by calculating the average distance from its centroid to the boundary. Since the blind zone range may vary across different monitoring periods, the largest equivalent radius needs to be selected as the overall regulatory blind zone range from all monitoring periods. For example, if the radii obtained in multiple periods are 150 meters, 180 meters, 170 meters, and 160 meters respectively, then 180 meters is selected as the final regulatory blind zone range. The migration path and blind zone range obtained in this way can fully describe the spatial movement trajectory and maximum impact range of the regulatory blind zone caused by concealed vessel activities.

[0079] In step S6, based on the migration path and scope of the regulatory blind spot, a traffic dynamic monitoring risk area is generated on the electronic nautical chart, and traffic monitoring early warning is issued.

[0080] After identifying the migration path of concealed targets within the regulatory blind zone, the migration path is spatially expanded to construct a dynamic risk area for traffic monitoring. All concealed targets identified in the list of suspected vessels are read, and the migration path curve for each target within the regulatory blind zone is extracted. This path curve is formed by connecting the centroid coordinates of the regulatory blind zone obtained from multiple consecutive monitoring periods in chronological order, reflecting the trajectory of the concealed target's movement within the communication blind zone during periods of AIS signal loss. Subsequently, the path is spatially expanded based on the obtained regulatory blind zone range parameters. In practice, the migration path is considered a centerline, and equidistant expansion is performed on both sides of this centerline, with the expansion distance set according to the regulatory blind zone range. For example, if the regulatory blind zone range is approximately 180 meters in a monitoring result, then an expansion of 180 meters is performed on both sides of the path centerline, forming a strip-shaped area covering the surrounding space of the path. When the migration path is curved, the connecting lines between each node of the path are buffered and expanded segment by segment, and adjacent buffer areas are merged to form a continuous strip-shaped area structure. For multiple migration paths generated by different concealed targets, the strip-shaped areas generated by each path are overlaid. When different areas overlap, they are unified into a single overall area through spatial merging. After this processing, a strip-shaped dynamic risk area distributed along the migration path is obtained on the electronic nautical chart. This area represents the spatial range in which concealed vessels may operate and affect surrounding traffic.

[0081] After constructing the dynamic risk zone, real-time position monitoring is performed on vessels normally operating AIS within the regulated waters to determine if they are about to enter the risk zone. The system continuously reads vessel position coordinates from the real-time vessel traffic status set, which includes the position, speed, and heading information of all vessels normally sending AIS messages within the regulated waters. Then, the spatial relationship between each vessel's current position coordinates and the spatial range of the dynamic risk zone is determined. When a vessel's current coordinates are already within the risk zone, it is immediately recorded as a vessel navigating within the risk zone; when a vessel's current coordinates are outside the risk zone but its heading points towards the risk zone, further assessment is made to determine if it is trending towards entering. The vessel's current heading is read and extended on the electronic chart to form a navigation direction line. When this direction line intersects the risk zone boundary and the distance between the intersection and the intersection is less than a certain warning distance, it is determined that the vessel is about to enter the dynamic risk zone. The warning distance is set based on statistical analysis of the average vessel speed. For example, in a congested waterway, the warning distance is set to 0.5 nautical miles. When a vessel is within 0.5 nautical miles of its current heading and is about to enter the risk zone, a warning is triggered. This method allows for the identification of potentially risky vessels before they enter risky areas.

[0082] After identifying vessels about to enter a dynamic risk area, traffic monitoring and early warning are issued to the relevant vessels. The identification code, current position coordinates, speed, and heading information of the identified vessel are recorded and linked to the dynamic risk area information. Subsequently, the vessel is highlighted in the traffic monitoring interface, and its spatial relationship with the risk area is displayed graphically on an electronic nautical chart. Simultaneously, its expected navigation trajectory is plotted according to its course, and the intersection of this trajectory and the risk area is marked, allowing regulators to intuitively determine the time and location of the vessel's entry into the risk area. When multiple vessels approach the risk area simultaneously, all vessels are uniformly sorted, with priority given to the vessel closest to the risk area boundary, so that regulators can focus on targets with higher potential risks. If a vessel continues to navigate and enters the dynamic risk area, its navigation position is continuously updated, and changes in its speed and heading are recorded in real time. When a vessel's heading changes significantly and it moves away from the risk area, it is removed from the warning status; if the vessel maintains its original heading and continues to approach the risk area, the warning marker remains until it leaves the risk area. This method enables continuous monitoring of traffic around areas where concealed vessels are active, and timely alerts to legitimate vessels that may enter the area, thereby reducing navigational safety risks in the presence of concealed vessels.

[0083] Example 2

[0084] The difference between Embodiment 2 and Embodiment 1 is that this embodiment introduces a ship transportation safety supervision system based on AI and big data.

