A ship anchoring state recognition method, device and equipment

Abstract

The application provides a ship anchoring state identification method, device and equipment, and relates to the technical field of ship state analysis. The method comprises the following steps: acquiring an AIS dynamic message of a target ship; starting from a current frame in at least one state point time sequence, state point data in a multi-scale time window is intercepted according to the multi-scale time window; target ship navigation track information is determined according to the state point data in the multi-scale time window; geometric consistency verification, directional drift judgment and double conversation sparse report processing are performed according to the target ship navigation track information; and the anchoring state of the target ship is determined according to a preset judgment condition of the processed result. The scheme improves the identification accuracy of the anchoring state of the ship.

Description

Technical Field

[0001] This invention relates to the field of marine vessel status analysis technology, and in particular to a method, apparatus and equipment for identifying the anchoring status of a vessel. Background Technology

[0002] The Automatic Identification System (AIS) is a core data source for maritime surveillance, waterway management, and multi-source trajectory fusion. Accurately identifying whether a vessel is anchored, stationary, moving, or in an unknown state is a fundamental capability for intelligent maritime surveillance. With the increasing number of vessels in inland waterways, ports, and the open sea, the reporting patterns of AIS devices are becoming increasingly diverse, including complex targets such as high-speed moving vessels, low-speed drifting vessels, and vessels reporting intermittently or infrequently. This places higher demands on the robustness, real-time performance, and scenario adaptability of vessel status detection algorithms. Traditional methods for detecting anchored vessels generally suffer from problems such as erroneous target state switching and delayed anchor return identification. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method, device, equipment and medium for identifying the anchoring status of ships, so as to improve the speed and accuracy of identifying the anchoring status of ships.

[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: In a first aspect, embodiments of the present invention provide a method for identifying the anchoring status of a ship, including: Acquire the AIS dynamic message of the target vessel, wherein the AIS dynamic message includes: a time-series frame sequence of at least one status point based on a timestamp; Based on the non-reusable multi-scale time windows, starting from the current frame in at least one state point time frame sequence, state point data within the multi-scale time window are extracted respectively. Based on the state point data within the multi-scale time window, the navigation trajectory information of the target vessel is determined; Based on the target vessel's navigation trajectory information, a geometric consistency check is performed to determine the first analysis result; Based on the target vessel's navigation trajectory information, directional drift is determined, and the second analysis result is confirmed. Based on the target vessel's navigation trajectory information, perform two-way snub reporting processing to determine the third analysis result; Based on the first analysis result, the second analysis result, the third analysis result, and the ship route information, a fourth analysis result is determined; Based on the preset judgment conditions and the results of the fourth analysis, the anchoring status of the target vessel is determined.

[0005] Optionally, after obtaining the AIS dynamic message of the target vessel, at least one of the following is included: Perform timing validity checks on AIS dynamic messages; Perform time-interruption detection on AIS dynamic messages; Filter AIS dynamic messages and rewind time; Invalid timestamp messages are processed for AIS dynamic messages.

[0006] Optionally, state point data within multiple time windows can be extracted, including: Based on the first duration of the main history window, the historical status point data of the first duration before the current status point of the target vessel is obtained from multiple status points of the AIS dynamic message, and used as the status point data of the main history window. Based on the mesoscopic constraint window of the second duration, a first preset number of state point data in the second duration are obtained from multiple state points of the AIS dynamic message, and used as the state point data of the mesoscopic constraint window. According to the third duration of the navigation segment window, from the multiple status points of the AIS dynamic message, less than a second preset number of status point data is extracted each time, until the data segment is greater than a third preset number of status point data, which is used as the status point data of the navigation segment window; the first duration is greater than the second duration, the second duration is greater than the third duration, and the first preset number is greater than the second preset number.

[0007] Optionally, determine the target vessel's navigation trajectory information, including: Calculate the first trajectory of the target vessel based on the status point data in the main history window; Calculate the second trajectory of the target vessel based on the state point data of the mesoscopic constraint window; Calculate the third trajectory of the target vessel based on the status point data of the navigation segment window; The target vessel's navigation trajectory information is determined based on the first trajectory, the second trajectory, and the third trajectory.

[0008] Optionally, a geometric consistency check is performed based on the target vessel's navigation trajectory information to determine a first analysis result, including: Based on the target vessel's status information and the multi-scale time window, the reference point and cluster information are determined; Based on the reference point, cluster information, and target vessel navigation trajectory information, a geometric consistency check is performed to determine the first analysis result.

[0009] Optionally, a geometric consistency check is performed based on the reference point, cluster information, and target vessel trajectory information to determine a first analysis result, including: Based on the preset step length calculation formula, the actual step length and theoretical step length of the target ship's current position are determined according to the reference point, cluster information and the target ship's navigation trajectory information. The actual step size and the theoretical step size are compared to determine the first analysis result.

[0010] Optionally, based on the target vessel's navigation trajectory information, a directional drift judgment is made to determine the second analysis result, including: Based on a preset first judgment rule, the target vessel's navigation trajectory information is subjected to low-speed directional propulsion judgment to obtain a first judgment result; Based on the preset second judgment rule, the target vessel's navigation trajectory information is subjected to multi-scale tail window forward movement judgment to obtain the second judgment result; Based on the preset third judgment rule, the navigation trajectory information of the target vessel is divided into navigation segments for judgment, and the third judgment result is obtained. Based on the first judgment result, the second judgment result, and the third judgment result, the second analysis result is determined.

[0011] Optionally, a third analysis result is determined based on the target vessel's navigation trajectory information through two-session sparse reporting, including: Establish a two-session sparse reporting model based on the target vessel's status information; The target ship's navigation trajectory information is processed based on the two-session sparse reporting model to obtain a third analysis result.

[0012] Secondly, embodiments of the present invention also provide a ship anchoring status identification device, comprising: The acquisition module is used to acquire the AIS dynamic message of the target vessel, wherein the AIS dynamic message includes: a time-series frame sequence of at least one status point based on a timestamp; The processing module is used to extract state point data within each of at least one state point time-series frame sequences, starting from the current frame, based on non-reused multi-scale time windows; determine the target vessel's navigation trajectory information based on the state point data within the multi-scale time windows; perform geometric consistency verification based on the target vessel's navigation trajectory information to determine a first analysis result; determine directional drift based on the target vessel's navigation trajectory information to determine a second analysis result; perform two-way communication reporting processing based on the target vessel's navigation trajectory information to determine a third analysis result; determine a fourth analysis result based on the first, second, and third analysis results and the vessel's path information; and determine the target vessel's anchoring state based on preset judgment conditions and the fourth analysis result.

[0013] Thirdly, embodiments of the present invention also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described in the first aspect.

