A water traffic control method based on video hotspot area identification, AIS trajectory prediction and VHF prompting

Abstract

The application discloses a water traffic control method based on video hotspot area identification, AIS trajectory prediction and VHF reminding, which collects data and carries out time benchmark unification; key hotspot areas are configured in a monitoring picture, ships in the areas are detected, tracked and classified, and area occupation, congestion and abnormal stay events are output; AIS data is preprocessed to obtain candidate control objects, predicted bridge arrival time, ship-ship collision risk, ship-bridge collision risk, bridge hole occupation conflict risk and navigation priority; traffic control strategies are generated according to risk assessment results, and broadcast reminding or directional reminding is carried out through VHF; control execution is fused, verified and archived for analysis. The application can realize traffic risk early warning, navigation priority evaluation and closed-loop traffic control in bridge area and complex channel scenes without relying on video side geodetic positioning and PTZ tracking, and can effectively reduce the workload of manual judgment and improve the efficiency of water traffic control.

Description

Technical Field

[0001] This invention belongs to the field of intelligent water traffic supervision and proactive risk control technology, specifically involving a water traffic control method based on video hotspot area identification, AIS trajectory prediction and VHF alerts. Background Technology

[0002] In waterway traffic management in inland waterways, bridge areas, port entrances and exits, and complex encounter areas, on-duty personnel typically need to simultaneously monitor multiple data sources, including video surveillance footage, AIS dynamic information, bridge and waterway traffic rules, and on-site communication and command information. Currently, video surveillance is primarily used for manual observation of the scene, while AIS is mainly used to display vessel identity, location, speed, and heading. The lack of a collaborative processing mechanism between the two for traffic control operations means that scenarios such as bridge passage, encounter avoidance, bridge span allocation, abnormal stop identification, and restricted area monitoring still rely on manual comprehensive judgment.

[0003] Existing methods mainly suffer from the following problems: First, although video surveillance can reflect the on-site situation, it is difficult to automatically convert key locations such as bridge entrances, bridge approach areas, bridge occupancy areas, waiting areas, conflict-sensitive areas, and restricted areas into structured traffic events, resulting in a large workload for manual screening. Second, although AIS can provide vessel identity and dynamic location, it is difficult to support timely navigation scheduling and traffic intervention in bridge areas if it lacks awareness of local occupancy status, regional congestion status, and abnormal dwelling status in bridge areas. Third, existing systems mostly remain at the level of risk warning, lacking a complete closed-loop mechanism of "risk identification - strategy generation - VHF alert - execution verification - archiving optimization", which makes it difficult to meet the needs of proactive traffic control in complex waterways and bridge area scenarios.

[0004] Therefore, there is an urgent need for a water traffic control method that does not rely on video to obtain the latitude and longitude of ships, nor on PTZ to track ships, but instead uses semantic perception of video hotspot areas as the front-end trigger, AIS trajectory prediction as the core calculation, and VHF graded alerts as the execution means, in order to reduce the workload of manual judgment and improve the proactive supervision and control capabilities in bridge areas and complex waterway scenarios. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a water traffic control method based on video hotspot area identification, AIS trajectory prediction and VHF alerts, so as to realize traffic risk warning, navigation priority assessment, strategy generation, voice alerts, execution verification and archiving optimization in bridge area and complex waterway scenarios.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts, comprising the following steps:

[0007] S1, Multi-source data acquisition and time benchmark unification: Acquire data from fixed cameras or preset monitoring screens, AIS data, electronic navigation charts, bridge and waterway rule bases, and VHF communication interface data, and perform time alignment;

[0008] S2, Key Hotspot Area Configuration in Waterways: Configure video hotspot areas such as bridge entrance area, bridge opening approach area, bridge opening occupancy area, waiting area, conflict-sensitive area, and restricted area in the monitoring screen, and assign traffic control semantics to each area;

[0009] S3, Ship Detection and Classification in Video Hotspot Areas: Detects ship targets in the video and determines which video hotspot area the target enters, leaves, or stays in, outputting the video hotspot area occupancy status, congestion status, abnormal loitering status, and ship type classification results.

[0010] S4, AIS preprocessing and continuous track reconstruction: Receive AIS messages and parse them to obtain MMSI, latitude and longitude, speed, and heading; perform time alignment, interpolation, frame interpolation, anomaly removal, and smoothing filtering on the AIS data to obtain a continuous AIS track sequence. and estimated arrival time at the bridge;

[0011] S5, AIS candidate screening based on video hotspot events: Based on the status of the hotspot area, time window, navigation direction, expected arrival order and vessel type information, screen candidate control objects related to the control area from the AIS vessel set;

[0012] S6, Traffic Risk Prediction and Priority Assessment: Based on AIS trajectory prediction, calculate ship-to-ship collision risk, ship-to-bridge collision risk, bridge span occupancy conflict risk and navigation priority, and correct the risk level by combining the status of video hotspot areas;

[0013] S7, Traffic Control Decision Generation: Based on the risk level and priority assessment results, generate traffic control strategies such as deceleration, waiting, yielding, bridge switching, prohibition of entry, and resumption of traffic.

[0014] S8, VHF graded alerts and closed-loop feedback: Generate broadcast or targeted alerts through the nearest VHF, and evaluate the intervention effect based on subsequent AIS trajectory changes and video hotspot area status changes, and upgrade the alarm level when necessary;

[0015] S9, Multi-source data fusion verification: Integrates AIS dynamic data, video structured information and VHF feedback logs to verify the consistency of control strategy execution in real time, identify abnormal response vessels and trigger secondary intervention;

[0016] S10, Control Effect Archiving Analysis: Automatically archives closed-loop data according to event type, risk level, intervention duration and success rate, supporting model iteration optimization and dynamic updates of control rules.

[0017] In a preferred embodiment, the multi-source data acquisition and time base unification in S1 includes the following steps:

[0018] S1-1: Receive the video stream from a fixed camera or a preset monitoring screen, and decode and extract frames according to the set frame rate to obtain a video frame sequence. and corresponding video timestamps Simultaneously, it accesses AIS message streams, electronic navigation chart data, bridge and waterway rule base data, and VHF communication interface data, recording AIS timestamps respectively. Rule call timestamp and VHF interaction log timestamp ;

[0019] S1-2, Establishing a unified time standard For the corresponding video timestamp AIS timestamp Rule call timestamp and VHF interaction log timestamp Perform a unified mapping to create a unified timeline;

[0020] S1-3, Set a time alignment window on a unified timeline For each video frame time Retrieve from AIS cache that meet the requirements AIS observations, The AIS time is uniformly mapped; when no original AIS observation value that meets the conditions is retrieved, time interpolation is performed on adjacent AIS messages to obtain the AIS state estimate value aligned with the video frame time.

[0021] S1-4 involves structured encapsulation of various input data to form standardized data packets. The data packet includes at least: video frame number, video timestamp, AIS vessel identification information, vessel position, vessel speed, vessel heading, vessel electronic navigation chart area identifier, bridge rule parameters, and VHF equipment status; wherein:

[0022]

[0023] in, Indicates the video frame number. Represents the AIS state set. Indicates electronic navigation chart information, Represents the set of rule parameters. Represents the VHF state set;

[0024] S1-5, the standardized data packets are output to the channel key hotspot area configuration in step S2, the video hotspot area ship detection and classification in step S3, the AIS preprocessing in step S4, the risk prediction and priority assessment in step S6, and the closed-loop feedback in step S8, respectively, so as to realize the unified calling and time-series consistency processing of multi-source data.

