A method and system for waterway traffic capacity prediction and ship navigation guidance

Abstract

The application provides a waterway traffic capacity prediction and ship navigation guidance method and system, and relates to the technical field of intelligent transportation and water traffic safety management. The guidance method specifically comprises the following steps: collecting on-site images of an accident water area and calculating a three-dimensional posture of an accident ship; generating a time-varying residual waterway width function representing the available width of a future waterway; constructing a space-time state matrix of a to-be-passing ship; constructing a space geometric constraint based on the time-varying residual waterway width function, and constructing a wave effect constraint for determining the maximum critical navigation speed of each to-be-passing ship based on on-site observation of the wave generated by the passing ship; and based on the space-time state matrix, the space geometric constraint and the wave effect constraint, performing global optimization solving of multi-objective navigation scheduling to generate and issue a navigation timetable containing the optimal passing navigation speed of each to-be-passing ship.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation and water traffic safety management technology, and in particular to a method and system for predicting waterway capacity and guiding ship navigation. Background Technology

[0002] Important international shipping lanes (such as the Suez Canal and the Persian Gulf deep-water channel) are the lifelines of global energy and trade. However, due to limitations imposed by natural conditions, complex meteorological and hydrological environments, and increasing vessel flow density, incidents of super-large vessels running aground (such as the 'Changci' incident), striking reefs, or silting up reducing navigable width occur frequently. When such emergencies or planned projects occupy part of the channel resources, the originally open navigable waters can instantly become navigation bottlenecks, easily causing congestion and delays for upstream and downstream vessels. If not properly managed, this can even lead to secondary accidents caused by congestion.

[0003] Currently, maritime authorities or waterway management departments primarily rely on Vessel Traffic Management Systems (VTS) combined with on-site patrol boats for manual traffic control when managing navigation in areas affected by accidents or construction. However, this traditional static control model has significant technical limitations in practical application, making it difficult to maximize waterway traffic efficiency while ensuring safety.

[0004] First, existing control measures often overlook the dynamic and time-varying characteristics of the hydrological environment. Unlike the fixed width of land routes, the actual navigable width and depth of waterways change dynamically over time (e.g., astronomical tides, storm surges). When water levels are high, some shallow areas around the accident vessel may temporarily be navigable, or the remaining effective width of the channel may be sufficient to support two-way passage for vessels of a certain draft. However, existing technologies typically use the worst-case scenario principle to delineate fixed no-navigation boundaries, resulting in the underutilization of valuable waterway resources during periods of favorable hydrological conditions, leading to unnecessary waste of transport capacity and queuing.

[0005] Secondly, existing technologies for setting constraints on ship behavior are singular and "one-size-fits-all." The most common behavioral constraint is issuing a uniform low-speed navigation instruction, such as a speed limit of 5 knots for all ships passing through. This singular constraint does not consider the individual differences between different ships and their complex interactions with the environment. On the one hand, for smaller ships with weak wave-making effects, excessively low speed limits unnecessarily sacrifice their passage efficiency; on the other hand, for large, heavily loaded ships, even if they comply with the uniform speed limit, the waves they generate during navigation may still cause impacts beyond the capacity of a grounded ship in a fragile equilibrium state, thus triggering significant safety hazards such as secondary slippage or capsizing. Existing technologies lack a means to quantify the physical impact online and set personalized, differentiated speed limits based on the individual characteristics of passing ships.

[0006] Furthermore, at the ship scheduling level, existing queuing and release mechanisms mostly follow a simple first-come, first-served logic, lacking intelligent planning based on global optimization. At both ends of restricted waterways, a large number of ships often accumulate disorderly, which not only increases fuel consumption and carbon emissions but also increases the risk of rear-end collisions or minor collisions due to the frequent starting, braking, and positioning of ships in narrow waters.

[0007] In summary, how to break through the traditional static and singular management and control thinking, establish a technical solution that can perceive changes in hydrological and navigation environment in real time, accurately quantify the hydrodynamic impact of navigation vessels on accident sites, and thereby achieve dynamic and refined navigation guidance, is a key issue that urgently needs to be addressed in the field of water transportation engineering. Summary of the Invention

[0008] This invention provides a method for predicting waterway capacity and guiding ship navigation, the method comprising: Collect on-site images of the accident area and calculate the three-dimensional attitude of the wrecked vessel; Generate a time-varying remaining channel width function characterizing the future available channel width; construct the spatiotemporal state matrix of the vessels to be navigated; construct spatial geometric constraints based on the time-varying remaining channel width function; and construct wave-making effect constraints to determine the maximum critical speed of each vessel to be navigated based on on-site observations of the waves generated by the passing vessels. Based on the spatiotemporal state matrix, spatial geometric constraints, and wave-making effect constraints, a global optimization solution for multi-objective navigation scheduling is performed to generate and distribute a navigation timetable containing the optimal passage speed for each vessel waiting to pass.

[0009] The specific steps for calculating the three-dimensional attitude of the wrecked vessel include: extracting the longitudinal and transverse structural lines of the vessel's hull from the on-site image, and calculating the intersection points of the longitudinal structural lines on the image plane. Intersection with the horizontal structural line Based on camera intrinsic parameter matrix The unit direction vectors of the longitudinal and transverse structural lines in the camera coordinate system can be calculated using the following formula. and : ; ; according to and Construct the rotation matrix of the wrecked ship relative to the camera coordinate system. And combined with the UAV's IMU attitude matrix ,Will Converting to the world coordinate system to obtain the roll angle of the wrecked ship Pitch angle and heading angle .

[0010] The steps for generating the time-varying remaining channel width function include calculating the future relative water level fluctuation function. ; and utilize Predict any future moment using the following formula Predicted roll angle of the accident vessel : in, The initial roll angle of the vessel involved in the accident. According to The stranding stability coefficient is set according to the grade. The preset draft of the vessel involved in the accident.