[0014] The above-described solution of the present invention has at least the following beneficial effects: The above-mentioned solution of the present invention completely eliminates pollution of long-distance voyage trajectory through multi-scale windows, significantly reduces the delay of ship anchoring identification, and improves identification accuracy; through geometric consistency verification, it achieves accurate filtering of false speeds, significantly reducing the false identification rate of anchoring status; through drift judgment, it achieves accurate differentiation between anchoring swaying and low-speed slippage, eliminating potential safety hazards in waterway supervision; through two-way dialogue reporting, it solves the problem of targets remaining in a fixed position for a long time; finally, through preset judgment conditions as a fallback, it further improves the identification accuracy of anchoring status. Attached Figure Description

[0015] Figure 1 This is a flowchart of an embodiment of the ship anchoring status identification method of the present invention; Figure 2 This is a schematic diagram of the multi-scale window processing flow of the present invention; Figure 3 This is a schematic diagram of the step size consistency verification process of the present invention; Figure 4 This is a schematic diagram of the drift determination process of the present invention; Figure 5 This is a schematic diagram of the two-session sniffing process of the present invention; Figure 6 This is a schematic diagram of the process for determining the anchoring status of a ship according to the present invention; Figure 7 This is a schematic diagram of an embodiment of the ship anchoring status identification device of the present invention. Detailed Implementation

[0016] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0017] like Figure 1 As shown, this invention proposes a method for identifying the anchoring status of a ship, the method comprising: Step 1: Obtain the AIS dynamic message of the target vessel. The AIS dynamic message includes: at least one time-series frame sequence of status points based on timestamps; multiple time-series frame sequences of status points may include: the target vessel's position (latitude and longitude), timestamps of each status point, speed (SOG), heading (COG), and other navigation status data. Step 2: Based on the non-reusable multi-scale time windows, starting from the current frame in at least one state point time frame sequence, extract the state point data within the multi-scale time window respectively; Step 3: Determine the target ship's navigation trajectory information based on the state point data within the multi-scale time window; Step 4: Perform geometric consistency verification based on the target vessel's navigation trajectory information to determine the first analysis result; Step 5: Determine the directional drift based on the target vessel's navigation trajectory information to determine the second analysis result; Step 6: Perform two-way call processing based on the target vessel's navigation trajectory information to determine the third analysis result; Step 7: Determine the fourth analysis result based on the first analysis result, the second analysis result, and the third analysis result; Step 8: Determine the anchoring status of the target vessel based on the preset judgment conditions and the results of the fourth analysis.

[0018] This embodiment of the invention completely eliminates pollution from long-distance voyage tracks through multi-scale windows, significantly reduces the delay in ship anchoring identification, and improves identification accuracy; through geometric consistency verification, it achieves accurate filtering of false speeds, significantly reducing the false identification rate of anchoring status; through drift judgment, it achieves accurate differentiation between anchoring swaying and low-speed slippage, eliminating potential safety hazards in waterway supervision; through two-way dialogue reporting, it solves the problem of targets remaining in a fixed position for a long time; finally, through preset judgment conditions as a fallback, it further improves the identification accuracy of anchoring status.

[0019] In an optional embodiment of the present invention, step 1, after obtaining the AIS dynamic message of the target vessel, may include at least one of the following: Step 11: Perform time sequence validity verification on AIS dynamic messages; check whether the order of AIS points conforms to normal time logic, initially filter out abnormal messages that are obviously disordered or out of order, and ensure that the points are basically ordered in the time dimension.

[0020] Step 12: Perform time gap detection on AIS dynamic messages.

[0021] The time interruption threshold gapSec is dynamically configured based on the AIS device's reporting mode, with a default of 120 seconds, which can be extended to 1800 seconds to accommodate long-interval reporting scenarios from ocean-going vessels. Specifically... 1. Calculate the step size using the preGapPoint and the current point, and determine the time difference before and after the timing interruption. If step size and Or, the target ship was in a moving state before the timing interruption and the step size was... If so, the ship is directly determined to be in motion; and the current state of the ship is directly output.

[0022] 2. When the AIS point cluster satisfies spatial aggregation (i.e., the points in the meso-window converge within a small area), the forwardMeso directional movement feature is used to determine whether the ship has sustained directional displacement. If the points are scattered and not clustered (mesoBounded=false), forwardMeso is not used as a decision criterion. Specifically, the forwardMeso directional movement feature is a derived indicator for determining whether points within the meso-window have sustained directional movement, and it can be obtained by combining the net displacement (net), total path (path), and COG consistency within the window.

[0023] 3. If the current ship's state is determined to be Moving, the anchor reference is cleared and the Moving segment is reset; if the current ship's state is determined to be stationary, the reference point is redefined, and the current ship's state is directly output.

[0024] This step identifies gaps in data reporting between two consecutive AIS messages that have not been reported for a long time: the threshold can be dynamically adjusted, with a base of 120 seconds and a maximum relaxation to 1800 seconds in ocean scenarios; at the same time, it combines the distance before and after the gap, the time difference, and the ship's historical movement status to intelligently determine whether the ship is currently sailing or stationary, and updates the motion segment and anchorage reference point synchronously.

[0025] Step 13: Filter AIS dynamic messages by time reversal; AIS devices may occasionally have out-of-order message timestamps or later received messages that are earlier than the previous ones. This step removes such time-reversal erroneous data.

[0026] Step 14 involves processing the AIS dynamic messages for invalid timestamps to obtain at least one state point time sequence frame. This step removes invalid messages with missing, illegal, or out-of-range timestamps. After the above four checks and filters, a continuous and timely sequence of ship state points time frames is finally generated for use in subsequent algorithms.

[0027] The sequence obtained by concatenating at least one state point time sequence obtained through steps 11 to 14 with the historical state point time sequence in chronological order is used as the input for step 2.

[0028] In some optional implementations of the present invention, step 2, which involves extracting state point data within a multi-scale time window, includes: Step 21: Based on the first duration of the main history window, obtain the historical status point data of the first duration before the current status point of the target vessel from multiple status points of the AIS dynamic message, and use it as the status point data of the main history window. Step 22: Based on the mesoscopic constraint window of the second duration, obtain a first preset number of state point data from multiple state points of the AIS dynamic message during the second duration, and use them as the state point data of the mesoscopic constraint window. Step 23: Based on the third duration of the navigation segment window, extract less than a second preset number of status point data from multiple status points in the AIS dynamic message each time, until the data segment exceeds the third preset number of status point data, and use this as the status point data for the navigation segment window; the first duration is greater than the second duration, the second duration is greater than the third duration, and the first preset number is greater than the second preset number.

[0029] The first duration can be 1800 seconds, the second duration can be less than or equal to 900 seconds, the third duration can be less than or equal to 0.5 seconds, the first preset quantity can be less than or equal to 200, and the second preset quantity can be less than or equal to 3. In this embodiment, by acquiring state point data of multi-scale isolation windows, window isolation for different business scenarios is achieved, eliminating the interference of long-distance voyage trajectory on anchor return determination and improving the speed and accuracy of anchor return recognition.

[0030] In an optional embodiment of the present invention, step 3, determining the target ship's navigation trajectory information based on state point data within a multi-scale time window, may include: Step 31: Calculate the first trajectory of the target ship based on the status point data of the main history window; Step 32: Calculate the second trajectory of the target vessel based on the state point data of the mesoscopic constraint window; Step 33: Calculate the third trajectory of the target vessel based on the status point data of the navigation segment window; Step 34: Determine the target vessel's navigation trajectory information based on the first trajectory, the second trajectory, and the third trajectory.

