A method for ship scheduling in maritime alliances based on dynamic knowledge graphs

Abstract

This invention provides a method for ship scheduling in maritime alliances based on dynamic knowledge graphs, relating to the field of maritime logistics management technology. The scheduling and sharing method of this invention specifically includes: converting the future voyage plans of each alliance member into daily graph snapshots with ports as nodes and vessels at sea as edges, constructing a local dynamic knowledge graph; merging the daily graph snapshots through temporal fusion to generate a globally fused dynamic knowledge graph for the alliance; performing constraint-based cross-time dynamic optimization based on static optimization and iterative search to minimize a global dynamic cost function; performing pruning verification of the overall scheduling plan, calculating the baseline cost of the dynamically optimized plan, and feeding back the alliance ship scheduling plan.

Description

Technical Field

[0001] This invention relates to the field of maritime logistics management technology, and more specifically, to a method for scheduling ships in maritime alliances based on dynamic knowledge graphs. This method is particularly suitable for realizing joint scheduling of ships among multiple shipping companies and dynamic optimization of container space resources. Background Technology

[0002] Global container shipping is a capital-intensive industry characterized by high market volatility and operating costs. To enhance service coverage, optimize asset utilization, and respond to market changes, shipping companies widely collaborate by forming maritime alliances. Under this alliance model, member companies share route networks, vessel capacity, and port facilities, improving overall operational efficiency through coordinated operations. However, the collaborative operation of alliances, particularly vessel scheduling and space sharing, faces significant technological challenges.

[0003] Vessel scheduling within shipping alliances is an extremely complex dynamic optimization problem involving multiple objectives and constraints. Alliances manage massive fleets with routes covering major ports worldwide, involving a wide variety of vessel call sequences. Traditional vessel scheduling largely relies on pre-established, long-term static schedules. However, actual navigation is fraught with uncertainties, such as port congestion, changes in berth windows, severe weather, traffic control, and sudden changes in the technical condition of vessels. These real-time disturbances can rapidly cause schedules to deviate from the planned schedule, creating a cascading effect that impacts the punctuality of multiple routes and vessels within the alliance. Existing technologies often exhibit lag in response to these high-frequency dynamic disturbances, making rapid and effective coordinated adjustments across the entire alliance difficult.

[0004] Space sharing is one of the core mechanisms of alliance operations, aiming to maximize the overall load factor of the alliance fleet, reduce ineffective empty container repositioning and empty voyages, thereby reducing energy consumption per unit of transportation. Achieving efficient space sharing presupposes that alliance members possess highly timely information exchange capabilities. However, in practice, alliance members typically use their own independent, heterogeneous IT systems to manage their own vessel, cargo, and contract data. This decentralized data management model leads to severe information silos. When a member needs to find shared space within the alliance for excess cargo, or when a member's vessel has surplus space, it is often difficult to accurately match real-time space supply and cargo demand across systems and companies in a short period. Current information exchange mechanisms often rely on batch processing or manual coordination, whose timeliness and granularity cannot meet the demands of the rapidly changing shipping market for refined management.

[0005] Therefore, the scheduling and sharing problem within shipping alliances is essentially a data-level challenge. Information such as ship dynamics, port status, cargo demand, and weather and sea conditions is characterized by its massive volume, heterogeneity, high dynamism, and strong correlation. Current technologies lack an effective framework to uniformly describe and process this dynamic time-series data scattered across different members and systems, making it impossible to effectively model and reason about the complex time-varying relationships between ships, routes, ports, and cargo from a global perspective. Therefore, how to integrate heterogeneous dynamic data within the alliance, construct a unified global operational view, and on this basis achieve agile, collaborative, and energy-optimized joint scheduling and space sharing is a pressing technical challenge that needs to be addressed in this field. Summary of the Invention

[0006] This invention provides a method for ship scheduling in maritime alliances based on dynamic knowledge graphs, the method comprising:

[0007] The future navigation plans of each alliance member are converted into daily graph snapshots with ports as nodes and ships at sea as edges, thus constructing a local dynamic knowledge graph;

[0008] By merging the daily graph snapshots through time-series fusion and performing data source annotation on the attributes of nodes and edges, a global fusion dynamic knowledge graph of the alliance is generated.

[0009] Constraint-based static optimization checks whether the daily global graph snapshots in the globally fused dynamic knowledge graph meet physical constraints, including port berth capacity constraints, and resolves conflicts; it also checks for cargo-ship type mismatches and generates mismatch annotation attributes.

[0010] Constraint-based time-series dynamic optimization is performed by iteratively searching to minimize a global dynamic cost function.

[0011] The overall scheduling plan is pruned and verified, the baseline cost of the dynamic optimization plan is calculated, and the average failure cost is calculated by simulating route failures and their cascading effects. The feasibility of the plan is determined based on the difference between the average failure cost and the baseline cost, and feedback is given to the alliance vessel scheduling plan.

[0012] Constraint-based static optimization, when checking port berth capacity constraints, includes: calculating the total daily berthing demand. :

[0013]

[0014] in For the port, For date, for Hong Kong Total daily docking demand This represents the number of ships that have docked that day. This represents the number of ships scheduled to arrive that day.

[0015] when At that time, it was determined that a berth conflict had occurred, among which For the port berth capacity.

[0016] Conflict resolution includes: calculating the number of ships that need to be delayed. ;

[0017] And to assemble the ships scheduled to arrive that day. ships in Calculate scheduling priority,

[0018] From the sorted Starting with the lowest priority ship in the list, select in order. The ships were grouped into the delayed set. its estimated arrival time Modified to .

[0019] Checking for cargo-ship type mismatches and generating mismatch annotation attributes includes: calculating the required cargo type for shipment.

[0020]

[0021] in For member index, For the total number of members, For the type of goods, For the quantity of goods, For members The planned inventory collection of goods, for Hong Kong Daily demand for cargo types; calculation of available vessel type supply:

[0022]

[0023] in Supply for available vessel types, For the types of vessels that have already docked, The type of vessel arriving on the same day after static optimization processing;

[0024] Calculate the set of mismatch types ;

[0025] when When it is a non-empty set, at the port Add mismatch annotation attributes to nodes , This represents the difference set.

