Method, device and electronic equipment for path planning of a ship

Abstract

The application provides a ship path planning method and device and electronic equipment. The method comprises the following steps: acquiring multi-source data of a target water area, determining a multi-dimensional situation data frame according to the multi-source data, determining a predicted traffic flow of a passing area according to the multi-dimensional situation data frame, determining a multi-constraint joint scheduling model according to ship arrival information in the multi-dimensional situation data frame, channel passing time window constraints, weather prediction condition constraints, tidal time window constraints, berth adaptation constraints and channel capacity constraints, determining scheduling decision data containing berth allocation and channel entering time according to the multi-constraint joint scheduling model, determining a dynamic navigation cost field according to the scheduling decision data, the predicted traffic flow and the multi-dimensional situation data frame, determining a candidate path under the condition of satisfying ship kinematics constraints according to the dynamic navigation cost field, and determining recommended path data according to a preset risk scoring rule. The application comprehensively considers ship physical motion characteristics and multi-dimensional navigation risks, and realizes safe path planning.

Description

Technical Field

[0001] This application relates to the field of shipping big data and port intelligent scheduling technology, and more specifically, to a ship path planning method, device and electronic equipment. Background Technology

[0002] In related technologies, large hub ports typically have a large daily number of ships entering and leaving the port and a large annual pilotage scale. Port scheduling systems usually adopt a multi-level scheduling mode, with core processes covering voyage application, plan acceptance, plan preparation, plan release, and dynamic adjustment. However, common scheduling rules are complex with multiple constraints and coupling, frequent sudden disturbances, high costs for dynamic adjustment, and insufficient utilization of waterway resources. Therefore, rule-based scheduling systems lack adaptability, single-objective optimization cannot adapt to multi-constraint coupled scenarios, and the level of intelligence of pilotage assistance systems is low. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the prior art or related technologies.

[0004] Therefore, the first aspect of this application proposes a method for ship path planning.

[0005] The second aspect of this application proposes a path planning device for ships.

[0006] A third aspect of this application proposes an electronic device.

[0007] In view of this, the first aspect of this application provides a ship path planning method, which is executed by a computer. The ship path planning method includes: acquiring multi-source data of a target waterway, including ship dynamic data, environmental data, and nautical chart data; determining a multi-dimensional situational data frame based on the multi-source data; determining the predicted flow of the passage area based on the multi-dimensional situational data frame; determining a multi-constraint joint scheduling model based on ship arrival information, channel passage window constraints, tidal window constraints, meteorological forecast constraints, berth adaptation constraints, and channel capacity constraints in the multi-constraint joint scheduling model; determining scheduling decision data including berth allocation and channel entry time by using combinatorial optimization and reinforcement learning to solve the multi-constraint joint scheduling model; determining a dynamic navigation cost field based on the scheduling decision data, predicted flow, and multi-dimensional situational data frame; determining candidate paths based on the dynamic navigation cost field while satisfying ship kinematic constraints, and determining recommended path data according to preset risk scoring rules.

[0008] This application's solution involves a complete technical chain: firstly, acquiring multi-source data (including vessel dynamics, environmental data, and nautical charts) of the target waterway and determining a multi-dimensional situational data frame; secondly, determining the predicted flow of the passage area based on the multi-dimensional situational data frame; thirdly, determining a multi-constraint joint scheduling model based on vessel arrival information, channel passage window constraints, tidal window constraints, meteorological forecast constraints, berth suitability constraints, and channel capacity constraints from the multi-dimensional situational data frame; fourthly, using combinatorial optimization and reinforcement learning to collaboratively solve the multi-constraint joint scheduling model to obtain scheduling decision data including berth allocation and channel entry time; fifthly, determining a dynamic navigation cost field based on the scheduling decision data, predicted flow, and multi-dimensional situational data frame; and finally, determining candidate paths based on the dynamic navigation cost field while satisfying vessel kinematic constraints and determining recommended paths according to preset risk scoring rules. This firstly breaks down data silos scattered across different systems, achieving unified integration and standardized output of multi-source data across the entire port area, forming a comprehensive port situational awareness capability covering vessels, environment, and geography. Secondly, by predicting future channel traffic flow, it provides forward-looking data support for subsequent scheduling and path planning, enabling proactive decision-making based on future situational awareness. Furthermore, by constructing a multi-constraint joint scheduling model integrating all core hard constraints and employing a combinatorial optimization and reinforcement learning collaborative solution mechanism, the challenge of global optimization due to multi-constraint coupling was solved. This enables the coordinated and optimized allocation of berth and waterway resources, improving waterway traffic efficiency by 15% to 20% in a single main waterway scenario, supporting over 100 vessel arrivals and departures per day, and significantly reducing the reliance on manual experience in scheduling. Simultaneously, by integrating scheduling plans, future traffic flow, and real-time environmental information to construct a dynamic navigation cost field, dynamic quantification and spatial representation of waterway traffic costs are achieved, adapting to dynamic traffic flow and real-time environmental changes. Finally, by comprehensively considering ship physical motion characteristics and multi-dimensional navigation risk generation and path optimization, quantitative decision support is provided for pilotage operations, improving the standardization and safety of pilotage operations.

[0009] In some technical solutions of this application, multi-dimensional situational data frames are determined based on multi-source data, including: determining abnormal data based on ship dynamic data; determining corrected ship dynamic data and data reliability level based on abnormal data; and performing unified timestamp alignment and coordinate system transformation processing based on corrected ship dynamic data, environmental data, and nautical chart data to determine multi-dimensional situational data frames.

[0010] In the above technical solution, abnormal data refers to inaccurate data in ship dynamic data caused by signal loss, drift, interference, or spoofing attacks. Data reliability level refers to a grading index used to identify the credibility of corrected ship dynamic data. Unified timestamp alignment refers to the process of synchronizing data from different data sources to the same time base. Coordinate system transformation refers to the process of converting geographic data in different coordinate systems to the same coordinate system. Based on ship dynamic data, abnormal data is identified, and based on the abnormal data, corrected ship dynamic data and data reliability level are determined. This effectively filters noise and errors in ship dynamic data, improves the accuracy of ship status data, and provides a credibility basis for subsequent data processing through data reliability level identification. Unified timestamp alignment and coordinate system transformation are performed on the corrected ship dynamic data, environmental data, and nautical chart data to determine multi-dimensional situational data frames. Unified timestamp alignment achieves time synchronization between different data sources, and coordinate system transformation achieves unification of different geographic coordinate systems, ensuring the spatiotemporal consistency of multi-dimensional situational data frames.

[0011] In some technical solutions of this application, the ship path planning method further includes: determining the nearest encounter distance and the time to reach the nearest encounter point for each moving ship in the target water area based on multi-dimensional situational data frames; generating collision warning information when the nearest encounter distance is less than a preset safe distance and the time to reach the nearest encounter point is less than a preset warning time; and / or determining the turning angular velocity of each moving ship based on multi-dimensional situational data frames, generating abnormal maneuvering warning information when the turning angular velocity exceeds a preset threshold.

[0012] In the above technical solution, the nearest encounter distance refers to the minimum distance expected to be reached between the two vessels during an encounter, and the time to reach the nearest encounter point refers to the time required to travel from the current moment to the nearest encounter point. The preset safety distance and preset warning time can be configured according to the actual conditions of the port. Collision warning information is pushed to the pilot's mobile terminal and the vessel traffic service workstation according to the level of urgency. The preset threshold for turning angular velocity is determined based on the statistical upper limit of historical data for vessels of the same type. When the turning angular velocity exceeds this threshold, it is judged as abnormal maneuvering behavior, triggering the tugboat emergency standby command.

[0013] In some technical solutions of this application, based on a multi-constraint joint scheduling model, combinatorial optimization and reinforcement learning are used to solve the scheduling decision data, including berth allocation and channel entry time. This includes: determining the solution method based on the comparison between the number of ships to be scheduled and a preset ship number threshold; when the number of ships to be scheduled is less than the preset ship number threshold, determining the global optimal solution based on the multi-constraint joint scheduling model; when the number of ships to be scheduled is greater than or equal to the preset ship number threshold, determining a fast feasible solution based on the multi-constraint joint scheduling model, and determining the scheduling decision data after constraint violation repair based on the fast feasible solution.