[0011] Generate time-varying residual channel width function The steps also include: calculating any future time. Effective obstacle width : in, The width of the ship involved in the accident is the pixel width on the image plane. This is the physical scale conversion factor. The safety buffer threshold; Prediction function of future total width of the waterway minus The time-varying residual channel width function is obtained. .

[0012] The steps for constructing wave-making effect constraints include setting a critical wave height threshold for the ship involved in the accident. The calculation formula is as follows: in, For safety reasons, This is the actual freeboard height. This represents the initial roll angle of the vessel involved in the accident.

[0013] The steps for constructing wave-making effect constraints also include online calibration of the speed-wave height model for passing vessels: when a vessel acting as a pathfinder... At actual speed When passing through the accident area, the physical wave height generated is measured using on-site imagery. Based on data points The coefficients of the speed-wave height relationship model applicable to the exploratory vessel are calibrated using the following formula. : For any subsequent vessel awaiting passage According to its captain Habune Hiro Compared to scout ships Captain Habune Hiro The difference is expressed by the following formula. Make corrections to obtain a result suitable for coefficient : .

[0014] The specific steps for determining the maximum critical speed of each vessel to be passed are as follows: based on any vessel to be passed... The maximum wave height generated must not exceed the critical wave height threshold. The constraints, namely Inverse solution applicable to Maximum critical speed : in, for The speed at which ships can pass.

[0015] The steps for solving the global optimization problem of multi-objective air traffic scheduling are to minimize the following multi-objective optimization function. For the goal: in, The total waiting time cost for all vessels waiting to pass. The total passage risk cost for all vessels waiting to pass. This is the risk aversion weighting coefficient; and The calculation formula is: in, For the queue of ships waiting to pass, For ships The moment of release, For ships The arrival time, For ships The speed at which ships can pass, For ships The maximum critical speed.

[0016] The specific steps for generating the time-varying remaining channel width function include: periodically observing the accident area and simultaneously recording the sequence of the total physical channel width. and water area pixel sequence Linear regression analysis was performed on paired data consisting of two sequences to solve the model. The average slope coefficient of the riverbank and tidal flats in China ; Linear regression analysis was performed on the total physical width sequence of the waterway to determine the average rate of change of the total waterway width. ; based on The width value obtained from the latest observation Construct a prediction function for the total width of the future waterway. and combined The future relative water level fluctuation function is calculated, and then the time-varying remaining channel width function is generated.

[0017] The present invention also provides a waterway capacity prediction and ship navigation guidance system, the system comprising: Acquisition module: Acquires on-site images of the accident area and calculates the three-dimensional attitude of the wrecked vessel; Constraint Module: Generates a time-varying remaining channel width function characterizing the future available channel width; constructs the spatiotemporal state matrix of vessels to be navigated; constructs spatial geometric constraints based on the time-varying remaining channel width function; and constructs wave-making effect constraints to determine the maximum critical speed of each vessel to be navigated based on on-site observations of the waves generated by the navigating vessels. Optimization module: Based on the spatiotemporal state matrix, spatial geometric constraints and wave-making effect constraints, it performs global optimization of multi-objective navigation scheduling to generate and issue a navigation timetable containing the optimal passage speed of each vessel waiting to pass.

[0018] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned method for predicting waterway capacity and guiding ship navigation.

[0019] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for predicting waterway capacity and guiding ship navigation.

[0020] The waterway capacity prediction and ship navigation guidance method and system provided by this invention have achieved significant technological progress and beneficial effects in handling emergency scenarios where waterways are restricted, compared with existing technologies. This invention breaks through the limitations of traditional static, single-control modes and establishes a dynamic closed-loop decision-making system based on real-time on-site perception, greatly improving the safety and efficiency of navigation management.

[0021] Unlike static management relying on outdated nautical charts or fixed navigation marks, this invention can accurately capture the entire scene at the moment an accident occurs, including the precise three-dimensional attitude of the vessel involved and the actual channel boundaries. More importantly, through observation of the waters, this invention can establish a data-driven, time-varying evolution model, thereby predicting the dynamic changes in channel width over the next few hours. This capability allows navigation decisions to move beyond a conservative response based on the worst-case scenario, enabling proactive identification and utilization of valuable navigation windows created by rising water levels.

[0022] Secondly, this invention constructs a dynamic, precise, and entirely observation-based personalized navigation constraint set, changing the existing technology's one-size-fits-all approach to setting behavioral constraints. The wave-making effect constraint proposed in this invention, through online calibration—that is, real-time observation of the actual waves generated by the first or a few safely passing vessels—establishes a speed-wave height response model that conforms to the characteristics of the current waterway. Based on this, the system can back-calculate a personalized maximum critical speed that must be adhered to for each subsequent vessel of different size and type to ensure no secondary impact on the malfunctioning vessel. This transforms a vague safety concern into a clear, quantifiable, and enforceable dynamic safety red line, achieving refined and differentiated control over navigation safety.

[0023] Finally, based on the aforementioned dynamic and precise constraint set, this invention achieves a high degree of unity between navigation efficiency and safety through multi-objective global optimization, solving the congestion and risks caused by the simple queuing logic of existing technologies. This invention no longer passively handles ships, but proactively plans an optimal scheduling scheme that minimizes the total waiting time and overall passage risk for all ships waiting to pass. The resulting timetable provides each ship with remote speed control instructions from its current location to the accident area, precise release time windows, and cross-sectional control speeds when passing through the core area, achieving virtual queuing and on-time arrival. This not only significantly reduces ship congestion and anchoring in narrow waterways, lowering fuel consumption and collision risks, but also transforms the ship traffic flow in the entire accident area from disordered waiting to orderly, efficient, and safe collaborative passage, significantly improving the channel capacity and overall transportation system operational efficiency in emergency scenarios. Attached Figure Description

[0024] Figure 1 This is a flowchart of the waterway capacity prediction and ship navigation guidance of the present invention. Detailed Implementation

[0025] This embodiment provides a basic hardware architecture and application scenario for a waterway capacity prediction and ship navigation guidance system. The system is configured in international maritime chokepoints (such as straits and canals) or large deep-water port access channels with navigation restrictions. The specific application scenario is in accident-prone waters where navigation width is reduced due to ship grounding, reef collisions, or damage to some waterway facilities. The system's physical architecture comprises four core components: a shore-based computing center subsystem, an aerial three-dimensional perception subsystem, a multi-source data access interface, and a ship-side interactive terminal. These components interact via a high-bandwidth, low-latency wireless communication network, collectively forming the physical platform for executing subsequent environmental modeling and scheduling commands.