[0026] Global dynamic cost function Defined as:

[0027] in, For an evaluation The Sky Project for Global dynamic cost; Cost of type mismatch; Cost of empty flight; To cover maintenance and repair costs; To manage volatility costs; This is a preset penalty conversion factor used to convert the original costs of each entity into a uniform, additive penalty metric.

[0028] Type mismatch cost Defined as:

[0029]

[0030] in Total number of days For port assembly, For static optimization in the port Added mismatched annotation attributes to nodes; maintenance conflict costs Defined as:

[0031] in For ship maintenance plan data set, For ships, For port maintenance, For maintenance date, For the maintenance satisfaction criterion function, when the ship exist The sun never stopped at Hong Kong time, A penalty is generated. .

[0032] Empty flight cost Defined as:

[0033]

[0034] in for exist The edge of the day, The route distance of the flight, This is the empty-run distance discrimination function, which is based on the actual cargo load of the ship. With rated cargo capacity The comparison is used to make a judgment; scheduling volatility costs Defined as:

[0035] in For ships, For the total number of ships in the alliance For scheduling difference discrimination function, This is the statically optimized plan sequence.

[0036] The iterative search employs the simulated annealing algorithm and includes: based on The cost structure triggers route redirection or maintenance scheduling adjustment operations to generate candidate solutions. ;

[0037] Will The changed parts are re-entered into the statically optimized PCC check and parsing and CTM check and annotation process for static re-verification;

[0038] Based on the acceptance probability of simulated annealing Decide whether to accept ,

[0039] ,in For cost difference, This is the current temperature.

[0040] The simulated route failure and its cascading effects include: a. randomly selecting a route alongside an in-flight vessel. And apply a random delay of days. To modify it :

[0041] in This is the original estimated arrival time. The number of days is a random delay.

[0042] b. Add all affected elements from the modified map in section a. in , for The daily global map snapshot is re-entered into the statically optimized PCC check and parsing and CTM check and annotation process for re-verification, resulting in a failed map. .

[0043] Determining the feasibility of the plan includes: calculating the baseline cost. , For dynamically optimized plans; through Sub-route failure simulation to calculate average failure cost. ; Calculate robust impact score ;

[0044] when Less than or equal to the preset robustness threshold At that time, the plan was deemed feasible and marked as .

[0045] This invention also provides a maritime alliance vessel scheduling system based on dynamic knowledge graphs, the system comprising:

[0046] Local dynamic knowledge graph construction module: Converts the future voyage plans of each alliance member into daily graph snapshots with ports as nodes and ships at sea as edges, and constructs a local dynamic knowledge graph;

[0047] Global Fusion Dynamic Knowledge Graph Task Module: Merges the daily graph snapshots through time-series fusion, and performs data source tracing and annotation on the attributes of nodes and edges to generate a global fusion dynamic knowledge graph for the alliance;

[0048] Static optimization module: Constraint-based static optimization checks whether the daily global graph snapshot in the globally fused dynamic knowledge graph meets physical constraints, including port berth capacity constraints, and resolves conflicts; it also checks for cargo-ship type mismatches and generates mismatch annotation attributes.

[0049] Dynamic optimization module: Performs constraint-based cross-time dynamic optimization by iteratively searching to minimize a global dynamic cost function;

[0050] Pruning module: Performs pruning verification of the overall scheduling plan, calculates the baseline cost of the dynamic optimization plan, calculates the average failure cost by simulating route failures and their cascading effects, determines the feasibility of the plan based on the difference between the average failure cost and the baseline cost, and feeds back the alliance vessel scheduling plan.

[0051] 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 scheduling ships in maritime alliances based on dynamic knowledge graphs.

[0052] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for scheduling ships in maritime alliances based on a dynamic knowledge graph.

[0053] The maritime alliance vessel scheduling method based on dynamic knowledge graphs provided by this invention has significant beneficial effects. This invention, through a local dynamic knowledge graph construction step, transforms the heterogeneous future voyage plans of each member within the alliance into a unified graph structure with ports as nodes and vessels at sea as edges. This standardized graph model is the foundation for achieving data interoperability and subsequent computation within the alliance, solving the problems of data heterogeneity and information silos in existing technologies. In the alliance-wide fusion dynamic knowledge graph generation step, this invention employs a data source annotation method. While creating a global view of the alliance, it ensures that the ownership information of each alliance member's assets and inventory is preserved through attribute aggregation annotation.

[0054] The present invention adopts a technical process of static optimization followed by dynamic optimization. Static optimization acts as a preprocessor and filter to resolve physical conflicts. The dynamic optimizer can run on a graph sequence without immediate conflicts and with diagnosis, allowing it to concentrate computing resources to specifically solve complex problems that must be coordinated on a global time scale, such as empty flight distance.

[0055] Furthermore, this invention introduces robust pruning verification of the overall scheduling plan, addressing the shortcomings of traditional optimization methods that only pursue theoretical optimality while neglecting the vulnerability of the plan. Through simulation of route failures and cascading effects, this method quantifies the average failure cost of the plan when encountering a single point of failure. Through acceptance decision-making, this invention ensures that the final output of the alliance's overall scheduling plan is not only close to optimal in cost but also possesses robustness against real-world unforeseen events. Attached Figure Description

[0056] Figure 1 This is a flowchart of the maritime alliance vessel scheduling process based on dynamic knowledge graphs, as presented in this invention. Detailed Implementation

[0057] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0058] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this application, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0059] Additionally, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that practice can be carried out without these specific details.

[0060] This embodiment provides a method for ship scheduling in maritime alliances based on dynamic knowledge graphs, the method comprising:

[0061] The future navigation plans of each alliance member are converted into daily graph snapshots with ports as nodes and ships at sea as edges, thus constructing a local dynamic knowledge graph;

[0062] By merging the daily graph snapshots through time-series fusion and performing data source annotation on the attributes of nodes and edges, a global fusion dynamic knowledge graph of the alliance is generated.