[0031] In this embodiment, each window independently calculates the target ship's navigation trajectory, including indicators such as SOG quantile, cluster radius R95, total trajectory path, and net displacement net, which are used for geometric consistency verification, directional drift judgment, and two-way session sparse reporting processing.

[0032] In an optional embodiment of the present invention, step 31, calculating the first trajectory of the target vessel based on the state point data of the main history window, may include: Historical position information exceeding 1800 seconds is removed. The mean center A of each point in the main historical window is determined by the arithmetic mean method. The distance from each verified position information in the main historical window to the mean center of the main historical window is determined. The 95th percentile of the distance sequence of all points in the main historical window relative to the mean center of the window is taken (the dynamic mean center of the main historical window is denoted as R95). The total trajectory path, net displacement, and other indicators of each point in the main historical window are also determined. Based on the dynamic mean center of the main historical window, the total trajectory path, net displacement, and other indicators of each point in the main historical window, the first trajectory of the target ship is obtained.

[0033] As an example, the determination process for R95 includes: (1) Discard points in the main history window whose duration exceeds 1800s; (2) Calculate the mean of the geographic coordinates of all points within the window using the arithmetic mean, and obtain the mean center A of the window; (3) Calculate the geographical distance between each point in the window and the mean center A; (4) Sort all distances from smallest to largest and take the 95th percentile value. This value is R95 (cluster radius).

[0034] In an optional embodiment of the present invention, step 32, calculating the second trajectory of the target vessel based on the state point data of the mesoscopic constraint window, may include: The meso-level constraint window includes the point location information from the verification location information. Duration Position information per second (points, i.e., the ship's position information obtained each time). Based on the position information in the mesoscopic constraint window, the following can be used: Anomaly point limiting, Calculate the maximum outer diameter of the trajectory The second trajectory is determined to be an "unbounded region" if it is greater than the preset value, and otherwise it is an "unbounded region".

[0035] In an optional embodiment of the present invention, step 33, calculating the third trajectory of the target vessel based on the state point data of the navigation segment window, may include: If the current vessel enters Moving state (i.e., the vessel enters a moving state), the navigation segment window includes the forced retention information when entering Moving state (i.e., the vessel enters a moving state) in the position verification information. Each point (point is location information); if the current vessel enters the anchoring state, the navigation segment window includes the 30 most recently collected points from the verification location information.

[0036] Based on the position information in the navigation segment window, the SOG quantile, cluster radius R95, and the total trajectory path and net displacement corresponding to the navigation segment window are calculated based on the 30 collected points to obtain the third trajectory.

[0037] In an optional embodiment of the present invention, in step 34, the target ship navigation trajectory information is constructed together based on the first trajectory, the second trajectory, and the third trajectory.

[0038] As an example, refer to Figure 2 , Figure 2 This demonstrates the process in steps 2 and 3 of extracting state point data within a multi-scale time window, and determining the target vessel's navigation trajectory information using the state point data within the multi-scale time window. The ship's AIS time-series location sequence is used as input, and the current AIS location is loaded frame by frame. It drives multi-window parallel computation. The algorithm maintains a main historical sliding window with a maximum duration of 1800 seconds to calculate global statistical features such as speed, point displacement, and center distance; simultaneously, it constructs a mesoscopic window with a maximum duration of 900 seconds and a maximum number of points of 200 to calculate the maximum scatter diameter of points and determine whether points are clustered in space; when the ship is in navigation, it activates a navigation segment window, adaptively selecting samples to calculate navigation features based on the number of points in each segment. Specifically, this includes: 1. The input is an ordered sequence of AIS points, including historical points. and the current processing frame Each new current frame This drives all subsequent window operations in parallel.

[0039] 2. Main history sliding window (m_points double-ended queue): Window eviction rule: Continuously prune the head of the queue to ensure that the time span between the latest and earliest points within the window does not exceed 1800 seconds (storageSec). Execution logic is as follows: (1) Call computeMetrics() to perform full-window metric statistics; (2) A set of global statistical features are calculated: median of ground speed SOG, P90 quantile of SOG, R95 feature, path statistics pathM, net displacement netM, and distance from point to window center point distToCenterM. The output statistical features are sent to the "path anomaly removal module" and the "R95 truncation upper limit calculation module".

[0040] 3. Mesoscopic window (meso window, configured to calculate mesoscopic diameter D_meso): dual constraints of the number of points and time duration are adopted: a maximum of 200 points are retained, the total time duration does not exceed 900 seconds, and data is selected by backtracking from the latest point. The execution process is as follows: (1) Screen points within the median in the window, supporting two sets of thresholds: 300m and 800m; (2) Calculate the maximum distance between every two points in the window to obtain the mesoscopic diameter D_meso; (3) Condition judgment: 0<D_meso ≤ 150m; if this condition is satisfied: mark mesoBounded=true (mesoscopic range convergence, regional aggregation); if this condition is not satisfied: mark mesoBounded=false (scattered points). The mesoscopic convergence marking result is output to the "anchoring reference point management module".

[0041] 4. Navigation segmentation window (Moving motion segmentation window) is only activated after a ship enters the Moving navigation state, and the execution process is as follows: (1) At the moment of switching to the navigation state, crop the queue to retain at most 3 points, and initialize the starting index of the navigation segment; (2) Continuously accumulate AIS points in the navigation state within the window; (3) Judge the number of points in the current navigation segment: if the number of points ≥ 30: only select the latest 30 points to perform navigation segment feature calculation; if the number of points < 30: use all points of the navigation segment for statistics; (4) Based on navigation segment indicators, perform static feature evaluation and navigation state admission feature evaluation; The final feature result is input into the "three-state finite state machine" for navigation or anchoring state judgment.

[0042] 5. Abnormal path elimination module (universal for all windows): filters abnormal navigation segments based on displacement and time difference data output by the main window. The judgment condition is: the distance between two points > 500m and the time interval < 300s (long distance in a short time, which is most likely an abnormal drift point); if the condition is satisfied, the displacement of this segment is not included in the path statistics pathM; if the condition is not satisfied, pathM is normally accumulated and updated; the processed effective path statistics are sent to the "anchoring reference point management module".

[0043] 6. R95 upper radius limit calculation module, configured to calculate the distance of each point relative to the window center point, set an upper limit of 400 meters for winsorization; obtain the 95th percentile of the distance distribution to obtain the radius95 aggregation radius; the aggregation radius result is sent to the "anchoring reference point management module".

[0044] In an optional embodiment of the present invention, step 4, performing a geometric consistency check based on the target ship's navigation trajectory to determine a first analysis result, includes: Step 41: Determine the reference point and cluster information based on the target ship's status information and the preset multi-scale window; Step 42: Perform geometric consistency verification based on the reference point, cluster information, and target ship navigation trajectory information to determine the first analysis result.