[0025] The configuration of key hotspot areas in the waterway mentioned in step S2 includes the following steps:

[0026] S2-1, based on the bridge structure, waterway boundaries, bridge arch locations, waiting waters, prohibited waters, and high-conflict encounter areas, pre-defines multiple key hotspot areas in the monitoring screen. The key hotspot areas are defined in the form of polygons, rectangles, polyline buffer zones, or area masks;

[0027] S2-2, the key hotspot areas include at least one or more of the following: bridge entrance area, bridge opening approach area, bridge opening occupancy area, waiting area, conflict-sensitive area, and restricted area, and each area is assigned a unique area number. ;

[0028] S2-3 assigns traffic control semantic labels to different key hotspot areas. The semantic tags include at least one or more of the following: allow passage, priority passage, slow down and observe, wait for yield, prohibit entry, and conflict warning.

[0029] S2-4, Configure event determination rules for each key hotspot area. These rules include at least entry events, exit events, sustained stay events, reverse crossing events, area congestion events, and abnormal aggregation events. Specifically, an entry event is defined when the target trajectory center point enters the area from outside the area. An entry event is defined when the target stays in the area for a certain duration. The event is determined to be a continuous stay event, among which, For the target in the region Duration of stay within, This is the dwell threshold for that area;

[0030] S2-5, Configure corresponding control parameter sets for each key hotspot area. The control parameter set includes at least the area code, control level, associated bridge opening code, acceptable vessel type, and maximum allowable number of vessels. Maximum permitted stay duration and corresponding VHF alert templates; area occupancy rate is defined as:

[0031]

[0032] in, For the target quantity in the current area, when A regional congestion event is triggered when the threshold is exceeded.

[0033] In a preferred embodiment, the video hotspot area ship detection and classification in step S3 includes the following steps:

[0034] S3-1, Perform ship target detection on video frames and obtain target bounding boxes. Detection confidence Including target category information, and removing those below the detection threshold. Candidate targets;

[0035] in, Indicates the first The x-coordinate of the bounding box of each target; Indicates the first The ordinate of the bounding box of each target; Indicates the first The width of the target bounding box; Indicates the first The height of the target bounding box;

[0036] S3-2, perform continuous frame correlation on the detected ship targets to form a video trajectory.

[0037] ;

[0038] in, Indicates the first A sequence of video trajectories formed by individual ship targets; Indicates the first The first goal in The bounding box of the target in the frame;

[0039] And calculate the target's entry time, exit time, and dwell time relative to each hotspot area based on the target's position changes in consecutive frames;

[0040] S3-3, classify the ship into categories based on its external dimensions, length-to-width ratio, outline features, superstructure features, or a trained classification model. The categories include at least one or more of the following: large ship, small ship, tugboat, engineering vessel, and non-AIS small target; wherein the length-to-width ratio... Defined as:

[0041]

[0042] The aspect ratio, area, and texture features are used together as classification input features;

[0043] S3-4 generates the area occupancy status, congestion status, and abnormal loitering status based on the number of targets, average dwell time, consistency of target movement direction, and area occupancy rate within each hotspot area; among which, the area congestion index... Defined as:

[0044]

[0045] in, The target quantity for the current area. The average length of stay, This serves as an indicator of the consistency of the direction of movement of targets within the region. These are the weighting coefficients; The maximum permitted stay duration; This represents the maximum allowed number of spaces;

[0046] S3-5, Constructing the Video Hotspot Event Result Package The video hotspot event result package includes at least: video target number, target category, hotspot area, area event type, entry / exit time, dwell time, area occupancy index, and abnormal status marker, and is output to steps S5, S6, and S9.

[0047] In a preferred embodiment, the AIS preprocessing and continuous track reconstruction in step S4 includes the following steps:

[0048] S4-1 receives AIS messages and parses them to obtain MMSI, ship name, latitude and longitude, speed, heading, ship type, ship size, and timestamp information, forming the original AIS observation sequence. :

[0049]

[0050] in, Indicates the ship's identification. , These represent longitude and latitude, respectively. The ship's current speed, The ship's current course. The timestamp corresponding to the AIS observation data;

[0051] S4-2 performs time sorting, duplicate message removal, abnormal jump point detection, and message segmentation on AIS messages; when the time interval between adjacent messages meets the requirement...

[0052]

[0053] in, Indicates the first The timestamp corresponding to each AIS message Indicates the first The timestamp corresponding to each AIS message; This indicates the preset time threshold for trajectory segmentation;

[0054] Or the distance between adjacent positions satisfies

[0055]

[0056] At that time, a new trajectory segment is initiated;

[0057] in, Indicates the first The ship's position point corresponding to each AIS message. Indicates the first The ship's location corresponding to each AIS message; This indicates the distance between two adjacent AIS location points; This indicates the preset threshold for trajectory segment distance;

[0058] S4-3, Project the AIS latitude and longitude coordinates onto a unified plane coordinate system to obtain the plane position coordinates. ;

[0059] S4-4 addresses the uneven update frequency of AIS by resampling and interpolating frames to generate a fixed time step. The following is a continuous AIS trajectory sequence; when When the position interpolation is used, it is expressed as:

[0060]

[0061] in, Indicates the time to be interpolated The estimated ship position point; Indicates the first The ship's location corresponding to each AIS message; Indicates the first The ship's position point corresponding to each AIS message, Indicates the first The timestamp corresponding to each AIS message; Indicates the first The timestamp corresponding to each AIS message; This indicates the intermediate time when position estimation is required;

[0062] S4-5, outlier removal and smoothing filtering are performed on the AIS trajectory to obtain a continuous sequence of position, velocity, and heading; and the heading-projected distance from the current position to the bridge control line is then used as the basis for this sequence. and the ship's current speed Estimated time of arrival of the vessel at the bridge :

[0063]

[0064] in, To prevent the smallest positive number with a denominator of zero;

[0065] S4-6, Generate AIS trajectory quality weights based on message update interval, interpolation ratio, filter residual, and positioning quality. The quality weight is expressed as:

[0066]

[0067] in, As a time continuity indicator, This is an interpolation ratio indicator. The residual index is the filtering performance index. To define quality indicators, to These are the weighting coefficients.

[0068] In a preferred embodiment, the AIS candidate filtering based on video hotspot events in step S5 includes the following steps:

[0069] S5-1, Read the hotspot area event result packet output in step S3, and determine the set of control areas that need to be focused on and controlled at present based on the hotspot area number, event type and event level;

[0070] S5-2, Within a unified time window, select a candidate vessel set from the AIS vessel set that is near the control area and whose navigation direction is consistent with the rules of that area. ;

[0071] S5-3, combining vessel category, estimated arrival time at the bridge, estimated entry time window into the bridge opening, course direction, current speed, and historical bridge crossing paths, assigns a relevance score to candidate vessels, obtaining a matching score between hotspot events and AIS candidate vessels. :

[0072]

[0073] in, The score represents the consistency of direction. For arrival time consistency score, The score represents the ship type consistency score. The historical path consistency score. For AIS trajectory quality weights, to These are the weighting coefficients;

[0074] S5-4, When a hotspot area is identified as being occupied by a bridge opening, congested in the bridge area, experiencing abnormal stops, or being in a conflict-sensitive state, the screening priority of candidate vessels related to that area is increased; the priority increase is achieved by increasing the weight coefficient of the corresponding event or decreasing the candidate score threshold;

[0075] S5-5, Output a set of candidate control objects. The set of candidate control objects includes at least: MMSI, ship name, ship type, current position, speed, heading, estimated time to bridge, candidate bridge span, candidate priority score, and corresponding hotspot area number, and is called by step S6.

[0076] In a preferred embodiment, the traffic risk prediction and priority assessment in step S6 includes the following steps:

[0077] S6-1, For each candidate ship, predict the future time domain Trajectory prediction is performed within the system to obtain a sequence of future discrete trajectory points.