[0063] Constraint-based static optimization checks whether the daily global graph snapshots in the globally fused dynamic knowledge graph meet physical constraints, including port berth capacity constraints, and resolves conflicts; it also checks for cargo-ship type mismatches and generates mismatch annotation attributes.

[0064] Constraint-based time-series dynamic optimization is performed by iteratively searching to minimize a global dynamic cost function.

[0065] The overall scheduling plan is pruned and verified, the baseline cost of the dynamic optimization plan is calculated, and the average failure cost is calculated by simulating route failures and their cascading effects. The feasibility of the plan is determined based on the difference between the average failure cost and the baseline cost, and feedback is given to the alliance vessel scheduling plan.

[0066] This embodiment details the member companies within the alliance of this invention. How to transform its future flight plans into a series of standardized, day-segmented local dynamic knowledge graphs ( The process outputs members. Local dynamic knowledge graph Local dynamic knowledge graph Its definition includes An ordered set of image snapshots: ,in Member of the alliance In the future Daily plan status diagram.

[0067] S1-1 Data Input and Preprocessing, for Building The system from alliance members Three types of planning data are obtained from the internal operations system: global static data ,member Static ship data ,member Future dynamic planning data .

[0068] Global static data Including all physical ports covered by the alliance set Any two ports and Standard route distance between (Unit: nautical miles) Standard route distance Fixed and unchanging.

[0069] member Static ship data Including members owned ships set Ships Type The categories include, but are not limited to: refrigerated ships (labeled as Reefer), dangerous goods ships (labeled as DG), and general container ships. Rated cargo capacity (Unit: TEU). Ships Historical route data This means using a set of (route, frequency) key-value pairs to represent historical route data. (Ships) average fuel consumption rate (Unit: tons / day). The vessel described. average fuel consumption rate Based on the acquisition of historical flight route data that has already been completed.

[0070] member Future dynamic planning data Including members Future voyage plans collection Each voyage is planned Each is defined by a sequence of tuples:

[0071]

[0072] in, Indicates the vessel performing this voyage , Indicates the first in the voyage One port of call, Indicates the planned arrival at the port. The date (day). Indicates plan to depart from port The date (day). Indicates plan to depart from port Fuel load (tons) at that time. Member A collection of future inventory plans. Each batch of goods... Defined as:

[0073]

[0074] Indicate the type of goods. Indicates the quantity of goods (TEU). Indicates the port of origin. Indicates the port of destination. Indicates that the goods are in Prepared date (days).

[0075] S1-2 Generate a local dynamic knowledge graph The system executes a loop. Traverse from 1 to Generate a snapshot of the graph for each day. , The specific number of days can be set to 28, 29, 30, or 31 depending on the month.

[0076] S1-2-1 Initialize Node Set Node set use The global port set in [the context]. This set is in [the context of]... arrive It remains unchanged during the period.

[0077] S1-2-2 Calculate Node Attributes ,exist In the daily cycle, the system traverses Each port in and generate dynamic attributes for it. The dynamic attributes Including planned berthing of ships and planned goods inventory collection

[0078] Generate a list of vessels scheduled to dock. :

[0079] initialization Traversal Each voyage plan : ,if and ,Right now The day is within the planned stop window, from Searching for ships static information ,and .

[0080] Generate a planned goods inventory set :

[0081] initialization Traversal Each batch of goods :

[0082] ,judge exist Is the sun at the port? :

[0083] Case 1 (Pending): IF AND AND Not yet Any voyage in The day before yesterday (inclusive) (Day) Loading.

[0084] Case 2 (Transfer): IF yes Transit port AND Sun in exist Between the planned unloading date and the planned loading date.

[0085] In either case, Add to To improve efficiency, the system will process identical... of Accumulate, so that There is only one record for each type of goods.

[0086] S1-2-3 Generate Edge Set and edge attributes ,exist In the cycle of days, traverse Each voyage plan This generates edge relationships between ships in transit, i.e., nodes.

[0087] Establish edges traversal segment .

[0088] IF AND ,Right now The day is within the planned sailing time window:

[0089] exist Establish a directed edge in the middle ,in yes The corresponding vessel. If multiple voyages (Corresponding ships) )exist Daily average The flight segment will establish multiple parallel edges. and .

[0090] Calculate edge attributes ,for Every edge established daily Calculate its attribute information, including flight path distance. Types of ships Estimated arrival time The estimated position of the vessel Estimated remaining oil volume of the vessel .

[0091] from Static route distance obtained directly from the source ,from Ships acquired from China Type Estimated arrival Time ,exist During the construction phase, it is estimated that it will reach... Time equal to the plan Right now Ships exist Predicted position of the day This refers to the percentage of the remaining route. The estimated location... The calculation process is as follows:

[0092] Assuming the ship travels at a constant speed within the voyage segment, calculate the total planned voyage days. and number of days sailed :

[0093]

[0094] Percentage of remaining routes for:

[0095] Ships exist Estimated remaining oil volume for the day Remaining fuel based on In and In Calculation yielded:

[0096] Calculation from Departure to Daily cumulative fuel consumption :

[0097]

[0098] but Estimated remaining oil volume for the day for:

[0099]

[0100] in yes In the plan Fuel level at departure.

[0101] Finally, a complete local dynamic knowledge graph is generated and output. .Should Member An operational plan is a structured, computable graphical representation managed by the members.

[0102] The following describes the construction members using an exemplary embodiment. of

[0103] This embodiment simulates alliance members. Constructing its next 3 days ( to Local dynamic knowledge graph The process.

[0104] S1-1 Data Input and Preprocessing, Company The system obtained the following plan data:

[0105] Global static data ( Including port collection : Distance of flight route : nautical miles nautical miles In the sea.

[0106] member Static ship data Including ship collection : Ship attributes, as shown in Table 1:

[0107]

[0108] member Future dynamic planning data ( This includes a collection of future voyage plans. :

[0109] (S1 voyage): ( , ( =PA, =1, =2, =500), ( =PB, =5, =5, =440)) 、 (S2 voyage): ( , ( =PA, =2, =3, =400), ( =PC, =7, =7, =340))

[0110] Future Goods Inventory Planning Collection :

[0111] :( , ='General', =50, =PA, =PB, =1) :( , ='Reefer', =20, =PA, =PC, =2) :( , ='General', =30, =PB, =PC, =4).