[0045] For step 41, when the ship is at anchor: When the target vessel's status is detected to switch from an unknown state to anchored, the mean center A of the main history window is used as the reference point. When the target vessel's status is detected to switch from Moving to Anchoring, the mean center of the last 30 points of the navigation segment window is used as the reference point. When the target vessel is detected to be stationary, according to Determine the reference point, where R_base represents the reference point and R95 represents the dynamic centering of the main history window.

[0046] according to To determine cluster information. This indicates the distance from the current point to the reference point, and determines whether the current point is within a cluster (the cluster includes historical location information that meets the criteria).

[0047] Additionally, if you are not currently at anchor, you can refer to... Figure 2 , Figure 2 The anchoring reference point management module receives four types of upstream inputs: effective path statistics pathM, radius95 clustering radius, mesoBounded convergence marker, and navigation segment features; and dynamically freezes the anchoring reference point or rebinds it to a new static reference datum based on the current features, providing a spatial reference for anchoring determination.

[0048] If the reference point and cluster information cannot be determined in this embodiment, the first analysis result will be empty, or other rules can be used to determine the result as needed.

[0049] In some optional implementations, step 42 involves performing a geometric consistency check based on the reference point, cluster information, and target vessel trajectory information to determine a first analysis result, including: Step 421: Based on the preset step length calculation formula, determine the actual step length and theoretical step length of the target ship's current position according to the reference point, cluster information and ship path information; Step 422: Compare the actual step size and the theoretical step size to determine the first analysis result.

[0050] In this embodiment, within the spatial domain of the anchorage cluster, the theoretical travel distance is calculated based on the current speed and frame time difference, and compared with the geographical step size of adjacent AIS (Automatic Identification System) points: For step 421, according to Determine the theoretical step size; in, Indicates the theoretical step size. This indicates the ship's current actual speed relative to the seabed / land. (Revolutions per second) The message time difference has a maximum limit of 600 seconds.

[0051] according to ,in, , , Represents the true step size, the great circle distance between two adjacent AIS points. Represents the average radius of the Earth's sphere. This indicates the geocentric angle (in radians) between the previous AIS point and the current AIS point. Indicates intermediate variables. Indicates the latitude (in radians) of the previous AIS point. This indicates the latitude (in radians) of the current AIS location. Indicates the longitude (in radians) of the previous AIS point. This indicates the longitude (in radians) of the current AIS point.

[0052] For step 422, if the actual step size is much smaller than the theoretical step size, the first analysis result is "false speed" and it is not included in the high-speed continuous count; if And implicit speed If so, the first analysis result is "single-frame hard anchoring"; if And single-frame displacement The first analysis result is "single-frame hard anchoring".

[0053] This embodiment identifies instances where the SOG reported a high speed but the actual distance traveled was not significant (false speed).

[0054] As an example, refer to Figure 3 , Figure 3 This is a flowchart of the speed-geographic step size consistency check algorithm. Figure 3The input is the result of the multi-scale window calculation in step 2, including the current point's speed (SOG), timestamp, latitude and longitude, the timestamp and latitude and longitude of the previous valid point, and the latitude and longitude of the frozen reference point and the baseline radius R_base. First, spatial domain determination is performed, calculating the distance between the current point and the frozen reference point. If the distance exceeds a threshold, consistency verification is disabled; if it is within the threshold range, the geographic step size and time interval between adjacent points are calculated. Then, forward filtering of false speeds is performed. Based on the time interval, measured step size, and the theoretical distance traveled corresponding to the speed, it is determined whether to set a speed blocking flag (geomSogBlock) to filter out abnormal messages that only report speed but have no actual displacement. If the verification is not blocked, reverse identification of real maneuvers is further performed: for anchor-like scenarios, on the one hand, the reported speed and the reverse-calculated equivalent speed are combined to determine real maneuvers, generating a B3 enhanced flag (highSogStepHard); on the other hand, scenarios with large coordinate jumps within a short period are identified, generating a C2 coordinate jump flag (coordJumpHard). The algorithm ultimately outputs three sets of judgment flags: geomSogBlock, highSogStepHard, and coordJumpHard, which provide a basis for the final judgment on the ship's navigation and anchoring status. Figure 3 Specifically, it includes: 1. Spatial Domain Determination. The distance between the current point and the frozen reference point is set to distToRef; set the distance threshold to... And make a judgment: (1) If distToRef exceeds the threshold: determine that the ship is outside the reference domain, disable the speed-step consistency check, and set... Speed ​​determination uses conventional logic; (2) If distToRef is within the threshold: the ship is in the vicinity of the reference point, the complete consistency verification process is activated, and the step size calculation is entered.

[0055] 2. Calculate the actual geographic step size. Determine the actual geographic distance *step* between the current point and the previous valid point. Calculate the time difference between two points The time interval between adjacent messages is obtained and sent to the positive filtering logic.

[0056] 3. Forward Filtering – False Speed ​​Suppression (Identifying Inflated SOG). When the vessel has barely moved, but the AIS reports an excessively high speed, the speed marking is invalidated, suppressing false speed detection. Branch logic includes: (1) If the time difference Then, based on the distance between the two points Determined as stationary shaking, set The flight speed in this frame will not be included in the sustained high-speed statistics; otherwise, it will be set to... (2) If Then when , directly place ;when Theoretical travel distance converted from AIS speed (SOG) ; Calculate the minimum reasonable step size If the actual measured step size This indicates that the geographical location movement is much less than the theoretical displacement at the speed, and is therefore judged as a false speed. ;on the contrary .

[0057] 4. Verification of actual movement. Only at the verification mark. Under the premise of execution, for suspected anchored stationary scenarios, two types of reliable real motion evidence are identified. First, it is determined whether the ship is in a static like-anchor scenario: (1) The maneuver enhancement flag is not triggered. This means that the anchorage condition is not met. (2) If the anchorage condition is met, continue with the identification branch: Branch B3 (i.e., speed-displacement matching real maneuver enhancement): If the AIS reports the speed The equivalent speed impliedKn is inversely calculated from the measured step size and time interval; when , turn on the sign (i.e., B3 rule takes effect), and is incorporated into hard navigation conditions, supporting single-frame triggering to enter navigation state, that is, the first analysis result is "single-frame hard anchor take-off"; if this condition is not met, B3 will not be triggered.

[0058] Branch C2 (Large Coordinate Jump Maneuver Decision): Must simultaneously meet three conditions: Step length and Then place (C2 coordinate jump flag is active), and is incorporated into hard navigation conditions, supporting single-frame triggering to enter navigation state, i.e., the first analysis result is "single-frame hard anchor takeoff"; if any condition is not met, C2 will not trigger.

[0059] In an optional embodiment of the present invention, step 5, determining the second analysis result based on the directional drift judgment of the target vessel's navigation trajectory, includes: Step 51: Based on the preset first judgment rule, perform low-speed directional propulsion judgment on the target ship's navigation trajectory information to obtain the first judgment result; Step 52: Based on the preset second judgment rule, perform multi-scale tail window forward movement judgment on the target ship's navigation trajectory information to obtain the second judgment result; Step 53: Based on the preset third judgment rule, the navigation trajectory information of the target vessel is divided into navigation segments for judgment to obtain the third judgment result; Step 54: Determine the second analysis result based on the first judgment result, the second judgment result, and the third judgment result.