[0026] The shore-based computing center subsystem, serving as the core of data processing and decision-making, is deployed in the computer room of the maritime management agency or emergency command center, preferably using high-performance industrial-grade graphics workstations or server clusters. Considering the need for 3D digital twin modeling and real-time hydrodynamic numerical simulation, the computing center is configured with a multi-core, high-frequency architecture for its central processing unit, with a clock speed of at least 3.0 GHz, and a minimum of 64 GB of error-correcting memory to support large-scale matrix operations and parallel processing tasks. To meet the requirements for visualization rendering of time-varying channel evolution models and acceleration of deep learning algorithms, the computing center integrates a high-performance independent graphics processing unit, preferably with at least 12 GB of dedicated video memory. For storage, the system is equipped with high-speed solid-state storage based on a RAID disk array to store the real-time texture data collected from the water area.

[0027] The aerial stereo perception subsystem mainly consists of an industrial-grade multi-rotor UAV platform and its payload, used for non-contact physical data acquisition of the accident area. The UAV platform uses a hexacopter or octacopter aircraft with real-time differential positioning (RTK) capabilities to ensure centimeter-level hovering positioning accuracy under complex electromagnetic environments and water surface wind interference. Specifically, the horizontal positioning accuracy is preferably between 1 cm and 3 cm, and the vertical accuracy is preferably between 2 cm and 5 cm. The aircraft has wind resistance capabilities exceeding level 5 and a payload endurance of at least 30 minutes, ensuring coverage of the typical accident area. The payload carried by the UAV is a five-lens or single-lens high-resolution oblique photography camera. The image sensor target surface size is preferably full-frame or APS-C, with a total effective pixel count of at least 40 megapixels. The lens supports three-axis gimbal stabilization control, enabling the acquisition of water surface and vessel attitude information at vertical and different tilt angles. In addition, this subsystem includes an automated hangar or mobile control station for UAV take-off and landing, maintaining real-time connection with the shore-based computing center via an image transmission link.

[0028] The multi-source data access interface serves as the connection channel between the system and existing maritime traffic management infrastructure. The system connects to the Vessel Traffic Management System (VTS) server and Automatic Identification System (AIS) base station data streams via a hardware firewall and dedicated network. This interface is equipped with a protocol parsing module capable of resolving NMEA-0183 or AIVDM format data packets, used to extract in real-time the MMSI code, latitude and longitude coordinates, heading, speed, and static declaration information (breadth, draft, cargo type) of passing vessels. This is crucial for obtaining accurate hydrological boundary conditions.

[0029] The shipboard interactive terminal is the execution end of the system for issuing guidance instructions to vessels awaiting passage. This terminal can be an extension module of the electronic chart display and information system integrated into the ship's bridge, or an industrial-grade ruggedized tablet or smart mobile terminal held by the crew. Equipped with a BeiDou / GPS dual-mode positioning module and a 4G / 5G / satellite communication module, this terminal can receive scheduling instruction packages from the shore-based computing center, including speed vectors, time windows, and trajectory planning. Through a graphical human-computer interaction interface, it visually overlays suggested speed, clearance countdown, and dynamic no-navigation zones onto the electronic nautical chart background for the navigator's reference and execution.

[0030] Next, this embodiment details the specific implementation method for constructing the basic digital twin scenario of the accident waters. The construction of the basic digital twin scenario specifically includes: S1.1 Construct a local projected coordinate system based on the UAV pose.

[0031] In the absence of external geographical references, the system establishes a local projection measurement system based on the hovering position of the UAV at the moment of the accident.

[0032] Control the drone to perform fixed-point hovering and filming above the accident area to obtain high-resolution orthophotos including the accident vessel and the shorelines on both sides. At this time, the flight parameters output by the UAV's onboard sensors are recorded simultaneously: air pressure altitude relative to the takeoff point. The gimbal pitch angle provided by the inertial measurement unit (IMU) of the fuselage. and yaw angle .

[0033] In order to obtain image pixel coordinates Converted to physical distance, based on camera sensor pixel size Lens focal length Calculate the physical scale conversion factor : Physical scale conversion factor Each pixel in the image represents the actual length of the water surface (meters per pixel).

[0034] S1.2 Channel boundary extraction and flow direction vector based on semantic segmentation.

[0035] To determine where the water comes from, where it flows to, and the width of the waterway, images were used. Perform semantic parsing.

[0036] Input Image The pre-trained DeepLabV3+ semantic segmentation network outputs a pixel-level classification mask. The mask contains three types of labels: . , , These labels represent the division into water areas, shorelines, and ship hulls, respectively.

[0037] extract and The system uses the set of boundary pixels to fit the physical boundary curves of the left and right banks. Considering the irregular characteristics of unstructured riverbanks, the system employs the RANSAC algorithm to fit the local centerline of the main channel.

[0038] Based on the centerline of the main channel, the centerline direction vector in the image coordinate system is obtained as follows: Combined with the drone's heading angle Calculate the main channel flow angle (Relative to due north): The effective obstruction width of a vessel in a waterway is not its own width or length, but rather the projected width of its hull profile in a direction perpendicular to the waterway flow direction. When calculating the static remaining waterway width, the main channel flow angle must be used as a reference for projection calculation. Ignoring this angle and simply calculating the width in a coordinate system will lead to serious misjudgments regarding air traffic. This main channel flow angle is a prerequisite for ensuring the accuracy of geometric measurements, especially in typical scenarios where the waterway curves or the vessel runs aground at an angle.