[0085] Figure 2 A schematic diagram of a ship transportation safety supervision system based on AI and big data is provided. The system includes:

[0086] The traffic status construction module is used to receive AIS ship dynamic data collected in real time by the base station, clean, sort and associate the ship dynamic navigation data, and generate a set of current real-time traffic status.

[0087] The concealed suspect identification module is used to monitor the continuity of AIS signals after a vessel enters a traffic-controlled waterway and to identify concealed suspect vessels that have lost signals and are in a maneuvering state.

[0088] The shadow position inversion module is used to extract the course time series and speed time series of ships around the concealed suspected ship, identify ships that turn to avoid them, and reverse cross-locate the concealment probability position domain.

[0089] The concealed target locking module is used to perform spatiotemporal correlation verification between the concealment probability location domain and the last position of the concealed suspected vessel, lock the concealed target, and infer its current speed range and heading range.

[0090] The blind spot monitoring and early warning module is used to identify the migration path and range of the blind spot based on the location domain of the hidden target within a continuous monitoring period, and to generate dynamic monitoring risk areas to implement traffic warnings.

[0091] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0092] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0093] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0094] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0095] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0096] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0097] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0099] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for supervising ship transportation safety based on AI and big data, characterized in that, Includes the following steps: S1. The base station collects real-time ship dynamic data from AIS and generates a set of current real-time traffic status data. S2. When a vessel loses its AIS signal after entering the traffic-controlled waters and its speed was greater than zero before the signal was lost, the corresponding vessel will be marked as a concealed suspected vessel and its last position and motion parameters will be recorded. S3. Extract the real-time traffic status set of all ships within a preset range around the last location of the concealed suspected ship, identify the ship clusters that synchronously turn to avoid it, perform reverse cross-location based on the spatial orientation of the ship clusters avoiding it, and mark the concealment probability location domain. S4. Perform spatiotemporal correlation verification between the concealment probability location domain and the last position of the concealed suspected vessel to lock the concealed target, and infer the current speed range and heading range of the concealed target based on the distribution of the avoidance direction of the vessel cluster within the concealed target location domain and the last motion parameters. S5. Based on the location sequence of the same concealed target within multiple consecutive monitoring periods, identify the location changes of the regulatory blind spot caused by the continuous movement of the concealed target, and generate the migration path and range of the regulatory blind spot. S6. Based on the migration path and scope of the regulatory blind spot, generate traffic dynamic monitoring risk areas on the electronic nautical chart and issue traffic monitoring early warnings. In step S3, reverse cross-location is performed based on the spatial orientation of the ship cluster avoidance, and the concealment probability location domain is marked specifically including: The search area is defined with the last location of the concealed suspected vessel as the center. The course and speed time sequences of all vessels in the search area within the preset time before and after the signal loss are extracted from the real-time vessel traffic status set. The heading and speed time sequences of each ship are input into a pre-trained collision avoidance behavior recognition model. The collision avoidance behavior recognition model identifies the current motion state of the ship by turning to avoid collision based on the ship type, heading change characteristics, speed change characteristics and time sequence characteristics. Starting from the position coordinates of each ship identified as turning to avoid an obstacle at the beginning of the avoidance maneuver, a ray is drawn in the opposite direction of its instantaneous course. The pairwise intersections of all rays are calculated and the coordinates of the intersections are spatially clustered. The cluster center is retained as the estimated location of the hidden interference source. Centered on the estimated locations of all concealed interference sources, and combined with the distribution range of all avoidant vessels, an elliptical region containing confidence boundaries is generated and marked as the concealment probability location domain.

2. The method for supervising ship transportation safety based on AI and big data according to claim 1, characterized in that, In step S1, generating the current real-time traffic status set specifically includes: The real-time ship dynamic navigation data collected by the base station is cleaned to remove abnormal data points with speeds exceeding the threshold corresponding to the ship type and cross-boundary data points with coordinates exceeding the jurisdiction boundary. Data with the same ship identification code is sorted by timestamp to form a sequence of ship motion trajectories. The ship motion trajectory sequence is associated with static data including ship length and ship type to generate a dynamic attribute vector for each ship, which includes real-time position, motion parameters and ship type. The motion parameters include instantaneous speed and instantaneous heading. Spatially synchronize the dynamic attribute vectors of all ships at the current moment to generate a real-time ship traffic status set.

3. The method for supervising ship transportation safety based on AI and big data according to claim 1, characterized in that, In S2, when a vessel loses its AIS signal after entering the traffic-controlled waters and its speed was greater than zero before the signal was lost, the corresponding vessel will be marked as a suspected concealed vessel and its last position and motion parameters will be recorded. Specifically, this includes: Real-time monitoring of the AIS message reception interval for each vessel; when the signal loss duration exceeds a preset threshold after a vessel enters the traffic-controlled waters, record the last position coordinates, last speed, last course, and last timestamp of the last frame of the message before the loss. Extract the historical speed sequence of the vessel before it was lost, and combine it with the last speed to determine whether the vessel was in a maneuvering state at the time of loss. If the vessel is in a maneuvering state, the type of the regulated waterway where the vessel is currently located is queried based on the last position coordinates, and the suspected weight of the vessel's covert behavior is generated based on the type of regulated waterway. By associating the last position coordinates, last speed, last course, last timestamp, maneuvering status determination results, and the weight of suspected concealment behavior, a concealment suspicion record of the vessel is generated and added to the list of suspected concealment vessels.