[0112] S1-2 Generate a local dynamic knowledge graph The system executes a loop. Iterate from 1 to 5

[0113] when Time: Generation .

[0114] S1-2-1 Node Set : ;

[0115] S1-2-2 Node Attributes : : Traversal , =PA and (Right now ) was established. .

[0116] : Traversal , =PA, (Right now ) is established, and Not yet loaded. .

[0117] Other port attribute sets are empty.

[0118] S1-2-3 edge set :

[0119] Traversal (PA PB): (Right now This statement is invalid.

[0120] Traversal (PA PC): (Right now This statement is invalid.

[0121] .

[0122] when Time: Generation .

[0123] Node set : Node attributes : : Traversal : =PA and Established. Traversal : =PA and (Right now ) was established.

[0124] .

[0125] : Traversal : =PA, (Right now ) is established, and Not yet loaded.

[0126] Edge set : Traversal : (Right now This statement is invalid.

[0127] Traversal : (Right now This statement is invalid.

[0128] .

[0129] when Time: Generation

[0130] Node set :

[0131] Node attributes : : Traversal : Not true. Traversal : Established. .

[0132] : Traversal :exist Japanese (S2) Loading, the condition of not being loaded is not met. .

[0133] Edge set : Traversal (PA PB): (Right now ) and (Right now Established. (Establishment of borders) Traversal (PA PC): (Right now This statement is invalid. .

[0134] edge attributes :for : In the sea, , , (Remaining percentage):

[0135] sky;

[0136] sky;

[0137] (66.7%)

[0138] (Remaining fuel):

[0139] ton;

[0140] ton.

[0141] Next, this embodiment will describe in detail the alliance-wide fusion dynamic knowledge graph. The generation method. Globally integrated dynamic knowledge graph. based on Each alliance member's local dynamic knowledge graph By combining time-series fusion with data source tracing and annotation, a unified global [database / database] is constructed. Specifically, local fusion dynamic knowledge graph based on A collection of local dynamic knowledge graphs ,in The global fusion dynamic knowledge graph in this embodiment Its structure is . This is a daily snapshot of the global graph, representing the alliance's state in the future. The complete daily plan status. Constructing a globally integrated dynamic knowledge graph. include:

[0142] S2-1: Timing Fusion Main Process Control: Based on As the basic unit of time, execution is based on... arrive The main loop ensures indivual Aligned by day.

[0143] S2-2: Daily Global Graph Snapshot Generation: Every day in the main loop ,Will Local map snapshot Merge into a global graph snapshot The daily global graph snapshot The generation is accomplished through two sub-steps: the union of graph elements and the aggregation of attribute labels.

[0144] Specifically, the S2-1 timing fusion main process control is to... A single time series graph ( ) are merged into a global time series graph ( The outer loop algorithm ensures the synchronization of the fusion process in the time dimension.

[0145] The input to the main process of time-series fusion is A collection of local dynamic knowledge graphs Its output is a globally integrated dynamic knowledge graph for the alliance. .

[0146] initialization (An empty, ordered set). The outer loop execution includes:

[0147] FOR TO DO:

[0148]

[0149]

[0150] ( (Indicates orderly addition)

[0151] END FOR

[0152] RETURN

[0153] The outer loop algorithm ensures The first in Snapshot of the image It is and only is indivual The Middle Snapshot of the image The fusion result.

[0154] Specifically, S2-2) Daily Global Graph Snapshot Generation Algorithm (), it receives As input, generate and its global attributes and .

[0155] S2-2-1 Node Set The determination

[0156] all Share the same Derived fixed and unchanging physical port node set Therefore, global graph snapshot node set No calculation required, simply reuse the existing data. .

[0157]

[0158] S2-2-2 Global Edge Set Fusion

[0159] Global edge set Through the Local edge set The union operation yields:

[0160]

[0161] Here, each edge By ship Unique identifier. Within maritime alliances, each vessel... Belongs to only one member (Right now and This physical constraint guarantees and ( The sets () are disjoint. Therefore, this invention utilizes the union operation. It can gather all the ships (sides) of the alliance members without conflict. middle.

[0162] S2-2-3 Global Node Attributes aggregation

[0163] This invention does not simply merge global node attributes during aggregation, but rather maintains data traceability while aggregating, that is, it clearly distinguishes between them. Which part of the port's assets (ships docked) and inventory (cargo) belong to which member? .to this end, China Port global node attributes Defined as a member Local attribute collection for indexing .

[0164] Global node attributes The aggregation specifically includes:

[0165] initialization .

[0166] FOR TO DO:

[0167]

[0168]

[0169]

[0170] END FOR

[0171] RETURN

[0172] The aggregation-annotation structure of the present invention ensures exist Hong Kong is subject to distinction of ownership. For example, when visiting... At that time, it is possible to know precisely what Company 1 is doing. Hong Kong has The inventory and Company 2 have This allows for precise matching and constraints between companies, such as Company 1's ships. Priority matching Goods.

[0173] S2-2-4 Global Edge Attributes annotation

[0174] Similar to node attributes, global edge attributes must also be annotated to indicate which member the edge belongs to. .

[0175] Global edge attribute annotation specifically includes:

[0176] Step a: Locate the vessel Define the ship ownership lookup function for the member i. This function queries ;

[0177]

[0178] Step b: Retrieve and enhance local attributes from Search Corresponding local edge attributes .

[0179]

[0180] Step c: Construct a global property containing the member identifier i;

[0181]

[0182] Step d Return

[0183] The alliance-wide fusion dynamic knowledge graph generation method provided in this embodiment will Individual, decentralized, and independent local plans ( This is transformed into a single, time-synchronized, globally dynamic knowledge graph. Key asset ownership information was retained, and the ownership of vessels and cargo inventories from different members was not lumped together. Instead, the ownership of each asset (vessel) and inventory (cargo) was clearly maintained. ).