[0039] Next, perpendicular to Calculate the Euclidean distance between the left and right boundary pixels in the direction of the boundary. (Number of pixels) is used to obtain the total physical width of the channel at the current water level. : S1.3 Calculation of the three-dimensional attitude of the accident vessel.

[0040] This application utilizes the inherent parallel-line geometry of ships to calculate their spatial attitude. The deck structure of ships involved in accidents (such as bulk carriers and container ships) exhibits obvious geometric patterns: hatch coamings, container edges, and hulls are either parallel or perpendicular to each other.

[0041] The basis for attitude calculation in this application is that parallel straight lines in three-dimensional space, after perspective projection, will inevitably intersect at a single point on the two-dimensional image plane, which this application refers to as the structural line intersection point. As a typical rigid body, the hull structure of a ship involved in an accident (such as the longitudinal stiffeners of the deck and the transverse frame) naturally contains two (or more) sets of mutually orthogonal parallel lines. By locating the intersection point of these parallel line clusters in the image, the orientation of these structural lines in three-dimensional space can be calculated in reverse, thereby determining the complete attitude of the hull.

[0042] In the segmented Within the region, edges are extracted using the Canny operator, and a set of line segments is extracted using the Hough transform.

[0043] The straight line segments are clustered into two main directions: one group corresponds to the longitudinal direction of the ship's hull (bow and stern direction), and the extensions of these lines on the image plane intersect at the longitudinal intersection point. The other set corresponds to the transverse direction of the hull (port and starboard), converging at the transverse intersection point. .

[0044] The intersection point is directly related to the camera's rotation matrix relative to the object. This is based on the camera's intrinsic parameter matrix. A system of equations was constructed to solve for the rotation matrix of the wrecked vessel relative to the camera coordinate system. .

[0045] Rotation matrix The specific calculations include: a) In the segmented Within the region, edges are extracted using the Canny operator, and a set of line segments is extracted using the Hough transform, corresponding to the longitudinal structural lines of the hull. and horizontal structural lines Calculate the intersection points of the longitudinal structural lines Intersection with horizontal structural lines The homogeneous coordinates.

[0046] b) A direction vector in three-dimensional space in the camera coordinate system. The coordinates of its intersection point in the image The following projection relationship exists between them: in, For the camera intrinsic parameter matrix, This is the scale factor. By inverting the matrix, the unit direction vector in the camera coordinate system is calculated from the image intersection points: in and These represent the three-dimensional directions of the longitudinal and transverse axes of the wrecked vessel in the camera coordinate system, respectively.

[0047] c) Due to the naturally orthogonal structure of the hull. and Theoretically, they should be perpendicular to each other. Considering computational errors, the system uses Schmidt orthogonalization to correct for these errors, resulting in a standard orthogonal basis. The normal direction of the ship's deck plane (i.e., the Z-axis direction of the ship) can be obtained through the cross product of vectors: .

[0048] These three mutually orthogonal unit vectors Together, they constitute the rotation matrix of the wrecked vessel relative to the camera coordinate system. : This rotation matrix describes the tilt state of the wrecked vessel. Combined with the IMU attitude matrix from the UAV... This can be converted to the world coordinate system and decomposed into roll angle, pitch angle, and yaw angle.

[0049] Using the IMU attitude matrix of the UAV Transform the relative rotation matrix to the world coordinate system to obtain the absolute attitude Euler angles of the wrecked vessel: roll angle. (Tilt) and Pitch Angle (Tip) and heading angle .

[0050] Among them, roll angle The calculation utilizes the offset of the horizontal intersection point in the vertical direction, and its approximate analytical expression is: ; in Principal point of the image, The equivalent focal length is longitudinal. When a ship capsizes, the originally horizontal transverse structural lines will exhibit significant perspective convergence in the image, directly reflecting the degree of danger. Specifically, when the ship is righted, its transverse structural lines (such as deck beams and container bottom edges) are parallel to the water surface, exhibiting almost no perspective convergence in orthophotos. The intersection of these transverse structural lines will be located at infinity, meaning the Y-axis coordinate in the image coordinate system approaches infinity. However, once the ship capsizes due to grounding or collision, generating a roll angle, these originally horizontal structural lines will tilt, causing their projection lines on the two-dimensional image to converge at a finite-distance intersection point. The larger the capsizing angle, the closer this intersection point is to the principal point of the image in the vertical direction. Therefore, this application calculates the attitude by calculating the intersection point position, essentially directly linking the ship's degree of danger, i.e., the tilt angle, to a measurable geometric feature in the image (intersection coordinates), achieving non-contact, high-precision quantification of the accident state.

[0051] S1.4 Calculation of remaining channel width.

[0052] At the accident site, the ship was not a rectangle on the nautical chart, but a tilted, three-dimensional obstacle that could potentially sink partially.

[0053] Calculate the minimum bounding rectangle of the wrecked vessel on the image plane and obtain its pixel width. .

[0054] Considering that ship roll causes the underwater hull (such as the keel and propeller) to protrude significantly to one side, and this part cannot be directly captured by the camera, visual width compensation is required. Define the effective obstacle width of the accident vessel. for: in: The projected width of the visible portion above the water surface; This is an estimated value for the lateral extension of the underwater portion due to the ship's roll. Draft depth; A safety buffer threshold is set to cope with water flow turbulence.

[0055] Current moment Static remaining channel width : like The system determines that the waterway is completely blocked; if This value will serve as the physical constraint boundary for calculating navigation capacity.