[0060] For step 51, in the main history window, count 10 points backward from the current location information. If each point satisfies net≥12m, net / path≥0.55, and average deviation of adjacent COGs≤15°, then the first judgment result is no anchorage; otherwise, the first judgment result is anchorage. For step 52, within the main history window, retrieve the location information of points 15, 30, and 40 counting backwards from the current location, and obtain the total trajectory path and net displacement index for each point. If point 15 satisfies: or ,and Point 30 is satisfied: and Point 40 is satisfied: and If the second judgment result is no anchorage, then the second judgment result is that the ship is anchored; otherwise, the second judgment result is that the ship is anchored. For step 53, within the navigation segment window, calculate the total trajectory path and net displacement corresponding to the navigation segment window based on the 30 collected points. If path ≥ 10m or net ≥ 5m, and net / path ≥ 0.7 / 0.85, then the third judgment result is no anchorage; otherwise, the third judgment result is anchorage.

[0061] For step 54, if any of the first, second, or third judgment results indicate that the ship is not anchored, then the second analysis result is "drifting"; otherwise, it is "no drifting".

[0062] This embodiment identifies whether the vessel is in a directional propulsion state (e.g., coasting or slow movement, i.e., drifting state) to prevent the vessel from being mistakenly identified as being in an anchored state.

[0063] As an example, refer to Figure 4 , Figure 4This is a flowchart of a three-layer directional drift rejection logic. The input is the AIS point sequence from the main historical window. Points are extracted from the end of the sequence and fed into three parallel decision-making logic layers. The first layer identifies short-term low-speed directional movement based on the most recent 10 points, combining net displacement, path proportion, and heading stability for judgment. The second layer constructs three sets of multi-scale sliding windows (15, 30, and 40 points) to identify recent continuous directional forward movement. The third layer extracts the features of the most recent 30 points within a navigation segment to determine if directional movement exists within that segment. The three-layer judgment results are summarized and judged hierarchically: if any layer determines directional movement, the rejection flag VETO is set to true, preventing the ship's state from switching to a near-anchored stationary state; only when all three layers fail to meet the directional movement condition is VETO set to false, allowing the anchoring judgment to be executed, effectively avoiding misidentifying low-speed directional drifting ships as anchored. Specifically, Figure 4 The specific steps include: 1. First layer, short-term low-speed directional propulsion determination: (1) Select the 10 most recent consecutive AIS points in time sequence; (2) Calculate the total path length and remove abnormal segments; (3) Calculate the net displacement net between the first and last points using the Haversine formula; (4) Conditions are judged in sequence: if the net displacement If the ratio of net displacement to total path length is not valid, then the first layer is invalid; If the distance condition is met, then the first layer is not valid; after that, calculate the average difference between adjacent headings COG; if the average heading difference is... And the number of effective heading comparisons reached This indicates that the course is stable and the movement has directionality; (5) All conditions are met: Mark (The first layer is true); otherwise, the first layer is false.

[0064] 2. Second layer: Multi-scale tail window forward shift determination, using three sets of parallel sliding window multi-scale verification, constructing 15-point windows, 30-point windows, and 40-point windows respectively to independently calculate path (total path length) and net (net of first and last displacement): (1) 15-point window determination criteria: and or and ; (2) 30-point window determination criteria: and ; (3) 40-point window determination criteria: and ; If any one of the three groups of windows meets the condition, mark it. (The second layer is valid); if none of the three groups are satisfied, then the second layer is invalid.

[0065] 3. Third layer: Global propulsion determination for navigation segments, which only evaluates segments that have already entered the moving state: (1) Select the 30 nearest points within the navigation segment and call the navigation segment index calculation function computeMovingSegmentMetrics(); (2) Obtain the total number of segmented paths seg.pathM and the net displacement seg.netM; (3) Judgment: and ,or and Or it satisfies the long path combination condition; If the condition is met, mark it. (The third layer is valid); otherwise, the third layer is invalid.

[0066] 4. Validate sequentially by level: (1) If the first layer ,but ; (2) If the first layer is not valid, then the second layer... ,but ; (3) If the first two layers are not true, then the third layer... ,but ; (4) If all three determinations are invalid, then ; Triggering a directional drift rejection prohibits switching the vessel to a static like-anchor state, meaning the second analysis result is "non-drift"; The lack of evidence of sustained directional movement allows for subsequent procedures to determine the anchorage status; the second analysis result is "drift".

[0067] In an optional embodiment of the present invention, step 6 involves performing two-way interrogation processing based on the target vessel's state information and its navigation trajectory to determine a third analysis result, including: Step 61: Establish a two-session sparse reporting model based on the state information of the target vessel; Step 62: Process the second analysis result according to the two-session sparse reporting model to obtain the third analysis result.

[0068] In this embodiment, the target ship's state information includes an unknown state (all other states of the ship besides its motion state can be considered unknown states) and a motion state. This embodiment establishes a two-way sparse reporting model for the ship's unknown position state and motion state: Table 1 Two-session sparse reporting model

[0069] In this embodiment, if the third analysis result is "session sparse reporting meets the standard", then proceed to step 7.

[0070] As an example, Figure 5 This is a flowchart of the dual-session sparse reporting process. The dual-session sparse reporting model is designed for AIS sparse reporting scenarios on ocean-going vessels. It consists of two identification branches: Unknown (unknown state) sessions and Moving (navigation state) sessions. A unified session reset rule is set, clearing session statistics when there is a prolonged message interruption or a significant state change. For the Unknown session, given insufficient samples in the main window, it determines whether the anchoring access conditions are met based on the spatial envelope range of the location, peak speed, and session duration / number of points. If met, it directly switches to a near-anchoring state. This branch does not verify directional drift features. For the Moving session, in addition to envelope, speed, and session sample conditions, it pre-verifies three layers of directional drift rejection conditions to exclude interference from continuous low-speed directional navigation. When all conditions are met, it can skip the traditional 4-frame counting and quickly switch from navigation to a near-anchoring state. The two branches complement each other, improving anchoring identification capabilities in scenarios with large AIS reporting intervals and scarce effective locations, while also distinguishing the risk boundary between unknown and navigation states, reducing the probability of misjudgment. For details, please refer to the dual-session sparse reporting process: 1. Session reset conditions. If any of the following conditions occur, execute `ResetUnknownSession()` and `resetMovingSession()` to clear the statistics for unknown sessions and navigation sessions: (1) The AIS message time interval gap exceeds the set threshold gapSec; (2) The ship's state changes to enter the navigation state (enterMovingState()); (3) The ship switches from sailing to stationary state: enterStaticFromMoving(); (4) The vessel switches to a static like anchor state via an Unknown path.