[0078]

[0079] in, Indicates the first Candidate ships in the future prediction time domain A sequence of future discrete trajectory points within; Indicates the candidate ship number; Represents the first term in the future prediction time domain. One predicted moment; This represents the total number of discrete prediction points within the prediction time domain; and They represent the first The candidate ships in The horizontal and vertical coordinates of the plane at each predicted time point; Indicates the first The candidate ships in Predicted speed at each predicted moment; Indicates the first The candidate ships in Predicted heading at each predicted moment;

[0080] And at the same time, an uncertainty parameter is given to each prediction point. The uncertainty parameter Here is the position error covariance matrix of the predicted point in the planar coordinate system:

[0081]

[0082] in, , This represents the standard deviation of the prediction error in the x and y directions. The correlation coefficient is used; the uncertainty parameter can be obtained from the statistical analysis of prediction residuals of historical data: inputting historical trajectories into the prediction model to obtain prediction results, and calculating residuals with the actual locations. And estimate using the variance / covariance of the residuals To reflect that the longer the prediction step size, the greater the error, Follow Monotonically increasing or amplifying according to a preset growth model;

[0083] S6-2, for any two candidate ships Calculate future relative distance :

[0084]

[0085] in, Indicates the first The candidate ships in the future The horizontal coordinate of the plane at each predicted time point; Indicates the first The candidate ships in the future The ordinate of the plane at each predicted time point; Indicates the first The candidate ships in the future The horizontal coordinate of the plane at each predicted time point; Indicates the first The candidate ships in the future The ordinate of the plane at each predicted time point;

[0086] And take the minimum value :

[0087]

[0088] Its corresponding time :

[0089]

[0090] To assess the risk of ship-to-ship collisions;

[0091] S6-3, based on the spatial relationship between the predicted trajectory and the bridge protection zone, pier protection zone, or bridge opening centerline, calculate the ship-bridge collision risk, pier abrasion risk, and bridge opening yaw risk; among these, when the lateral deviation of the ship's predicted position from the bridge opening centerline is... Then the risk of bridge sloping. Represented as:

[0092]

[0093] in, The allowable lateral deviation threshold;

[0094] S6-4 calculates the bridge occupancy conflict risk based on the bridge opening occupancy zone status, bridge opening approach zone status, expected overlap of entry time windows, and the number of candidate vessels; when two or more vessels enter the same bridge opening, the predicted time window overlap duration is... At that time, the risk of bridge arch conflict Represented as:

[0095]

[0096] in, This refers to the occupancy status of the bridge opening. The number of candidate vessels for the bridge arches. Provide a reference passage time window for the bridge arch; to These are the weighting coefficients; It indicates the first The maximum number of vessels allowed to pass through or participate in scheduling at the same time in each bridge opening;

[0097] S6-5 calculates the vessel priority score based on vessel type, vessel size, hazard class, navigation direction, estimated arrival time at the bridge, bridge span compatibility, whether the vessel has entered the waiting area, and current traffic density. :

[0098]

[0099] in, to This represents the weight coefficient corresponding to each evaluation factor; Indicates the ship type priority factor; Indicates the ship size priority factor; Indicates the hazard level factor; This represents the estimated time factor to reach the bridge. Indicates the bridge arch adaptability or navigation direction matching factor; Indicates the waiting state factor; This represents the current traffic density factor;

[0100] S6-6, adjust the risk level based on the status of video hotspot areas; when a hotspot area is congested, abnormally occupied, or restricted, the corresponding vessel's business risk level is adjusted as follows:

[0101]

[0102] in, Indicates the first The business risk value or business risk level of a vessel after correction of its status in video hotspot areas; The base risk value is obtained from AIS prediction. This represents the incremental risk introduced by the status of video trending topics;

[0103] S6-7, Output the risk assessment result package, which includes at least: MMSI, ship-to-ship risk level, ship-to-bridge risk level, bridge span conflict risk level, priority score, and recommended control action.

[0104] In a preferred embodiment, the traffic control decision generation in step S7 includes the following steps:

[0105] S7-1, Based on the risk assessment result package output in step S6, determine the type of traffic control strategy to be executed and construct a control decision set. ;

[0106] S7-2, the traffic control strategy includes at least one or more of the following: deceleration, maintaining course, yielding in advance, entering the waiting area, prohibiting entry into the designated bridge opening, switching to the backup bridge opening, and restoring traffic; different strategies are triggered by the corresponding risk level range and priority range.

[0107] S7-3, When multiple vessels are competing for the same bridge span, priority scores will be used as the basis for determining the appropriate bridge span. Determine the single priority passage object based on the occupancy status of the bridge opening. :

[0108]

[0109] And designate the remaining vessels as waiting targets or avoidance targets;

[0110] S7-4: When there is a risk of overlap between the predicted trajectory of a vessel and the bridge pier protection zone or restricted area, priority should be given to generating control decisions such as emergency deceleration, course correction, or suspension of entry into the bridge area; among which, when the comprehensive risk value... satisfy

[0111]

[0112] Emergency control strategies are triggered at certain times, among which, This is the emergency control threshold;

[0113] S7-5, output the control decision in the form of a structured control message, which includes at least: target vessel identity, risk type, target bridge opening, control level, control action, effective time window and alert priority, for VHF broadcast in step S8.

[0114] In a preferred embodiment, the VHF graded alert and closed-loop feedback in step S8 includes the following steps:

[0115] S8-1 Select the broadcast alert or targeted alert mode according to the control level of the control message; broadcast alerts are used for low-level risks, and targeted alerts based on the vessel name or MMSI are used for high-level risks.

[0116] S8-2, based on different risk types, call the corresponding voice templates to automatically generate VHF broadcast content; the broadcast content includes at least: target vessel identity, risk description, control requirements, suggested actions and execution time limits;

[0117] S8-3, the tiered alerts include at least one or more of the following: situation alert level, early warning alert level, control intervention level, and emergency warning level; and are based on risk values. The alert level will be automatically selected based on the current location.

[0118] S8-4, After the VHF alert is issued, continuously monitor the changes in the target vessel's AIS trajectory and the status changes in video hotspot areas to assess whether the vessel has executed the aforementioned control requirements; execution deviation. Defined as:

[0119]

[0120] in, and The first The actual speed and course of the vessel after receiving the VHF alert. and These are respectively controlling the target's speed and controlling the target's heading. The deviation between the target and the specified waiting area or the specified bridge opening passage path; to These are the weighting coefficients;

[0121] S8-5: When the risk decreases or the vessel's actions conform to the control strategy after the alert, it is recorded as effective intervention; when there is still no response after the alert or the risk continues to rise, the alarm level is automatically escalated, triggering a second alert or manual intervention; among these, when continuous... Satisfaction within each evaluation period

[0122]

[0123] When this occurs, it is determined that the control requirements have not been effectively implemented; among them, Indicates the first Control execution deviation values ​​for each vessel; This indicates the preset control execution deviation threshold.

[0124] In a preferred embodiment, the multi-source data fusion verification in step S9 includes the following steps:

[0125] S9-1 integrates AIS dynamic data, video structured event data, and VHF alert logs to construct a control policy execution verification link and establish multi-source association records according to event number;

[0126] S9-2 performs consistency verification on changes in ship speed, course, bridge approach status, and hotspot area occupancy status before and after the issuance of control commands; consistency score. Represented as:

[0127]

[0128] in, As a speed consistency indicator, For heading consistency indicators. This serves as an indicator of regional consistency. For VHF execution log consistency metrics; to These are the weighting coefficients;

[0129] S9-3, when AIS shows that a vessel has not slowed down as instructed, has not waited as instructed, or continues to approach the collision bridge opening, it is identified as an abnormal response vessel; when the consistency score meets the requirements...

[0130]

[0131] When this occurs, an abnormality in control execution is determined; among which, Indicates the first Consistency score of control execution of the vessels; This indicates the preset consistency judgment threshold;

[0132] S9-4, triggering secondary intervention for abnormal response vessels, the secondary intervention includes at least raising the VHF alert level, extending the waiting time window, transferring to manual duty confirmation or linking with law enforcement modules;

[0133] S9-5, Output multi-source verification results, which include at least: target vessel identity, original control strategy, execution status, response delay, anomaly markers, and secondary intervention records, as input for the archiving analysis in step S10.