[0056] In this application, considering the urgency and unstructured nature of accident scenes, directly measuring the real-time draft of the vessel involved in the accident is often not feasible. Therefore, this invention employs a method combining ship type classification and preset engineering safety values ​​to obtain the draft. For bulk carriers or general cargo ships, typically 150-190 meters in length and with a deadweight of 20,000-40,000 tons, the preset... The design draft is 10.5 meters. These vessels are the mainstay of international regional maritime transport, and their design draft is typically limited by the depth of the entrance channels in small and medium-sized ports. Under full load, their design draft is usually between 9.5 and 10.5 meters. This application selects 10.5 meters as a conservative typical value, which covers the heavy-load draft of the vast majority of such vessels operating in secondary channels, ensuring sufficient safety margin when calculating the increase in underwater transverse beam due to capsizing. Panamax bulk carriers or Aframax tankers, typically 200-250 meters in length, with a deadweight tonnage of 60,000-120,000 tons, are pre-designed... The design draft is 14.0 meters. These vessels operate extensively on major transoceanic routes and busy straits worldwide, forming the core capacity for international bulk commodity transportation. The full-load design draft of this class of vessels generally ranges from 12 to 14.5 meters. Using 14.0 meters as a preset value, for ultra-large container ships, typically 350-400 meters in length, with a capacity exceeding 15,000 TEU, the preset... The draft is 16.5 meters. These giant vessels are key hubs connecting Asia and Europe, and trans-Pacific shipping routes, and are also the main vessels navigating vital waterways such as the Persian Gulf and the Suez Canal. Although the actual draft of container ships varies depending on the container load and stowage, their enormous depth and bulbous bow structure pose a serious threat to shipping lanes should an accident (such as listing) occur.

[0057] The remaining channel width provides the system While real-time channel conditions are important, in actual emergency scenarios, the aquatic environment, especially water level and current velocity, is constantly changing. Furthermore, the aquatic environment exhibits significant time-varying characteristics, and the stability of stranded vessels is not constant. If dispatching decisions rely solely on… Static data at any given moment might miss valuable navigation windows created by rising water levels, or fail to anticipate the risks of narrowing channels and instability of vessels due to falling water levels. Therefore, this step constructs a dynamic prediction model for a future period, evolving static channel snapshots into dynamic time functions, providing time-dimensional constraints for subsequent vessel scheduling.

[0058] S2.1 Acquire temporal images and perform spatial reference alignment.

[0059] To capture changing trends in the aquatic environment, the system controls the drone to perform fixed-point, periodic patrol missions. Specifically, after completing... After initial modeling of the time intervals, the drone operates at preset time intervals. The system repeatedly hovers at roughly the same position over the accident area, collecting a series of orthophotos with timestamps. This results in a time-series image set. ,in .here, This represents the i-th data collection time, and k is the number of data collections.

[0060] Since the hovering position of a drone will not be exactly the same each time, spatial reference alignment is necessary for accurate comparison of images from different times. (Select...) Images of moments As a reference frame. For each subsequent frame of imagery. The system utilizes the ORB feature matching algorithm to extract and match feature points in static regions of the image. Based on the matched feature point pairs, a homography transformation matrix is ​​calculated. This matrix can transform images Mapping the coordinates of any pixel in the reference frame to the base frame Under this coordinate system, all subsequent visual measurements are ensured to be performed within a unified spatial coordinate system, eliminating measurement errors caused by drone position drift.

[0061] S2.2 Extraction of water width change rate based on visual observation.

[0062] Each frame of the image after alignment By applying the S1.2 semantic segmentation method for channel boundary extraction, a series of timestamped total channel width measurements are obtained. : .here This is the measured value of the total width of the channel at time i.

[0063] For each frame of image The semantic segmentation mask is analyzed directly, and the segments are labeled as The number of pixels representing the water area is counted. This count represents the total pixel area occupied by the water area on the image plane at the current moment. The resulting set of total pixel areas representing the water area is then obtained. .here Let be the total pixel area occupied by the water area on the image plane at the i-th time.

[0064] Therefore, the present invention obtains a comprising A two-dimensional paired dataset of data points: .

[0065] This is a discrete dataset showing the change in channel width over time. The core objective of this invention is not to provide precise hydrological forecasts for 12 or 24 hours, but rather to provide a reliable, monotonic trend for ship scheduling over the next 2 to 6 hours. In a complete tidal cycle, approximately 6 hours from high to low tide, if the total duration of the observation and prediction windows is much shorter than half a cycle, then the water level change curve segment within that time period can be locally approximated by a linear function. That is, a complete hydrological cycle is mathematically equivalent to a sine function, but a first-order Taylor series expansion of a sine function at a certain point results in a linear approximation in its neighborhood. This invention utilizes this principle, using a linear model to fit the most significant current trend—whether it is rising or falling, and its approximate rate.

[0066] Therefore, this invention employs the least squares method to perform linear regression analysis on the dataset in order to solve for the average rate of change of the total channel width. The goal is to find the optimal one. and initial width Such that the following residual sum of squares Minimum: By solving and ,get The solution. This parameter It objectively represents the actual rate of rise and fall of water level at the accident site during the observation period. This indicates that the tide is rising and the waterway is widening.

[0067] S2.3 Construction of the future total width of the waterway and the relative water level fluctuation function.

[0068] Based on the calculated rate of change Building a future-oriented Hourly channel physical total width prediction function : in, It is the width value obtained from the latest observation, which serves as the baseline for prediction.

[0069] Furthermore, in order to drive the hull attitude drift model in S2.4, the system requires a relative water level fluctuation function. Establish an approximate relationship between channel width changes and water level changes, and predict future relative water level fluctuation functions. It can be derived from the width variation function: Average slope of riverbank This is a geometric coefficient representing the linear ratio of the total physical width of the channel to the change in the pixel area of ​​the water area. Within the local area of ​​the accident waters, how many meters will the total physical width of the channel increase for every unit increase in the pixel area of ​​the water area? The system uses a standard linear least squares regression method to fit the above two-dimensional paired dataset, solving for the slope parameter of the following linear model. : in, This is the regression intercept term.