[0071] 2. The Unknown session branch unknownSessionSparseAnchorEligible() is executed when the ship is in the Unknown state. (1) If the number of points in the main history window This supplementary logic does not intervene; the ship's status is determined using the standard determination path. (2) If the main window has a number of points Continue to verify: the unknown session is active and the session has accumulated at least 5 valid points; if not, the standard path will be enabled. (3) If the condition is met, calculate the unknown session point spatial envelope unknownSessionEnvelopeM(); (4) Determine the envelope range in sequence. Peak ground speed during the conversation ; (5) Simultaneously satisfy the session points requirement Or session duration Second; If all conditions are met, mark eligible=true, directly determine that the ship has entered the StaticLikeAnchor anchoring state, update the static reference, and bind the average point value of the main window as the anchoring reference center.

[0072] The Unknown session branch does not verify the three-layer directional drift rejection condition, and mainly relies on the spatial clustering range of points and low speed characteristics to complete the judgment.

[0073] 3. The Moving state session branch movingSessionSparseAnchorEligible() is executed when the ship is in the Moving state. (1) If the points within the navigation segment The supplementary logic will not intervene; the existing standard 4-frame judgment process will be used. (2) Number of navigation segment points At that time, first confirm that there are no strong navigational characteristics. ; (3) Further verification showed that all three-layer directional drift rejections were invalid: there was no low-speed directional propulsion, no recent continuous forward movement, and no directional propulsion in the navigation segments. ; (4) Verify that the navigation session is active and the session has accumulated no less than 5 points; (5) After the preconditions are met, calculate the spatial envelope of the navigation session point, movingSessionEnvelopeM(); (6) Sequential verification: envelope range Peak conversations ; (7) Meet the session points Or session duration Second; If all conditions are met The `enterStaticFromMoving()` function is executed directly, immediately switching to a stationary, anchor-like state, binding the average of the most recent 30 points of the navigation segment as the reference center, and skipping the original 4 frames while waiting to count.

[0074] Moving sessions enforce a three-layer drift rejection check beforehand to prevent vessels that are continuously moving at low speeds from being incorrectly identified as anchored.

[0075] 4. Unknown Session Branch: No need to verify directional drift rejection conditions; the core judgment criteria are the degree of point envelope clustering and low speed. Moving Session Branch: Three layers of directional drift rejection verification must be passed in advance to filter out continuous directional slow navigation scenarios and prevent ships from being misjudged as anchoring when sailing at low speeds.

[0076] In an optional embodiment of the present invention, step 7, determining the fourth analysis result based on the first analysis result, the second analysis result, and the third analysis result, may include: Based on the first analysis result, the second analysis result, the third analysis result, and the ship route information, a fourth analysis result is determined. In this embodiment, the fourth analysis result is determined according to the following table: Table 2 Ship Status Inquiry Table

[0077] As an example, the fourth analysis result, namely the determination of the ship's state, includes three processing branches: Unknown state, StaticLikeAnchor state, and Moving state. The Unknown branch combines sample size, motion characteristics, and sparse alarm identification logic to complete the initial state differentiation. All branches converge to a common path to uniformly calculate mesoscopic aggregation features, speed consistency features, and dynamic / static judgment evidence. The anchor-like static branch relies on a navigation feature counter to switch to the moving state. The moving branch supports two static switching paths: in the normal scenario, switching is triggered by waiting for four consecutive frames of static candidate signals; in the sparse alarm scenario, switching can be skipped immediately after the directional drift verification passes. After the state machine completes the determination, it uniformly outputs the ship's state, state change flag, and various dynamic / static identification features, achieving smooth switching between the three states of anchoring, moving, and unknown. For the specific determination process of the fourth analysis result, please refer to [link / reference]. Figure 6 : 1. Entry process: update (point) pre-processing: (1) Time sequence guard verification: if the timestamp is 0 or time backtracking occurs, directly maintain the original state and return, without performing subsequent calculations; (2) Gap detection: determine whether the interval between adjacent packets exceeds the gapSec threshold; regardless of whether it times out, queue pruning is performed: update the double-ended queue, clean up the statistical sequence, and save the preGapPoint of the gap; (3) call computeMetrics() to complete the basic feature statistics of the main historical window; 2. Long-duration gap decision: The decision condition is that a gap appears in the message and the state of the previous frame is not equal to the Unknown state. If the condition is met: execute computeGapEvidence(), and fuse the meso-level feature forwardMeso to generate gap motion evidence. If the previous state is the Moving state: maintain the moving state and unbind the anchor reference point. If the previous state is stationary: determine whether there is strong moving evidence movingHard. If there is, switch to moving. If there is no moving evidence, switch to anchor-like stationary and update the stationary reference. After the decision is completed, return directly and do not enter the subsequent regular branch. If the condition is not met: skip the gap branch and enter the Unknown state branch; 3. Unknown state branch processUnknownBranch(). Only enter this branch when the current state is Unknown: (1) Write diagnostic features such as mesoscopic diameter, aggregation identifier, and distance, and record the current point in the unknown session; (2) Determine: If strong motion features appear in the unknown state and the number of valid points is ≥5, then switch to the navigation state, trim the historical point queue, and reset the unknown session; (3) If the navigation switch is not triggered, determine whether the sample is sufficient (number of points ≥30 or duration ≥600s): If the sample is insufficient, maintain the Unknown state and fill the output results; if the sample is sufficient, first verify that the initial window has anchor aggregation features and no directional low-speed drift. If it meets the requirements, switch to anchor-like stationary; if it does not meet the requirements, further call the sparse report recognition logic unknownSessionSparseAnchorEligible; if the sparse report condition is met, switch to anchor-like stationary as well; (4) After the state is switched, re-execute the feature calculation and uniformly merge into the common path; if the state is not switched, directly output the results.

[0078] 4. Common path (unified feature calculation after all state branches converge): (1) Calculate the diameter of the mesoscopic window and obtain the point aggregation mark mesoBounded; (2) Construct the frame context: calculate the distance to the anchor reference point, the speed consistency check mark geomSogBlock, and the basic speed judgment mark speedHard; (3) Execute the comprehensive evidence solution computeEvidence(), output movingHard hard navigation mark, movingForward forward motion mark, and staticCandidate static candidate mark, and then enter the state branch decision.

[0079] 5. The current state is anchored and stationary. The following logic is executed: (1) Determine whether there is strong navigation evidence ev.movingStrong: If there is no strong navigation feature, the round trip switching counter is cleared and the anchoring reference radius R_base is updated slightly; the stationary state is maintained; if there is strong navigation feature, the navigation counter toMovingCounter is started to increment and the stationary counter is cleared. (2) If hard navigation evidence appears, or the navigation counter accumulates to 3 frames, then the execution state switch is performed to enter the navigation state; the point queue is trimmed and the navigation session is reset; if the conditions are not met, the anchor-like stationary state is maintained.

[0080] 6. Enter the location into the navigation session, and then make two-level branch decisions: (1) Level 1: If there is a static candidate feature and there is no strong navigation feature, the static counter toStaticCounter is incremented; when the counter reaches 4 frames, the navigation is normally switched to anchor-like static, and the average of the last 30 points of the navigation segment is used to bind a new reference benchmark; (2) If Level 1 is not met, enter Level 2: if there is no strong navigation feature, all three layers of directional drift rejection are not triggered, and the navigation sparse report identification condition is met; if it is met, there is no need to wait for 4 frames to count, and the navigation is immediately switched to anchor-like static; (3) If all switching conditions are not met: clear the static counter and continue to maintain the navigation state.