[0184] Next, this embodiment uses a daily analysis and correction approach to identify and resolve issues without crossing the time dimension. The system identifies physical conflicts and resource mismatches on a given day and outputs a statically optimized plan sequence that resolves the day's physical conflicts. .in yes After today ( (Daily) Optimized and corrected version.

[0185] The static optimization specifically includes:

[0186] S3-1: Initialize static constraint parameters, define the required global static parameters, including the physical port berth capacity. Static priority weights for ship types .

[0187] S3-2: Static optimization of main process control: Execution is based on days as the basic time unit. arrive The main cycle, daily Perform static optimization.

[0188] S3-3: Port berth capacity constraint check and analysis, abbreviated as PCC check and analysis: In During the daily cycle, check Each port in China Does the total number of ships scheduled to dock exceed its physical capacity? If the priority exceeds the limit, then the lower priority vessels will be deferred according to their vessel priority. .

[0189] S3-4: Cargo-Vessel Type Mismatch Inspection and Marking (CTM Inspection and Marking): In During the daily cycle, check Each port in China Goods awaiting shipment Required cargo type Whether it can be booked by ships that dock or arrive on the same day. Satisfaction is achieved. If not, then at the port. Add a mismatch annotation to the node attributes .

[0190] Specifically, to perform static optimization, two key global static parameters are first introduced, which serve as... A portion of it was loaded.

[0191] Port berth capacity Ports are represented by integers. In one The maximum number of vessels that can be handled simultaneously (both berthing and loading / unloading) within a day (24 hours).

[0192] Ship type priority weight : for ship type A static, dimensionless weight value is assigned to calculate the ship scheduling priority. In a preferred embodiment, this is set... (Refrigerated ships, highest priority). (Dangerous goods vessel) (Ordinary ship).

[0193] S3-2 Static optimization of main process control, in In each iteration of the day, from Extract the global graph snapshot for the current day. As the current processing object, this process consists of two serial steps, S3-3 and S3-4.

[0194] After PCC checking and parsing in S3-3, an intermediate image snapshot is generated; this intermediate image snapshot then undergoes CTM checking and annotation in S3-4, finally generating... Daily static optimization graph snapshot The system will Store in order until The loop has ended.

[0195] Specifically, port berth capacity constraint checking and analysis, i.e., PCC checking and analysis, for Each port in First, check its total parking demand for the day. The total berthing demand consists of two parts: the number of vessels that have already berthed that day. and the number of ships scheduled to arrive that day By querying global node attributes Extract and merge all members Planned docking vessel collection Calculate its base: By querying global edge set Filter out all destinations as And estimated arrival time Exactly Japanese ships at sea Forming a fleet of arriving ships And calculate its base:

[0196] Total docking demand for:

[0197] Conflict checking and resolution will Ports as defined in S3-1 Physical berth capacity Compare them.

[0198] if ,but Hong Kong There were no berth conflicts during the day, so no action was required.

[0199] if Then determine Hong Kong A berth conflict occurred. Calculate the number of vessels that need to be delayed. .

[0200] To assemble the ships scheduled to arrive on the same day China has determined For vessels requiring delays, this invention employs a phased conditional judgment method, which strictly adheres to the business logic that prioritizes vessel type over voyage position. This method involves... This can be achieved through multi-level sorting:

[0201] First-level sorting (primary condition): by ship type Sort:

[0202] according to Ship type priority weighting will ships in according to Sort in ascending order.

[0203] 'General' (0.5) < 'DG' (0.8) < 'Reefer' (1.0). Therefore, 'General' (ordinary ships) will be ranked first, becoming the lowest priority ship group.

[0204] Secondary sorting (sub-condition): by flight path position Sort: In each order determined by the first-level sort Within the type group, based on the estimated position of the vessel. Sort in descending order.

[0205] The larger the value, the more distance remains, meaning a lower completion rate. Therefore, according to... Sort in descending order, meaning that in the same order... Under this type, the flight completion rate is the lowest (i.e. The largest ship has the lowest priority and is listed first.

[0206] After the above two-stage sorting process The ships in the set are arranged into a lexicographical list. The top of this list (i.e., the lowest priority, the ship most deserving of delay) is a... For 'General' and The largest vessel (i.e., an ordinary ship that has just set sail).

[0207] At the very bottom of the list (i.e., the highest priority, the ship least likely to be delayed) is a For 'Reefer' and The smallest vessel (a refrigerated ship about to arrive at the port).

[0208] From the sorted Starting from the top of the list (i.e., the lowest priority ship), select in sequence. The ships were grouped into the delayed set. .

[0209] Revise To generate This refers to an intermediate snapshot. For Each edge in Set its edge attributes Modified to That is, postponed to Arrival date.

[0210] Cargo-Vessel Type Mismatch Inspection and Labelling, also known as CTM inspection and labelling, identifies... Port on that day Goods to be shipped With available vessels The issue is a type mismatch. For Each port in Check the types of goods required for shipment on that day. Query traversal global node attributes All members of Set, and for all Find the union of (goods types):

[0211]

[0212] Query Hong Kong Available ship types for Japan The supply It consists of two parts: the types of vessels already docked. : 2. Type of vessel arriving on the same day This set is the result of processing S3-3. still exist Arrival in Japan ( ) ships of gather. ,in (This is the edge set after S3-3 processing)

[0213] Available ship types supply for:

[0214]

[0215] Mismatch checking and labeling, calculation and The difference between the sets of the two sets yields the set of mismatch types. :

[0216] if If the set is non-empty, meaning there are types of goods to be shipped that cannot be satisfied by the available ship types for the day, the system will... port Add a new mismatch annotation attribute to the node. and will Assign the value to this property.

[0217] Through PCC parsing in S3-3, this method quantifies the port congestion problem abstracted in maritime scheduling into... and A comparison. By introducing a range-based comparison. and ship type priority function This enables intelligent decision-making regarding ship delays, rather than random delays. Plans for conflict Revised to A conflict-free plan This significantly improves the feasibility of the plan. Through CTM annotations in S3-4, resource mismatch issues, such as ports having refrigerated cargo, are addressed. But there are no refrigerated ships. Transform into a specific data tag .