[0070] S2.4 Prediction of attitude drift risk of the accident vessel and generation of time-varying remaining channel width function.

[0071] Using relative water level fluctuation function To predict the attitude drift risk of the ship involved in the accident. At any future moment... Predicted roll angle The computational model remains unchanged: Generate the final time-varying remaining channel width function At any point in the future Effective obstacle width for: The predicted total width of the future shipping channel Subtract the predicted effective obstacle width ,get: This function This constitutes a three-dimensional spatial variation of the future general aviation environment over time.

[0072] After generating the time-varying remaining channel width function Subsequently, this invention constructs a navigation constraint set based on field observation data.

[0073] Construct the state matrix of the vessel to be passed: The purpose of constructing the state matrix is ​​to transform the disordered and dynamic ship traffic flow in the waterway into a queue of structured data objects that the system can process.

[0074] S3.1 Establishment of the main queue for air traffic demand.

[0075] Utilizing multi-source data access interfaces, the upstream and downstream areas of the accident site are monitored in real time. When the Automatic Identification System (AIS) data indicates that a vessel is about to enter the accident area, it is automatically added to the candidate list of vessels waiting to pass. A unique identifier is assigned to the vessel, and it is officially added to the upstream or downstream waiting queue. or middle.

[0076] S3.2 Extraction of attribute vectors.

[0077] For each ship that enters the queue The system parses its AIS messages and, in conjunction with its built-in shipbuilding engineering knowledge base, constructs a standardized multidimensional spatiotemporal state vector for it. .

[0078] in, For the captain, For the width of the boat, The draft depth value is preferentially obtained from the AIS static data message. If this field is missing or unreliable, the system will assign a conservative draft depth value based on the ship type classification and preset engineering safety values ​​method in S1.4. For maximum speed, This is the initial position. To estimate displacement: This application employs an engineering estimation method based on the ship's main dimensions, according to... and Query the typical block coefficient for this ship type from the built-in database. Then calculate the estimated drainage volume. : in This is the density of water.

[0079] Calculate the navigation constraint set: S4.1 Spatial geometric constraints are constructed.

[0080] Single vessel passage constraint: For any vessel At any moment The condition for allowing passage in one direction is: in, To ensure safe lateral clearance for one-way traffic, the preferred option is... .

[0081] Two-ship encounter constraints: For the plan at time... The two vessels that met in the accident section and The conditions for determining whether a meeting can be safe are: ,in, To ensure sufficient redundancy for safe lateral clearance during two-way passing, the preferred option is... .

[0082] S4.2 Wave-making effect constraint based on UAV observation.

[0083] By observing the waves generated by the first (or first few) ships that pass safely in real time using drones, this invention calibrates a speed-wave height response model online, and uses this model to constrain the passing speed of subsequent ships.

[0084] S4.2.1 Critical wave height threshold setting for accident vessels.

[0085] For a stranded vessel, its stability is severely threatened by waves. It is necessary to determine the maximum wave height it can withstand, i.e., the critical wave height threshold. The threshold is set based on the calculated attitude of the ship involved in the accident: in, This is the actual freeboard height. It is a safety factor, ranging from 0 to 1, set by on-site commanders based on factors such as the importance of the ship and cargo and environmental sensitivity.

[0086] The maximum wave height that the vessel can withstand cannot exceed the critical height at which the deck would be wetted or even cause water to enter from the upper deck under its current attitude. This is an engineering threshold that can be directly calculated based on geometric relationships and safety considerations.

[0087] Regarding the actual freeboard height, two key horizontal lines were simultaneously identified through image processing: one is the boundary line between the hull and the water surface, i.e., the instantaneous waterline; the other is the main deck line of the hull.

[0088] The vertical pixel distance between the main deck line and the instantaneous waterline in the midship region of the ship is measured and recorded as the freeboard pixel height. Using known physical scale conversion factors Convert pixel distance to actual freeboard height . S4.2.2 Online calibration of the transit vessel speed-wave height model.

[0089] When the first (or first few) vessels are waiting to pass (Referred to as a scout boat) at a safe speed confirmed by the on-site commander. As the drone passes through the waters of the accident, it will hover continuously and film the bow area of ​​the vessel.

[0090] The system performs image processing on the video stream captured by the drone, and uses edge detection and Hough transform to identify passing vessels in real time. The crests of the Kelvin wave system generated on both sides of the bow.

[0091] This involves measuring the change in brightness or color gradient of the wave crest relative to the surrounding calm water surface. Under specific lighting conditions, wave crests appear brighter due to reflection, while wave troughs appear darker. This difference in visual intensity is analyzed. It is possible to establish its relationship with the actual wave height. The correspondence.

[0092] To complete the calibration, the actual speed at the time of passage by the scout vessel is required to be reported. Preferably, the physical wave height is directly measured using the ship's hull as a scale to identify passing vessels. The waterline on the side of the ship was measured, and the pixel height of the rising wave crests on its hull was measured. Combined with physical scale conversion coefficient To obtain the physical wave height .

[0093] According to the theory of ship wave-making, the relationship between ship speed and the maximum wave height is approximately quadratic. This is achieved using data points from the scout vessel. The coefficients in this relational model are calibrated. : This coefficient is related to the ship type, especially the bow's width. For other vessels awaiting passage... The system can be based on its length Habune Hiro Compared to scout ships The difference is used to correct the coefficient, resulting in a value suitable for... coefficient : The wider the hull (the greater the width-to-height ratio), the higher the waves it generates at the same speed.

[0094] Constraint establishment and safe speed solution: To ensure safety, any subsequent passing vessels... Maximum wave height generated None of them can exceed the critical wave height threshold of the accident vessel. .Right now: Therefore, the solution applicable to ships can be derived from this. Maximum critical speed : This embodiment elaborates on the final decision-making and instruction generation steps of the application process. After a complete set of navigation constraints, including time-varying spatial geometric constraints and online calibrated wave-making effect constraints, has been generated, the goal of this step is to solve a multi-objective optimization problem that can take into account both global navigation efficiency and individual traffic safety.