[0081] 7. Finally, call fillResult() to package and output all results, including: current status, whether the status has changed, navigation-related features, stationary candidate markers, count variables, point cluster radius, mesoscopic diameter, spatial cluster identifiers and other complete set of feature parameters, for upper-layer business to call.

[0082] In an optional embodiment of the present invention, step 8, determining the state of the target vessel based on preset judgment conditions and the fourth analysis result, may include: If the result of the fourth analysis is not "anchoring", the result of the fourth analysis can be directly output as the result.

[0083] In this embodiment, if the fourth analysis result is "anchored," an "anchorage holding circle" with a radius of 500 m is established with the current point as the center. If the current point is within the "anchorage holding circle" and the following release conditions are not met, the business layer continues to output "anchored," even if the detector has briefly determined that the ship is underway: 1. High-speed matching maneuver (whether the B3 rule is valid). The B3 rule is to determine whether SOG ≥ 3kn and at the same time, the equivalent speed calculated by the geographically measured step length is ≥ 0.5kn. 2. The first frame of hard motion by the detector (i.e., judging from detJustEnteredMoving&&movingHard&&!movingForward whether the detector has just detected the ship entering navigation and whether it is a sudden real motion, not a slow directional drift); where detJustEnteredMoving indicates that the underlying state machine has just switched to the Moving navigation state for the first time, movingHard indicates that there is evidence of hard navigation (reliable motion characteristics, excluding small jitters), and !movingForward indicates that the continuous directional forward movement characteristics are not met (no slow drift interference). 3. The ship's coordinates have left the hold circle (i.e., whether the current point is more than 500 meters from the center of the hold circle). When the ship's geographical location is far from the initial anchorage, it means that the ship has left the anchorage significantly and is no longer suitable to continue anchoring.

[0084] The above embodiments of the present invention relate to a ship AIS anchoring three-state detection method, which executes eight steps in a fixed time sequence: time sequence preprocessing → multi-scale three-window independent maintenance → frozen reference point binding → speed-geographic step size verification → three-layer directional drift rejection → dual-session sparse report discrimination → three-state state machine switching → layered business hysteresis output; data flow between modules is unidirectionally transmitted, forming a streaming processing without batch waiting.

[0085] The specific configuration of the multi-scale three-window system includes: a main history window with a wall clock limit of 1800 seconds; a meso-level constraint window with a maximum of 200 points and a maximum of 900 seconds, employing a 300m / 800m median inlier mechanism; and a navigation segment Moving window that is clipped to ≤3 points when entering navigation mode.

[0086] Among them, the binding and updating rules for frozen reference points are as follows: when Unknown changes to anchor, it is bound to the mean center of the main history window; when Moving changes to anchor, it is bound to the mean center of the most recent 30 points of the navigation segment window; and R_base = max(20m, R95) is dynamically updated during the stationary period.

[0087] The threshold parameters for rejection of directional drift in three layers are as follows: the first layer is 10 consecutive points with net ≥ 12m and net / path ≥ 0.55 and COG deviation ≤ 15°; the second layer is 15 / 30 / 40 parallel tail windows with net / path thresholds of 0.85 / 0.75, 0.65 and 0.70 respectively; the third layer is navigation segment path ≥ 10m or net ≥ 5m with net / path ≥ 0.70 / 0.85.

[0088] Among them, the constraints for the two-session sparse reporting judgment are: uniform constraints are envelope ≤ 1852m, peak SOG < 3kn, number of points ≥ 5, duration ≥ 600s or number of points ≥ 10; Unknown sessions can directly switch anchors after meeting the criteria, while Moving sessions must be rejected by three-layer drift after meeting the criteria.

[0089] Figure 7 This is a schematic diagram of an embodiment of a ship anchoring status identification device proposed in this invention. The device 700 includes: The acquisition module 701 is used to acquire the AIS dynamic message of the target vessel, wherein the AIS dynamic message includes: at least one time-series frame sequence of status points based on timestamps; The processing module 702 is used for multi-scale time windows. Starting from the current frame in at least one state point time-series frame, it extracts state point data within the multi-scale time window. Based on the state point data within the multi-scale time window, it determines the target ship's navigation trajectory information. Based on the target ship's navigation trajectory information, it performs geometric consistency verification to determine a first analysis result. Based on the target ship's navigation trajectory information, it performs directional drift judgment to determine a second analysis result. Based on the target ship's navigation trajectory information, it performs two-way communication reporting processing to determine a third analysis result. Based on the first analysis result, the second analysis result, the third analysis result, and the ship's path information, it determines a fourth analysis result. Based on preset judgment conditions and the fourth analysis result, it determines the anchoring state of the target ship.

[0090] Optionally, after obtaining the AIS dynamic message of the target vessel, at least one of the following is included: Perform timing validity checks on AIS dynamic messages; Perform time-interruption detection on AIS dynamic messages; Filter AIS dynamic messages and rewind time; Invalid timestamp messages are processed for AIS dynamic messages.

[0091] Optionally, state point data within multiple time windows can be extracted, including: Based on the first duration of the main history window, the historical status point data of the first duration before the current status point of the target vessel is obtained from multiple status points of the AIS dynamic message, and used as the status point data of the main history window. Based on the mesoscopic constraint window of the second duration, a first preset number of state point data in the second duration are obtained from multiple state points of the AIS dynamic message, and used as the state point data of the mesoscopic constraint window. According to the third duration of the navigation segment window, from the multiple status points of the AIS dynamic message, less than a second preset number of status point data is extracted each time, until the data segment is greater than a third preset number of status point data, which is used as the status point data of the navigation segment window; the first duration is greater than the second duration, the second duration is greater than the third duration, and the first preset number is greater than the second preset number.

[0092] Optionally, determine the target vessel's navigation trajectory information, including: Calculate the first trajectory of the target vessel based on the status point data in the main history window; Calculate the second trajectory of the target vessel based on the state point data of the mesoscopic constraint window; Calculate the third trajectory of the target vessel based on the status point data of the navigation segment window; The target vessel's navigation trajectory information is determined based on the first trajectory, the second trajectory, and the third trajectory.

[0093] Optionally, a geometric consistency check is performed based on the target vessel's navigation trajectory information to determine a first analysis result, including: Based on the target vessel's status information and the multi-scale time window, the reference point and cluster information are determined; Based on the reference point, cluster information, and target vessel navigation trajectory information, a geometric consistency check is performed to determine the first analysis result.

[0094] Optionally, a geometric consistency check is performed based on the reference point, cluster information, and target vessel trajectory information to determine a first analysis result, including: Based on the preset step length calculation formula, the actual step length and theoretical step length of the target ship's current position are determined according to the reference point, cluster information and the target ship's navigation trajectory information. The actual step size and the theoretical step size are compared to determine the first analysis result.