[0014] In the above technical solution, the preset ship number threshold refers to a pre-set critical value for the number of ships used to distinguish between small-scale and large-scale scheduling scenarios. The global optimal solution refers to the solution with the optimal objective function value among all feasible solutions. The fast feasible solution refers to a solution that satisfies all constraints and is obtained in a short time. Constraint violation repair refers to the process of correcting solutions that violate constraints during the solution process to satisfy all constraints. The solution method is determined based on the comparison between the number of ships to be scheduled and the preset ship number threshold, achieving adaptive solution matching for scheduling scenarios of different scales. When the number of ships to be scheduled is less than the preset ship number threshold, the global optimal solution is determined based on the multi-constraint joint scheduling model, ensuring the global optimality of the scheduling scheme in small-scale scenarios. When the number of ships to be scheduled is greater than or equal to the preset ship number threshold, the fast feasible solution is determined based on the multi-constraint joint scheduling model, and the scheduling decision data after constraint violation repair is determined based on the fast feasible solution. This process significantly shortens the solution time while ensuring the feasibility of the scheduling scheme, meeting the response requirements of large-scale real-time scheduling.

[0015] In some technical solutions of this application, a dynamic navigation cost field is determined based on scheduling decision data, predicted flow, and multi-dimensional situational data frames. This includes: determining a real-time traffic flow density field based on predicted flow; determining a meteorological risk field and a water depth margin field based on environmental data; determining a superimposed basic cost field based on nautical chart data, real-time traffic flow density field, meteorological risk field, and water depth margin field; and determining a dynamically updated dynamic navigation cost field based on the basic cost field and the normalized weighted results of channel location, time, risk level, and water depth margin.

[0016] In the above technical solution, the real-time traffic flow density field refers to the spatial distribution field reflecting the real-time vessel density at different locations within the waterway. The meteorological risk field refers to the spatial distribution field reflecting the meteorological risk level at different locations within the waterway. The water depth margin field refers to the spatial distribution field reflecting the difference between water depth and vessel draft at different locations within the waterway. The basic cost field refers to the initial passage cost distribution field obtained by overlaying electronic charts and various risk fields. Normalized weighting refers to the process of converting influencing factors of different dimensions into a unified dimension and then performing weighted summation according to preset weights. The real-time traffic flow density field is determined based on the predicted flow rate, and the real-time traffic flow density field quantitatively expresses the degree of traffic congestion at different locations within the waterway in the future. The meteorological risk field and water depth margin field are determined based on environmental data, and the meteorological risk field quantitatively expresses the meteorological navigation risk at different locations, while the water depth margin field quantitatively expresses the shallow water navigation risk at different locations. The overlaid basic cost field is determined based on chart data, the real-time traffic flow density field, the meteorological risk field, and the water depth margin field, integrating multi-dimensional navigation influencing factors such as geography, traffic, meteorology, and water depth. Based on the normalized weighted results of the basic cost field and the channel location, time, risk level, and water depth margin, a dynamic navigation cost field that is dynamically updated over time is determined. This process realizes the multi-dimensional integration and dynamic updating of navigation costs, providing accurate quantitative cost basis for route planning.

[0017] In some technical solutions of this application, candidate paths are determined based on the dynamic navigation cost field while satisfying ship kinematic constraints. This includes: determining path search constraints based on preset ship kinematic constraints, which include minimum turning radius constraints, restricted area avoidance constraints, and maximum cross current constraints; performing dynamic path search based on the dynamic navigation cost field and path search constraints; updating path costs for affected local areas and replanning based on changes in waterway conditions; and determining multiple candidate paths.

[0018] In the above technical solution, the minimum turning radius constraint refers to the minimum radius limit for a ship to turn safely. The restricted area avoidance constraint refers to the restriction that a ship must not enter a prohibited navigation area. The maximum cross-current constraint refers to the maximum lateral current speed limit for a ship to navigate safely. Dynamic path search refers to the process of searching for a path on a dynamically changing cost field and updating the path in real time according to changes in the cost field. Path search constraints are determined based on preset ship kinematic constraints, including the minimum turning radius constraint, restricted area avoidance constraint, and maximum cross-current constraint. This process ensures that the searched path conforms to the ship's physical motion characteristics and navigation safety regulations, guaranteeing the physical feasibility of the path. Dynamic path search is performed based on the dynamic navigation cost field and path search constraints. The path cost is updated and replanned for affected local areas according to changes in the waterway situation, determining multiple candidate paths. This process achieves dynamic path updating and local replanning, ensuring the real-time feasibility of the path when the waterway situation changes.

[0019] In some technical solutions of this application, recommended path data is determined according to preset risk scoring rules, including: determining the comprehensive risk score of multiple candidate paths based on the degree of deviation between the collision risk index of multiple candidate paths and historical reference trajectories, the degree of meteorological influence, and the difficulty of tugboat coordination; and determining recommended path data based on the ranking of the comprehensive risk scores of multiple candidate paths from low to high.

[0020] In the aforementioned technical solution, the collision risk index refers to an indicator that quantitatively assesses the likelihood of collisions between vessels. The historical reference trajectory refers to the optimal navigation trajectory extracted from historical pilotage trajectories. The comprehensive risk score refers to the overall quantitative value of the path risk obtained by integrating multiple risk dimensions. Based on the deviation of the collision risk index of multiple candidate paths from the historical reference trajectory, the degree of weather impact, and the difficulty of tugboat coordination, the comprehensive risk score of multiple candidate paths is determined. This process comprehensively considers risk factors from multiple dimensions, including navigation safety, pilotage experience, environmental impact, and operational coordination, to achieve a comprehensive quantitative assessment of path risk. Recommended path data is determined based on the ranking of the comprehensive risk scores of multiple candidate paths from low to high. This process provides pilots with the optimal navigation path selection, improving the scientific nature and safety of pilotage decisions.

[0021] In some technical solutions of this application, the route planning method for ships further includes: obtaining emergency event signals based on a preset event monitoring data source; determining the set of affected ships based on the dependency graph between the emergency event signals and the preset scheduling plan; and determining updated scheduling decision data based on the set of affected ships and the existing plans of unaffected ships.

[0022] In the above technical solution, the preset event monitoring data source refers to various data sources pre-configured for monitoring port emergencies. An emergency event signal indicates a signal indicating an emergency event affecting normal port operations. The preset scheduling plan dependency graph refers to a pre-constructed graph structure representing the dependencies between ship scheduling plans. The affected ship set refers to the set of ships whose scheduling plans need adjustment due to the emergency event. By acquiring the emergency event signal from the preset event monitoring data source, automatic detection and identification of emergencies are achieved. The affected ship set is determined based on the emergency event signal and the preset scheduling plan dependency graph, quickly and accurately identifying the scope of the emergency's impact and its cascading propagation chain. Updated scheduling decision data is determined based on the affected ship set and the existing plans of unaffected ships. This process only performs local re-optimization on affected ships, keeping the plans of unaffected ships unchanged, minimizing the impact of the emergency on the overall port scheduling, and improving the port's ability to respond to emergencies.

[0023] A second aspect of this application provides a route planning device for a ship, comprising:

[0024] The data processing unit is used to acquire multi-source data of the target water area, including ship dynamic data, environmental data and nautical chart data, and to determine multi-dimensional situational data frames based on the multi-source data.

[0025] The traffic prediction unit is used to determine the predicted traffic flow in the passage area based on multi-dimensional situational data frames.

[0026] The scheduling decision unit is used to determine a multi-constraint joint scheduling model based on ship arrival information, channel passage time window constraints, tidal time window constraints, weather forecast constraints, berth adaptation constraints, and channel capacity constraints in the multi-dimensional situational data frame; and based on the multi-constraint joint scheduling model, it uses combinatorial optimization and reinforcement learning to solve the problem together to determine scheduling decision data including berth allocation and channel entry time.