4. The method for supervising ship transportation safety based on AI and big data according to claim 1, characterized in that, In step S4, the process of inferring the current speed range and heading range of the concealed target based on the distribution of avoidance directions of the ship cluster within the concealed target's location domain and the last motion parameters specifically includes: Extract the last timestamp of the concealed suspected vessel, and at the same time obtain the average start time of the avoidance action of all avoidance vessels within the concealment probability location domain, and calculate the time difference between the two as the duration of the movement. The final course of the concealed suspected vessel is compared with the main peak direction of the statistical distribution of the heading direction of the avoidance vessel within the concealment probability location domain. When the angle difference between the two is less than a preset threshold, the course consistency is determined. Extract the final speed and combine it with the duration of motion to estimate the theoretical sailing distance. At the same time, calculate the distance between the centroid of the concealment probability location domain and the line connecting the final position. When the ratio of the theoretical sailing distance to the line connecting the final position is within a preset correlation range, the spatiotemporal correlation is determined to be established, the concealment suspected vessel is locked as the concealment target, and the corresponding concealment probability location domain is marked as the concealment target location domain. Extract the distance between the centroid of the hidden target's location domain and its last location, estimate the estimated speed by combining the average duration of the avoidance maneuvers of the ship cluster within the range, and take the main peak direction of the avoidance ship's course as the estimated course to generate the current motion parameter range of the hidden target.

5. The method for supervising ship transportation safety based on AI and big data according to claim 1, characterized in that, In step S5, generating the migration path and scope of the regulatory blind spot specifically includes: Extract the location domain of the same hidden target locked within multiple consecutive monitoring periods, and obtain the deployment location and signal coverage spatial distribution of all base stations in the traffic monitoring water area; For each monitoring period, the ratio of the area of ​​intersection between the target location and the base station signal coverage to the total area of ​​the target location is calculated as the signal coverage. The target location with signal coverage below a preset threshold is marked as a regulatory blind spot. Based on the current speed range and heading range of the concealed target obtained in the previous monitoring cycle, calculate the theoretical displacement range of the centroid coordinates of the monitoring blind zone between adjacent monitoring cycles, eliminate abnormal jump points that do not conform to the theoretical displacement range, and retain the monitoring blind zone sequence with continuous motion. Connect the centroid coordinates of the regulatory blind spots in chronological order to generate the migration path curve of the regulatory blind spots, calculate the cumulative distance between adjacent centroid coordinates as the cumulative migration distance, extract the spatial radius of the regulatory blind spots in each cycle, and take the maximum value of all cycle radii as the range of the regulatory blind spots.

6. The method for supervising ship transportation safety based on AI and big data according to claim 1, characterized in that, In step S6, based on the migration path and scope of the regulatory blind spot, the generation of a dynamic traffic monitoring risk area on the electronic nautical chart and the issuance of traffic monitoring early warning specifically include: Based on the migration path curves of the regulatory blind spots corresponding to all locked hidden targets in the list of hidden suspected vessels, the regulatory blind spot range is expanded equidistantly to both sides to generate a strip-shaped dynamic risk area that includes the entire path. The coordinates of vessels with AIS normally enabled are extracted from the real-time vessel traffic status set, and traffic supervision warnings are issued for vessels entering the dynamic risk area.

7. A ship transportation safety supervision system based on AI and big data, used to implement the ship transportation safety supervision method based on AI and big data as described in any one of claims 1-6, characterized in that, include: The traffic status construction module is used to receive AIS ship dynamic data collected in real time by the base station, clean, sort and associate the ship dynamic navigation data, and generate a set of current real-time traffic status. The concealed suspect identification module is used to monitor the continuity of AIS signals after a vessel enters a traffic-controlled waterway and to identify concealed suspect vessels that have lost signals and are in a maneuvering state. The shadow position inversion module is used to extract the course time series and speed time series of ships around the concealed suspected ship, identify ships that turn to avoid them, and reverse cross-locate the concealment probability position domain. The concealed target locking module is used to perform spatiotemporal correlation verification between the concealment probability location domain and the last position of the concealed suspected vessel, lock the concealed target, and infer its current speed range and heading range. The blind spot monitoring and early warning module is used to identify the migration path and range of the blind spot based on the location domain of the hidden target within a continuous monitoring period, and to generate dynamic monitoring risk areas to implement traffic warnings.

Images