[0095] Optionally, based on the target vessel's navigation trajectory information, a directional drift judgment is made to determine the second analysis result, including: Based on a preset first judgment rule, the target vessel's navigation trajectory information is subjected to low-speed directional propulsion judgment to obtain a first judgment result; Based on the preset second judgment rule, the target vessel's navigation trajectory information is subjected to multi-scale tail window forward movement judgment to obtain the second judgment result; Based on the preset third judgment rule, the navigation trajectory information of the target vessel is divided into navigation segments for judgment, and the third judgment result is obtained. Based on the first judgment result, the second judgment result, and the third judgment result, the second analysis result is determined.

[0096] Optionally, a third analysis result is determined based on the target vessel's navigation trajectory information through two-session sparse reporting, including: Establish a two-session sparse reporting model based on the target vessel's status information; The target ship's navigation trajectory information is processed based on the two-session sparse reporting model to obtain a third analysis result.

[0097] It should be noted that this device is the same as the method described above. All implementations in the above method embodiments are applicable to the embodiments of this device and can achieve the same technical effect.

[0098] An electronic device includes: a processor, a memory, and a bus, wherein the processor and the memory communicate with each other via the bus; The memory stores program instructions that can be executed by the processor, and the processor can execute the steps of the method described above by calling the program instructions.

[0099] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform a synchronous playback method based on multimodal data of maritime targets provided in the above embodiments.

[0100] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules; that is, 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. Those skilled in the art can understand and implement this without any creative effort.

[0101] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0102] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for identifying the anchoring status of a ship, characterized in that, include: Acquire the AIS dynamic message of the target vessel, wherein the AIS dynamic message includes: a time-series frame sequence of at least one status point based on a timestamp; Based on the non-reusable multi-scale time windows, starting from the current frame in at least one state point time frame sequence, state point data within the multi-scale time window are extracted respectively. Based on the state point data within the multi-scale time window, the navigation trajectory information of the target vessel is determined; Based on the target vessel's navigation trajectory information, a geometric consistency check is performed to determine the first analysis result; Based on the target vessel's navigation trajectory information, directional drift is determined, and the second analysis result is confirmed. Based on the target vessel's navigation trajectory information, perform two-way snub reporting processing to determine the third analysis result; Based on the first analysis result, the second analysis result, the third analysis result, and the ship route information, a fourth analysis result is determined; Based on the preset judgment conditions and the results of the fourth analysis, the anchoring status of the target vessel is determined.

2. The method for identifying the anchoring status of a ship according to claim 1, characterized in that, After obtaining the AIS dynamic message of the target vessel, at least one of the following must be included: Perform timing validity checks on AIS dynamic messages; Perform time-interruption detection on AIS dynamic messages; Filter AIS dynamic messages and rewind time; Invalid timestamp messages are processed for AIS dynamic messages.

3. The method for identifying a ship's anchoring status according to claim 1, characterized in that, State point data within multiple time windows are extracted, including: Based on the first duration of the main history window, the historical status point data of the first duration before the current status point of the target vessel is obtained from multiple status points of the AIS dynamic message, and used as the status point data of the main history window. Based on the mesoscopic constraint window of the second duration, a first preset number of state point data in the second duration are obtained from multiple state points of the AIS dynamic message, and used as the state point data of the mesoscopic constraint window. According to the third duration of the navigation segment window, from the multiple status points of the AIS dynamic message, less than a second preset number of status point data is extracted each time, until the data segment is greater than a third preset number of status point data, which is used as the status point data of the navigation segment window; the first duration is greater than the second duration, the second duration is greater than the third duration, and the first preset number is greater than the second preset number.

4. The method for identifying the anchoring status of a ship according to claim 3, characterized in that, Determine the target vessel's navigation trajectory information, including: Calculate the first trajectory of the target vessel based on the status point data in the main history window; Calculate the second trajectory of the target vessel based on the state point data of the mesoscopic constraint window; Calculate the third trajectory of the target vessel based on the status point data of the navigation segment window; The target vessel's navigation trajectory information is determined based on the first trajectory, the second trajectory, and the third trajectory.

5. The method for identifying a ship's anchoring status according to claim 1, characterized in that, Based on the target vessel's navigation trajectory information, a geometric consistency check is performed to determine the first analysis result, including: Based on the target vessel's status information and the multi-scale time window, the reference point and cluster information are determined; Based on the reference point, cluster information, and target vessel navigation trajectory information, a geometric consistency check is performed to determine the first analysis result.

6. The method for identifying a ship's anchoring status according to claim 5, characterized in that, Based on the reference point, cluster information, and target vessel trajectory information, a geometric consistency check is performed to determine the first analysis result, including: Based on the preset step length calculation formula, the actual step length and theoretical step length of the target ship's current position are determined according to the reference point, cluster information and the target ship's navigation trajectory information. The actual step size and the theoretical step size are compared to determine the first analysis result.

7. The method for identifying a ship's anchoring status according to claim 1, characterized in that, Based on the target vessel's navigation trajectory information, a directional drift judgment is made to determine the second analysis result, including: Based on a preset first judgment rule, the target vessel's navigation trajectory information is subjected to low-speed directional propulsion judgment to obtain a first judgment result; Based on the preset second judgment rule, the target vessel's navigation trajectory information is subjected to multi-scale tail window forward movement judgment to obtain the second judgment result; Based on the preset third judgment rule, the navigation trajectory information of the target vessel is divided into navigation segments for judgment, and the third judgment result is obtained. Based on the first judgment result, the second judgment result, and the third judgment result, the second analysis result is determined.

8. The method for identifying a ship's anchoring status according to claim 1, characterized in that, Based on the target vessel's navigation trajectory information, two-way sparse reporting is performed to determine the third analysis result, including: Establish a two-session sparse reporting model based on the target vessel's status information; The target ship's navigation trajectory information is processed based on the two-session sparse reporting model to obtain a third analysis result.

9. A ship anchoring status identification device, characterized in that, include: The acquisition module is used to acquire the AIS dynamic message of the target vessel, wherein the AIS dynamic message includes: a time-series frame sequence of at least one status point based on a timestamp; The processing module is configured to: extract state point data within a multi-scale time window, starting from the current frame in at least one state point time-series frame sequence, based on non-reused multi-scale time windows; determine the target vessel's navigation trajectory information based on the state point data within the multi-scale time window; perform geometric consistency verification based on the target vessel's navigation trajectory information to determine a first analysis result; determine directional drift based on the target vessel's navigation trajectory information to determine a second analysis result; perform two-way communication reporting processing based on the target vessel's navigation trajectory information to determine a third analysis result; determine a fourth analysis result based on the first analysis result, the second analysis result, the third analysis result, and the vessel path information; and determine the target vessel's anchoring state based on preset judgment conditions and the fourth analysis result.

10. An electronic device, characterized in that, include: A processor, a memory, and a bus, wherein the processor and the memory communicate with each other via the bus; The memory stores program instructions that can be executed by the processor, which can invoke the program instructions to perform the steps of the method as described in any one of claims 1 to 8.

Images