[0027] The cost field construction unit is used to determine the dynamic navigation cost field based on scheduling decision data, predicted traffic, and multi-dimensional situational data frames.

[0028] The path planning unit is used to determine candidate paths based on the dynamic navigation cost field while satisfying ship kinematic constraints, and to determine recommended path data according to preset risk scoring rules.

[0029] A third aspect of this application provides an electronic device comprising: a processor and a memory, the memory storing a computer program, the processor executing the computer program in the memory to implement the ship path planning method as described in any of the above-described technical solutions. Therefore, the electronic device possesses all the beneficial effects of the ship path planning method as described in any of the above-described technical solutions.

[0030] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0031] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0032] Figure 1 This is one of the flowcharts illustrating a ship path planning method according to an embodiment of this application;

[0033] Figure 2 This is a schematic block diagram of one embodiment of a system for implementing the above-described ship path planning method according to an embodiment of this application;

[0034] Figure 3 This is a schematic block diagram of a ship path planning device according to an embodiment of this application;

[0035] Figure 4This is a schematic block diagram of an electronic device according to an embodiment of the present application. Detailed Implementation

[0036] To better understand the above-mentioned objectives, features, and advantages of this application, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0037] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Therefore, the scope of protection of this application is not limited to the specific embodiments disclosed below.

[0038] The following reference Figures 1 to 4 This application describes a method, apparatus, and electronic device for ship route planning according to some embodiments.

[0039] like Figure 1 As shown, an embodiment of this application provides a ship path planning method, which is executed by a computer. The ship path planning method includes the following steps:

[0040] Step 102: Obtain multi-source data for the target water area. The multi-source data includes vessel dynamic data, environmental data, and nautical chart data. Based on the multi-source data, determine the multi-dimensional situational data frame.

[0041] Step 104: Determine the predicted traffic flow in the passage area based on the multi-dimensional situational data frame;

[0042] Step 106: Determine the multi-constraint joint scheduling model based on the ship arrival information, waterway passage time window constraints, tidal time window constraints, weather forecast constraints, berth adaptation constraints, and waterway capacity constraints in the multi-dimensional situational data frame.

[0043] Step 108: Based on the multi-constraint joint scheduling model, combinatorial optimization and reinforcement learning are used to solve the scheduling decision data, which includes berth allocation and channel entry time.

[0044] Step 110: Determine the dynamic navigation cost field based on scheduling decision data, predicted traffic flow, and multi-dimensional situational data frames;

[0045] Step 112: Based on the dynamic navigation cost field, candidate paths are determined under the constraints of ship kinematics, and recommended path data is determined according to the preset risk scoring rules.

[0046] In this application, multi-source data refers to port waterway-related data from different acquisition systems and data formats, including vessel dynamic data, environmental data, and nautical chart data. Multi-dimensional situational data frames refer to standardized data structures that integrate multi-source heterogeneous data into a unified time base and coordinate system, containing multi-dimensional information such as vessel status, meteorological and hydrological data, and channel geography. Multi-dimensional situational data frames integrate data scattered across different systems into a unified format, achieving unified integration and standardized output of multi-source data across the entire port area, forming a comprehensive port situational awareness capability covering vessels, environment, and geography. Predicted flow refers to the number of vessels passing through a channel section within a future period, predicted based on historical and real-time situational data. Predicted flow reflects future vessel traffic trends in the channel, providing forward-looking data support for subsequent scheduling decisions and route planning, enabling proactive decision-making based on future situations, and improving the timeliness and accuracy of decision-making. Multi-constraint joint scheduling model refers to a mathematical optimization model that integrates various constraints in port scheduling to solve berth allocation and channel traffic sequences. The multi-constraint joint scheduling model incorporates all core hard constraints in port scheduling, achieving unified modeling and coupled processing of multiple constraints to construct a complete port scheduling constraint system. Scheduling decision data refers to the results obtained from solving the multi-constraint joint scheduling model, including berth allocation information and channel entry time information. The combined optimization and reinforcement learning collaborative solution mechanism balances the optimality and real-time performance of the scheduling scheme, achieving coordinated and optimized allocation of berth and channel resources, and improving the overall resource utilization rate and vessel passage efficiency of the port.

[0047] A dynamic navigation cost field refers to a channel passage cost distribution map generated by overlaying real-time traffic, weather, and water depth information on electronic nautical charts, and updated over time. The dynamic navigation cost field integrates real-time scheduling plans, future traffic flows, and real-time environmental information to achieve dynamic quantification and spatial representation of channel passage costs, providing comprehensive and accurate cost data for route planning.

[0048] Candidate paths refer to multiple feasible navigation paths that satisfy ship kinematic constraints. Preset risk scoring rules refer to pre-defined quantitative standards used to assess the comprehensive risk of navigation paths. Recommended path data refers to the optimal navigation path information obtained after ranking by risk scores. This process comprehensively considers the ship's physical motion characteristics and multi-dimensional navigation risks to achieve a path planning that balances safety and efficiency, providing quantitative decision support for pilotage operations and improving the standardization and safety of pilotage operations. Meteorological forecast constraints refer to mandatory restrictions imposed on ship entry and exit times, channel passage permissions, and operational methods based on meteorological and hydrological forecast data for port waters within a preset time period. These are core dynamic hard constraints in the multi-constraint joint scheduling model of this application, alongside channel passage window constraints and tidal window constraints.

[0049] This application's solution involves a complete technical chain: firstly, acquiring multi-source data (including vessel dynamics, environmental data, and nautical charts) of the target waterway and determining a multi-dimensional situational data frame; secondly, determining the predicted flow of the passage area based on the multi-dimensional situational data frame; thirdly, determining a multi-constraint joint scheduling model based on vessel arrival information, channel passage window constraints, tidal window constraints, berth adaptation constraints, and channel capacity constraints from the multi-dimensional situational data frame; fourthly, using combinatorial optimization and reinforcement learning to collaboratively solve the multi-constraint joint scheduling model to obtain scheduling decision data including berth allocation and channel entry time; fifthly, determining a dynamic navigation cost field based on the scheduling decision data, predicted flow, and multi-dimensional situational data frame; and finally, determining candidate paths based on the dynamic navigation cost field while satisfying vessel kinematic constraints and determining recommended paths according to preset risk scoring rules. This firstly breaks down data silos scattered across different systems, achieving unified integration and standardized output of multi-source data across the entire port area, forming a comprehensive port situational awareness capability covering vessels, environment, and geography. Secondly, by predicting future channel traffic flow, it provides forward-looking data support for subsequent scheduling and path planning, enabling proactive decision-making based on future situational awareness. Furthermore, by constructing a multi-constraint joint scheduling model integrating all core hard constraints and employing a combinatorial optimization and reinforcement learning collaborative solution mechanism, the challenge of global optimization due to multi-constraint coupling was solved. This enables the coordinated and optimized allocation of berth and waterway resources, improving waterway traffic efficiency by 15% to 20% in a single main waterway scenario, supporting over 100 vessel arrivals and departures per day, and significantly reducing the reliance on manual experience in scheduling. Simultaneously, by integrating scheduling plans, future traffic flow, and real-time environmental information to construct a dynamic navigation cost field, dynamic quantification and spatial representation of waterway traffic costs are achieved, adapting to dynamic traffic flow and real-time environmental changes. Finally, by comprehensively considering ship physical motion characteristics and multi-dimensional navigation risk generation and path optimization, quantitative decision support is provided for pilotage operations, improving the standardization and safety of pilotage operations.

[0050] In some embodiments of this application, determining a multi-dimensional situational data frame based on multi-source data includes: determining abnormal data based on ship dynamic data; determining corrected ship dynamic data and data reliability level based on the abnormal data; and performing unified timestamp alignment and coordinate system transformation processing based on the corrected ship dynamic data, environmental data, and nautical chart data to determine the multi-dimensional situational data frame.