[0134] In a preferred embodiment, the control effect archiving analysis in step S10 includes the following steps:

[0135] S10-1, Archive a complete traffic control process according to the event number. The archived content shall include at least: the original AIS trajectory, video hotspot events, risk assessment results, control decisions, VHF broadcast logs, execution verification results, and final handling results.

[0136] S10-2, based on indicators such as event type, risk level, intervention duration, response duration, execution success rate, and false alarm rate, statistical analysis is conducted on the control effectiveness; among which, the execution success rate... Represented as:

[0137]

[0138] in, Indicates the number of events that successfully executed the control request; This represents the total number of all traffic control events within the statistical period;

[0139] False alarm rate Represented as:

[0140]

[0141] in, Indicates the false alarm rate; Indicates the total number of warning events;

[0142] S10-3 Update hotspot area threshold parameters, risk assessment weight parameters, bridge hole priority parameters, and VHF voice template priority based on the archived results; parameter updates can be performed using sliding window statistical updates or incremental correction methods.

[0143] S10-4 utilizes the archived analysis results for model iterative optimization and dynamic updating of control rules to improve the accuracy and closed-loop efficiency of subsequent water traffic control; among which, the updated parameter set... Represented as

[0144]

[0145] in, Indicates the first The parameter set after the update; indicating the first... The original parameter set used in the next update.

[0146] Compared with the prior art, the present invention has the following beneficial effects:

[0147] (1) This invention transforms video surveillance from simple image observation into semantic perception of hotspot areas for traffic control. It can identify key business statuses such as bridge entrance, bridge hole occupancy, waiting congestion and restricted area occupancy without directly obtaining the latitude and longitude of ships through video.

[0148] (2) This invention utilizes AIS to provide dynamic information such as ship identity, location, speed and heading, and achieves quantitative assessment of ship-to-ship collision risk, ship-to-bridge collision risk and bridge span conflict risk through continuous track reconstruction and future trajectory prediction, making risk analysis more in line with traffic control business needs.

[0149] (3) This invention triggers AIS candidate screening and key calculation through video hot events, and performs detailed analysis only on ships related to key areas, which can effectively reduce the total amount of system calculation and manual screening workload.

[0150] (4) The present invention introduces a bridge area navigation priority assessment and traffic control decision generation mechanism, which can realize deceleration, waiting, yielding, bridge opening switching and entry prohibition control, thereby improving the navigation organization capability in bridge areas and complex waterway scenarios.

[0151] (5) This invention combines VHF graded alerts, multi-source fusion verification and control effect archiving analysis to form a closed-loop control process from risk identification to command execution and then to feedback optimization, which significantly improves the timeliness, accuracy and feasibility of water traffic control. Attached Figure Description

[0152] Figure 1 This is a schematic diagram of the overall process of the method of the present invention. The diagram shows the complete closed-loop process of the present invention, starting from multi-source data acquisition and time benchmark unification, through waterway key hotspot area configuration, video hotspot area ship detection and classification, AIS preprocessing and continuous track reconstruction, AIS candidate screening based on video hotspot events, traffic risk prediction and priority assessment, traffic control decision generation, VHF graded alerts and closed-loop feedback, multi-source data fusion verification, and control effect archiving analysis.

[0153] Figure 2 This diagram illustrates the process of cross-view association and risk management of ships under multiple preset PTZ camera images in this invention. The diagram shows that, based on the input of multiple preset PTZ camera image streams, the system first detects and tracks the ship target from a single viewpoint, then extracts ship features and performs single-view trajectory analysis. Subsequently, through cross-preset position identity association and maintenance, continuous identity recognition and trajectory association of the same ship under different monitoring views are achieved. Based on this, the system further performs continuous associated trajectory input, collision risk assessment, abnormal behavior detection, and boundary crossing alarms, and generates risk management command issuance results. The diagram also demonstrates the process of cross-preset position identity matching, abnormal area identification, collision risk area judgment, and risk alarm and risk management command output between different PTZ camera monitoring areas in the application scenario.

[0154] Figure 3This is a schematic diagram illustrating the configuration of key hotspot areas in the waterway according to the present invention. The diagram shows the method of delineating and semantically labeling hotspot areas, such as bridge entrance areas, bridge opening approach areas, bridge opening occupancy areas, waiting areas, conflict-sensitive areas, and restricted areas, based on bridge structure, waterway boundaries, bridge opening locations, waiting areas, conflict-sensitive areas, and restricted areas, using fixed cameras or pre-set monitoring screens. Here, "Semantics" represents the traffic control semantics corresponding to the hotspot area, "Occupancy" indicates area occupancy status, "Risk Gauge" indicates risk indication, "High" indicates high risk, "Boundary Alert" indicates boundary warning, and Z1 to Z6 represent the numbers of different key hotspot areas in the waterway.

[0155] Figure 4 This diagram illustrates the traffic risk prediction and navigation priority assessment methods used in this invention. It shows how the system calculates ship-to-ship collision risk, ship-to-bridge collision risk, and bridge occupancy conflict risk based on AIS continuous track reconstruction and future trajectory prediction. The system then adjusts the risk level by incorporating the status of video hotspot areas and generates a navigation priority score for the target vessel. In this diagram, DCPA represents the nearest encounter distance, TCPA represents the nearest encounter time, DCPA / TCPA represents the collision risk assessment method based on the nearest encounter distance and time, nm represents nautical miles, min represents minutes, and VHF represents very high frequency communication.

[0156] Figure 5 This diagram illustrates the VHF graded alerts and closed-loop feedback mechanism in this invention. It shows the system's process of generating broadcast or directional alerts based on traffic control decisions, and then verifying, escalating alarms, conducting secondary interventions, and archiving analyses based on subsequent AIS trajectory changes, video hotspot area status changes, and VHF alert logs. In this diagram, MMSI represents the Maritime Mobile Service Identifier, AIS represents the Automatic Identification System for Ships, VHF represents Very High Frequency Communication, Speed ​​represents ship speed, Heading represents ship heading, kn represents knots, Δv represents speed deviation, Δψ represents heading deviation, and min represents minutes. Detailed Implementation

[0157] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0158] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0159] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations according to this application; as used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise; furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0160] This embodiment discloses a water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts.

[0161] I. Overall Process

[0162] like Figure 1-5 As shown, this invention proposes a water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts. This method is applicable to scenarios such as bridge areas, complex waterways, and encounter-sensitive areas. It outputs regional-level traffic events from the video side, outputs vessel identity, location, and future trajectory from the AIS side, and completes a traffic control closed loop by combining VHF alerts.

[0163] This method includes the following steps: First, it integrates fixed cameras or pre-set monitoring footage, AIS data, electronic navigation chart data, bridge and waterway rule base data, and VHF communication interface data, and establishes a unified time reference. Second, it delineates key hotspot areas in the monitoring footage, such as bridge entrance areas, bridge opening approach areas, bridge opening occupancy areas, waiting areas, conflict-sensitive areas, and restricted areas. Then, it detects, tracks, and classifies vessel targets in the video footage, and outputs structured events such as area occupancy, area congestion, and abnormal stops. Simultaneously, it preprocesses AIS messages, reconstructs trajectories, and predicts future trajectories to obtain candidate control objects and their estimated arrival times at the bridge. Subsequently, it filters and correlates video hotspot events with AIS candidate objects, calculates ship-to-ship collision risk, ship-to-bridge collision risk, bridge opening conflict risk, and navigation priority, and generates traffic control strategies. Finally, it broadcasts or provides targeted alerts via VHF, and verifies and archives the control execution based on changes in AIS trajectories, changes in video hotspot status, and log records after the alerts.