[0218] Without addressing the scheduling volatility during the static optimization process, this invention next performs cross-time optimization adjustments:

[0219] The input for cross-time optimization adjustment is and members Ship maintenance plan data It outputs a dynamically optimized alliance-level overall scheduling plan. .

[0220] Cross-time optimization adjustments specifically include: constructing a global multi-objective cost function. This function is used for quantization. The total cost of scheduling volatility, empty flight distance, maintenance conflicts, and type mismatch in the plan.

[0221] Using iterative optimization methods As the initial solution, generate scheduling change operations and minimize... To achieve the goal, search and generate the optimal . Defined as:

[0222] in, For an evaluation The Tian Plan, for example Or its optimization. For type mismatch cost, For empty flight cost, To cover maintenance conflict costs, To manage volatility costs. Using predefined penalty factors, the original costs are converted into a unified, additive penalty metric. Specifically, the units of these penalty factors are defined as follows: [Penalty Units / Number of Mismatch Types], [Punishment Unit / Sea Miles], [Penalty Unit / Number of Unsatisfied Events] [Penalty Unit / Number of Fluctuations], in a preferred embodiment, those skilled in the art can set the relative magnitudes of these conversion factors according to business needs, for example, setting... The corresponding punishment is higher than Corresponding penalties are imposed to reflect the priority of different optimization objectives.

[0223] Iterative optimization specifically employs the annealing algorithm, with As the initial solution, an iterative search is performed to find a solution that makes... Minimum optimal solution .

[0224] The specific implementation of iterative optimization includes:

[0225] Set the initial solution: As the current optimal solution .

[0226] Call the function defined in S4-1 to calculate global dynamic cost .

[0227] Set the SA parameters and initialize the control parameters for simulated annealing, including: initial temperature. Cooling rate Minimum termination temperature .make, .

[0228] Iterative search loop, the loop in This process continues under the given conditions. In each iteration, the following operations are performed:

[0229] Scheduling change operations generate candidate solutions Through the Apply a scheduling change operation to generate a neighborhood solution. This invention defines two specific non-random change operations, based on the system's... Based on the cost structure, prioritize the appropriate operations:

[0230] Route redirection operation: Triggering condition: when In or This operation is triggered first when the cost percentage is high. (For...) :from Randomly select one port That is, there are ports with type mismatches. Search for a ship The type of the ship Able to meet Hong Kong's mismatched demand, namely ,and Current route For low priority, for example Empty run, or The penalties are relatively low. (Targeting...) Randomly select one (Empty) side and in Searching for an existence Ports with backlog of goods As a new destination.

[0231] generate ,in The route was changed to ( for Or a newly found port of origin.

[0232] Maintenance and dispatch adjustment operation: Triggering condition: when In When the cost is high (i.e.) This operation is triggered first. From Select one unmet maintenance requirement. .generate By modifying exist A flight route prior to the date of ,For example The day, to make time for it to be in The day can dock at port.

[0233] Because changes may lead to new static conflicts. It must be reprocessed. The changed parts were re-entered into the PCC inspection and parsing and S3-4CTM inspection and annotation processes. This ensured... Always meet the physical constraints of the day.

[0234] Calculate the new cost: Calculate the cost after static revalidation. global dynamic cost Calculate the cost difference: .

[0235] if ,Right now This is a better solution, and the system unconditionally accepts the change: and .

[0236] if The system calculates an acceptance probability. : Generate a arrive Random numbers between .if The system still accepts this poor solution: and .

[0237] After completing one iteration, the system measures the temperature. Cooling:

[0238]

[0239] when When the loop terminates, the iterative search loop terminates. This is the optimal solution found in this invention. .

[0240] Cost function Used for quantification The severity of the mismatch problem.

[0241]

[0242] in, exist of In the snapshot, The set of mismatch types marked on the port. This represents the number of mismatch types.

[0243] Cost function This is used to quantify the cost of empty voyages in a plan, i.e., the voyage of a ship when it is not carrying cargo or has a very low cargo load. An empty voyage discrimination function is defined. Used to determine in The edge of the sun Is it an empty run?

[0244]

[0245] in, For ships exist Daily actual cargo volume For ships Rated cargo capacity. A very low threshold (preferably 0.05, or 5%) is used to define empty runs.

[0246] Total cost of empty flight Defined as the sum of the distances traveled during all empty flights:

[0247] in, for edge set, This represents the route distance for this voyage.

[0248] Cost function Used to quantify the penalty costs of unmet ship maintenance plans. Ship maintenance plan data. A set of tuples , indicating ships Must Date docked at Port maintenance is being carried out.

[0249] Defined as the total penalty for all unmet maintenance requirements:

[0250] in It is a maintenance satisfaction criterion function:

[0251] For when In the plan, ships exist The ship did indeed dock at Port, i.e. Existing in the port on that day In the set, maintenance is satisfied, and the cost is 0.

[0252] This indicates that if the condition is not met, a penalty will be incurred. .

[0253] Cost function Used for quantification Compared to The scheduling volatility cost serves as a stabilizer to balance the first three radical optimizations.

[0254]

[0255] in It is the total collection of ships in the alliance. It is a scheduling difference discrimination function:

[0256] Indicates a ship exist Daily plan status. if exist If the daily plan differs from the original output plan, it is counted as one fluctuation.

[0257] The cross-time dynamic optimization method provided in this embodiment, through... Optimize, and move from passive labeling The mismatch problem is transformed into a proactive optimization objective. Through proxy search, resource mismatch is eliminated at the planning level. Simultaneously, the three key dynamic objectives of resolving mismatch, reducing empty runs, and meeting maintenance requirements are balanced with maintaining stability, and unified under [a framework / mechanism]. Within this framework, it is ensured that the present invention does not blindly optimize at the expense of stability, but rather seeks the optimal solution from a global perspective.