[0051] In the above embodiments, abnormal data refers to inaccurate data in ship dynamic data caused by signal loss, drift, interference, or spoofing attacks. Data reliability level refers to a grading index used to identify the credibility of corrected ship dynamic data. Unified timestamp alignment refers to the process of synchronizing data from different data sources to the same time base. Coordinate system transformation refers to the process of converting geographic data in different coordinate systems to the same coordinate system. Based on the ship dynamic data, abnormal data is identified, and based on the abnormal data, corrected ship dynamic data and data reliability level are determined. This effectively filters noise and errors in the ship dynamic data, improves the accuracy of ship status data, and provides a credibility basis for subsequent data processing through data reliability level identification. Unified timestamp alignment and coordinate system transformation are performed on the corrected ship dynamic data, environmental data, and nautical chart data to determine multi-dimensional situational data frames. Unified timestamp alignment achieves time synchronization between different data sources, and coordinate system transformation achieves unification of different geographic coordinate systems, ensuring the spatiotemporal consistency of multi-dimensional situational data frames. Through the above three steps, noise and errors in ship dynamic data are first filtered and corrected, then spatiotemporal unification of multi-source data is achieved, and finally standardized multi-dimensional situational data frames are generated, providing accurate, consistent and reliable basic data input for the entire system.

[0052] In one embodiment, the ship dynamic data is AIS (Automatic Identification System) data. Abnormal data includes gaps caused by signal loss, position jumps caused by signal drift, garbled data caused by electromagnetic interference, and false position data caused by GNSS (Global Navigation Satellite System) spoofing. A Kalman filter combined with historical trajectory similarity matching is used to correct the abnormal data. Kalman filtering is used for interpolation prediction of short-term signal loss, while historical trajectory similarity matching is used to correct position anomalies caused by long-term signal loss or GNSS spoofing. The reliability level of the corrected data is divided into three levels: high, medium, and low. The original uncorrected data is of high reliability, the Kalman-filtered corrected data is of medium reliability, and the historical trajectory matching corrected data is of low reliability. The temporal resolution of the multi-dimensional situational data frame is no less than 30 seconds, and the spatial resolution is no less than 1 meter.

[0053] In some embodiments of this application, the ship path planning method further includes: determining the nearest encounter distance and the time to reach the nearest encounter point for each moving ship in the target water area based on a multi-dimensional situational data frame; generating collision warning information when the nearest encounter distance is less than a preset safe distance and the time to reach the nearest encounter point is less than a preset warning time; and / or determining the turning angular velocity of each moving ship based on the multi-dimensional situational data frame, generating abnormal maneuvering warning information when the turning angular velocity exceeds a preset threshold.

[0054] In the above embodiments, the nearest encounter distance refers to the minimum distance expected to be reached between two ships during an encounter, and the time to reach the nearest encounter point refers to the time required to travel from the current moment to the nearest encounter point. These two parameters are core indicators for quantifying ship collision risk. By calculating the nearest encounter distance and the time to reach the nearest encounter point between all moving ships in real time, the collision risk situation in the target waters can be comprehensively grasped. When the nearest encounter distance is less than the preset safe distance and the time to reach the nearest encounter point is less than the preset warning time, it indicates that there is a risk of collision between the two ships. At this time, a collision warning message is generated and pushed to relevant parties, allowing the pilot to take evasive measures in advance. The turning angular velocity refers to the angle turned by a ship per unit time when turning. There is a statistical range for the normal turning angular velocity of a ship. When the turning angular velocity exceeds this range, it usually indicates that the ship is experiencing abnormal maneuvering, such as steering gear failure or pilot error. By monitoring the turning angular velocity of a ship in real time, abnormal maneuvering behavior of the ship can be detected in time, and emergency measures can be taken in advance to avoid accidents.

[0055] This solution achieves comprehensive monitoring of ship navigation safety through a proactive safety mechanism that combines collision risk quantification and early warning with abnormal maneuver detection. It transforms safety management from post-event handling to pre-event prevention, significantly improving the safety of port navigation.

[0056] In one embodiment, the nearest encounter distance and time to the nearest encounter point are calculated in 30-second intervals. For scenarios with a large number of vessels, R-tree spatial index pruning is used to reduce computational load. The preset safe distance is 1 nautical mile, and the preset warning time is 20 minutes. Warnings are categorized into three levels of urgency, pushed to pilot mobile devices, VTS (Vessel Traffic Service) workstations, and dispatch centers 20 minutes, 15 minutes, and 10 minutes in advance, respectively. Simultaneously, based on the relative bearing and velocity vectors of the vessels, the system determines whether the current encounter situation falls under Articles 12 to 18 of the International Regulations for Preventing Collisions at Sea (ICMLS) as overtaking, crossing, or head-on encounters. It automatically verifies whether the current maneuvering behavior of each vessel complies with the collision avoidance rules and issues immediate warnings for violations. The preset threshold for turning angular velocity is three times the standard deviation of the historical average turning angular velocity of vessels of the same type. When the turning angular velocity exceeds this threshold, it is determined to be abnormal maneuvering behavior, automatically triggering a tugboat emergency standby command.

[0057] In some embodiments of this application, based on a multi-constraint joint scheduling model, combinatorial optimization and reinforcement learning are used to collaboratively solve the problem and determine scheduling decision data including berth allocation and channel entry time. This includes: determining the solution method based on the comparison between the number of ships to be scheduled and a preset ship number threshold; when the number of ships to be scheduled is less than the preset ship number threshold, determining the global optimal solution based on the multi-constraint joint scheduling model; when the number of ships to be scheduled is greater than or equal to the preset ship number threshold, determining a fast feasible solution based on the multi-constraint joint scheduling model, and determining the scheduling decision data after constraint violation repair based on the fast feasible solution.

[0058] In the above embodiments, the port scheduling problem is a typical NP-hard problem, and the solution time increases exponentially with the number of ships to be scheduled. Using only a single exact solution method will not meet real-time requirements in large-scale scenarios; using only a single heuristic solution method cannot guarantee the optimality of the solution. This scheme adopts a strategy of combinatorial optimization and reinforcement learning to solve the problem collaboratively. It adaptively selects the solution method based on the number of ships to be scheduled, perfectly balancing the optimality of the solution and the real-time performance. When the number of ships to be scheduled is small, the exact solution method is used to obtain the globally optimal solution, ensuring the optimality of the scheduling scheme. When the number of ships to be scheduled is large, a deep reinforcement learning method is used to quickly obtain a high-quality feasible solution. Then, the exact solution method is used to post-process the solution, correcting any potential constraint violations, thus ensuring both solution speed and feasibility.

[0059] This solution overcomes the limitations of local optima based on human experience through an adaptive collaborative solution mechanism, achieving joint optimization across three dimensions: berth, time window, and waterway, effectively improving the overall resource utilization rate and vessel passage efficiency of the port.

[0060] In one embodiment, the preset threshold for the number of ships is 20. When the number of ships is less than or equal to 20, a Mixed Integer Programming (MIP) solver (such as Gurobi or CPLEX) is used to find the global optimum, with the objective function being a weighted sum of minimizing the total ship waiting time and maximizing the channel utilization rate. When the number of ships is greater than 20, a deep reinforcement learning agent trained using the Proximal Policy Optimization (PPO) algorithm is used to find a fast feasible solution, with the reward function consisting of a weighted sum of constraint satisfaction, total ship waiting time, and channel utilization rate. After the reinforcement learning agent outputs an initial solution, a mixed integer programming post-processing module corrects for constraint violations, ensuring that the final scheduling decision data satisfies all hard constraints.

[0061] In some embodiments of this application, a dynamic navigation cost field is determined based on scheduling decision data, predicted traffic flow, and multi-dimensional situational data frames. This includes: determining a real-time traffic flow density field based on predicted traffic flow; determining a meteorological risk field and a water depth margin field based on environmental data; determining a superimposed basic cost field based on nautical chart data, the real-time traffic flow density field, the meteorological risk field, and the water depth margin field; and determining a dynamically updated dynamic navigation cost field based on the basic cost field and the normalized weighted results of channel location, time, risk level, and water depth margin.