[0164] II. Configuration of Key Hotspot Areas in Waterways

[0165] like Figure 2 As shown in this embodiment, the system first performs regional semantic division of the monitoring screen based on the bridge structure, bridge opening layout, channel centerline, boundary line, waiting water area, prohibited water area, and historical high conflict encounter area.

[0166] The hotspot areas include at least the following types:

[0167] (1) Bridge entrance area, used to identify vessels about to enter the controlled area of ​​the bridge area;

[0168] (2) Bridge approach zone, used to identify vessels that are about to enter a specific bridge approach path;

[0169] (3) Bridge opening occupancy area, used to determine whether the bridge opening is currently occupied by a ship;

[0170] (4) Waiting area, used to identify whether vessels are waiting or queuing in accordance with traffic control requirements;

[0171] (5) Conflict-sensitive areas are used to identify high-risk situations such as encounters, overtaking, and crossings;

[0172] (6) Restricted area, used to identify vessels that have mistakenly entered the edge of the bridge area, the vicinity of non-navigable bridge openings, or other areas that are not permitted to enter.

[0173] Each hotspot area can be defined in the form of a rectangle, polygon, polyline buffer zone, or area mask, and each area can be configured with parameters such as area number, control level, associated bridge hole number, acceptable ship type, maximum allowable number of ships, maximum allowable stay duration, and VHF alert template.

[0174] Furthermore, for the target quantity within the region is The maximum allowed number of spaces is area Its regional occupancy rate can be defined as:

[0175]

[0176] when When the threshold is exceeded, the area is determined to be in a congested state.

[0177] III. Ship Inspection and Classification in Video Hotspot Areas

[0178] In this embodiment, the system performs ship target detection on the input video frames to obtain the bounding boxes, categories, and detection confidence of candidate targets; and performs trajectory association on the same target in consecutive frames to form video target trajectories.

[0179] For each detected vessel target, the system determines whether an entry event, departure event, sustained stay event, or reverse crossing event has occurred based on its position changes across consecutive frames. Specifically, if the target is in the area... The number of consecutive frames within the time limit is The video frame rate is The duration of stay is:

[0180]

[0181] when When this occurs, it is determined to be a continuous stay event, among which, This represents the dwell threshold for the corresponding area.

[0182] Simultaneously, the system identifies the target's ship type based on the ship's bounding box dimensions, aspect ratio, outline features, superstructure features, or a trained classification model. The identified ship types include at least large ships, small ships, tugboats, engineering vessels, and non-AIS small targets.

[0183] Furthermore, the system combines the number of targets within the area, average dwell time, consistency of target movement direction, and area occupancy rate to generate area occupancy status, congestion status, and abnormal dwell status. The area congestion index can be expressed as:

[0184]

[0185] in, For the target quantity in the region, The average length of stay, This serves as an indicator of the consistency of the direction of movement of targets within the region. These are the weighting coefficients.

[0186] Finally, the system outputs a video hotspot event result package, which includes at least: video target number, target category, hotspot area, area event type, entry time, exit time, dwell time, area occupancy index, and abnormal status marker.

[0187] IV. AIS Preprocessing and Continuous Track Reconstruction

[0188] In this embodiment, the system receives AIS messages and parses them to obtain information such as MMSI, ship name, latitude and longitude, speed, heading, ship type, ship size and timestamp, forming the original AIS observation sequence.

[0189] Subsequently, the AIS messages are sorted by time, duplicate messages are removed, abnormal jump points are detected, and trajectory segments are processed. When the time interval between adjacent messages exceeds a set threshold, or the distance between adjacent locations exceeds a set threshold, a new trajectory segment is initiated.

[0190] To facilitate subsequent distance and risk calculations, the system projects AIS latitude and longitude onto a unified plane coordinate system to obtain planar position coordinates. To address the issue of uneven AIS update frequency, the system further resamples and interpolates the trajectory. For each time interval... Its position interpolation can be expressed as:

[0191]

[0192] in, Indicates the time to be interpolated The estimated ship position point; Indicates the first The ship's location corresponding to each AIS message; Indicates the first The ship's position point corresponding to each AIS message, Indicates the first The timestamp corresponding to each AIS message; Indicates the first The timestamp corresponding to each AIS message; This indicates the intermediate time when position estimation is required;

[0193] Furthermore, the system performs outlier removal and smoothing filtering on the AIS trajectory to obtain a continuous sequence of position, velocity, and heading. This is based on the heading distance from the current position to the bridge control line. With current speed Estimated time of arrival of the vessel at the bridge:

[0194]

[0195] in, To prevent the smallest positive number with a denominator of zero.

[0196] In addition, the system generates AIS trajectory quality weights based on message update intervals, interpolation ratios, filtering residuals, and positioning quality, which serve as the basis for credibility in candidate selection and risk assessment.

[0197] V. AIS Candidate Selection Based on Video Hot Topics

[0198] In this embodiment, the system reads the video hotspot event result packet output in step three and determines the current key control area based on the hotspot area number, event type, and event level.

[0199] Within a unified time window, a pool of candidate vessels is selected from the AIS vessel ensemble. These vessels are located near the control area, their navigation direction conforms to the rules of that area, and they are expected to enter the area. The selection of candidates considers not only spatial proximity but also factors such as expected arrival time at the bridge, expected time window for entering the bridge opening, current course, vessel type, and historical bridge crossing routes.

[0200] For candidate ships Its relevance score can be defined as:

[0201]

[0202] in, The score represents the consistency of direction. For arrival time consistency score, The score represents the ship type consistency score. The historical path consistency score. This represents the quality weight of the AIS trajectory.

[0203] When a hotspot area is identified as having bridge opening occupancy, bridge congestion, abnormal stops, or being in a conflict-sensitive state, the system increases the screening priority of AIS vessels related to that area so that traffic risk calculations and control decisions can be implemented more effectively in the future.

[0204] VI. Traffic Risk Prediction and Air Traffic Priority Assessment

[0205] like Figure 3 As shown, in this embodiment, the system predicts the trajectory of each candidate ship in the future prediction time domain, and obtains the future discrete trajectory point sequence and the corresponding uncertainty parameters.

[0206] For any two candidate ships and Their future relative distance can be expressed as:

[0207]

[0208] Take the minimum distance among all predicted times as the nearest encounter distance (DCPA):

[0209]

[0210] The corresponding time is taken as the nearest meeting time (TCPA).

[0211]

[0212] The system calculates ship-to-ship collision risk based on DCPA and TCPA; simultaneously, it calculates ship-to-bridge collision risk and bridge span yaw risk based on the spatial relationship between the predicted trajectory and the pier protection zone, bridge boundary, and bridge span centerline. If the lateral deviation of the ship's predicted position from the bridge span centerline is... The bridge yaw risk can then be expressed as:

[0213]

[0214] in, The threshold for allowing lateral deviation.

[0215] Furthermore, if two or more vessels are predicted to compete for the same bridge opening within the same time window, the bridge opening occupancy conflict risk is calculated. The bridge opening conflict risk can be expressed as:

[0216]

[0217] in, This refers to the occupancy status of the bridge opening. The number of candidate vessels for the bridge arches. This serves as a reference time window for passage through the bridge arches.

[0218] In addition, the system calculates the vessel's navigation priority score by combining factors such as vessel type, vessel size, hazard level, navigation direction, estimated arrival time at the bridge, bridge span compatibility, whether it has entered the waiting area, and current traffic density.

[0219]

[0220] Meanwhile, the status of video hotspot areas serves as a business risk correction item, adjusting the base risk value obtained from AIS predictions.

[0221]

[0222] in, This refers to the incremental risks introduced by video events such as congestion, abnormal lingering, or restricted access.