[0095] S5.1 Mathematical modeling of the scheduling optimization problem.

[0096] This application constructs the ship scheduling problem as a constrained combinatorial optimization problem, determining the upstream and downstream waiting queues. and The unique sequence of passage for all vessels and the precise release time for each vessel entering the accident section. .

[0097] To minimize a multi-objective optimization function of a weighted combination To solve the objective: in, This represents the total waiting time cost and indicates navigation efficiency. It is calculated based on the total waiting time cost of all vessels. The sum of waiting times: in For ships The moment of arrival at the waiting area. The total cost of navigational risk represents navigational safety. This application defines risk as the actual speed of the vessel traveling through the accident section. Its safe speed limit The closer it is to the upper limit, the higher the potential risk.

[0098] The present invention uses the square of the speed ratio to impose a greater penalty on high-speed passage, thereby guiding the optimization results toward a more conservative and safer speed selection. The risk aversion weighting coefficient is a dimensionless parameter that can be dynamically adjusted by on-site commanders based on the actual situation. (Improvement) This will make the optimization results more safety-oriented, but will sacrifice some traffic efficiency.

[0099] Any feasible scheduling scheme must strictly adhere to all generated constraints, including time-varying spatial geometric constraints and wave-making effect constraints based on field observations.

[0100] S5.2 Global optimization solution and output of the optimal solution.

[0101] Genetic algorithms are used to minimize the objective function. Guided by the principle of satisfying all hard constraints, the optimal solution is searched from a massive number of potential scheduling schemes.

[0102] The input to this optimization process is a time-varying residual channel width function. Spatiotemporal state matrix of ships waiting to pass List of navigation windows and wave-making constraints for each ship (i.e., maximum critical speed) ).

[0103] The output is an optimal scheduling scheme, which is an optimal passage sequence containing all vessels waiting to pass, and a schedule for each vessel in this sequence. The planned optimal release time and optimal speed .

[0104] S5.2.1 Input and Chromosome Encoding for Genetic Algorithms

[0105] The input to this optimization algorithm is the time-varying remaining channel width function. Spatiotemporal state matrix of ships waiting to pass The wave-making effect constraint for each ship, i.e., its unique maximum critical speed. .

[0106] To enable genetic algorithms to handle this scheduling problem, a potential solution is encoded using chromosomes. This application encodes a complete scheduling scheme (i.e., the passage sequence of all ships) as a chromosome arranged by natural numbers. If there are currently a total of For a number of ships waiting to pass, a chromosome is a chromosome of length... Of, including from 1 to An array of all unique integers. For example, the chromosome [3, 1, 4, 2] represents the passage sequence: the 3rd ship in the queue passes first, then the 1st, then the 4th, and finally the 2nd. This encoding ensures that any offspring chromosome generated by a genetic operator directly corresponds to a valid, conflict-free passage sequence.

[0107] S5.2.2 Population initialization and fitness function evaluation.

[0108] At the start of the algorithm, the system randomly generates a set of chromosomes in the above encoding format to form the initial population.

[0109] For each chromosome in the population, i.e., each undetermined travel sequence, its quality must be evaluated using a fitness function. The fitness function is then used to perform a full-process travel simulation to calculate the objective function value corresponding to that sequence. (The smaller the value, the higher the fitness).

[0110] The specific process of general aviation simulation is as follows: Ship sequences defined according to chromosomes The passage is planned for each ship in turn.

[0111] For the first ship in the sequence Its release time The time of its arrival in the waiting area The system is based on its maximum critical speed. Calculate its actual speed for the route it traversed. and travel time Thus, we can obtain its departure time. .

[0112] For each subsequent ship ( (its earliest possible release time) This depends on the departure time of the preceding vessel traveling in the same direction and whether there is a chance of encountering an oncoming vessel. The system will query the navigation window list to determine... Is there any permission during the time period? A two-way window for encountering oncoming vessels.

[0113] Under the premise of satisfying all spatial geometric constraints and wave-making effect constraints, for Determine an optimal actual release time. and passing speed In order to minimize its own waiting time and risk costs.

[0114] Repeat the above process until all ships in the sequence have been simulated. Finally, sum up the results to calculate the total for the sequence. and and according to The final evaluation value of the chromosome is obtained.

[0115] S5.2.3 Genetic Operators and Iterative Optimization.

[0116] The population is evolved by iteratively applying genetic operators to gradually approach the optimal solution: A roulette wheel or tournament selection method is used to prioritize chromosomes with high fitness for the next generation. For the selected chromosome pairs, sequential crossover is used to generate offspring. A small probability is used to perform crossover mutations on the offspring chromosomes, randomly swapping two ship IDs to maintain population diversity. This iterative process terminates when a preset maximum number of generations is reached or when the optimal solution in the population no longer improves over multiple generations.

[0117] S5.2.4 Decoding and output of the optimal solution.

[0118] When the genetic algorithm terminates, the population with the highest fitness (i.e., the lowest fitness) will have a significant number of cells. The chromosome with the given value is the optimal scheduling scheme found by the system. Decoding this optimal chromosome outputs the final result: S5.3 Generation and distribution of air traffic timetables.

[0119] After obtaining the optimal scheduling plan, the system automatically converts it into a list for each vessel waiting to pass. All vessels possess a navigation timetable with executable commands. This timetable is distributed to the corresponding vessels via shipboard interactive terminals in both graphical and textual formats, and its specific content includes: Remote speed control instructions: based on the calculated optimal release time. and the ship's current position The directive plans a recommended speed for vessels from their current position to the entrance of the accident area. This directive aims to guide vessels to arrive on time, replacing physical anchoring with virtual queuing, thereby reducing vessel congestion in narrow waters, lowering fuel consumption, and reducing the risk of collisions.