[0062] In the above embodiments, the real-time traffic flow density field refers to the spatial distribution field reflecting the real-time vessel density at different locations within the waterway. The meteorological risk field refers to the spatial distribution field reflecting the meteorological risk level at different locations within the waterway. The water depth margin field refers to the spatial distribution field reflecting the difference between water depth and vessel draft at different locations within the waterway. The basic cost field refers to the initial passage cost distribution field obtained by overlaying electronic charts and various risk fields. Normalized weighting refers to the process of converting influencing factors of different dimensions into a unified dimension and then performing weighted summation according to preset weights. The real-time traffic flow density field is determined based on the predicted flow rate, and the real-time traffic flow density field quantitatively expresses the degree of traffic congestion at different locations within the waterway in the future. The meteorological risk field and water depth margin field are determined based on environmental data, and the meteorological risk field quantitatively expresses the meteorological navigation risk at different locations, while the water depth margin field quantitatively expresses the shallow water navigation risk at different locations. The overlaid basic cost field is determined based on chart data, the real-time traffic flow density field, the meteorological risk field, and the water depth margin field, integrating multi-dimensional navigation influencing factors such as geography, traffic, meteorology, and water depth. Based on the normalized weighted results of the basic cost field and the channel location, time, risk level, and water depth margin, a dynamic navigation cost field that is dynamically updated over time is determined. This process realizes the multi-dimensional integration and dynamic updating of navigation costs, providing accurate quantitative cost basis for route planning.

[0063] In one embodiment, the dynamic navigation cost field is updated every 30 seconds, and the normalized weights are adaptively adjusted based on the actual port operation data to ensure that the cost field can accurately reflect real-time navigation conditions.

[0064] In some embodiments of this application, candidate paths are determined based on a dynamic navigation cost field while satisfying ship kinematic constraints. This includes: determining path search constraints based on preset ship kinematic constraints, which include minimum turning radius constraints, restricted area avoidance constraints, and maximum cross current constraints; performing dynamic path search based on the dynamic navigation cost field and path search constraints; updating path costs for affected local areas and replanning based on changes in waterway conditions; and determining multiple candidate paths.

[0065] In the above embodiments, the minimum turning radius constraint refers to the minimum radius limit for a ship to turn safely. The restricted area avoidance constraint refers to the restriction that a ship must not enter a prohibited navigation area. The maximum crossflow constraint refers to the maximum lateral current speed limit for a ship to navigate safely. Dynamic path search refers to the process of searching for a path on a dynamically changing cost field and updating the path in real time according to changes in the cost field. Path search constraints are determined based on preset ship kinematic constraints, including the minimum turning radius constraint, restricted area avoidance constraint, and maximum crossflow constraint. This process ensures that the searched path conforms to the ship's physical motion characteristics and navigation safety regulations, guaranteeing the physical feasibility of the path. Dynamic path search is performed based on the dynamic navigation cost field and path search constraints. The path cost is updated and replanned for affected local areas according to changes in the waterway situation, determining multiple candidate paths. This process achieves dynamic path updating and local replanning, ensuring the real-time feasibility of the path when the waterway situation changes.

[0066] In one embodiment, an improved dynamic A-Star algorithm is used to perform dynamic path search. When sudden changes in ships or weather occur in the waterway, the path cost is recalculated only for the affected local area, which greatly shortens the path update time.

[0067] In some embodiments of this application, recommended path data is determined according to preset risk scoring rules, including: determining the comprehensive risk score of multiple candidate paths based on the degree of deviation of the collision risk index of multiple candidate paths from historical reference trajectories, the degree of weather impact, and the difficulty of tugboat coordination; and determining recommended path data based on the ranking of the comprehensive risk scores of multiple candidate paths from low to high.

[0068] In the above embodiments, the collision risk index refers to an indicator that quantitatively assesses the probability of collisions between vessels. The historical reference trajectory refers to the optimal navigation trajectory extracted from historical pilotage trajectories. The comprehensive risk score refers to the overall quantitative value of the path risk obtained by integrating multiple risk dimensions. Based on the deviation of the collision risk index of multiple candidate paths from the historical reference trajectory, the degree of weather impact, and the difficulty of tugboat coordination, the comprehensive risk score of multiple candidate paths is determined. This process comprehensively considers risk factors from multiple dimensions such as navigation safety, pilotage experience, environmental impact, and operational coordination, achieving a comprehensive quantitative assessment of path risk. Recommended path data is determined based on the ranking of the comprehensive risk scores of multiple candidate paths from low to high. This process provides pilots with the optimal navigation path selection, improving the scientific nature and safety of pilotage decisions.

[0069] In one embodiment, the collision risk index is calculated based on the nearest encounter distance and the time to the nearest encounter point. The historical reference trajectory is obtained by clustering pilotage trajectories from the past three years using five-dimensional features: starting point, ending point, vessel type, draft, and weather conditions. The weight of the comprehensive risk score can be adjusted according to the actual needs of the port. By constructing a pilotage experience knowledge base, personal experience is transformed into structured digital assets, reducing the risk of a generational gap in pilotage and shortening the training cycle for new pilots.

[0070] In some embodiments of this application, the method for planning the route of a ship further includes: obtaining a sudden event signal based on a preset event monitoring data source; determining the set of affected ships based on the dependency graph between the sudden event signal and a preset scheduling plan; and determining updated scheduling decision data based on the set of affected ships and the existing plans of unaffected ships.

[0071] In the above embodiments, the preset event monitoring data source refers to various data sources pre-configured for monitoring port emergencies. An emergency event signal indicates a signal indicating an emergency event occurring at the port that affects normal scheduling. The preset scheduling plan dependency graph refers to a pre-constructed graph structure representing the dependencies between ship scheduling plans. The affected ship set refers to the set of ships whose scheduling plans need adjustment due to the emergency event. By acquiring the emergency event signal from the preset event monitoring data source, automatic detection and identification of emergencies are achieved. The affected ship set is determined based on the emergency event signal and the preset scheduling plan dependency graph, quickly and accurately identifying the scope of the emergency's impact and its cascading propagation chain. Updated scheduling decision data is determined based on the affected ship set and the existing plans of unaffected ships. This process only performs local re-optimization on affected ships, keeping the plans of unaffected ships unchanged, minimizing the impact of the emergency on the overall port scheduling, and improving the port's ability to respond to emergencies.

[0072] In one embodiment, the preset event monitoring data sources include ship status data, meteorological sensor data, terminal operation data, and manually entered data. The types of emergencies include ship mechanical failures, insufficient visibility, port inspection delays, and berth equipment malfunctions. The target response time for local re-optimization is less than 3 minutes. Updated scheduling decision data is automatically synchronized to the ship agency system, pilot station, tugboat company, terminal, and maritime authorities. Through this mechanism, a rapid response can be achieved after an emergency occurs, automatically generating structured change notifications and pushing them to multiple stakeholders, significantly reducing the cascading impact and collaborative communication costs of emergencies.

[0073] The method provided in this application automatically generates scheduling plans that meet all constraints through an intelligent scheduling engine, significantly reducing reliance on manual labor and efficiently supporting port operations of more than 100 vessels per day. Through minute-level precise time slot allocation scheduling, it theoretically improves channel utilization by 15% to 20% in a single main channel scenario. Collision warnings and abnormal maneuver detection provide proactive safety warnings 15 to 20 minutes in advance, transforming passive safety into proactive prevention. Through combinatorial optimization and reinforcement learning collaborative solving, it overcomes the limitations of local optima in human experience, achieving joint optimization across berths, time windows, and channels. A pilotage experience knowledge base transforms personal experience into structured digital assets, reducing the risk of pilot generational gaps and shortening the training cycle for new pilots. A standard interface architecture enables low-intrusive integration and deployment without replacing existing systems, reducing promotion resistance and implementation risks. A rapid response mechanism for emergencies completes rescheduling and automatically notifies multiple parties within 3 minutes of an event being triggered, significantly reducing the cascading impact of emergencies.