[0223] VII. Traffic Control Decision Generation

[0224] In this embodiment, the system generates a traffic control strategy for the target vessel based on traffic risk prediction and priority assessment results. The control strategy includes at least one or more of the following: deceleration, maintaining course, giving way in advance, entering the waiting area, prohibiting entry into the designated bridge opening, switching to the backup bridge opening, and restoring traffic flow.

[0225] When multiple vessels are competing for the same bridge opening, the system determines the single priority passage target based on priority scores:

[0226]

[0227] The remaining vessels were designated as waiting targets or avoidable targets.

[0228] When the overall risk value exceeds a preset emergency threshold, the system prioritizes generating control strategies such as emergency deceleration, course correction, or suspension of entry into the bridge area. Ultimately, the system organizes the control decisions into structured control messages, which include at least: target vessel identity, risk type, target bridge opening, control level, control action, effective time window, and alert priority.

[0229] VIII. VHF Classification Alerts and Closed-Loop Feedback

[0230] like Figure 4 As shown, in this embodiment, the system selects between broadcast alerts and targeted alerts based on the control level in the control message. For low-level risks, broadcast alerts are used; for high-level risks, targeted alerts based on the vessel name or MMSI are used.

[0231] The system automatically generates VHF broadcast content by calling corresponding voice templates based on different risk types. The broadcast content must include at least the target vessel's identity, risk description, control requirements, recommended actions, and execution time limits. The alert levels must include at least situational awareness, early warning, control intervention, and emergency alert.

[0232] After the VHF alert is issued, the system continuously monitors changes in the target vessel's AIS trajectory and the status of video hotspot areas to assess whether the vessel has complied with the control requirements. The execution deviation can be defined as:

[0233]

[0234] in, and These are respectively controlling the target's speed and controlling the target's heading. The deviation between the target and the specified waiting area or the specified bridge opening passage path.

[0235] When the risk decreases or the vessel's actions meet control requirements after the alert, it is recorded as effective intervention; when there is still no response after the alert or the risk continues to rise, the alarm level is automatically upgraded, and a second alert or manual intervention is triggered.

[0236] IX. Multi-source fusion verification and control effect archiving analysis

[0237] In this embodiment, the system integrates AIS dynamic data, video structured event data, and VHF alert logs to construct a control strategy execution verification link. Consistency verification is performed on changes in ship speed, course, bridge approach status, and hotspot area occupancy status before and after the issuance of control commands.

[0238] The consistency score can be expressed as:

[0239]

[0240] in, As a speed consistency indicator, For heading consistency indicators. This serves as an indicator of regional consistency. This is a consistency metric for VHF execution logs.

[0241] When the consistency score is lower than the preset threshold, the vessel is deemed to have failed to effectively execute control commands, and secondary intervention is triggered, including raising the VHF alert level, extending the waiting time window, transferring to manual duty confirmation, or linking with the law enforcement module.

[0242] Finally, the system archives a complete traffic control process by event number. The archived content includes at least: the original AIS trajectory, video hotspot events, risk assessment results, control decisions, VHF broadcast logs, execution verification results, and final handling results.

[0243] For archived events, the system further performs statistical analysis based on event type, risk level, intervention duration, response time, execution success rate, and false alarm rate. The execution success rate can be expressed as:

[0244]

[0245] The false alarm rate can be expressed as:

[0246]

[0247] Based on the archived analysis results, the system updates the threshold parameters for hotspot areas, risk assessment weight parameters, bridge hole priority parameters, and VHF voice template priority, thereby improving the accuracy and closed-loop efficiency of subsequent water traffic control.

Claims

1. A water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts, characterized in that, Includes the following steps: S1, Multi-source data acquisition and time benchmark unification: Acquire data from fixed cameras or preset monitoring screens, AIS data, electronic navigation charts, bridge and waterway rule bases, and VHF communication interface data, and perform time alignment; S2, Key Hotspot Area Configuration in Waterways: Configure video hotspot areas such as bridge entrance area, bridge opening approach area, bridge opening occupancy area, waiting area, conflict-sensitive area, and restricted area in the monitoring screen, and assign traffic control semantics to each area; S3, Ship Detection and Classification in Video Hotspot Areas: Detects ship targets in the video and determines which video hotspot area the target enters, leaves, or stays in, outputting the video hotspot area occupancy status, congestion status, abnormal loitering status, and ship type classification results. S4, AIS preprocessing and continuous track reconstruction: Receive AIS messages and parse them to obtain MMSI, latitude and longitude, speed, and heading; perform time alignment, interpolation, frame interpolation, anomaly removal, and smoothing filtering on the AIS data to obtain a continuous AIS track sequence. and estimated arrival time at the bridge; S5, AIS candidate screening based on video hotspot events: Based on the status of the hotspot area, time window, navigation direction, expected arrival order and vessel type information, screen candidate control objects related to the control area from the AIS vessel set; S6, Traffic Risk Prediction and Priority Assessment: Based on AIS trajectory prediction, calculate ship-to-ship collision risk, ship-to-bridge collision risk, bridge span occupancy conflict risk and navigation priority, and correct the risk level by combining the status of video hotspot areas; S7, Traffic Control Decision Generation: Based on the risk level and priority assessment results, generate traffic control strategies such as deceleration, waiting, yielding, bridge switching, prohibition of entry, and resumption of traffic. S8, VHF graded alerts and closed-loop feedback: Generate broadcast or targeted alerts through the nearest VHF, and evaluate the intervention effect based on subsequent AIS trajectory changes and video hotspot area status changes, and upgrade the alarm level when necessary; S9, Multi-source data fusion verification: Integrates AIS dynamic data, video structured information and VHF feedback logs to verify the consistency of control strategy execution in real time, identify abnormal response vessels and trigger secondary intervention; S10, Control Effect Archiving Analysis: Automatically archives closed-loop data according to event type, risk level, intervention duration and success rate, supporting model iteration optimization and dynamic updates of control rules.

2. The water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, The multi-source data acquisition and time base unification described in S1 include the following steps: S1-1: Receive the video stream from a fixed camera or a preset monitoring screen, and decode and extract frames according to the set frame rate to obtain a video frame sequence. and corresponding video timestamps Simultaneously, it accesses AIS message streams, electronic navigation chart data, bridge and waterway rule base data, and VHF communication interface data, recording AIS timestamps respectively. Rule call timestamp and VHF interaction log timestamp ; S1-2, Establishing a unified time standard For the corresponding video timestamp AIS timestamp Rule call timestamp and VHF interaction log timestamp Perform a unified mapping to create a unified timeline; S1-3, Set a time alignment window on a unified timeline For each video frame time Retrieve from AIS cache that meet the requirements AIS observations, The AIS time is uniformly mapped; when no original AIS observation value that meets the conditions is retrieved, time interpolation is performed on adjacent AIS messages to obtain the AIS state estimate value aligned with the video frame time. S1-4 involves structured encapsulation of various input data to form standardized data packets. The data packet includes at least: video frame number, video timestamp, AIS vessel identification information, vessel position, vessel speed, vessel heading, vessel electronic navigation chart area identifier, bridge rule parameters, and VHF equipment status; wherein: in, Indicates the video frame number. Represents the AIS state set. Indicates electronic navigation chart information, Represents the set of rule parameters. Represents the VHF state set; S1-5, the standardized data packets are output to the channel key hotspot area configuration in step S2, the video hotspot area ship detection and classification in step S3, the AIS preprocessing in step S4, the risk prediction and priority assessment in step S6, and the closed-loop feedback in step S8, respectively, so as to realize the unified calling and time-series consistency processing of multi-source data. The configuration of key hotspot areas in the waterway mentioned in step S2 includes the following steps: S2-1, based on the bridge structure, waterway boundaries, bridge arch locations, waiting waters, prohibited waters, and high-conflict encounter areas, pre-defines multiple key hotspot areas in the monitoring screen. The key hotspot areas are defined in the form of polygons, rectangles, polyline buffer zones, or area masks; S2-2, the key hotspot areas include at least one or more of the following: bridge entrance area, bridge opening approach area, bridge opening occupancy area, waiting area, conflict-sensitive area, and restricted area, and each area is assigned a unique area number. ; S2-3 assigns traffic control semantic labels to different key hotspot areas. The semantic tags include at least one or more of the following: allow passage, priority passage, slow down and observe, wait for yield, prohibit entry, and conflict warning. S2-4, Configure event determination rules for each key hotspot area. These rules include at least entry events, exit events, sustained stay events, reverse crossing events, area congestion events, and abnormal aggregation events. Specifically, an entry event is defined when the target trajectory center point enters the area from outside the area. An entry event is defined when the target stays in the area for a certain duration. The event is determined to be a continuous stay event, among which, For the target in the region Duration of stay within, This is the dwell threshold for that area; S2-5, Configure corresponding control parameter sets for each key hotspot area. The control parameter set includes at least the area code, control level, associated bridge opening code, acceptable vessel type, and maximum allowable number of vessels. Maximum permitted stay duration and corresponding VHF alert templates; area occupancy rate is defined as: in, For the target quantity in the current area, when A regional congestion event is triggered when the threshold is exceeded.