[0258] The scheduling of shipping alliances is a complex problem that combines physical constraints with long-term strategic constraints. If all constraints are mixed and tried... Figure 1 A one-time solution leads to an excessively large search space that is computationally difficult to converge. This invention solves this problem through a two-stage decomposition. The static optimization of this invention acts as a preprocessor and filter, specifically handling and resolving local physical conflicts, such as port berth capacity constraints. This step first ensures the physical feasibility of the alliance plan on each day, revising an original plan that might have many physical conflicts into a plan without immediate conflicts. The beneficial effect of this is that it greatly reduces the computational complexity of subsequent dynamic optimization. The dynamic optimizer no longer needs to be distracted by handling these local berth conflicts, but can instead focus entirely on solving complex problems that cannot be solved and must be coordinated across time. Even more beneficially, static optimization not only resolves PCC conflicts, but also proactively identifies and labels all resource mismatch problems that require cross-time scheduling through cargo-ship type mismatch checks. Finally, dynamic optimization can concentrate all its computational resources on a clean and diagnosed input to solve the highest-value cross-time problems, such as empty voyages, thereby achieving global, dynamic optimization while ensuring physical feasibility.

[0259] exist Before formal implementation, stress tests are conducted to quantify its vulnerability and robustness. In the context of maritime alliances, a theoretically optimal plan may heavily rely on the timeliness of a few key shipping routes (edges) or ports (nodes). This embodiment simulates the random failure of these shipping routes (edges) and evaluates the resulting cascading penalties to determine whether the plan can maintain acceptable performance in the event of a single point of failure, thereby avoiding systemic collapse due to over-optimization.

[0260] The output of this embodiment is a validated scheduling plan. .

[0261] The implementation of the verification method specifically includes:

[0262] S5-1: Set the baseline cost and robustness threshold, and calculate... Baseline cost And set a maximum acceptable performance degradation threshold for the alliance. .

[0263] S5-2: Simulation of route failure and cascading effects, defining a pruning operation to simulate... Any one of the ships in transit Random delay occurs The subsequent cascading effect.

[0264] S5-3: Robustness Verification: Using the Monte Carlo simulation method, execute... The second pruning operation of S5-2 was performed, and calculations were made. Average penalty in sub-failure scenarios .

[0265] S5-4: Acceptance and Decision-Making: Calculating Robust Impact Score , and Compare to decide whether to accept the plan's output. Or refuse and return to S4.

[0266] First, call the global dynamic cost function. Calculate the input plan The total penalty, defined as the baseline cost. .

[0267] This represents the optimal cost of the plan under ideal conditions, i.e., without any unexpected failures.

[0268] Set robustness threshold The system loads a preset static robustness acceptance threshold. This parameter is set by the alliance planner and is measured in [penalty units], representing the upper limit of the average increase in penalty that the alliance is willing to bear due to a single point of random failure.

[0269] The purpose of pruning algorithms is not simply to remove an edge, but to simulate the full cost that the system must pay to correct the failure after the edge fails.

[0270] The input to the pruning algorithm is The output is a simulated failure. .

[0271] copy Get a modifiable copy .

[0272] Load a random delay number of days For example, 5 days.

[0273] from of One day is randomly selected in the sky. and from edge set Randomly select a side of a ship in transit Modify this edge. Attributes :

[0274]

[0275] Changes can trigger a series of cascading static conflicts, such as In Arrival on the same day Hong Kong, may lead to Hong Kong on that day .

[0276] To simulate this cascading effect, this step will All affected ,(in ), and re-enter into the PCC check and parsing and S3-4CTM check and annotation process for re-verification.

[0277] PCC inspection will be handled. Delayed arrival The resulting congestion may force delays. Other vessels in the port Meanwhile, CTM checks will reveal... Unable to be scheduled Arrival on the day, resulting in Hong Kong may see new... Mismatch.

[0278] Returning to the impact and re-verification .Should This represents the alliance's optimal passive response plan after encountering a random failure.

[0279] Monte Carlo simulation was used to perform the pruning algorithm. This process is repeated to obtain a statistically significant average failure cost. The average failure cost is calculated.

[0280] Calculate robust impact score The robustness impact score is obtained by calculating the difference between the average failure cost and the baseline cost. .

[0281]

[0282] Will With the set robustness threshold Compare them. The assessment concluded that the average failure cost of the plan was within the acceptable range for the alliance. It was deemed robust. Marked as .

[0283] The plan was deemed to have an excessively high average failure cost. Deemed vulnerable. Output feedback rejecting the plan.

[0284] Without changing The confirmed shipping routes and estimated arrival times (in China) Under the premise of [missing information], the final step is to optimize efficiency. This involves precisely matching cargo to be shipped with available space among alliance members to maximize the overall load factor of the alliance fleet.

[0285] by Based on this, iterate through each day of the plan. and every port Scan global node attributes ( This is used to identify position supply and position demand.

[0286] The identified supply and demand are matched across companies, and the matching process strictly adheres to the technical constraints of the cargo and the vessel.

[0287] Successful matching results will be integrated into position-sharing instructions and incorporated into... In the middle, forming .

[0288] A maritime alliance vessel scheduling system based on dynamic knowledge graphs, the system comprising:

[0289] Local dynamic knowledge graph construction module: Converts the future voyage plans of each alliance member into daily graph snapshots with ports as nodes and ships at sea as edges, and constructs a local dynamic knowledge graph;

[0290] Global Fusion Dynamic Knowledge Graph Task Module: Merges the daily graph snapshots through time-series fusion, and performs data source tracing and annotation on the attributes of nodes and edges to generate a global fusion dynamic knowledge graph for the alliance;

[0291] Static optimization module: Constraint-based static optimization checks whether the daily global graph snapshot in the globally fused dynamic knowledge graph meets physical constraints, including port berth capacity constraints, and resolves conflicts; it also checks for cargo-ship type mismatches and generates mismatch annotation attributes.

[0292] Dynamic optimization module: Performs constraint-based cross-time dynamic optimization by iteratively searching to minimize a global dynamic cost function;

[0293] Pruning module: Performs pruning verification of the overall scheduling plan, calculates the baseline cost of the dynamic optimization plan, calculates the average failure cost by simulating route failures and their cascading effects, determines the feasibility of the plan based on the difference between the average failure cost and the baseline cost, and feeds back the alliance vessel scheduling plan.

[0294] 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 scheduling ships in maritime alliances based on dynamic knowledge graphs.

[0295] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for scheduling ships in maritime alliances based on a dynamic knowledge graph.