[0074] In some embodiments of this application, determining the predicted flow rate of the passage area based on multidimensional situational data frames includes: determining the channel cross-sectional interaction characteristics and tidal cycle and periodicity characteristics of operation periods based on the multidimensional situational data frames; and determining the predicted flow rate for a preset future period based on the channel cross-sectional interaction characteristics and tidal cycle, weather forecast, and periodicity characteristics of operation periods.

[0075] In the above embodiments, the channel cross-section interaction characteristics refer to the mutual influence relationship of ship traffic between different channel cross-sections. The tidal cycle refers to the periodic rise and fall of seawater. The periodic characteristics of the operating period refer to the periodic variation of port ship operating volume over time. Based on multi-dimensional situational data frames, the channel cross-section interaction characteristics, tidal cycle, weather forecast, and operating period periodic characteristics are determined. The channel cross-section interaction characteristics characterize the spatial correlation of the channel network, while the tidal cycle and operating period periodic characteristics characterize the temporal regularity of port traffic, achieving a comprehensive extraction of the spatiotemporal characteristics of port traffic. Based on the channel cross-section interaction characteristics, tidal cycle, and operating period periodic characteristics, the predicted traffic flow for a preset future period is determined. This process fully utilizes the spatiotemporal regularity of port traffic, improving the accuracy of traffic flow prediction and achieving precise prediction of channel traffic conditions. Weather forecast specifically includes forecasts of elements directly affecting ship navigation safety and passage efficiency, such as wind speed, wind direction, visibility, wave height, ocean current speed and direction, and tidal height.

[0076] In one embodiment, a spatiotemporal prediction model based on a converter architecture extracts the aforementioned features. This model includes a spatial attention module for modeling the interactive effects of waterway cross-sections and a temporal attention module for capturing tidal cycles and operational period patterns. This enables rolling predictions of waterway traffic flow over the next 6 to 72 hours, while simultaneously outputting the predicted mean and confidence interval. A congestion warning is triggered when the predicted traffic flow exceeds a preset proportion of the waterway's capacity. In one specific embodiment, the preset proportion is 85%, and the confidence interval is a 95% confidence interval.

[0077] like Figure 2 As shown, in one embodiment of this application, a system for implementing the above-described ship path planning method is provided. The system includes the following functional modules:

[0078] The data fusion module 210 is used to access multi-source heterogeneous data, including Automatic Identification System (AIS) data, meteorological and hydrological data, electronic nautical chart data, ship traffic service radar data, and terminal production data. It performs anomaly cleaning, Kalman filtering correction, unified timestamp alignment, and coordinate system transformation on the multi-source heterogeneous data, outputting standardized multi-dimensional port situational data frames. The data fusion module 210 achieves unified integration and standardized output of multi-source data across the entire port area, forming a comprehensive port situational awareness capability covering ships, environment, and geography.

[0079] The traffic flow prediction module 220 is used to perform rolling predictions of ship traffic in each waterway for the next 6 to 72 hours based on multi-dimensional port situation data frames and a spatiotemporal sequence prediction network. The spatiotemporal sequence prediction network includes a spatial feature extraction layer for modeling the interaction effects of each cross section of the waterway network, and a temporal feature extraction layer for capturing the patterns of tidal cycles and peak operating periods. The traffic flow prediction module 220 is also used to output congestion warning information when the predicted traffic exceeds 85% of the waterway's capacity.

[0080] The multi-constraint scheduling optimization module 230 is used to construct a multi-constraint joint scheduling model based on ship arrival information, channel passage time window constraints, tidal time window constraints, weather forecast constraints, berth adaptation constraints, and channel capacity constraints. It then employs combinatorial optimization and reinforcement learning to collaboratively solve the multi-constraint joint scheduling model, outputting scheduling decision data including berth allocation and channel entry times. In small-scale scenarios, the multi-constraint scheduling optimization module 230 uses mixed-integer programming to find the global optimal solution; in large-scale scenarios, it uses policy gradient reinforcement learning to find a fast feasible solution and corrects constraint violations through post-processing, balancing the optimality and real-time performance of the scheduling scheme.

[0081] The pilotage route recommendation module 240 is used to construct a dynamic navigation cost field based on scheduling decision data, predicted flow, and multi-dimensional port situation data frames. It then performs incremental heuristic search on this dynamic navigation cost field to determine multiple candidate routes while satisfying constraints such as minimum turning radius, restricted area avoidance, and maximum cross-current. Recommended routes are then output and sorted from low to high based on comprehensive risk scores. The pilotage route recommendation module 240 also constructs a pilotage route knowledge base based on five-dimensional feature clustering of historical pilotage trajectories, providing historical experience support for route planning.

[0082] The collision risk warning module 250 is used to calculate the nearest encounter distance and the time to reach the nearest encounter point for each moving vessel in the port waters at a preset period. When the preset collision warning conditions are met, a graded collision warning message is generated. The collision risk warning module 250 is also used to monitor the turning angular velocity of each moving vessel in real time. When the turning angular velocity exceeds the historical statistical threshold of the same type of vessel, an abnormal maneuver warning message is generated and the warning message is pushed to the pilot's mobile terminal and the vessel traffic service workstation.

[0083] The emergency rescheduling module 260 is used to acquire emergency signals based on preset event monitoring data sources, determine the set of affected vessels based on a preset scheduling plan dependency graph, perform local rolling re-optimization on the affected vessel set, output updated scheduling decision data, and automatically synchronize the updated scheduling decision data to the external collaborative system. The target response time of the emergency rescheduling module 260 is less than 3 minutes.

[0084] The Pilotage Experience Knowledge Base 270 is used to structure and archive historical pilotage records. It clusters historical pilotage trajectories according to five-dimensional features: starting point, ending point, ship type, draft, and weather conditions, and extracts representative reference paths. It supports experience scenario retrieval and reuse, as well as training assistance for new pilots and digital inheritance of experience.

[0085] The dispatcher workbench 280 provides a visual interface for plan arrangement, a comprehensive display of port dispatch status, manual review and overriding of optimization suggestions, plan release and change management functions, and automated distribution of multi-party collaborative notifications.

[0086] The pilot mobile terminal 290 is used to receive recommended path data and collision warning information, and provides real-time path navigation, warning information display, tugboat status monitoring and operation record reporting functions, and supports offline degraded operation mode when the network is interrupted.

[0087] The Open Interface Platform 295 provides standard descriptive state transition interfaces and full-duplex communication protocol interfaces, supporting data interaction and business integration with existing port scheduling platforms, pilot station scheduling systems, maritime vessel traffic service systems, ship agency systems, and terminal production systems.

[0088] The aforementioned modules work together to achieve intelligent scheduling and pilotage decisions for port vessel arrivals and departures across the entire process, from data fusion, traffic forecasting, scheduling optimization, route planning, safety early warning, emergency response to experience accumulation. The system employs a hybrid architecture combining a cloud computing engine and local backup, automatically degrading to local mode in the event of a network outage to ensure the continuous availability of core functions.

[0089] like Figure 3 As shown, an embodiment of this application provides a ship path planning device 300, including: a data processing unit 310, a flow prediction unit 320, a scheduling decision unit 330, a cost field construction unit 340, and a path planning unit 350. The data processing unit 310 is used to acquire multi-source data of the target water area, including ship dynamic data, environmental data, and nautical chart data, and to determine a multi-dimensional situational data frame based on the multi-source data. The flow prediction unit 320 is used to determine the predicted flow of the passage area based on the multi-dimensional situational data frame. The scheduling decision unit 330 is used to determine the predicted flow of the ship in the multi-dimensional situational data frame based on the ship... Arrival information, channel passage time window constraints, tidal time window constraints, berth adaptation constraints, and channel capacity constraints are used to determine a multi-constraint joint scheduling model. Based on the multi-constraint joint scheduling model, combinatorial optimization and reinforcement learning are used to solve the problem collaboratively to determine scheduling decision data including berth allocation and channel entry time. The cost field construction unit 340 is used to determine a dynamic navigation cost field based on the scheduling decision data, predicted flow, and multi-dimensional situational data frames. The path planning unit 350 is used to determine candidate paths based on the dynamic navigation cost field while satisfying ship kinematic constraints, and to determine recommended path data according to preset risk scoring rules.