3. The water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, The video hotspot area ship detection and classification described in step S3 includes the following steps: S3-1, Perform ship target detection on video frames and obtain target bounding boxes. Detection confidence Including target category information, and removing those below the detection threshold. Candidate targets; in, Indicates the first The x-coordinate of the bounding box of each target; Indicates the first The ordinate of the bounding box of each target; Indicates the first The width of the target bounding box; Indicates the first The height of the target bounding box; S3-2, perform continuous frame correlation on the detected ship targets to form a video trajectory. ； in, Indicates the first A sequence of video trajectories formed by individual ship targets; Indicates the first The first goal in The bounding box of the target in the frame; And calculate the target's entry time, exit time, and dwell time relative to each hotspot area based on the target's position changes in consecutive frames; S3-3, classify the ship into categories based on its external dimensions, length-to-width ratio, outline features, superstructure features, or a trained classification model. The categories include at least one or more of the following: large ship, small ship, tugboat, engineering vessel, and non-AIS small target; wherein the length-to-width ratio... Defined as: The aspect ratio, area, and texture features are used together as classification input features; S3-4 generates the area occupancy status, congestion status, and abnormal loitering status based on the number of targets, average dwell time, consistency of target movement direction, and area occupancy rate within each hotspot area; among which, the area congestion index... Defined as: in, The target quantity for the current area. The average length of stay, This serves as an indicator of the consistency of the direction of movement of targets within the region. These are the weighting coefficients; The maximum permitted stay duration; This represents the maximum allowed number of spaces; S3-5, Constructing the Video Hotspot Event Result Package The video hotspot event result package includes at least: video target number, target category, hotspot area, area event type, entry / exit time, dwell time, area occupancy index, and abnormal status marker, and is output to steps S5, S6, and S9.

4. The water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, Step S4, AIS preprocessing and continuous track reconstruction, includes the following steps: S4-1 receives AIS messages and parses them to obtain MMSI, ship name, latitude and longitude, speed, heading, ship type, ship size, and timestamp information, forming the original AIS observation sequence. : in, Indicates the ship's identification. , These represent longitude and latitude, respectively. The ship's current speed, The ship's current course. The timestamp corresponding to the AIS observation data; S4-2 performs time sorting, duplicate message removal, abnormal jump point detection, and message segmentation on AIS messages; when the time interval between adjacent messages meets the requirement... in, Indicates the first The timestamp corresponding to each AIS message Indicates the first The timestamp corresponding to each AIS message; This indicates the preset time threshold for trajectory segmentation; Or the distance between adjacent positions satisfies At that time, a new trajectory segment is initiated; in, Indicates the first The ship's position point corresponding to each AIS message. Indicates the first The ship's location corresponding to each AIS message; This indicates the distance between two adjacent AIS location points; This indicates the preset threshold for trajectory segment distance; S4-3, Project the AIS latitude and longitude coordinates onto a unified plane coordinate system to obtain the plane position coordinates. ; S4-4 addresses the uneven update frequency of AIS by resampling and interpolating frames to generate a fixed time step. The following is a continuous AIS trajectory sequence; when When the position interpolation is used, it is expressed as: in, Indicates the time to be interpolated The estimated ship position point; Indicates the first The ship's location corresponding to each AIS message; Indicates the first The ship's position point corresponding to each AIS message, Indicates the first The timestamp corresponding to each AIS message; Indicates the first The timestamp corresponding to each AIS message; This indicates the intermediate time when position estimation is required; S4-5, outlier removal and smoothing filtering are performed on the AIS trajectory to obtain a continuous sequence of position, velocity, and heading; and the heading-projected distance from the current position to the bridge control line is then used as the basis for this sequence. and the ship's current speed Estimated time of arrival of the vessel at the bridge : in, To prevent the smallest positive number with a denominator of zero; S4-6, Generate AIS trajectory quality weights based on message update interval, interpolation ratio, filter residual, and positioning quality. The quality weight is expressed as: in, As a time continuity indicator, This is an interpolation ratio indicator. The residual index is the filtering performance index. To define quality indicators, to These are the weighting coefficients.

5. A water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, Step S5, which involves AIS candidate filtering based on video hotspot events, includes the following steps: S5-1, Read the hotspot area event result packet output in step S3, and determine the set of control areas that need to be focused on and controlled at present based on the hotspot area number, event type and event level; S5-2, Within a unified time window, select a candidate vessel set from the AIS vessel set that is near the control area and whose navigation direction is consistent with the rules of that area. ; S5-3, combining vessel category, estimated arrival time at the bridge, estimated entry time window into the bridge opening, course direction, current speed, and historical bridge crossing paths, assigns a relevance score to candidate vessels, obtaining a matching score between hotspot events and AIS candidate vessels. : in, The score represents the consistency of direction. For arrival time consistency score, The score represents the ship type consistency score. The historical path consistency score. For AIS trajectory quality weights, to These are the weighting coefficients; S5-4 When a hotspot area is identified as being occupied by a bridge span, congested in the bridge area, experiencing abnormal stops, or being in a conflict-sensitive state, the screening priority of candidate vessels related to that area is increased; the priority increase is achieved by increasing the weight coefficient of the corresponding event or lowering the candidate score threshold; S5-5, Output a set of candidate control objects. The set of candidate control objects includes at least: MMSI, ship name, ship type, current position, speed, heading, estimated time to bridge, candidate bridge span, candidate priority score, and corresponding hotspot area number, and is called by step S6.