[0120] Release window: Issues a time window, accurate to the minute, for a vessel to enter the accident segment.

[0121] Passage mode instructions: clearly inform the vessel whether the passage is a one-way exclusive passage or a two-way escort passage with a specific oncoming vessel, providing the navigator with a clear expectation of the encounter.

[0122] Cross-section speed control: This clearly specifies the maximum speed the vessel must adhere to when passing through the core area of ​​the accident section. This speed is the optimal passage speed determined through optimization. It was clearly stated that the speed must not exceed its safe speed limit. .

[0123] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0124] In this specification, the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the descriptions of the embodiments described later are relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments.

[0125] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting waterway capacity and guiding ship navigation, characterized in that, The guidance methods include: Collect on-site images of the accident area and calculate the three-dimensional attitude of the wrecked vessel; Generate a time-varying remaining channel width function characterizing the future available channel width; construct the spatiotemporal state matrix of the vessels to be navigated; construct spatial geometric constraints based on the time-varying remaining channel width function; and construct wave-making effect constraints to determine the maximum critical speed of each vessel to be navigated based on on-site observations of the waves generated by the passing vessels. Based on the spatiotemporal state matrix, spatial geometric constraints, and wave-making effect constraints, a global optimization solution for multi-objective navigation scheduling is performed to generate and distribute a navigation timetable containing the optimal passage speed for each vessel waiting to pass.

2. The method according to claim 1, characterized in that, The specific steps for calculating the three-dimensional attitude of the wrecked vessel include: extracting the longitudinal and transverse structural lines of the vessel's hull from the on-site image, and calculating the intersection points of the longitudinal structural lines on the image plane. Intersection with the horizontal structural line Based on camera intrinsic parameter matrix The unit direction vectors of the longitudinal and transverse structural lines in the camera coordinate system can be calculated using the following formula. and : ； ； according to and Construct the rotation matrix of the wrecked ship relative to the camera coordinate system. And combined with the UAV's IMU attitude matrix ,Will Converting to the world coordinate system to obtain the roll angle of the wrecked ship Pitch angle and heading angle .

3. The method according to claim 1, characterized in that, The steps for generating the time-varying remaining channel width function include calculating the future relative water level fluctuation function. ; and utilize Predict any future moment using the following formula Predicted roll angle of the accident vessel : in, The initial roll angle of the vessel involved in the accident. According to The stranding stability coefficient is set according to the grade. The preset draft of the vessel involved in the accident.

4. The method according to claim 3, characterized in that, Generate time-varying residual channel width function The steps also include: calculating any future time. Effective obstacle width : in, The width of the ship involved in the accident is the pixel width on the image plane. This is the physical scale conversion factor. The safety buffer threshold; Prediction function for the total width of the future waterway minus The time-varying residual channel width function is obtained. .

5. The method according to claim 1, characterized in that, The steps for constructing wave-making effect constraints include setting a critical wave height threshold for the ship involved in the accident. The calculation formula is as follows: in, For safety reasons, This is the actual freeboard height. This represents the initial roll angle of the vessel involved in the accident.

6. The method according to claim 5, characterized in that, The steps for constructing wave-making effect constraints also include online calibration of the speed-wave height model for passing vessels: when a vessel acting as a pathfinder... At actual speed When passing through the accident area, the physical wave height generated is measured using on-site imagery. Based on data points The coefficients of the speed-wave height relationship model applicable to the exploratory vessel are calibrated using the following formula. : For any subsequent vessel awaiting passage According to its captain Habune Hiro Compared to scout ships Captain Habune Hiro The difference is expressed by the following formula. Make corrections to obtain a result suitable for coefficient : 。 7. The method according to claim 6, characterized in that, The specific steps for determining the maximum critical speed of each vessel to be passed are as follows: based on any vessel to be passed... The maximum wave height generated must not exceed the critical wave height threshold. The constraints, namely Inverse solution applicable to Maximum critical speed : in, for The speed at which ships can pass.

8. The method according to claim 1, characterized in that, The steps for solving the global optimization problem of multi-objective air traffic scheduling are to minimize the following multi-objective optimization function. For the goal: in, The total waiting time cost for all vessels waiting to pass. The total passage risk cost for all vessels waiting to pass. This is the risk aversion weighting coefficient; and The calculation formula is: in, For the queue of ships waiting to pass, For ships The moment of release, For ships The arrival time, For ships The speed at which ships can pass, For ships The maximum critical speed.

9. The method according to claim 1, characterized in that, The specific steps for generating the time-varying remaining channel width function include: periodically observing the accident area and simultaneously recording the sequence of the total physical channel width. and water area pixel sequence Linear regression analysis was performed on paired data consisting of two sequences to solve the model. The average slope coefficient of the riverbank and tidal flats in China ; Linear regression analysis was performed on the total physical width sequence of the waterway to determine the average rate of change of the total waterway width. ; based on The width value obtained from the latest observation Construct a prediction function for the total width of the future waterway. and combined The future relative water level fluctuation function is calculated, and then the time-varying remaining channel width function is generated.

10. A waterway capacity prediction and ship navigation guidance system, characterized in that, The system includes: Acquisition module: Acquires on-site images of the accident area and calculates the three-dimensional attitude of the wrecked vessel; Constraint Module: Generates a time-varying remaining channel width function characterizing the future available channel width; constructs the spatiotemporal state matrix of vessels to be navigated; constructs spatial geometric constraints based on the time-varying remaining channel width function; and constructs wave-making effect constraints to determine the maximum critical speed of each vessel to be navigated based on on-site observations of the waves generated by the navigating vessels. Optimization module: Based on the spatiotemporal state matrix, spatial geometric constraints and wave-making effect constraints, it performs global optimization of multi-objective navigation scheduling to generate and issue a navigation timetable containing the optimal passage speed of each vessel waiting to pass.

Images