[0090] The ship path planning device 300 provided in this application embodiment integrates multi-source heterogeneous data through a data processing unit 310 to generate a spatiotemporally unified multi-dimensional situational data frame, achieving standardized integration of ship, environmental, and geographic information across the entire port area; predicts future channel traffic flow through a flow prediction unit 320, providing forward-looking data support for subsequent decision-making; performs multi-constraint joint optimization of berth allocation and channel traffic sequence through a scheduling decision-making unit 330, balancing the global optimality and real-time responsiveness of the scheduling scheme; superimposes scheduling decisions, predicted flow, and real-time environmental information into a dynamic navigation cost field through a cost field construction unit 340, realizing a multi-dimensional quantitative expression of navigation costs; and generates multiple candidate paths under the constraints of ship kinematics, ranks them by risk scoring, and outputs the optimal recommended path data through a path planning unit 350. The collaborative work of these units effectively reduces the reliance on manual experience in port scheduling, improves channel resource utilization and ship traffic efficiency, and provides quantitative safety decision-making basis for pilotage operations, realizing the intelligentization and standardization of port ship entry and exit scheduling and pilotage decision-making.

[0091] like Figure 4 As shown, an embodiment of this application provides an electronic device 400, including a processor 402 and a memory 404. The memory 404 stores a computer program, and the processor 402 implements the ship path planning method as described in any of the above embodiments when executing the computer program in the memory 404. Therefore, the electronic device 400 possesses all the beneficial effects of the ship path planning method as described in any of the above embodiments.

[0092] Embodiments of this application provide a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the ship path planning method as described in any of the above embodiments. Therefore, the readable storage medium possesses all the beneficial effects of the ship path planning method as described in any of the above embodiments.

[0093] One embodiment of this application addresses one of the highly complex extreme operating conditions in large hub ports: a scenario involving the special management and control of LNG carrier arrivals combined with sudden fog. This scenario simultaneously involves the rigid safety constraints of high-risk special vessels and the global plan disruptions caused by sudden weather. Manual scheduling often struggles to balance safety and efficiency, easily leading to channel congestion, operational delays, and even safety hazards. The system first completes the full-dimensional acquisition and preprocessing of multi-source data from the target waters. It integrates dynamic data of LNG carriers, including IMO number, vessel type, draft, and estimated arrival time, environmental data including hourly visibility forecasts, wind speed, and tidal data, as well as nautical chart data of the corresponding sea area. Kalman filtering is used to correct abnormal data caused by AIS signal loss and drift. After labeling the data reliability level, unified timestamp alignment and WGS-84 coordinate system transformation are performed, outputting standardized multi-dimensional situational data frames with a time resolution of 30 seconds. Based on this data frame, the Transformer model with channel spatial attention mask and tidal time bias is used to predict the ship traffic and passage risk of the main channel in the next 24 hours. When it is predicted that the traffic flow of LNG ships entering the port exceeds 70% of the channel capacity, a pre-schedule warning is triggered 24 hours in advance, so as to reserve sufficient time for subsequent planning and resource coordination.

[0094] During the multi-constraint joint scheduling phase, the system integrates the first-come-first-served soft constraint, the channel VTS time window hard constraint, the tidal time window hard constraint, the berth adaptation constraint, the channel capacity constraint, and the special control constraint for LNG ship one-way passage into the scheduling model. It adopts a strategy of joint solution using mixed integer programming and near-end strategy optimization. A total of 18 ships were scheduled in this case. The system called the mixed integer programming solver and performed global optimization with the objective function of "minimizing the total waiting time of ships plus absolute priority passage of LNG ships". In the final generated scheduling plan, a 20-minute safety buffer time is reserved before and after the LNG ship enters the port. The corresponding channel is set to one-way passage during the time period. The berthing and departure times and channel passage order of other ordinary ships are optimized and arranged around the operation window of the LNG ship. This not only strictly meets the safety control requirements of the LNG ship, but also minimizes the impact on the operation of other ships.

[0095] Based on the generated scheduling decision data, the system constructs a dynamic navigation cost field by combining predicted traffic flow and real-time multi-dimensional situational data. The traffic flow density field has a weight of 0.4, the meteorological risk field has a weight of 0.3, and the water depth margin field has a weight of 0.3. The weight of the traffic flow density field in the LNG ship passageway section is temporarily increased to 0.6 to enhance the perception of navigation risks in this area. Subsequently, an improved D* algorithm is invoked to generate three candidate pilotage routes for LNG ships, under the premise of satisfying the minimum turning radius constraint, restricted area avoidance constraint, and maximum crossflow constraint. The collision risk index and historical data for each route are calculated. The system calculates the deviation from the optimal pilotage trajectory, the meteorological risk level, and the difficulty of tugboat coordination, and outputs the optimal pilotage route with the lowest comprehensive risk score. At the same time, the system continuously calculates the nearest encounter distance and arrival time of all moving vessels in the port area every 30 seconds. It automatically verifies the compliance of the encounter situation in conjunction with the International Maritime Collision Avoidance Rules. When it is detected that the nearest encounter distance between the LNG carrier and other vessels is less than 1 nautical mile and the arrival time of the nearest encounter point is less than 20 minutes, a red warning is pushed to the pilot terminal and VTS command system 12 minutes in advance to reserve sufficient operation time for emergency avoidance.

[0096] When the system detects a sudden drop in visibility in the main channel to below 0.5 nautical miles through environmental data monitoring, it immediately activates the emergency adaptive rescheduling mechanism. Within 30 seconds, based on a preset scheduling plan dependency graph, it quickly identifies a group of 7 vessels affected by the fog. The rescheduling process strictly follows the priority principle of LNG carriers > deep-draft vessels > ordinary container ships > bulk carriers. Only the berthing and departure times and channel passage sequences of the affected vessels are partially and continuously optimized, while the berthing and departure times and channel passage sequences of the LNG carriers remain unchanged throughout the process, avoiding safety risks to LNG carrier operations caused by significant adjustments to the plan. After the rescheduling plan is generated, the system automatically pushes the updated scheduling decision data to the ship agency system, tugboat scheduling system, and terminal operation system through standardized interfaces, replacing the traditional manual telephone notification mode and achieving automated synchronization of multi-party collaboration.

[0097] Specifically, actual operational data from the above embodiments show that, under extreme scenarios involving special management and control of LNG ships combined with sudden fog, the system's response time for completing the entire process of rescheduling is only 2.5 minutes, which is 83% shorter than traditional manual scheduling; the main channel's traffic efficiency remains at 91%, and no large-scale channel congestion occurs; collision warnings are accurately pushed 12 minutes in advance, which can effectively avoid at least 3 potential ship encounter risks. This fully verifies that the system has both strong robustness and efficient response capabilities under complex scenarios with multiple constraints coupled and sudden disturbances superimposed, filling the technical gap in port scheduling systems under extreme conditions.