6. A water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, The traffic risk prediction and priority assessment described in step S6 includes the following steps: S6-1, For each candidate ship, predict the future time domain Trajectory prediction is performed within the system to obtain a sequence of future discrete trajectory points. in, Indicates the first Candidate ships in the future prediction time domain A sequence of future discrete trajectory points within; Indicates the candidate ship number; Represents the first term in the future prediction time domain. One predicted moment; This represents the total number of discrete prediction points within the prediction time domain; and They represent the first The candidate ships in The horizontal and vertical coordinates of the plane at each predicted time point; Indicates the first The candidate ships in Predicted speed at each predicted moment; Indicates the first The candidate ships in Predicted heading at each predicted moment; And at the same time, an uncertainty parameter is given to each prediction point. The uncertainty parameter Here is the position error covariance matrix of the predicted point in the planar coordinate system: in, , This represents the standard deviation of the prediction error in the x and y directions. The correlation coefficient is used; the uncertainty parameter can be obtained from the statistical analysis of prediction residuals of historical data: inputting historical trajectories into the prediction model to obtain prediction results, and calculating residuals with the actual locations. And estimate using the variance / covariance of the residuals To reflect that the longer the prediction step size, the greater the error, Follow Monotonically increasing or amplifying according to a preset growth model; S6-2, for any two candidate ships Calculate future relative distance : in, Indicates the first The candidate ships in the future The horizontal coordinate of the plane at each predicted time point; Indicates the first The candidate ships in the future The ordinate of the plane at each predicted time point; Indicates the first The candidate ships in the future The horizontal coordinate of the plane at each predicted time point; Indicates the first The candidate ships in the future The ordinate of the plane at each predicted time point; And take the minimum value : Its corresponding time : To assess the risk of ship-to-ship collisions; S6-3, based on the spatial relationship between the predicted trajectory and the bridge protection zone, pier protection zone, or bridge opening centerline, calculate the ship-bridge collision risk, pier abrasion risk, and bridge opening yaw risk; among these, when the lateral deviation of the ship's predicted position from the bridge opening centerline is... Then the risk of bridge sloping. Represented as: in, The allowable lateral deviation threshold; S6-4 calculates the bridge occupancy conflict risk based on the bridge opening occupancy zone status, bridge opening approach zone status, expected overlap of entry time windows, and the number of candidate vessels; when two or more vessels enter the same bridge opening, the predicted time window overlap duration is... At that time, the risk of bridge arch conflict Represented as: in, This refers to the occupancy status of the bridge opening. The number of candidate vessels for the bridge arches. Provide a reference passage time window for the bridge arch; to These are the weighting coefficients; It indicates the first The maximum number of vessels allowed to pass through or participate in scheduling at the same time in each bridge opening; S6-5 calculates the vessel priority score based on vessel type, vessel size, hazard class, navigation direction, estimated arrival time at the bridge, bridge span compatibility, whether the vessel has entered the waiting area, and current traffic density. : in, to This represents the weight coefficient corresponding to each evaluation factor; Indicates the ship type priority factor; Indicates the ship size priority factor; Indicates the hazard level factor; This represents the estimated time factor to reach the bridge. Indicates the bridge arch adaptability or navigation direction matching factor; Indicates the waiting state factor; This represents the current traffic density factor; S6-6, adjust the risk level based on the status of video hotspot areas; when a hotspot area is congested, abnormally occupied, or restricted, the corresponding vessel's business risk level is adjusted as follows: in, Indicates the first The business risk value or business risk level of a vessel after correction of its status in video hotspot areas; The base risk value is obtained from AIS prediction. This represents the incremental risk introduced by the status of video trending topics; S6-7, Output the risk assessment result package, which includes at least: MMSI, ship-to-ship risk level, ship-to-bridge risk level, bridge span conflict risk level, priority score, and recommended control action.

7. A water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, The traffic control decision generation in step S7 includes the following steps: S7-1, Based on the risk assessment result package output in step S6, determine the type of traffic control strategy to be executed and construct a control decision set. ; S7-2, the traffic control strategy includes at least one or more of the following: deceleration, maintaining course, yielding in advance, entering the waiting area, prohibiting entry into the designated bridge opening, switching to the backup bridge opening, and restoring traffic; different strategies are triggered by the corresponding risk level range and priority range. S7-3, When multiple vessels are competing for the same bridge span, priority scores will be used as the basis for determining the appropriate bridge span. Determine the single priority passage object based on the occupancy status of the bridge opening. : And designate the remaining vessels as waiting targets or avoidance targets; S7-4: When there is a risk of overlap between the predicted trajectory of a vessel and the bridge pier protection zone or restricted area, priority should be given to generating control decisions such as emergency deceleration, course correction, or suspension of entry into the bridge area; among which, when the comprehensive risk value... satisfy Emergency control strategies are triggered at certain times, among which, This is the emergency control threshold; S7-5, output the control decision in the form of a structured control message, which includes at least: target vessel identity, risk type, target bridge opening, control level, control action, effective time window and alert priority, for VHF broadcast in step S8.

8. A water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, The VHF graded alerts and closed-loop feedback mentioned in step S8 include the following steps: S8-1 Select the broadcast alert or targeted alert mode according to the control level of the control message; broadcast alerts are used for low-level risks, and targeted alerts based on the vessel name or MMSI are used for high-level risks. S8-2, based on different risk types, call the corresponding voice templates to automatically generate VHF broadcast content; the broadcast content includes at least: target vessel identity, risk description, control requirements, suggested actions and execution time limits; S8-3, the tiered alerts include at least one or more of the following: situation alert level, early warning alert level, control intervention level, and emergency warning level; and are based on risk values. The alert level will be automatically selected based on the current location. S8-4, After the VHF alert is issued, continuously monitor the changes in the target vessel's AIS trajectory and the status changes in video hotspot areas to assess whether the vessel has executed the aforementioned control requirements; execution deviation. Defined as: in, and The first The actual speed and course of the vessel after receiving the VHF alert. and These are respectively controlling the target's speed and controlling the target's heading. The deviation between the target and the specified waiting area or the specified bridge opening passage path; to These are the weighting coefficients; S8-5: When the risk decreases or the vessel's actions conform to the control strategy after the alert, it is recorded as effective intervention; when there is still no response after the alert or the risk continues to rise, the alarm level is automatically escalated, triggering a second alert or manual intervention; among these, when continuous... Satisfaction within each evaluation period When this occurs, it is determined that the control requirements have not been effectively implemented; among them, Indicates the first Control execution deviation values ​​for each vessel; This indicates the preset control execution deviation threshold.

9. A water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, The multi-source data fusion verification in step S9 includes the following steps: S9-1 integrates AIS dynamic data, video structured event data, and VHF alert logs to construct a control policy execution verification link and establish multi-source association records according to event number; S9-2 performs consistency verification on changes in ship speed, course, bridge approach status, and hotspot area occupancy status before and after the issuance of control commands; consistency score. Represented as: in, As a speed consistency indicator, For heading consistency indicators. This serves as an indicator of regional consistency. For VHF execution log consistency metrics; to These are the weighting coefficients; S9-3, when AIS shows that a vessel has not slowed down as instructed, has not waited as instructed, or continues to approach the collision bridge opening, it is identified as an abnormal response vessel; when the consistency score meets the requirements... When this occurs, an abnormality in control execution is determined; among which, Indicates the first Consistency score of control execution of the vessels; This indicates the preset consistency judgment threshold; S9-4, triggering secondary intervention for abnormal response vessels, the secondary intervention includes at least raising the VHF alert level, extending the waiting time window, transferring to manual duty confirmation or linking with law enforcement modules; S9-5, Output multi-source verification results, which include at least: target vessel identity, original control strategy, execution status, response delay, anomaly markers, and secondary intervention records, as input for the archiving analysis in step S10.

10. A water traffic control method based on video hotspot area identification, AIS trajectory prediction, and VHF alerts according to claim 1, characterized in that, The control effect archiving analysis described in step S10 includes the following steps: S10-1. Archive a complete traffic control process by event number. The archived content shall include at least: original AIS trajectory, video hotspot events, risk assessment results, control decisions, VHF broadcast logs, execution verification results, and final handling results. S10-2, based on indicators such as event type, risk level, intervention duration, response duration, execution success rate, and false alarm rate, statistical analysis is conducted on the control effectiveness; among which, the execution success rate... Represented as: in, Indicates the number of events that successfully executed the control request; This represents the total number of all traffic control events within the statistical period; False alarm rate Represented as: in, Indicates the false alarm rate; Indicates the total number of warning events; S10-3 Update hotspot area threshold parameters, risk assessment weight parameters, bridge hole priority parameters, and VHF voice template priority based on the archived results; parameter updates can be performed using sliding window statistical updates or incremental correction methods. S10-4 utilizes the archived analysis results for model iterative optimization and dynamic updating of control rules to improve the accuracy and closed-loop efficiency of subsequent water traffic control; among which, the updated parameter set... Represented as in, Indicates the first The parameter set after the update; indicating the first... The original parameter set used in the next update.

Images