[0296] 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, and when executed, it 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 memory bus dynamic RAM (RDRAM), etc.

[0297] 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.

[0298] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the 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 ship scheduling in maritime alliances based on dynamic knowledge graphs, characterized in that, The method includes: The future navigation plans of each alliance member are converted into daily graph snapshots with ports as nodes and ships at sea as edges, thus constructing a local dynamic knowledge graph; By merging daily graph snapshots through time-series fusion and performing data source annotation on the attributes of nodes and edges, a global fusion dynamic knowledge graph of the alliance is generated. Constraint-based static optimization checks whether the daily global graph snapshot in the globally fused dynamic knowledge graph satisfies physical constraints, including port berth capacity constraints, and resolves conflicts; it also checks for cargo-ship type mismatches and generates mismatch annotation attributes. Constraint-based time-series dynamic optimization is performed by iteratively searching to minimize a global dynamic cost function. The overall scheduling plan is pruned and verified, the baseline cost of the dynamic optimization plan is calculated, and the average failure cost is calculated by simulating route failures and their cascading effects. The feasibility of the plan is determined based on the difference between the average failure cost and the baseline cost, and feedback is given to the alliance vessel scheduling plan. Constraint-based static optimization, when checking port berth capacity constraints, includes: calculating the total daily berthing demand. : ; in For the port, For date, for Hong Kong Total daily docking demand This represents the number of ships that have docked that day. This represents the number of ships scheduled to arrive that day. when At that time, it was determined that a berth conflict had occurred, among which For the port berth capacity.

2. The method for ship scheduling in maritime alliances based on dynamic knowledge graphs according to claim 1, characterized in that, Conflict resolution includes: calculating the number of ships that need to be delayed. ; And to assemble the ships scheduled to arrive that day. ships in Calculate scheduling priority, From the sorted Starting with the lowest priority ship in the list, select in order. The ships were grouped into the delayed set. its estimated arrival time Modified to .

3. The method for ship scheduling in maritime alliances based on dynamic knowledge graphs according to claim 2, characterized in that, Checking for cargo-ship type mismatches and generating mismatch annotation attributes includes: calculating the required cargo type for shipment. ; in For member index, For the total number of members, For the type of goods, For the quantity of goods, For members The planned inventory collection of goods, for Hong Kong Daily demand for cargo types; calculation of available vessel type supply: ; in Supply for available vessel types, For the types of vessels that have already docked, The type of vessel arriving on the same day after static optimization processing; Calculate the set of mismatch types ; when When it is a non-empty set, at the port Add mismatch annotation attributes to nodes , This represents the difference set.

4. The method for scheduling ships in maritime alliances based on dynamic knowledge graphs according to claim 3, characterized in that, Global dynamic cost function Defined as: ； in, For an evaluation The Sky Project for Global dynamic cost; Cost of type mismatch; Cost of empty flight; To cover maintenance and repair costs; To manage volatility costs; This is a preset penalty conversion factor used to convert the original costs of each entity into a uniform, additive penalty metric.

5. A method for scheduling ships in maritime alliances based on dynamic knowledge graphs according to claim 4, characterized in that: Type mismatch cost Defined as: ; in Total number of days For port assembly, For static optimization in the port Added mismatch annotation attributes to nodes The number of mismatch types; maintenance conflict costs. Defined as: ; ;in, For ship maintenance plan data set, For ships, For port maintenance, For maintenance date, For the maintenance satisfaction criterion function, when the ship exist The sun never stopped at When in Hong Kong, a penalty is imposed. , For the assembly of ships scheduled to dock.

6. A method for scheduling ships in maritime alliances based on dynamic knowledge graphs according to claim 5, characterized in that: Empty flight cost Defined as: ;in for exist The edge of the day, The route distance of the flight, This is the empty-run distance discrimination function, which is based on the actual cargo load of the ship. With rated cargo capacity The comparison is used to make a judgment; ;in, For ships exist Daily actual cargo volume For ships Rated cargo capacity; A very low threshold is used to define empty runs; scheduling volatility costs. Defined as: ; ; Indicates a ship exist Daily plan status; Indicates if exist If the daily plan differs from the original output plan, it is counted as one fluctuation. For ships, For the total number of ships in the alliance (...) is the scheduling difference discrimination function. This is the statically optimized plan sequence.

7. A method for scheduling ships in maritime alliances based on dynamic knowledge graphs according to claim 6, characterized in that, Iterative search employs simulated annealing algorithm, with As the initial solution, an iterative search is performed to find a solution that makes... Minimum optimal solution Set the initial solution: As the current optimal solution ,calculate global dynamic cost , and includes, according to The cost function is used to trigger route redirection or maintenance scheduling adjustment operations to generate candidate solutions. ;Will The changed parts are re-entered into the static optimization PCC check and parsing and CTM check and annotation process for static re-validation; based on the acceptance probability of simulated annealing. Decide whether to accept , ,in For cost difference, The current temperature; if, , This is a better solution, and the system unconditionally accepts the change: and ;if The system calculates an acceptance probability. : Generate a arrive Random numbers between ;if The system still accepts this poor solution: and After completing one iteration, the system cools down. hour, The iterative search loop terminates at the minimum termination temperature; upon loop termination, To find the optimal .

8. A method for scheduling ships in maritime alliances based on dynamic knowledge graphs according to claim 7, characterized in that, Simulated route failures and their cascading effects include: a. Randomly selecting a route alongside an in-flight vessel. And apply a random delay of days. To modify it : ;in This is the original estimated arrival time. b. The number of days of random delay; c. All affected elements in the modified map in a. in , for The daily global map snapshot is re-entered into the statically optimized PCC check and parsing and CTM check and annotation process for re-verification, resulting in a failed map. .

9. A method for scheduling ships in maritime alliances based on dynamic knowledge graphs according to claim 8, characterized in that, Determining the feasibility of the plan includes: calculating the baseline cost. , For dynamically optimized plans; through Sub-route failure simulation to calculate average failure cost. ; Calculate robust impact score ;when Less than or equal to the preset robustness threshold At that time, the plan was deemed feasible and marked as .

Images