[0098] Specifically, this embodiment details the construction and implementation of the multi-constraint joint scheduling model of this application, providing mathematical modeling support for the global scheduling of ships entering and leaving ports and the optimization of waterway passage sequences. First, it clearly defines four types of core decision variables: berth allocation variable x_{ij} (x_{ij}∈{0,1}, indicating whether ship i is allocated to berth j), berthing time variable t_i^{berth} (t_i^{berth}∈R) + Define the planned berthing time of vessel i and the channel entry time variable t_i^{entry} (t_i^{entry}∈R). +Let y_{ik} be the planned entry time of vessel i into the main channel. In the multi-channel port scenario, a channel selection variable y_{ik} (y_{ik}∈{0,1}, indicating whether vessel i enters the port via channel k) is added. Based on the above variables, this embodiment constructs a complete system covering soft and hard constraints, including: first-come-first-served soft constraint, that is, when the expected arrival time of vessel i is earlier than that of vessel j, vessel i is given priority to berth if berth resources allow, and adjustments exceeding the priority require additional manual confirmation; channel VTS time window constraint (corresponding to channel passage time window constraint), which limits the time interval t_{i+1}^{entry}-t_i^{entry} ≥ Δt_min (e.g., Δt_min=10 minutes for the core channel of port A), and the passage sequence follows the VTS release rhythm; tidal time window constraint, which limits the berthing time t_i of specific vessel types such as oil tankers, ore carriers, and LNG carriers. ^{berth} must fall within the allowable time window [T_tide_start(d),T_tide_end(d)] calculated by the real-time tide model; berth adaptation constraints, limiting the ship's draft d_i ≤ the maximum allowable draft of the berth D_j, and the ship's total length LOA_i ≤ the effective length of the berth L_j×1.1; special ship control constraints, limiting the corresponding channel to one-way passage during the LNG ship's entry into the port, and other ships to retain a 20-minute safety buffer time T_LNG_buffer before and after its passage; channel capacity constraints, limiting the number of ships passing in the same channel at any time t to not exceed the channel capacity limit N_channel_max. Based on the above variables and constraints, a multi-constraint joint scheduling model is constructed to achieve joint optimization of berth allocation, berthing time, channel passage sequence, and channel selection. This solves the problems of berth and channel scheduling being disconnected and multi-constraint coupling making global optimization difficult in existing technologies. It can be adapted to the above-mentioned combined optimization and reinforcement learning collaborative solution mechanism. In a single main channel scenario, it can improve channel passage efficiency by 15% to 20%, reduce the constraint violation rate of the scheduling scheme to below 1%, and take into account the safety, compliance, and resource utilization of port operations.

[0099] In the claims, description, and accompanying drawings of this application, the term "plural" refers to two or more objects. Unless otherwise explicitly defined, the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used solely for the convenience of describing this application and simplifying the descriptive process, and are not intended to indicate or imply that the device or element referred to must have the described specific orientation, or be constructed and operated in a specific orientation. Therefore, these descriptions should not be construed as limitations on this application. The terms "connection," "installation," "fixing," etc., should be interpreted broadly. For example, "connection" can be a fixed connection between multiple objects, a detachable connection between multiple objects, or an integral connection; it can be a direct connection between multiple objects or an indirect connection between multiple objects through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in this application can be understood based on the specific circumstances described above.

[0100] In the claims, description, and accompanying drawings of this application, the terms "one embodiment," "some embodiments," "specific embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of this application. In the claims, description, and accompanying drawings of this application, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0101] The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for ship path planning, characterized in that, The ship's path planning method is executed by a computer, and the ship's path planning method includes: Acquire multi-source data of the target water area, including vessel dynamic data, environmental data and nautical chart data, and determine multi-dimensional situational data frames based on the multi-source data; Based on the multi-dimensional situational data frame, the predicted traffic flow in the passage area is determined; Based on the ship arrival information, waterway passage window constraints, tidal window constraints, weather forecast constraints, berth adaptation constraints, and waterway capacity constraints in the multi-dimensional situational data frame, a multi-constraint joint scheduling model is determined. Based on the multi-constraint joint scheduling model, combinatorial optimization and reinforcement learning are used to solve the scheduling decision data, which includes berth allocation and channel entry time. The dynamic navigation cost field is determined based on the scheduling decision data, the predicted flow, and the multi-dimensional situational data frame. Based on the dynamic navigation cost field, candidate paths are determined under the constraints of ship kinematics, and recommended path data is determined according to the preset risk scoring rules.

2. The ship path planning method according to claim 1, characterized in that, The step of determining the multi-dimensional situation data frame based on the multi-source data includes: Based on the aforementioned ship dynamic data, abnormal data is identified; Based on the abnormal data, determine the corrected ship dynamic data and the data reliability level; Based on the corrected ship dynamic data, the environmental data, and the nautical chart data, a unified timestamp alignment and coordinate system transformation process is performed to determine the multidimensional situational data frame.

3. The ship path planning method according to claim 1, characterized in that, Also includes: Based on the multidimensional situational data frame, determine the nearest encounter distance and the time of arrival at the nearest encounter point for each moving vessel in the target waters; When the nearest encounter distance is less than a preset safe distance and the time to reach the nearest encounter point is less than a preset warning time, a collision warning message is generated; And / or, Based on the multidimensional situational data frame, the turning angular velocity of each moving vessel is determined. When the turning angular velocity exceeds a preset threshold, an abnormal maneuver warning message is generated.

4. The ship path planning method according to claim 1, characterized in that, The method, based on the multi-constraint joint scheduling model, employs combinatorial optimization and reinforcement learning to collaboratively solve the problem, determining scheduling decision data including berth allocation and channel entry times, including: The solution method is determined based on the comparison between the number of ships to be dispatched and the preset ship number threshold. When the number of ships to be scheduled is less than the preset ship number threshold, the global optimal solution is determined according to the multi-constraint joint scheduling model; When the number of ships to be scheduled is greater than or equal to the preset ship number threshold, a fast feasible solution is determined according to the multi-constraint joint scheduling model, and the scheduling decision data after constraint violation repair is determined according to the fast feasible solution.

5. The ship path planning method according to claim 1, characterized in that, The step of determining the dynamic navigation cost field based on the scheduling decision data, the predicted traffic flow, and the multi-dimensional situational data frame includes: Based on the predicted traffic flow, determine the real-time traffic flow density field; Based on the environmental data, the meteorological risk field and water depth margin field are determined; Based on the nautical chart data, the real-time traffic flow density field, the meteorological risk field, and the water depth margin field, the superimposed basic cost field is determined; Based on the normalized weighted results of the basic cost field and the channel location, time, risk level, and water depth margin, the dynamic navigation cost field that is dynamically updated over time is determined.

6. The ship path planning method according to claim 1, characterized in that, The step of determining candidate paths based on the dynamic navigation cost field, while satisfying ship kinematic constraints, includes: Based on preset ship kinematic constraints, path search constraints are determined. These preset ship kinematic constraints include minimum turning radius constraints, restricted area avoidance constraints, and maximum cross current constraints. Based on the dynamic navigation cost field and the path search constraints, a dynamic path search is performed, and the path cost of the affected local area is updated and replanned according to the changes in the waterway situation, thereby determining multiple candidate paths.

7. The method for ship path planning according to any one of claims 1 to 6, characterized in that, The step of determining the recommended path data according to the preset risk scoring rules includes: Based on the degree of deviation of the collision risk index of the candidate paths from the historical reference trajectory, the degree of meteorological influence, and the difficulty of tugboat coordination, a comprehensive risk score for the candidate paths is determined. The recommended path data is determined based on the ranking of the comprehensive risk scores of the candidate paths from low to high.

8. The method for ship path planning according to any one of claims 1 to 6, characterized in that, Also includes: Acquire emergency event signals based on preset event monitoring data sources; Based on the dependency diagram between the emergency event signal and the preset scheduling plan, the set of affected vessels is determined; The updated scheduling decision data is determined based on the existing plans of the affected vessel group and the unaffected vessels.

9. A path planning device for ships, characterized in that, include: The data processing unit is used to acquire multi-source data of the target water area, including ship dynamic data, environmental data and nautical chart data, and to determine multi-dimensional situational data frames based on the multi-source data. The traffic prediction unit is used to determine the predicted traffic flow in the passage area based on the multi-dimensional situational data frame. The scheduling decision unit is used to determine a multi-constraint joint scheduling model based on the ship arrival information, channel passage time window constraints, tidal time window constraints, meteorological forecast conditions constraints, berth adaptation constraints, and channel capacity constraints in the multi-dimensional situational data frame; and based on the multi-constraint joint scheduling model, to determine scheduling decision data including berth allocation and channel entry time by using combinatorial optimization and reinforcement learning to solve the problem. The cost field construction unit is used to determine the dynamic navigation cost field based on the scheduling decision data, the predicted traffic, and the multi-dimensional situation data frame. The path planning unit is used to determine candidate paths based on the dynamic navigation cost field while satisfying ship kinematic constraints, and to determine recommended path data according to preset risk scoring rules.

10. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method according to any one of claims 1 to 8.

Images